<?xmlversion="1.0" encoding="US-ASCII"?> <!DOCTYPE rfc SYSTEM "rfc2629.dtd" [ <!ENTITY rfc2119 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.2119.xml"> <!ENTITY rfc5170 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.5170.xml"> <!-- ldpc --> <!ENTITY rfc5510 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.5510.xml"> <!-- r-s --> <!ENTITY rfc6363 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.6363.xml"> <!ENTITY rfc6364 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.6364.xml"> <!ENTITY rfc6726 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.6726.xml"> <!-- flute --> <!ENTITY rfc6681 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.6681.xml"> <!-- raptor/q for ff --> <!ENTITY rfc6816 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.6816.xml"> <!-- ldpc-staircase for ff --> <!ENTITY rfc6865 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.6865.xml"> <!-- r-s for ff --> <!ENTITY rfc8174 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.8174.xml"> <!ENTITY rfc8406 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.8406.xml"> ]> <?rfc toc="yes" ?> <?rfc compact="yes" ?> <?rfc subcompact="yes" ?> <?rfc symrefs="yes" ?> <?rfc sortrefs="yes" ?> <?rfc rfcedstyle="yes" ?>version='1.0' encoding='utf-8'?> <rfc xmlns:xi="http://www.w3.org/2001/XInclude" version="3" category="std" consensus="true" docName="draft-ietf-tsvwg-rlc-fec-scheme-16"ipr="trust200902">indexInclude="true" ipr="trust200902" number="8681" prepTime="2020-01-08T16:08:50" scripts="Common,Latin" sortRefs="true" submissionType="IETF" symRefs="true" tocDepth="3" tocInclude="true" xml:lang="en"> <link href="https://datatracker.ietf.org/doc/draft-ietf-tsvwg-rlc-fec-scheme-16" rel="prev"/> <link href="https://dx.doi.org/10.17487/rfc8681" rel="alternate"/> <link href="urn:issn:2070-1721" rel="alternate"/> <front> <title abbrev="RLC FEC Scheme">Sliding Window Random Linear Code (RLC) Forward Erasure Correction (FEC) Schemes for FECFRAME</title> <seriesInfo name="RFC" value="8681" stream="IETF"/> <author fullname="Vincent Roca" initials="V" surname="Roca"><organization>INRIA</organization><organization showOnFrontPage="true">INRIA</organization> <address> <postal><street></street> <city>Univ.<street/> <city/> <code/> <extaddr>Univ. GrenobleAlpes</city> <code></code>Alpes</extaddr> <country>France</country> </postal> <email>vincent.roca@inria.fr</email> </address> </author> <author fullname="Belkacem Teibi" initials="B" surname="Teibi"><organization>INRIA</organization><organization showOnFrontPage="true">INRIA</organization> <address> <postal><street></street> <city>Univ.<street/> <city/> <code/> <extaddr>Univ. GrenobleAlpes</city> <code></code>Alpes</extaddr> <country>France</country> </postal> <email>belkacem.teibi@gmail.com</email> </address> </author><!--<datemonth="February" year="2017" /> --> <date/>month="01" year="2020"/> <workgroup>TSVWG</workgroup><abstract> <t><keyword>RLC</keyword> <keyword>FEC</keyword> <keyword>FECFRAME</keyword> <keyword>packet loss recovery</keyword> <keyword>reliability</keyword> <abstract pn="section-abstract"> <t pn="section-abstract-1"> This document describes twofully-specifiedfully specified Forward Erasure Correction (FEC) Schemes for Sliding Window Random Linear Codes (RLC), one for RLC over the Galois Field(A.K.A.(a.k.a., Finite Field) GF(2), a second one for RLC over the Galois FieldGF(2^^8),GF(2<sup>8</sup>), each time with the possibility of controlling the code density. They can protect arbitrary media streams along the lines defined by FECFRAME extended tosliding windowSliding Window FECcodes.Codes. Thesesliding windowSliding Window FECcodesCodes rely on an encoding window that slides over the source symbols, generating new repair symbols whenever needed. Compared to block FEC codes, thesesliding windowSliding Window FECcodesCodes offer key advantages with real-time flows in terms of reduced FEC-related latency while often providing improved packet erasure recovery capabilities. </t> </abstract></front> <middle><boilerplate> <sectionanchor="introduction" title="Introduction"> <!-- ====================== --> <t> Application-Level Forward Erasure Correction (AL-FEC) codes, or simply FEC codes, are a key element of communication systems. They are used to recover from packet losses (or erasures) during content delivery sessions to a potentially large numberanchor="status-of-memo" numbered="false" removeInRFC="false" toc="exclude" pn="section-boilerplate.1"> <name slugifiedName="name-status-of-this-memo">Status ofreceivers (multicast/broadcast transmissions).This Memo</name> <t pn="section-boilerplate.1-1"> This is an Internet Standards Track document. </t> <t pn="section-boilerplate.1-2"> This document is a product of thecase withInternet Engineering Task Force (IETF). It represents theFLUTE/ALC protocol <xref target="RFC6726"/> when usedconsensus of the IETF community. It has received public review and has been approved forreliable file transfers over lossy networks,publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841. </t> <t pn="section-boilerplate.1-3"> Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at <eref target="https://www.rfc-editor.org/info/rfc8681" brackets="none"/>. </t> </section> <section anchor="copyright" numbered="false" removeInRFC="false" toc="exclude" pn="section-boilerplate.2"> <name slugifiedName="name-copyright-notice">Copyright Notice</name> <t pn="section-boilerplate.2-1"> Copyright (c) 2020 IETF Trust and theFECFRAME protocol <xref target="RFC6363"/> when used for reliable continuous media transfers over lossy networks.persons identified as the document authors. All rights reserved. </t><t> The present<t pn="section-boilerplate.2-2"> This documentonly focusesis subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (<eref target="https://trustee.ietf.org/license-info" brackets="none"/>) in effect on theFECFRAME protocol, useddate of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described inmulticast/broadcast delivery mode,Section 4.e of the Trust Legal Provisions and are provided without warranty as described inparticular for contents that feature stringent real-time constraints: each source packet has a maximum validity period after which it will not be considered bythedestination application.Simplified BSD License. </t> </section> </boilerplate> <toc> <sectionanchor="intro:block_codes" title="Limitsanchor="toc" numbered="false" removeInRFC="false" toc="exclude" pn="section-toc.1"> <name slugifiedName="name-table-of-contents">Table of Contents</name> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1"> <li pn="section-toc.1-1.1"> <t keepWithNext="true" pn="section-toc.1-1.1.1"><xref derivedContent="1" format="counter" sectionFormat="of" target="section-1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-introduction">Introduction</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.1.2"> <li pn="section-toc.1-1.1.2.1"> <t keepWithNext="true" pn="section-toc.1-1.1.2.1.1"><xref derivedContent="1.1" format="counter" sectionFormat="of" target="section-1.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-limits-of-block-codes-with-">Limits of Block Codes with Real-TimeFlows"> <!-- ====================== --> <t> With FECFRAME, there is a single FEC encoding point (either an end-host/server (source) or a middlebox) and a single FEC decoding point per receiver (either an end-host (receiver) or middlebox). In this context, currently standardized AL-FEC codes for FECFRAME like Reed-Solomon <xref target="RFC6865"/>, LDPC-Staircase <xref target="RFC6816"/>, or Raptor/RaptorQ <xref target="RFC6681"/>, are all linear block codes: they require the data flow to be segmented into blocks of a predefined maximum size. </t> <t> To define this block size, it is required to find an appropriate balance between robustness and decoding latency: the larger the block size, the higher the robustness (e.g., in case of long packet erasure bursts), but also the higher the maximum decoding latency (i.e., the maximum time required to recover a lost (erased) packet thanks to FEC protection). Therefore, with a multicast/broadcast session where different receivers experience different packet loss rates, the block size should be chosen by considering the worst communication conditions one wants to support, but without exceeding the desired maximum decoding latency. This choice then impacts the FEC-related latency of all receivers, even those experiencing a good communication quality, since no FEC encoding can happen until all the source data of the block is available at the sender, which directly depends on the block size. </t> </section> <section anchor="intro:conv_codes" title="Lower LatencyFlows</xref></t> </li> <li pn="section-toc.1-1.1.2.2"> <t keepWithNext="true" pn="section-toc.1-1.1.2.2.1"><xref derivedContent="1.2" format="counter" sectionFormat="of" target="section-1.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-lower-latency-and-better-pr">Lower Latency and Better Protection of Real-Time Flows with the Sliding Window RLCCodes"> <!-- ====================== --> <t> This document introduces two fully-specified FEC Schemes that do not follow the block code approach:Codes</xref></t> </li> <li pn="section-toc.1-1.1.2.3"> <t keepWithNext="true" pn="section-toc.1-1.1.2.3.1"><xref derivedContent="1.3" format="counter" sectionFormat="of" target="section-1.3"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-small-transmission-overhead">Small Transmission Overheads with the Sliding WindowRandom Linear Codes (RLC) over either Galois Fields (A.K.A. Finite Fields) GF(2) (the "binary case") or GF(2^^8), each time with the possibility of controlling the code density. These FEC Schemes are used to protect arbitrary media streams along the lines defined by FECFRAME extended to sliding window FEC codes <xref target="fecframe-ext"/>. TheseRLC FECSchemes, and more generally SlidingScheme</xref></t> </li> <li pn="section-toc.1-1.1.2.4"> <t keepWithNext="true" pn="section-toc.1-1.1.2.4.1"><xref derivedContent="1.4" format="counter" sectionFormat="of" target="section-1.4"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-document-organization">Document Organization</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.2"> <t keepWithNext="true" pn="section-toc.1-1.2.1"><xref derivedContent="2" format="counter" sectionFormat="of" target="section-2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-definitions-and-abbreviatio">Definitions and Abbreviations</xref></t> </li> <li pn="section-toc.1-1.3"> <t keepWithNext="true" pn="section-toc.1-1.3.1"><xref derivedContent="3" format="counter" sectionFormat="of" target="section-3"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-common-procedures">Common Procedures</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.3.2"> <li pn="section-toc.1-1.3.2.1"> <t keepWithNext="true" pn="section-toc.1-1.3.2.1.1"><xref derivedContent="3.1" format="counter" sectionFormat="of" target="section-3.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-codec-parameters">Codec Parameters</xref></t> </li> <li pn="section-toc.1-1.3.2.2"> <t keepWithNext="true" pn="section-toc.1-1.3.2.2.1"><xref derivedContent="3.2" format="counter" sectionFormat="of" target="section-3.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-adu-adui-and-source-symbols">ADU, ADUI, and Source Symbols Mappings</xref></t> </li> <li pn="section-toc.1-1.3.2.3"> <t keepWithNext="true" pn="section-toc.1-1.3.2.3.1"><xref derivedContent="3.3" format="counter" sectionFormat="of" target="section-3.3"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-encoding-window-management">Encoding Window Management</xref></t> </li> <li pn="section-toc.1-1.3.2.4"> <t keepWithNext="true" pn="section-toc.1-1.3.2.4.1"><xref derivedContent="3.4" format="counter" sectionFormat="of" target="section-3.4"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-source-symbol-identificatio">Source Symbol Identification</xref></t> </li> <li pn="section-toc.1-1.3.2.5"> <t keepWithNext="true" pn="section-toc.1-1.3.2.5.1"><xref derivedContent="3.5" format="counter" sectionFormat="of" target="section-3.5"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-pseudorandom-number-generat">Pseudorandom Number Generator (PRNG)</xref></t> </li> <li pn="section-toc.1-1.3.2.6"> <t keepWithNext="true" pn="section-toc.1-1.3.2.6.1"><xref derivedContent="3.6" format="counter" sectionFormat="of" target="section-3.6"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-coding-coefficients-generat">Coding Coefficients Generation Function</xref></t> </li> <li pn="section-toc.1-1.3.2.7"> <t keepWithNext="true" pn="section-toc.1-1.3.2.7.1"><xref derivedContent="3.7" format="counter" sectionFormat="of" target="section-3.7"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-finite-field-operations">Finite Field Operations</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.3.2.7.2"> <li pn="section-toc.1-1.3.2.7.2.1"> <t keepWithNext="true" pn="section-toc.1-1.3.2.7.2.1.1"><xref derivedContent="3.7.1" format="counter" sectionFormat="of" target="section-3.7.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-finite-field-definitions">Finite Field Definitions</xref></t> </li> <li pn="section-toc.1-1.3.2.7.2.2"> <t keepWithNext="true" pn="section-toc.1-1.3.2.7.2.2.1"><xref derivedContent="3.7.2" format="counter" sectionFormat="of" target="section-3.7.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-linear-combination-of-sourc">Linear Combination of Source Symbol Computation</xref></t> </li> </ul> </li> </ul> </li> <li pn="section-toc.1-1.4"> <t keepWithNext="true" pn="section-toc.1-1.4.1"><xref derivedContent="4" format="counter" sectionFormat="of" target="section-4"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-sliding-window-rlc-fec-sche">Sliding Window RLC FECcodes, are recommendedScheme over GF(2<sup>8</sup>) forinstance, with media that feature real-time constraints sent within a multicast/broadcast session <xref target="Roca17"/>. </t> <t> TheArbitrary Packet Flows</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.4.2"> <li pn="section-toc.1-1.4.2.1"> <t keepWithNext="true" pn="section-toc.1-1.4.2.1.1"><xref derivedContent="4.1" format="counter" sectionFormat="of" target="section-4.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-formats-and-codes">Formats and Codes</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.4.2.1.2"> <li pn="section-toc.1-1.4.2.1.2.1"> <t keepWithNext="true" pn="section-toc.1-1.4.2.1.2.1.1"><xref derivedContent="4.1.1" format="counter" sectionFormat="of" target="section-4.1.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-fec-framework-configuration">FEC Framework Configuration Information</xref></t> </li> <li pn="section-toc.1-1.4.2.1.2.2"> <t keepWithNext="true" pn="section-toc.1-1.4.2.1.2.2.1"><xref derivedContent="4.1.2" format="counter" sectionFormat="of" target="section-4.1.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-explicit-source-fec-payload">Explicit Source FEC Payload ID</xref></t> </li> <li pn="section-toc.1-1.4.2.1.2.3"> <t keepWithNext="true" pn="section-toc.1-1.4.2.1.2.3.1"><xref derivedContent="4.1.3" format="counter" sectionFormat="of" target="section-4.1.3"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-repair-fec-payload-id">Repair FEC Payload ID</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.4.2.2"> <t keepWithNext="true" pn="section-toc.1-1.4.2.2.1"><xref derivedContent="4.2" format="counter" sectionFormat="of" target="section-4.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-procedures">Procedures</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.5"> <t keepWithNext="true" pn="section-toc.1-1.5.1"><xref derivedContent="5" format="counter" sectionFormat="of" target="section-5"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-sliding-window-rlc-fec-schem">Sliding Window RLCcodes belongFEC Scheme over GF(2) for Arbitrary Packet Flows</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.5.2"> <li pn="section-toc.1-1.5.2.1"> <t keepWithNext="true" pn="section-toc.1-1.5.2.1.1"><xref derivedContent="5.1" format="counter" sectionFormat="of" target="section-5.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-formats-and-codes-2">Formats and Codes</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.5.2.1.2"> <li pn="section-toc.1-1.5.2.1.2.1"> <t keepWithNext="true" pn="section-toc.1-1.5.2.1.2.1.1"><xref derivedContent="5.1.1" format="counter" sectionFormat="of" target="section-5.1.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-fec-framework-configuration-">FEC Framework Configuration Information</xref></t> </li> <li pn="section-toc.1-1.5.2.1.2.2"> <t keepWithNext="true" pn="section-toc.1-1.5.2.1.2.2.1"><xref derivedContent="5.1.2" format="counter" sectionFormat="of" target="section-5.1.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-explicit-source-fec-payload-">Explicit Source FEC Payload ID</xref></t> </li> <li pn="section-toc.1-1.5.2.1.2.3"> <t keepWithNext="true" pn="section-toc.1-1.5.2.1.2.3.1"><xref derivedContent="5.1.3" format="counter" sectionFormat="of" target="section-5.1.3"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-repair-fec-payload-id-2">Repair FEC Payload ID</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.5.2.2"> <t keepWithNext="true" pn="section-toc.1-1.5.2.2.1"><xref derivedContent="5.2" format="counter" sectionFormat="of" target="section-5.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-procedures-2">Procedures</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.6"> <t keepWithNext="true" pn="section-toc.1-1.6.1"><xref derivedContent="6" format="counter" sectionFormat="of" target="section-6"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-fec-code-specification">FEC Code Specification</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.6.2"> <li pn="section-toc.1-1.6.2.1"> <t keepWithNext="true" pn="section-toc.1-1.6.2.1.1"><xref derivedContent="6.1" format="counter" sectionFormat="of" target="section-6.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-encoding-side">Encoding Side</xref></t> </li> <li pn="section-toc.1-1.6.2.2"> <t keepWithNext="true" pn="section-toc.1-1.6.2.2.1"><xref derivedContent="6.2" format="counter" sectionFormat="of" target="section-6.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-decoding-side">Decoding Side</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.7"> <t keepWithNext="true" pn="section-toc.1-1.7.1"><xref derivedContent="7" format="counter" sectionFormat="of" target="section-7"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-security-considerations">Security Considerations</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.7.2"> <li pn="section-toc.1-1.7.2.1"> <t keepWithNext="true" pn="section-toc.1-1.7.2.1.1"><xref derivedContent="7.1" format="counter" sectionFormat="of" target="section-7.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-attacks-against-the-data-fl">Attacks Against the Data Flow</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.7.2.1.2"> <li pn="section-toc.1-1.7.2.1.2.1"> <t keepWithNext="true" pn="section-toc.1-1.7.2.1.2.1.1"><xref derivedContent="7.1.1" format="counter" sectionFormat="of" target="section-7.1.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-access-to-confidential-cont">Access to Confidential Content</xref></t> </li> <li pn="section-toc.1-1.7.2.1.2.2"> <t keepWithNext="true" pn="section-toc.1-1.7.2.1.2.2.1"><xref derivedContent="7.1.2" format="counter" sectionFormat="of" target="section-7.1.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-content-corruption">Content Corruption</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.7.2.2"> <t keepWithNext="true" pn="section-toc.1-1.7.2.2.1"><xref derivedContent="7.2" format="counter" sectionFormat="of" target="section-7.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-attacks-against-the-fec-par">Attacks Against thebroad class of sliding-window AL-FEC codes (A.K.A. convolutional codes) <xref target="RFC8406"/>. The encoding process is based on an encoding window that slides overFEC Parameters</xref></t> </li> <li pn="section-toc.1-1.7.2.3"> <t keepWithNext="true" pn="section-toc.1-1.7.2.3.1"><xref derivedContent="7.3" format="counter" sectionFormat="of" target="section-7.3"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-when-several-source-flows-a">When Several Source Flows are to be Protected Together</xref></t> </li> <li pn="section-toc.1-1.7.2.4"> <t keepWithNext="true" pn="section-toc.1-1.7.2.4.1"><xref derivedContent="7.4" format="counter" sectionFormat="of" target="section-7.4"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-baseline-secure-fec-framewo">Baseline Secure FEC Framework Operation</xref></t> </li> <li pn="section-toc.1-1.7.2.5"> <t keepWithNext="true" pn="section-toc.1-1.7.2.5.1"><xref derivedContent="7.5" format="counter" sectionFormat="of" target="section-7.5"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-additional-security-conside">Additional Security Considerations for Numerical Computations</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.8"> <t keepWithNext="true" pn="section-toc.1-1.8.1"><xref derivedContent="8" format="counter" sectionFormat="of" target="section-8"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-operations-and-management-c">Operations and Management Considerations</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.8.2"> <li pn="section-toc.1-1.8.2.1"> <t keepWithNext="true" pn="section-toc.1-1.8.2.1.1"><xref derivedContent="8.1" format="counter" sectionFormat="of" target="section-8.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-operational-recommendations">Operational Recommendations: Finite Field GF(2) Versus GF(2<sup>8</sup>)</xref></t> </li> <li pn="section-toc.1-1.8.2.2"> <t keepWithNext="true" pn="section-toc.1-1.8.2.2.1"><xref derivedContent="8.2" format="counter" sectionFormat="of" target="section-8.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-operational-recommendations-">Operational Recommendations: Coding Coefficients Density Threshold</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.9"> <t keepWithNext="true" pn="section-toc.1-1.9.1"><xref derivedContent="9" format="counter" sectionFormat="of" target="section-9"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-iana-considerations">IANA Considerations</xref></t> </li> <li pn="section-toc.1-1.10"> <t keepWithNext="true" pn="section-toc.1-1.10.1"><xref derivedContent="10" format="counter" sectionFormat="of" target="section-10"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-references">References</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.10.2"> <li pn="section-toc.1-1.10.2.1"> <t keepWithNext="true" pn="section-toc.1-1.10.2.1.1"><xref derivedContent="10.1" format="counter" sectionFormat="of" target="section-10.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-normative-references">Normative References</xref></t> </li> <li pn="section-toc.1-1.10.2.2"> <t keepWithNext="true" pn="section-toc.1-1.10.2.2.1"><xref derivedContent="10.2" format="counter" sectionFormat="of" target="section-10.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-informative-references">Informative References</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.11"> <t keepWithNext="true" pn="section-toc.1-1.11.1"><xref derivedContent="Appendix A" format="default" sectionFormat="of" target="section-appendix.a"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-tinymt32-validation-criteri">TinyMT32 Validation Criteria (Normative)</xref></t> </li> <li pn="section-toc.1-1.12"> <t keepWithNext="true" pn="section-toc.1-1.12.1"><xref derivedContent="Appendix B" format="default" sectionFormat="of" target="section-appendix.b"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-assessing-the-prng-adequacy">Assessing thesetPRNG Adequacy (Informational)</xref></t> </li> <li pn="section-toc.1-1.13"> <t keepWithNext="true" pn="section-toc.1-1.13.1"><xref derivedContent="Appendix C" format="default" sectionFormat="of" target="section-appendix.c"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-possible-parameter-derivati">Possible Parameter Derivation (Informational)</xref></t> <ul bare="true" empty="true" indent="2" spacing="compact" pn="section-toc.1-1.13.2"> <li pn="section-toc.1-1.13.2.1"> <t keepWithNext="true" pn="section-toc.1-1.13.2.1.1"><xref derivedContent="C.1" format="counter" sectionFormat="of" target="section-c.1"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-case-of-a-cbr-real-time-flo">Case ofsource packets (in fact source symbols as we will see in <xref target="CommonProc_adui_creation"/>), this window being eithera CBR Real-Time Flow</xref></t> </li> <li pn="section-toc.1-1.13.2.2"> <t keepWithNext="true" pn="section-toc.1-1.13.2.2.1"><xref derivedContent="C.2" format="counter" sectionFormat="of" target="section-c.2"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-other-types-of-real-time-fl">Other Types offixed sizeReal-Time Flow</xref></t> </li> <li pn="section-toc.1-1.13.2.3"> <t keepWithNext="true" pn="section-toc.1-1.13.2.3.1"><xref derivedContent="C.3" format="counter" sectionFormat="of" target="section-c.3"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-case-of-a-non-real-time-flo">Case of a Non-Real-Time Flow</xref></t> </li> </ul> </li> <li pn="section-toc.1-1.14"> <t keepWithNext="true" pn="section-toc.1-1.14.1"><xref derivedContent="Appendix D" format="default" sectionFormat="of" target="section-appendix.d"/>. <xref derivedContent="" format="title" sectionFormat="of" target="name-decoding-beyond-maximum-lat">Decoding Beyond Maximum Latency Optimization (Informational)</xref></t> </li> <li pn="section-toc.1-1.15"> <t keepWithNext="true" pn="section-toc.1-1.15.1"><xref derivedContent="" format="none" sectionFormat="of" target="section-appendix.e"/><xref derivedContent="" format="title" sectionFormat="of" target="name-acknowledgments">Acknowledgments</xref></t> </li> <li pn="section-toc.1-1.16"> <t keepWithNext="true" pn="section-toc.1-1.16.1"><xref derivedContent="" format="none" sectionFormat="of" target="section-appendix.f"/><xref derivedContent="" format="title" sectionFormat="of" target="name-authors-addresses">Authors' Addresses</xref></t> </li> </ul> </section> </toc> </front> <middle> <section anchor="introduction" numbered="true" toc="include" removeInRFC="false" pn="section-1"> <name slugifiedName="name-introduction">Introduction</name> <t pn="section-1-1"> Application-Level Forward Erasure Correction (AL-FEC) codes, orvariable size (A.K.A. an elastic window). Repair symbolssimply FEC codes, aregenerated on-the-fly, by computingarandom linear combinationkey element ofthe source symbols present in the current encoding window, and passedcommunication systems. They are used to recover from packet losses (or erasures) during content delivery sessions tothe transport layer. </t> <t> At the receiver,alinear systempotentially large number of receivers (multicast/broadcast transmissions). This ismanaged fromtheset of received source and repair packets. New variables (representing source symbols) and equations (representingcase with thelinear combination carried by each repair symbol received) are added upon receiving new packets. VariablesFile Delivery over Unidirectional Transport (FLUTE)/Asynchronous Layered Coding (ALC) protocol <xref target="RFC6726" format="default" sectionFormat="of" derivedContent="RFC6726"/> when used for reliable file transfers over lossy networks, and theequations they are involved in are removedFECFRAME protocol <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/> whenthey are too old with respect to their validity period (real-time constraints). Lost source symbols are then recovered thanks to this linear system whenever its rank permits to solve it (at least partially).used for reliable continuous media transfers over lossy networks. </t><t><t pn="section-1-2"> Theprotection of any multicast/broadcast session needs to be dimensioned by consideringpresent document only focuses on theworst communication conditions one wants to support. ThisFECFRAME protocol, which isalso true with RLC (more generally any sliding window) code. However, the receivers experiencing a good to medium communication quality will observe a reduced FEC-related latency compared to block codes <xref target="Roca17"/> since an isolated lostused in multicast/broadcast delivery mode, particularly for content that features stringent real-time constraints: each source packetis quickly recovered with the following repair packet. On the opposite, withhas ablock code, recovering an isolated lost source packet always requires waiting for the first repair packet to arrivemaximum validity period after which it will not be considered by theenddestination application. </t> <section anchor="intro_block_codes" numbered="true" toc="include" removeInRFC="false" pn="section-1.1"> <name slugifiedName="name-limits-of-block-codes-with-">Limits ofthe block. Additionally, under certain situations (e.g.,Block Codes witha limited FEC-related latency budgetReal-Time Flows</name> <t pn="section-1.1-1"> With FECFRAME, there is a single FEC encoding point (either an end host/server (source) or a middlebox) andwith constant bitrate transmissions after FECFRAME encoding), sliding window codes can more efficiently achieveatarget transmission quality (e.g., measured by the residual loss aftersingle FECdecoding) by sending fewer repair packets (i.e., higher code rate) than block codes. <!--decoding point per receiver (either an end host (receiver) or middlebox). In this context, currently standardized AL-FEC codes for FECFRAME like Reed-Solomon <xreftarget="Roca17"/> --> </t> </section> <section anchor="intro:low_tx_overhead" title="Small Transmission Overheads withtarget="RFC6865" format="default" sectionFormat="of" derivedContent="RFC6865"/>, LDPC-Staircase <xref target="RFC6816" format="default" sectionFormat="of" derivedContent="RFC6816"/>, or Raptor/RaptorQ <xref target="RFC6681" format="default" sectionFormat="of" derivedContent="RFC6681"/>, are all linear block codes: they require theSliding Window RLC FEC Scheme"> <!-- ====================== --> <t> The Sliding Window RLC FEC Scheme is designeddata flow tolimit the packet header overhead. The main requirement is that each repair packet header must enablebe segmented into blocks of areceiverpredefined maximum size. </t> <t pn="section-1.1-2"> To define this block size, it is required toreconstructfind an appropriate balance between robustness and decoding latency: theset of source symbols pluslarger theassociated coefficients used duringblock size, theencoding process. In order to minimize packet overhead,higher theset of source symbolsrobustness (e.g., inthe encoding window as well as the setcase ofcoefficients over GF(2^^m) (where m is 1 or 8, depending onlong packet erasure bursts), but also theFEC Scheme) used inhigher thelinear combination are not individually listed inmaximum decoding latency (i.e., therepairmaximum time required to recover a lost (erased) packetheader. Instead, eachthanks to FECRepair Packet header contains: <list style="symbols"> <t>the Encoding Symbol Identifier (ESI) of the first source symbol in the encoding window as well as the number of symbols (since this number may varyprotection). Therefore, with avariable size, elastic window). These two pieces of information enable each receiver to reconstruct the set of source symbols considered during encoding,multicast/broadcast session where different receivers experience different packet loss rates, theonly constraint being that there cannotblock size should beany gap;</t> <t>the seed and density threshold parameters usedchosen bya coding coefficients generation function (<xref target="CommonProc_coef_generation_func"/>). These two pieces of information enable each receiver to generateconsidering thesame set of coding coefficients over GF(2^^m) asworst communication conditions one wants to support, but without exceeding thesender;</t> </list> </t> <t> Therefore, no matterdesired maximum decoding latency. This choice then impacts thenumberFEC-related latency ofsource symbols present in the encoding window, each FEC Repair Packet features a fixed 64-bit long header, called Repair FEC Payload ID (<xref target="fig_repair_fpi"/>). Similarly, each FEC Source Packet featuresall receivers, even those experiencing afixed 32-bit long trailer, called Explicit Sourcegood communication quality, since no FECPayload ID (<xref target="fig_src_fpi"/>), that containsencoding can happen until all theESIsource data of thefirst source symbol (<xref target="CommonProc_adui_creation"/>).block is available at the sender, which directly depends on the block size. </t> </section> <sectiontitle="Document Organization"> <!-- ====================== --> <t> This fully-specified FEC Scheme followsanchor="intro_conv_codes" numbered="true" toc="include" removeInRFC="false" pn="section-1.2"> <name slugifiedName="name-lower-latency-and-better-pr">Lower Latency and Better Protection of Real-Time Flows with thestructure required by <xref target="RFC6363"/>, section 5.6. "FEC Scheme Requirements", namely: <list style="hanging"> <t hangText="3. Procedures:"> This section describes procedures specific to this FEC Scheme, namely:Sliding Window RLCparameters derivation, ADUI and source symbols mapping, pseudo-random number generator, and coding coefficients generation function;</t>Codes</name> <thangText="4. Formats and Codes:">pn="section-1.2-1"> Thissection defines the Source FEC Payload ID and Repairdocument introduces two fully specified FECPayload ID formats, carryingschemes that do not follow thesignaling information associated to each source or repair symbol. It also definesblock code approach: theFEC Framework Configuration Information (FFCI) carrying signaling information forSliding Window Random Linear Codes (RLC) over either Galois Fields (a.k.a., Finite Fields) GF(2) (the "binary case") or GF(2<sup>8</sup>), each time with thesession;</t> <t hangText="5. FEC Code Specification:"> Finally this section providespossibility of controlling the codespecification.</t> </list> </t> </section> </section> <section anchor="definitionsAndAbbreviations" title="Definitions and Abbreviations"> <!-- ====================== --> <t> The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this documentdensity. These FEC schemes are used tobe interpreted as described in BCP 14protect arbitrary media streams along the lines defined by FECFRAME extended to Sliding Window FEC Codes <xreftarget="RFC2119"/>target="RFC8680" format="default" sectionFormat="of" derivedContent="RFC8680"/>. These FEC schemes and, more generally, Sliding Window FEC Codes are recommended, for instance, with media that feature real-time constraints sent within a multicast/broadcast session <xreftarget="RFC8174"/> when, and only when, they appear in all capitals, as shown here.target="Roca17" format="default" sectionFormat="of" derivedContent="Roca17"/>. </t><t>This document uses the following definitions and abbreviations: <list style="hanging"><thangText="a^^b"> apn="section-1.2-2"> The RLC codes belong to thepowerbroad class ofb</t> <t hangText="GF(q)"> denotes a finite field (also known as the Galois Field) with q elements. We assumeSliding Window AL-FEC Codes (a.k.a., convolutional codes) <xref target="RFC8406" format="default" sectionFormat="of" derivedContent="RFC8406"/>. The encoding process is based on an encoding window thatq = 2^^m in this document</t> <t hangText="m"> definesslides over thelengthset ofthe elements in the finite field,source packets (in fact source symbols as we will see inbits. In<xref target="CommonProc_adui_creation" format="default" sectionFormat="of" derivedContent="Section 3.2"/>), thisdocument, m is equal to 1window being either of fixed size or8</t> <t hangText="ADU:"> Application Data Unit</t> <t hangText="ADUI:"> Application Data Unit Information (includesvariable size (a.k.a., an elastic window). Repair symbols are generated on-the-fly, by computing a random linear combination of theF, L and padding fieldssource symbols present inaddition totheADU)</t> <t hangText="E:"> size of ancurrent encodingsymbol (i.e., source or repair symbol), assumed fixed (in bytes)</t> <t hangText="br_in:"> transmission bitrate at the input ofwindow, and passed to theFECFRAME sender, assumed fixed (in bits/s)</t>transport layer. </t> <thangText="br_out:"> transmission bitrate at the output ofpn="section-1.2-3"> At theFECFRAME sender, assumed fixed (in bits/s)</t> <t hangText="max_lat:"> maximum FEC-related latency within FECFRAME (a decimal number expressed in seconds)</t> <t hangText="cr:"> RLC coding rate, ratio betweenreceiver, a linear system is managed from thetotal numberset of received sourcesymbolsandthe total number of source plusrepairsymbols</t> <!-- <t hangText="plr:"> packet loss rate during packet communications</t> --> <t hangText="ew_size:"> encoding window current size at a sender (in symbols)</t> <t hangText="ew_max_size:"> encoding window maximum size at a sender (in symbols)</t> <t hangText="dw_max_size:"> decoding window maximum size at a receiver (in symbols)</t> <t hangText="ls_max_size:"> linear system maximum size (or width) at a receiver (in symbols)</t> <t hangText="WSR:"> window size ratio parameter used to derive ew_max_size (encoder) and ls_max_size (decoder).</t> <t hangText="PRNG:"> pseudo-random number generator</t> <t hangText="TinyMT32:"> PRNG used in this specification.</t> <t hangText="DT:"> coding coefficients density threshold, an integer between 0packets. New variables (representing source symbols) and15 (inclusive) the controlsequations (representing thefraction of coefficients thatlinear combination carried by each repair symbol received) arenon zero</t> </list></t> </section> <!-- =========================================================================================== --> <section anchor="CommonProcedures" title="Common Procedures"> <!-- ================ --> <t> This section introducesadded upon receiving new packets. Variables and theprocedures thatequations they areused by these FEC Schemes. </t> <section anchor="CommonProc_rlcParameters" title="Codec Parameters"> <!-- ====================== --> <t> A codec implementing the Sliding Window RLC FEC Scheme relies on several parameters: <list style="hanging"> <t hangText="Maximum FEC-related latency budget, max_lat (a decimal number expressedinvolved inseconds)are removed when they are too old withreal-time flows:"> a source ADU flow can have real-time constraints, and therefore any FECFRAME related operation should take place within therespect to their validity periodof each ADU (<xref target="decodingBeyondMaxLatency"/> describes an exception to this rule). When there are multiple flows with different real-time constraints, we consider the most stringent constraints (see <xref target="RFC6363"/>, Section 10.2, item 6, for recommendations when several flows(real-time constraints). Lost source symbols areglobally protected). The maximum FEC-related latency budget, max_lat, accounts for all sources of latency added by FEC encoding (at a sender) and FEC decodingthen recovered thanks to this linear system whenever its rank permits to solve it (ata receiver). Other sources of latency (e.g., added by network communications) are outleast partially). </t> <t pn="section-1.2-4"> The protection ofscope and must be considered separately (said differently, they have already been deducted from max_lat). max_lat canany multicast/broadcast session needs to beregarded asdimensioned by considering thelatency budget permitted for all FEC-related operations.worst communication conditions one wants to support. This isan input parameter that enablesalso true with RLC (more generally, any sliding window) code. However, the receivers experiencing aFECFRAME sendergood toderive other internal parameters (see <xref target="possible_param_derivation"/>); </t> <t hangText="Encoding window current (resp. maximum) size, ew_size (resp. ew_max_size) (in symbols):"> at a FECFRAME sender, during FEC encoding,medium communication quality will observe arepair symbolreduced FEC-related latency compared to block codes <xref target="Roca17" format="default" sectionFormat="of" derivedContent="Roca17"/> since an isolated lost source packet iscomputed as a linear combination ofquickly recovered with theew_sizefollowing repair packet. On the opposite, with a block code, recovering an isolated lost sourcesymbols present inpacket always requires waiting for theencoding window. The ew_max_size isfirst repair packet to arrive after themaximum sizeend ofthis window, while ew_size is the current size. For example, inthecommon case at session start, upon receiving new source ADUs, the ew_size progressively increases until it reaches its maximum value, ew_max_size. We have: <list style="none"> <t> 0 < ew_size ≤ ew_max_size </t> </list></t> <t hangText="Decoding window maximum size, dw_max_size (in symbols):"> atblock. Additionally, under certain situations (e.g., with a limited FEC-related latency budget and with constant bitrate transmissions after FECFRAMEreceiver, dw_max_size isencoding), Sliding Window Codes can more efficiently achieve a target transmission quality (e.g., measured by themaximum number of received or lost source symbols that are still within their latency budget;residual loss after FEC decoding) by sending fewer repair packets (i.e., higher code rate) than block codes. </t> </section> <section anchor="intro_low_tx_overhead" numbered="true" toc="include" removeInRFC="false" pn="section-1.3"> <name slugifiedName="name-small-transmission-overhead">Small Transmission Overheads with the Sliding Window RLC FEC Scheme</name> <thangText="Linear system maximum size, ls_max_size (in symbols):"> at a FECFRAME receiver,pn="section-1.3-1"> The Sliding Window RLC FEC scheme is designed to limit thelinear system maximum size, ls_max_size,packet header overhead. The main requirement is that each repair packet header must enable a receiver to reconstruct themaximum numberset ofreceived or lostsource symbolsinplus thelinear system (i.e.,associated coefficients used during thevariables). It SHOULD NOT be smaller than dw_max_size since it would mean that, even after receiving a sufficient number of FEC Repair Packets, a lost ADU may not be recovered just becauseencoding process. In order to minimize packet overhead, theassociatedset of source symbolshave been prematurely removed fromin thelinear system, whichencoding window as well as the set of coefficients over GF(2<sup>m</sup>) (where m isusually counter-productive. On1 or 8, depending on theopposite,FEC scheme) used in the linearsystem MAY grow beyond the dw_max_size (<xref target="decodingBeyondMaxLatency"/>); <!-- with old source symbols keptcombination are not individually listed in thelinear system whereas their associated ADUs timed-out -->repair packet header. Instead, each FEC Repair Packet header contains: </t><t hangText="Symbol size, E (in bytes):"> the E parameter determines<ul spacing="normal" bare="false" empty="false" pn="section-1.3-2"> <li pn="section-1.3-2.1">the Encoding Symbol Identifier (ESI) of the first sourceand repairsymbolsizes (necessarily equal). This is an input parameter that enables a FECFRAME sender to derive other internal parameters,in the encoding window asexplained below. An implementation atwell as the number of symbols (since this number may vary with asender MUST fixvariable size, elastic window). These two pieces of information enable each receiver to reconstruct theE parameter and MUST communicate it as partset of source symbols considered during encoding, theFEC Scheme-Specific Informationonly constraint being that there cannot be any gap;</li> <li pn="section-1.3-2.2">the seed and density threshold parameters used by a coding coefficients generation function (<xreftarget="ArbitraryFlows_fssi"/>). </t> <t hangText="Code rate, cr:"> The code rate parameter determines the amounttarget="CommonProc_coef_generation_func" format="default" sectionFormat="of" derivedContent="Section 3.6"/>). These two pieces ofredundancy addedinformation enable each receiver to generate theflow. More precisely the cr issame set of coding coefficients over GF(2<sup>m</sup>) as theratio betweensender;</li> </ul> <t pn="section-1.3-3"> Therefore, no matter thetotalnumber of source symbolsandpresent in thetotal numberencoding window, each FEC Repair Packet features a fixed 64-bit long header, called Repair FEC Payload ID (<xref target="fig_repair_fpi" format="default" sectionFormat="of" derivedContent="Figure 8"/>). Similarly, each FEC Source Packet features a fixed 32-bit long trailer, called Explicit Source FEC Payload ID (<xref target="fig_src_fpi" format="default" sectionFormat="of" derivedContent="Figure 6"/>), that contains the ESI of the first sourceplus repair symbols andsymbol (<xref target="CommonProc_adui_creation" format="default" sectionFormat="of" derivedContent="Section 3.2"/>). </t> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-1.4"> <name slugifiedName="name-document-organization">Document Organization</name> <t pn="section-1.4-1"> This fully-specified FEC scheme follows the structure required bydefinition: 0 < cr ≤ 1.<xref target="RFC6363" format="default" sectionFormat="comma" section="5.6" derivedLink="https://rfc-editor.org/rfc/rfc6363#section-5.6" derivedContent="RFC6363"/> ("FEC Scheme Requirements"), namely: </t> <ol type="1" start="3" spacing="normal" pn="section-1.4-2"> <li pn="section-1.4-2.1" derivedCounter="3.">Procedures: Thisis an input parameter that enables a FECFRAME sendersection describes procedures specific toderive other internal parameters, as explained below. However, there is no need to communicatethis FEC scheme, namely: RLC parameters derivation, ADUI and source symbols mapping, pseudorandom number generator, and coding coefficients generation function;</li> <li pn="section-1.4-2.2" derivedCounter="4.">Formats and Codes: This section defines thecr parameter per see (it's not requiredSource FEC Payload ID and Repair FEC Payload ID formats, carrying the signaling information associated toprocess aeach source or repairsymbol at a receiver). Thissymbol. It also defines the FEC Framework Configuration Information (FFCI) carrying signaling information for the session;</li> <li pn="section-1.4-2.3" derivedCounter="5.">FEC Code Specification: Finally this section provides the coderate parameter canspecification.</li> </ol> </section> </section> <section anchor="definitionsAndAbbreviations" numbered="true" toc="include" removeInRFC="false" pn="section-2"> <name slugifiedName="name-definitions-and-abbreviatio">Definitions and Abbreviations</name> <t pn="section-2-1"> The key words "<bcp14>MUST</bcp14>", "<bcp14>MUST NOT</bcp14>", "<bcp14>REQUIRED</bcp14>", "<bcp14>SHALL</bcp14>", "<bcp14>SHALL NOT</bcp14>", "<bcp14>SHOULD</bcp14>", "<bcp14>SHOULD NOT</bcp14>", "<bcp14>RECOMMENDED</bcp14>", "<bcp14>NOT RECOMMENDED</bcp14>", "<bcp14>MAY</bcp14>", and "<bcp14>OPTIONAL</bcp14>" in this document are to bestatic. However,interpreted as described inspecific use-cases (e.g., with unicast transmissionsBCP 14 <xref target="RFC2119" format="default" sectionFormat="of" derivedContent="RFC2119"/> <xref target="RFC8174" format="default" sectionFormat="of" derivedContent="RFC8174"/> when, and only when, they appear inpresenceall capitals, as shown here. </t> <t pn="section-2-2">This document uses the following definitions and abbreviations: </t> <dl newline="false" spacing="normal" pn="section-2-3"> <dt pn="section-2-3.1">a<sup>b</sup></dt> <dd pn="section-2-3.2"> a to the power of b</dd> <dt pn="section-2-3.3">GF(q)</dt> <dd pn="section-2-3.4"> denotes afeedback mechanismfinite field (also known as the Galois Field) with q elements. We assume thatestimatesq = 2<sup>m</sup> in this document</dd> <dt pn="section-2-3.5">m</dt> <dd pn="section-2-3.6"> defines thecommunication quality, out of scopelength ofFECFRAME),thecode rate may be adjusted dynamically. </t> </list> </t> <t> <xref target="possible_param_derivation"/> proposes non normative techniqueselements in the finite field, in bits. In this document, m is equal toderive those parameters, depending on1 or 8</dd> <dt pn="section-2-3.7">ADU:</dt> <dd pn="section-2-3.8"> Application Data Unit</dd> <dt pn="section-2-3.9">ADUI:</dt> <dd pn="section-2-3.10"> Application Data Unit Information (includes theuse-case specificities. </t> </section> <section anchor="CommonProc_adui_creation" title="ADU, ADUIF, L andSource Symbols Mappings"> <!-- ==================================== --> <t> At a sender, an ADU coming from the application is not directly mappedpadding fields in addition tosource symbols. When multiple source flows (e.g., media streams) are mapped ontothesame FECFRAME instance, each flow is assigned its own Flow ID value (see below). This Flow ID is then prepended to each ADU before FEC encoding. This way, FEC decodingADU)</dd> <dt pn="section-2-3.11">E:</dt> <dd pn="section-2-3.12"> size of an encoding symbol (i.e., source or repair symbol), assumed fixed (in bytes)</dd> <dt pn="section-2-3.13">br_in:</dt> <dd pn="section-2-3.14"> transmission bitrate ata receiver also recovers this Flow ID andtherecovered ADU can be assigned toinput of theright source flow (note thatFECFRAME sender, assumed fixed (in bits/s)</dd> <dt pn="section-2-3.15">br_out:</dt> <dd pn="section-2-3.16"> transmission bitrate at the5-tuple used to identifyoutput of therightFECFRAME sender, assumed fixed (in bits/s)</dd> <dt pn="section-2-3.17">max_lat:</dt> <dd pn="section-2-3.18"> maximum FEC-related latency within FECFRAME (a decimal number expressed in seconds)</dd> <dt pn="section-2-3.19">cr:</dt> <dd pn="section-2-3.20"> RLC coding rate, ratio between the total number of sourceflowsymbols and the total number of source plus repair symbols</dd> <dt pn="section-2-3.21">ew_size:</dt> <dd pn="section-2-3.22"> encoding window current size at areceived ADU is absent withsender (in symbols)</dd> <dt pn="section-2-3.23">ew_max_size:</dt> <dd pn="section-2-3.24"> encoding window maximum size at arecovered ADU since it is not FEC protected). <!-- Indeed,sender (in symbols)</dd> <dt pn="section-2-3.25">dw_max_size:</dt> <dd pn="section-2-3.26"> decoding window maximum size at alost ADU recoveredreceiver (in symbols)</dd> <dt pn="section-2-3.27">ls_max_size:</dt> <dd pn="section-2-3.28"> linear system maximum size (or width) at a receivermust contain enough information to be assigned(in symbols)</dd> <dt pn="section-2-3.29">WSR:</dt> <dd pn="section-2-3.30"> window size ratio parameter used tothe right application flow (UDP port numbersderive ew_max_size (encoder) andIP addresses cannot bels_max_size (decoder).</dd> <dt pn="section-2-3.31">PRNG:</dt> <dd pn="section-2-3.32"> pseudorandom number generator</dd> <dt pn="section-2-3.33">TinyMT32:</dt> <dd pn="section-2-3.34"> PRNG usedtoin this specification.</dd> <dt pn="section-2-3.35">DT:</dt> <dd pn="section-2-3.36"> coding coefficients density threshold, an integer between 0 and 15 (inclusive) the controls the fraction of coefficients thatpurpose as theyarenot protected by FEC encoding).nonzero</dd> </dl> </section> <section anchor="CommonProcedures" numbered="true" toc="include" removeInRFC="false" pn="section-3"> <name slugifiedName="name-common-procedures">Common Procedures</name> <t pn="section-3-1"> Thisrequires addingsection introduces theflow identifier to each ADU before doingprocedures that are used by these FECencoding. -->schemes. </t><t> Additionally, since ADUs are of variable size, padding is needed so that each ADU (with its flow identifier) contribute to an integral number of source symbols. This requires adding<section anchor="CommonProc_rlcParameters" numbered="true" toc="include" removeInRFC="false" pn="section-3.1"> <name slugifiedName="name-codec-parameters">Codec Parameters</name> <t pn="section-3.1-1"> A codec implementing theoriginal ADU length to each ADU before doingSliding Window RLC FECencoding. Because of these requirements, an intermediate format, the ADUI, or ADU Information, is considered <xref target="RFC6363"/>.scheme relies on several parameters: </t><t> For each incoming ADU, an ADUI MUST created as follows. First of all, 3 bytes are prepended (<xref target="fig_adui_creation"/>): <list style="hanging"> <t hangText="Flow ID (F) (8-bit field):"> this unsigned byte contains the integer identifier associated to the<dl newline="true" spacing="normal" pn="section-3.1-2"> <dt pn="section-3.1-2.1">Maximum FEC-related latency budget, max_lat (a decimal number expressed in seconds) with real-time flows:</dt> <dd pn="section-3.1-2.2"> a source ADU flowto which this ADU belongs. It is assumed that a single byte is sufficient, which implies that no more than 256 flows will be protected by a singlecan have real-time constraints, and therefore any FECFRAMEsession instance.</t> <t hangText="Length (L) (16-bit field):"> this unsigned integer containsrelated operation should take place within thelengthvalidity period ofthis ADU, in network byte order (i.e., big endian). This length is for theeach ADUitself and does not include the F, L, or Pad fields. </t> </list> </t> <t> Then, zero padding is added(<xref target="decodingBeyondMaxLatency" format="default" sectionFormat="of" derivedContent="Appendix D"/> describes an exception tothe ADU if needed: <list style="hanging"> <t hangText="Padding (Pad) (variable size field):">thisfield contains zero padding to align the F, L, ADU and padding up to a size that isrule). When there are multipleof E bytes (i.e., the source and repair symbol length). </t> </list> The data unit resulting from the ADU and the F, L, and Pad fields is called ADUI. Since ADUs can haveflows with differentsizes, this is alsoreal-time constraints, we consider thecasemost stringent constraints (see item 6 in <xref target="RFC6363" format="default" sectionFormat="of" section="10.2" derivedLink="https://rfc-editor.org/rfc/rfc6363#section-10.2" derivedContent="RFC6363"/>, forADUIs. However, an ADUI always contributes to an integral number of source symbols. </t> <figure anchor="fig_adui_creation" title="ADUI Creation example (here 3 source symbolsrecommendations when several flows arecreatedglobally protected). The maximum FEC-related latency budget, max_lat, accounts forthis ADUI)."> <artwork><![CDATA[ symbol length, E E E < ------------------ >< ------------------ >< ------------------ > +-+--+---------------------------------------------+-------------+ |F| L| ADU | Pad | +-+--+---------------------------------------------+-------------+]]></artwork> </figure> <t> Note that neither the initial 3 bytes nor the optional padding are sent over the network. However, they are considered duringall sources of latency added by FECencoding, and a receiver who lostencoding (at acertainsender) and FECSource Packetdecoding (at a receiver). Other sources of latency (e.g.,the UDP datagram containing this FEC Source Packet when UDP is usedadded by network communications) are out of scope and must be considered separately (said differently, they have already been deducted from max_lat). max_lat can be regarded as thetransport protocol) will be ablelatency budget permitted for all FEC-related operations. This is an input parameter that enables a FECFRAME sender torecover the ADUI ifderive other internal parameters (see <xref target="possible_param_derivation" format="default" sectionFormat="of" derivedContent="Appendix C"/>); </dd> <dt pn="section-3.1-2.3">Encoding window current (resp. maximum) size, ew_size (resp. ew_max_size) (in symbols):</dt> <dd pn="section-3.1-2.4"> <t pn="section-3.1-2.4.1"> at a FECFRAME sender, during FECdecoding succeeds. Thanks to the initial 3 bytes, this receiver will get ridencoding, a repair symbol is computed as a linear combination of thepadding (if any) and identify the corresponding ADU flow. </t> </section> <section anchor="encodingWindowManagement" title="Encoding Window Management"> <!-- ====================== --> <t> Sourceew_size source symbolsand the corresponding ADUs are removed frompresent in the encodingwindow: <list style="symbols"> <t> whenwindow. The ew_max_size is thesliding encoding window has reached itsmaximumsize, ew_max_size. In that case the oldest symbol MUST be removed before adding a new symbol, so that the current encoding windowsizealways remains inferior or equal to the maximum size: ew_size ≤ ew_max_size;</t> <t> when an ADU has reached its maximum validity duration in caseofa real-time flow. Whenthishappens, all source symbols corresponding to the ADUI that expired SHOULD be removed from the encoding window; </t> </list> </t> <t> Source symbols are added to the sliding encoding window each time a new ADU arrives, oncewindow, while ew_size is theADU-to-source symbols mapping has been performed (<xref target="CommonProc_adui_creation"/>). Thecurrentsize ofsize. For example, in theencoding window, ew_size, is updated after addingcommon case at session start, upon receiving new sourcesymbols. This process may require to remove old source symbols so that:ADUs, the ew_size≤progressively increases until it reaches its maximum value, ew_max_size. We have: </t><t> Note that<ul spacing="normal" empty="true" bare="false" pn="section-3.1-2.4.2"> <li pn="section-3.1-2.4.2.1"> 0 < ew_size <= ew_max_size </li> </ul> </dd> <dt pn="section-3.1-2.5">Decoding window maximum size, dw_max_size (in symbols):</dt> <dd pn="section-3.1-2.6"> at aFEC codec may feature practical limits inFECFRAME receiver, dw_max_size is the maximum number of received or lost source symbolsin the encoding window (e.g., for computational complexity reasons). This factor may further limit the ew_max_size value, in addition to the maximum FEC-related latency budget (<xref target="CommonProc_rlcParameters"/>). </t> <!-- <t> Limitations MAY existthatimpact the encoding window management. For instance: <list style="symbols"> <t>are still within their latency budget; </dd> <dt pn="section-3.1-2.7">Linear system maximum size, ls_max_size (in symbols):</dt> <dd pn="section-3.1-2.8"> at a FECFRAME receiver, theFEC Framework level: the source flows can have real-time constraints that limitlinear system maximum size, ls_max_size, is the maximum number ofADUsreceived or lost source symbols in theencoding window;</t> <t> atlinear system (i.e., the variables). It <bcp14>SHOULD NOT</bcp14> be smaller than dw_max_size since it would mean that, even after receiving a sufficient number of FECScheme level: thereRepair Packets, a lost ADU may not betheoretical or practical limitations (e.g.,recovered just becauseof computational complexity aspect or field size limits inthesignaling headers)associated source symbols have been prematurely removed from the linear system, which is usually counter-productive. On the opposite, the linear system <bcp14>MAY</bcp14> grow beyond the dw_max_size (<xref target="decodingBeyondMaxLatency" format="default" sectionFormat="of" derivedContent="Appendix D"/>); </dd> <dt pn="section-3.1-2.9">Symbol size, E (in bytes):</dt> <dd pn="section-3.1-2.10"> the E parameter determines the source and repair symbol sizes (necessarily equal). This is an input parameter thatlimitenables a FECFRAME sender to derive other internal parameters, as explained below. An implementation at a sender <bcp14>MUST</bcp14> fix thenumberE parameter and <bcp14>MUST</bcp14> communicate it as part ofADUs intheencoding window.</t> </list>FEC Scheme-Specific Information (<xref target="ArbitraryFlows_fssi" format="default" sectionFormat="of" derivedContent="Section 4.1.1.2"/>). </dd> <dt pn="section-3.1-2.11">Code rate, cr:</dt> <dd pn="section-3.1-2.12"> Themost stringent limitation definescode rate parameter determines themaximum encoding window size, either in termsamount of redundancy added to the flow. More precisely the cr is the ratio between the total number of source symbolsorand the total number ofADUs, whichever applies. </t> --> </section> <section anchor="CommonProc_esi" title="Source Symbol Identification"> <!-- ================ --> <t> Each source symbol is identified by an Encoding Symbol ID (ESI), an unsigned integer. The ESI ofsource plus repair symbolsMUST start with value 0 for the first source symbolandMUST be managed sequentially. Wrapping to zero happens after reaching the maximum value made possiblebythe ESI field size (this maximum valuedefinition: 0 < cr <= 1. This isFEC Scheme dependant, for instance, 2^32-1 with FEC Schemes XXX and YYY). </t> <t> No such consideration appliesan input parameter that enables a FECFRAME sender torepair symbols. </t> </section> <section anchor="CommonProc_prng" title="Pseudo-Random Number Generator (PRNG)"> <!-- ====================== --> <t> In orderderive other internal parameters, as explained below. However, there is no need tocompute coding coefficients (see <xref target="CommonProc_coef_generation_func"/>), the RLC FEC Schemes rely oncommunicate theTinyMT32 PRNG defined in <xref target="tinymt32"/> with two additional functions defined in this section. </t> <t> This PRNG MUST first be initialized with a 32-bit unsigned integer, used ascr parameter per see (it's not required to process aseed, with: <list style="empty"> <t>void tinymt32_init (tinymt32_t * s, uint32_t seed);</t> </list> With the FEC Schemes definedrepair symbol at a receiver). This code rate parameter can be static. However, inthis document, the seed isspecific use-cases (e.g., with unicast transmissions inpractice restricted topresence of avalue between 0 and 0xFFFF inclusive (notefeedback mechanism thatthis PRNG accepts a seed value equal to 0), since this isestimates theRepair_Key 16-bit field valuecommunication quality, out of scope of FECFRAME), theRepair FEC Payload ID (<xref target="ArbitraryFlows_repair_fpi"/>). In practice, howcode rate may be adjusted dynamically. </dd> </dl> <t pn="section-3.1-3"><xref target="possible_param_derivation" format="default" sectionFormat="of" derivedContent="Appendix C"/> proposes non-normative techniques tomanagederive those parameters, depending on theseeduse-case specificities. </t> </section> <section anchor="CommonProc_adui_creation" numbered="true" toc="include" removeInRFC="false" pn="section-3.2"> <name slugifiedName="name-adu-adui-and-source-symbols">ADU, ADUI, andRepair_Key values (both are equal)Source Symbols Mappings</name> <t pn="section-3.2-1"> At a sender, an ADU coming from the application isleftnot directly mapped to source symbols. When multiple source flows (e.g., media streams) are mapped onto theimplementer, using a monotonically increasing counter being one possibility (<xref target="ArbitraryFlows_FECCodeSpecification_encoding"/>). In additionsame FECFRAME instance, each flow is assigned its own Flow ID value (see below). This Flow ID is then prepended tothe seed, this function takes as parametereach ADU before FEC encoding. This way, FEC decoding at apointerreceiver also recovers this Flow ID and the recovered ADU can be assigned toan instance of a tinymt32_t structurethe right source flow (note thatisthe 5-tuple used tokeepidentify theinternal stateright source flow ofthe PRNG. </t> <t> Then, each timeanew pseudo-random integer between 0 and 15 inclusive (4-bit pseudo-random integer)received ADU isneeded, the following function is used: <list style="empty"> <t>uint32_t tinymt32_rand16 (tinymt32_t * s);</t> </list> This function takes as parameter a pointer to the same tinymt32_t structure (that is left unchanged between successive calls to the function). </t> <t> Similarly, each timeabsent with anew pseudo-random integer between 0 and 255 inclusive (8-bit pseudo-random integer) is needed, the following functionrecovered ADU since it isused: <list style="empty"> <t>uint32_t tinymt32_rand256 (tinymt32_t * s);</t> </list>not FEC protected). </t><t> These two functions keep respectively the 4 or 8 less significant bits<t pn="section-3.2-2"> Additionally, since ADUs are ofthe 32-bit pseudo-randomvariable size, padding is needed so that each ADU (with its flow identifier) contribute to an integral numbergenerated by the tinymt32_generate_uint32() functionof<xref target="tinymt32"/>.source symbols. Thisis done by computingrequires adding theresultoriginal ADU length to each ADU before doing FEC encoding. Because ofa binary AND between the tinymt32_generate_uint32() output and respectivelythese requirements, an intermediate format, the0xFADUI, or0xFF constants, using 32-bitADU Information, is considered <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/>. </t> <t pn="section-3.2-3"> For each incoming ADU, an ADUI <bcp14>MUST</bcp14> be created as follows. First of all, 3 bytes are prepended (<xref target="fig_adui_creation" format="default" sectionFormat="of" derivedContent="Figure 1"/>): </t> <dl newline="false" spacing="normal" pn="section-3.2-4"> <dt pn="section-3.2-4.1">Flow ID (F) (8-bit field):</dt> <dd pn="section-3.2-4.2"> this unsigned byte contains the integeroperations. <xref target="fig_tinymt32_mapping"/> shows a possible implementation. Thisidentifier associated to the source ADU flow to which this ADU belongs. It is assumed that aC language implementation, written for C99 <xref target="C99"/>. Test results discussed in <xref target="annex_assessing_prng"/> showsingle byte is sufficient, which implies that no more than 256 flows will be protected by a single FECFRAME session instance.</dd> <dt pn="section-3.2-4.3">Length (L) (16-bit field):</dt> <dd pn="section-3.2-4.4"> thissimple technique, applied tounsigned integer contains the length of thisPRNG, isADU, inline withnetwork byte order (i.e., big endian). This length is for theRLC FEC Schemes needs. </t> <figure anchor="fig_tinymt32_mapping" title="4-bit and 8-bit mapping functions for TinyMT32"> <artwork><![CDATA[ <CODE BEGINS> /** * This function outputs a pseudo-random integer in [0 .. 15] range. * * @param s pointer to tinymt internal state. * @return unsigned integer between 0ADU itself and15 inclusive. */ uint32_t tinymt32_rand16(tinymt32_t *s) { return (tinymt32_generate_uint32(s) & 0xF); } /** * This function outputs a pseudo-random integer in [0 .. 255] range. * * @param s pointerdoes not include the F, L, or Pad fields. </dd> </dl> <t pn="section-3.2-5"> Then, zero padding is added totinymt internal state. * @return unsigned integer between 0 and 255 inclusive. */ uint32_t tinymt32_rand256(tinymt32_t *s) { return (tinymt32_generate_uint32(s) & 0xFF); } <CODE ENDS> ]]></artwork> </figure> <t> Any implementation ofthe ADU if needed: </t> <dl newline="false" spacing="normal" pn="section-3.2-6"> <dt pn="section-3.2-6.1">Padding (Pad) (variable size field):</dt> <dd pn="section-3.2-6.2"> thisPRNG MUST havefield contains zero padding to align thesame output asF, L, ADU and padding up to a size thatprovided by the reference implementationis multiple of<xref target="tinymt32"/>. In order to increase the compliancy confidence, three criteria are proposed: the one described in <xref target="tinymt32"/> (forE bytes (i.e., theTinyMT32 32-bit unsigned integer generator),source and repair symbol length). </dd> </dl> <t pn="section-3.2-7"> The data unit resulting from thetwo others detailed in <xref target="annex_tinymt32_validation"/> (for the mapping to 4-bitADU and8-bit intervals). Because of the waythemapping functions work, itF, L, and Pad fields isunlikely that an implementation that fulfillscalled ADUI. Since ADUs can have different sizes, this is also thefirst criterion failscase for ADUIs. However, an ADUI always contributes tofulfill the two others.an integral number of source symbols. </t></section> <section anchor="CommonProc_coef_generation_func" title="Coding Coefficients Generation Function"> <!-- ====================== --> <t> The coding coefficients, used during the encoding process, are generated at the RLC encoder by the generate_coding_coefficients() function each time a new repair<figure anchor="fig_adui_creation" align="left" suppress-title="false" pn="figure-1"> <name slugifiedName="name-adui-creation-example-resul">ADUI Creation Example, Resulting in Three Source Symbols</name> <artwork name="" type="" align="left" alt="" pn="section-3.2-8.1"> symbolneeds to be produced. The fraction of coefficientslength, E E E < ------------------ >< ------------------ >< ------------------ > +-+--+---------------------------------------------+-------------+ |F| L| ADU | Pad | +-+--+---------------------------------------------+-------------+ </artwork> </figure> <t pn="section-3.2-9"> Note thatare non zero (i.e.,neither thedensity) is controlled byinitial 3 bytes nor theDT (Density Threshold) parameter. DT has values between 0 (the minimum value) and 15 (the maximum value), andoptional padding are sent over theaverage probability of havingnetwork. However, they are considered during FEC encoding, and anon zero coefficient equals (DT + 1) / 16. In particular,receiver that lost a certain FEC Source Packet (e.g., the UDP datagram containing this FEC Source Packet whenDT equals 15UDP is used as thefunction guaranties that all coefficients are non zero (i.e., maximum density). </t> <t> These considerations applytransport protocol) will be able toboth the RLC over GF(2) and RLC over GF(2^^8),recover theonly difference beingADUI if FEC decoding succeeds. Thanks to thevalueinitial 3 bytes, this receiver will get rid of them parameter. Withpadding (if any) and identify theRLC over GF(2) FEC Scheme (<xref target="ArbitraryFlows_RLC_GF_2"/>), m is equal to 1. With RLC over GF(2^^8) FEC Scheme (<xref target="ArbitraryFlows_RLC_GF_28"/>), m is equal to 8.corresponding ADU flow. </t><t> <xref target="fig_coef_generation_func"/> shows</section> <section anchor="encodingWindowManagement" numbered="true" toc="include" removeInRFC="false" pn="section-3.3"> <name slugifiedName="name-encoding-window-management">Encoding Window Management</name> <t pn="section-3.3-1"> Source symbols and thereference generate_coding_coefficients() implementation. This is a C language implementation, written for C99 <xref target="C99"/>. </t> <figure anchor="fig_coef_generation_func" title="Coding Coefficients Generation Function Reference Implementation"> <artwork><![CDATA[ <CODE BEGINS> #include <string.h> /* * Fills incorresponding ADUs are removed from thetable of coding coefficients (ofencoding window: </t> <ul spacing="normal" bare="false" empty="false" pn="section-3.3-2"> <li pn="section-3.3-2.1"> when theright size) * provided with the appropriate number of coding coefficients to * use forsliding encoding window has reached its maximum size, ew_max_size. In that case therepairoldest symbolkey provided. * * (in) repair_key key associated to this repair symbol. This * parameter is ignored (useless) if m=1 and dt=15 * (in/out) cc_tab pointer to<bcp14>MUST</bcp14> be removed before adding atable ofnew symbol, so that therightcurrent encoding window size always remains inferior or equal tostore * coding coefficients. All coefficients are * stored as bytes, regardless ofthem parameter, * upon returnmaximum size: ew_size <= ew_max_size;</li> <li pn="section-3.3-2.2"> when an ADU has reached its maximum validity duration in case of a real-time flow. When thisfunction. * (in) cc_nb number of entries in the cc_tab table. This * value is equalhappens, all source symbols corresponding to thecurrent encoding window * size. * (in) dt integer between 0 and 15 (inclusive)ADUI that* controlsexpired <bcp14>SHOULD</bcp14> be removed from thedensity. With value 15, all * coefficientsencoding window; </li> </ul> <t pn="section-3.3-3"> Source symbols areguaranteed to be non zero * (i.e. equaladded to1 with GF(2) and equalthe sliding encoding window each time a new ADU arrives, once the ADU-to-source symbols mapping has been performed (<xref target="CommonProc_adui_creation" format="default" sectionFormat="of" derivedContent="Section 3.2"/>). The current size of the encoding window, ew_size, is updated after adding new source symbols. This process may require to remove old source symbols so that: ew_size <= ew_max_size. </t> <t pn="section-3.3-4"> Note that a* valueFEC codec may feature practical limits in{1,... 255} with GF(2^^8)), otherwise * a fractionthe number ofthem will be 0. * (in) m Finite Field GF(2^^m) parameter. In thissource symbols in the encoding window (e.g., for computational complexity reasons). This factor may further limit the ew_max_size value, in addition to the maximum FEC-related latency budget (<xref target="CommonProc_rlcParameters" format="default" sectionFormat="of" derivedContent="Section 3.1"/>). </t> </section> <section anchor="CommonProc_esi" numbered="true" toc="include" removeInRFC="false" pn="section-3.4"> <name slugifiedName="name-source-symbol-identificatio">Source Symbol Identification</name> <t pn="section-3.4-1"> Each source symbol is identified by an Encoding Symbol ID (ESI), an unsigned integer. The ESI of source symbols <bcp14>MUST</bcp14> start with value 0 for the first source symbol and <bcp14>MUST</bcp14> be managed sequentially. Wrapping to zero happens after reaching the maximum value made possible by the ESI field size (this maximum value is FEC scheme dependent, for instance, 2<sup>32</sup>-1 with FEC schemes 9 and 10). </t> <t pn="section-3.4-2"> No such consideration applies to repair symbols. </t> </section> <section anchor="CommonProc_prng" numbered="true" toc="include" removeInRFC="false" pn="section-3.5"> <name slugifiedName="name-pseudorandom-number-generat">Pseudorandom Number Generator (PRNG)</name> <t pn="section-3.5-1"> In order to compute coding coefficients (see <xref target="CommonProc_coef_generation_func" format="default" sectionFormat="of" derivedContent="Section 3.6"/>), the RLC FEC schemes rely on the TinyMT32 PRNG defined in <xref target="RFC8682" format="default" sectionFormat="of" derivedContent="RFC8682"/> with two additional functions defined in this section. </t> <t pn="section-3.5-2"> This PRNG <bcp14>MUST</bcp14> first be initialized with a 32-bit unsigned integer, used as a seed, with: </t> <sourcecode type="c" markers="false" pn="section-3.5-3"> void tinymt32_init (tinymt32_t * s, uint32_t seed); </sourcecode> <t pn="section-3.5-4"> With the FEC schemes defined in this document, the seed is in practice restricted to a value between 0 and 0xFFFF inclusive (note that this PRNG accepts a seed value equal to 0), since this is the Repair_Key 16-bit field value of the Repair FEC Payload ID (<xref target="ArbitraryFlows_repair_fpi" format="default" sectionFormat="of" derivedContent="Section 4.1.3"/>). In practice, how to manage the seed and Repair_Key values (both are equal) is left to the implementer, using a monotonically increasing counter being one possibility (<xref target="ArbitraryFlows_FECCodeSpecification_encoding" format="default" sectionFormat="of" derivedContent="Section 6.1"/>). In addition to the seed, this function takes as parameter a pointer to an instance of a tinymt32_t structure that is used to keep the internal state of the PRNG. </t> <t pn="section-3.5-5"> Then, each time a new pseudorandom integer between 0 and 15 inclusive (4-bit pseudorandom integer) is needed, the following function is used: </t> <sourcecode type="c" markers="false" pn="section-3.5-6"> uint32_t tinymt32_rand16 (tinymt32_t * s); </sourcecode> <t pn="section-3.5-7"> This function takes as parameter a pointer to the same tinymt32_t structure (that is left unchanged between successive calls to the function). </t> <t pn="section-3.5-8"> Similarly, each time a new pseudorandom integer between 0 and 255 inclusive (8-bit pseudorandom integer) is needed, the following function is used: </t> <sourcecode type="c" markers="false" pn="section-3.5-9"> uint32_t tinymt32_rand256 (tinymt32_t * s); </sourcecode> <t pn="section-3.5-10"> These two functions keep respectively the 4 or 8 less significant bits of the 32-bit pseudorandom number generated by the tinymt32_generate_uint32() function of <xref target="RFC8682" format="default" sectionFormat="of" derivedContent="RFC8682"/>. This is done by computing the result of a binary AND between the tinymt32_generate_uint32() output and respectively the 0xF or 0xFF constants, using 32-bit unsigned integer operations. <xref target="fig_tinymt32_mapping" format="default" sectionFormat="of" derivedContent="Figure 2"/> shows a possible implementation. This is a C language implementation, written for C99 <xref target="C99" format="default" sectionFormat="of" derivedContent="C99"/>. Test results discussed in <xref target="annex_assessing_prng" format="default" sectionFormat="of" derivedContent="Appendix B"/> show that this simple technique, applied to this PRNG, is in line with the RLC FEC schemes needs. </t> <figure anchor="fig_tinymt32_mapping" align="left" suppress-title="false" pn="figure-2"> <name slugifiedName="name-4-bit-and-8-bit-mapping-fun">4-bit and 8-bit Mapping Functions for TinyMT32</name> <sourcecode name="" type="c" markers="true" pn="section-3.5-11.1"> /** * This function outputs a pseudorandom integer in [0 .. 15] range. * * @param s pointer to tinymt internal state. * @return unsigned integer between 0 and 15 inclusive. */ uint32_t tinymt32_rand16(tinymt32_t *s) { return (tinymt32_generate_uint32(s) & 0xF); } /** * This function outputs a pseudorandom integer in [0 .. 255] range. * * @param s pointer to tinymt internal state. * @return unsigned integer between 0 and 255 inclusive. */ uint32_t tinymt32_rand256(tinymt32_t *s) { return (tinymt32_generate_uint32(s) & 0xFF); } </sourcecode> </figure> <t pn="section-3.5-12"> Any implementation of this PRNG <bcp14>MUST</bcp14> have the same output as that provided by the reference implementation of <xref target="RFC8682" format="default" sectionFormat="of" derivedContent="RFC8682"/>. In order to increase the compliance confidence, three criteria are proposed: the one described in <xref target="RFC8682" format="default" sectionFormat="of" derivedContent="RFC8682"/> (for the TinyMT32 32-bit unsigned integer generator), and the two others detailed in <xref target="annex_tinymt32_validation" format="default" sectionFormat="of" derivedContent="Appendix A"/> (for the mapping to 4-bit and 8-bit intervals). Because of the way the mapping functions work, it is unlikely that an implementation that fulfills the first criterion fails to fulfill the two others. </t> </section> <section anchor="CommonProc_coef_generation_func" numbered="true" toc="include" removeInRFC="false" pn="section-3.6"> <name slugifiedName="name-coding-coefficients-generat">Coding Coefficients Generation Function</name> <t pn="section-3.6-1"> The coding coefficients used during the encoding process are generated at the RLC encoder by the generate_coding_coefficients() function each time a new repair symbol needs to be produced. The fraction of coefficients that are nonzero (i.e., the density) is controlled by the DT (Density Threshold) parameter. DT has values between 0 (the minimum value) and 15 (the maximum value), and the average probability of having a nonzero coefficient equals (DT + 1) / 16. In particular, when DT equals 15 the function guaranties that all coefficients are nonzero (i.e., maximum density). </t> <t pn="section-3.6-2"> These considerations apply to both the RLC over GF(2) and RLC over GF(2<sup>8</sup>), the only difference being the value of the m parameter. With the RLC over GF(2) FEC scheme (<xref target="ArbitraryFlows_RLC_GF_2" format="default" sectionFormat="of" derivedContent="Section 5"/>), m is equal to 1. With RLC over GF(2<sup>8</sup>) FEC scheme (<xref target="ArbitraryFlows_RLC_GF_28" format="default" sectionFormat="of" derivedContent="Section 4"/>), m is equal to 8. </t> <t pn="section-3.6-3"><xref target="fig_coef_generation_func" format="default" sectionFormat="of" derivedContent="Figure 3"/> shows the reference generate_coding_coefficients() implementation. This is a C language implementation, written for C99 <xref target="C99" format="default" sectionFormat="of" derivedContent="C99"/>. </t> <figure anchor="fig_coef_generation_func" align="left" suppress-title="false" pn="figure-3"> <name slugifiedName="name-reference-implementation-of">Reference Implementation of the Coding Coefficients Generation Function</name> <sourcecode name="" type="c" markers="true" pn="section-3.6-4.1"> #include <string.h> /* * Fills in the table of coding coefficients (of the right size) * provided with the appropriate number of coding coefficients to * use for the repair symbol key provided. * * (in) repair_key key associated to this repair symbol. This * parameter is ignored (useless) if m=1 and dt=15 * (in/out) cc_tab pointer to a table of the right size to store * coding coefficients. All coefficients are * stored as bytes, regardless of the m parameter, * upon return of this function. * (in) cc_nb number of entries in the cc_tab table. This * value is equal to the current encoding window * size. * (in) dt integer between 0 and 15 (inclusive) that * controls the density. With value 15, all * coefficients are guaranteed to be nonzero * (i.e., equal to 1 with GF(2) and equal to a * value in {1,... 255} with GF(2^^8)), otherwise * a fraction of them will be 0. * (in) m Finite Field GF(2^^m) parameter. In this * document only values 1 and 8 are considered. * (out) returns 0 in case of success, an error code * different than 0 otherwise. */ int generate_coding_coefficients (uint16_t repair_key, uint8_t* cc_tab, uint16_t cc_nb, uint8_t dt, uint8_t m) { uint32_t i; tinymt32_t s; /* PRNG internal state */ if (dt>> 15) { return -1; /* error, bad dt parameter */ } switch (m) { case 1: if (dt == 15) { /* all coefficients are 1 */ memset(cc_tab, 1, cc_nb); } else { /* here coefficients are either 0 or 1 */tinymt32_init(&s,tinymt32_init(&s, repair_key); for (i = 0 ; i<< cc_nb ; i++) { cc_tab[i] =(tinymt32_rand16(&s) <=(tinymt32_rand16(&s) <= dt) ? 1 : 0; } } break; case 8:tinymt32_init(&s,tinymt32_init(&s, repair_key); if (dt == 15) { /* coefficient 0 is avoided here in order to include * all the source symbols */ for (i = 0 ; i<< cc_nb ; i++) { do { cc_tab[i] = (uint8_t)tinymt32_rand256(&s);tinymt32_rand256(&s); } while (cc_tab[i] == 0); } } else { /* here a certain number of coefficients should be 0 */ for (i = 0 ; i<< cc_nb ; i++) { if(tinymt32_rand16(&s) <=(tinymt32_rand16(&s) <= dt) { do { cc_tab[i] = (uint8_t)tinymt32_rand256(&s);tinymt32_rand256(&s); } while (cc_tab[i] == 0); } else { cc_tab[i] = 0; } } } break; default: return -2; /* error, bad parameter m */ } return 0; /* success */ }<CODE ENDS> ]]></artwork></sourcecode> </figure> </section> <section anchor="CommonProc_gf_specificiation"title="Finite Fields Operations"> <!-- ====================== -->numbered="true" toc="include" removeInRFC="false" pn="section-3.7"> <name slugifiedName="name-finite-field-operations">Finite Field Operations</name> <sectiontitle="Finitenumbered="true" toc="include" removeInRFC="false" pn="section-3.7.1"> <name slugifiedName="name-finite-field-definitions">Finite FieldDefinitions"> <!-- ====================== --> <t>Definitions</name> <t pn="section-3.7.1-1"> The two RLC FECSchemesschemes specified in this document reuse the Finite Fields defined in <xreftarget="RFC5510"/>, section 8.1.target="RFC5510" format="default" section="8.1" sectionFormat="comma" derivedLink="https://rfc-editor.org/rfc/rfc5510#section-8.1" derivedContent="RFC5510"/>. More specifically, the elements of the fieldGF(2^^m)GF(2<sup>m</sup>) are represented by polynomials with binary coefficients (i.e., over GF(2)) and degree lower or equal to m-1. The addition between two elements is defined as the addition of binary polynomials in GF(2), which is equivalent to a bitwise XORoperation on the binary representation of these elements. </t> <t> With GF(2^^8), multiplication between two elementsoperation on the binary representation of these elements. </t> <t pn="section-3.7.1-2"> With GF(2<sup>8</sup>), multiplication between two elements is the multiplication modulo a given irreducible polynomial of degree 8. The following irreducible polynomial is used for GF(2<sup>8</sup>): </t> <ul empty="true" spacing="normal" bare="false" pn="section-3.7.1-3"> <li pn="section-3.7.1-3.1">x<sup>8</sup> + x<sup>4</sup> + x<sup>3</sup> + x<sup>2</sup> + 1 </li> </ul> <t pn="section-3.7.1-4"> With GF(2), multiplication corresponds to a logical AND operation. </t> </section> <section anchor="CommonProc_linear_combination_computation" numbered="true" toc="include" removeInRFC="false" pn="section-3.7.2"> <name slugifiedName="name-linear-combination-of-sourc">Linear Combination of Source Symbol Computation</name> <t pn="section-3.7.2-1"> The two RLC FEC schemes require the computation of a linear combination of source symbols, using the coding coefficients produced by the generate_coding_coefficients() function and stored in the cc_tab[] array. </t> <t pn="section-3.7.2-2"> With the RLC over GF(2<sup>8</sup>) FEC scheme, a linear combination of the ew_size source symbol present in the encoding window, say src_0 to src_ew_size_1, in order to generate a repair symbol, is computed as follows. For each byte of position i in each source and the repair symbol, where i belongs to [0; E-1], compute: </t> <sourcecode type="pseudocode" markers="false" pn="section-3.7.2-3"> repair[i] = cc_tab[0] * src_0[i] XOR cc_tab[1] * src_1[i] XOR ... XOR cc_tab[ew_size - 1] * src_ew_size_1[i] </sourcecode> <t pn="section-3.7.2-4"> where * is the multiplicationmodulo a given irreducible polynomial of degree 8. The following irreducible polynomial isover GF(2<sup>8</sup>). In practice various optimizations need to be usedfor GF(2^^8): <list style="empty"> <t>x^^8 + x^^4 + x^^3 + x^^2 + 1 </t> </list> </t> <t> With GF(2), multiplication correspondsin order toa logical AND operation.make this computation efficient (see in particular <xref target="PGM13" format="default" sectionFormat="of" derivedContent="PGM13"/>). </t></section> <section anchor="CommonProc_linear_combination_computation" title="Linear Combination of Source Symbols Computation"> <!-- ====================== --> <t> The two<t pn="section-3.7.2-5"> With the RLC over GF(2) FECSchemes require the computation ofscheme (binary case), a linear combination is computed as follows. The repair symbol is the XOR sum of all the sourcesymbols, usingsymbols corresponding to a coding coefficient cc_tab[j] equal to 1 (i.e., the source symbols corresponding to zero coding coefficientsproduced byare ignored). The XOR sum of thegenerate_coding_coefficients() functionbyte of position i in each source is computed and stored in thecc_tab[] array.corresponding byte of the repair symbol, where i belongs to [0; E-1]. In practice, the XOR sums will be computed several bytes at a time (e.g., on 64 bit words, or on arrays of 16 or more bytes when using SIMD CPU extensions). </t><t><t pn="section-3.7.2-6"> With both FEC schemes, the details of how to optimize the computation of these linear combinations are of high practical importance but out of scope of this document. </t> </section> </section> </section> <section anchor="ArbitraryFlows_RLC_GF_28" numbered="true" toc="include" removeInRFC="false" pn="section-4"> <name slugifiedName="name-sliding-window-rlc-fec-sche">Sliding Window RLC FEC Scheme overGF(2^^8)GF(2<sup>8</sup>) for Arbitrary Packet Flows</name> <t pn="section-4-1"> This fully-specified FECScheme, a linear combination ofscheme defines theew_size source symbol present inSliding Window Random Linear Codes (RLC) over GF(2<sup>8</sup>). </t> <section anchor="ArbitraryFlows_formatsAndCodes" numbered="true" toc="include" removeInRFC="false" pn="section-4.1"> <name slugifiedName="name-formats-and-codes">Formats and Codes</name> <section numbered="true" toc="include" removeInRFC="false" pn="section-4.1.1"> <name slugifiedName="name-fec-framework-configuration">FEC Framework Configuration Information</name> <t pn="section-4.1.1-1"> Following the guidelines of <xref target="RFC6363" format="default" sectionFormat="of" section="5.6" derivedLink="https://rfc-editor.org/rfc/rfc6363#section-5.6" derivedContent="RFC6363"/>, this section provides theencoding window, say src_0FEC Framework Configuration Information (or FFCI). This FCCI needs tosrc_ew_size_1,be shared (e.g., using SDP) between the FECFRAME sender and receiver instances in order togeneratesynchronize them. It includes arepair symbol, is computed as follows. For each byte of position i in each source and the repair symbol, where i belongs to [0; E-1], compute: <list style="none"> <t> repair[i] = cc_tab[0] * src_0[i] XOR cc_tab[1] * src_1[i] XOR ... XOR cc_tab[ew_size - 1] * src_ew_size_1[i]</t> </list> where * is the multiplication over GF(2^^8). In practice various optimizations needFEC Encoding ID, mandatory for any FEC scheme specification, plus scheme-specific elements. </t> <section numbered="true" toc="exclude" removeInRFC="false" pn="section-4.1.1.1"> <name slugifiedName="name-fec-encoding-id">FEC Encoding ID</name> <dl newline="false" spacing="normal" pn="section-4.1.1.1-1"> <dt pn="section-4.1.1.1-1.1">FEC Encoding ID:</dt> <dd pn="section-4.1.1.1-1.2">the value assigned to this fully specified FEC scheme <bcp14>MUST</bcp14> be 10, as assigned by IANA (<xref target="iana" format="default" sectionFormat="of" derivedContent="Section 9"/>).</dd> </dl> <t pn="section-4.1.1.1-2"> When SDP is usedin ordertomakecommunicate the FFCI, thiscomputation efficient (seeFEC Encoding ID is carried inparticular <xref target="PGM13"/>).the 'encoding-id' parameter. </t><t> With</section> <section anchor="ArbitraryFlows_fssi" numbered="true" toc="exclude" removeInRFC="false" pn="section-4.1.1.2"> <name slugifiedName="name-fec-scheme-specific-informa">FEC Scheme-Specific Information</name> <t pn="section-4.1.1.2-1"> The FEC Scheme-Specific Information (FSSI) includes elements that are specific to theRLC over GF(2)present FECScheme (binary case), a linear combination is computed as follows. The repairscheme. More precisely: </t> <dl newline="false" spacing="normal" pn="section-4.1.1.2-2"> <dt pn="section-4.1.1.2-2.1">Encoding symbolissize (E):</dt> <dd pn="section-4.1.1.2-2.2"> a non-negative integer that indicates theXOR sumsize ofall the source symbols corresponding toeach encoding symbol in bytes;</dd> <dt pn="section-4.1.1.2-2.3">Window Size Ratio (WSR) parameter: </dt> <dd pn="section-4.1.1.2-2.4"> acoding coefficient cc_tab[j] equalnon-negative integer between 0 and 255 (both inclusive) used to initialize window sizes. A value of 0 indicates this parameter is not considered (e.g., a fixed encoding window size may be chosen). A value between 1(i.e., the source symbols corresponding to zero coding coefficients are ignored). The XOR sumand 255 inclusive is required by certain of thebyte of position iparameter derivation techniques described ineach source<xref target="possible_param_derivation" format="default" sectionFormat="of" derivedContent="Appendix C"/>;</dd> </dl> <t pn="section-4.1.1.2-3"> This element iscomputedrequired both by the sender (RLC encoder) andstoredthe receiver(s) (RLC decoder). </t> <t pn="section-4.1.1.2-4"> When SDP is used to communicate the FFCI, this FEC Scheme-Specific Information is carried in thecorresponding byte of'fssi' parameter in textual representation as specified in <xref target="RFC6364" format="default" sectionFormat="of" derivedContent="RFC6364"/>. For instance: </t> <sourcecode type="sdp" markers="false" pn="section-4.1.1.2-5"> fssi=E:1400,WSR:191 </sourcecode> <t pn="section-4.1.1.2-6"> In that case therepair symbol, where i belongsname values "E" and "WSR" are used to[0; E-1]. In practice,convey theXOR sums will be computed several bytes at a time (e.g., on 64 bit words, or on arrays of 16 or more bytes when using SIMD CPU extensions).E and WSR parameters respectively. </t><t> With both FEC Schemes,<t pn="section-4.1.1.2-7"> If another mechanism requires thedetails of howFSSI tooptimizebe carried as an opaque octet string, thecomputation of these linear combinations are of high practical importance but out of scopeencoding format consists ofthis document.the following three octets, where the E field is carried in "big-endian" or "network order" format, that is, most significant byte first: </t></section><dl newline="false" spacing="normal" pn="section-4.1.1.2-8"> <dt pn="section-4.1.1.2-8.1"/> <dd pn="section-4.1.1.2-8.2"> Encoding symbol length (E): 16-bit field;</dd> <dt pn="section-4.1.1.2-8.3"/> <dd pn="section-4.1.1.2-8.4"> Window Size Ratio Parameter (WSR): 8-bit field.</dd> </dl> <t pn="section-4.1.1.2-9"> These three octets can be communicated as such, or for instance, be subject to an additional Base64 encoding. </t> <figure anchor="fig_ArbitraryFlows_fssi_binary" align="left" suppress-title="false" pn="figure-4"> <name slugifiedName="name-fssi-encoding-format">FSSI Encoding Format</name> <artwork name="" type="" align="left" alt="" pn="section-4.1.1.2-10.1"> 0 1 2 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Encoding Symbol Length (E) | WSR | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ </artwork> </figure> </section> </section> <sectionanchor="ArbitraryFlows_RLC_GF_28" title="Sliding Window RLCanchor="ArbitraryFlows_src_fpi" numbered="true" toc="include" removeInRFC="false" pn="section-4.1.2"> <name slugifiedName="name-explicit-source-fec-payload">Explicit Source FEC Payload ID</name> <t pn="section-4.1.2-1"> A FECScheme over GF(2^^8) for ArbitrarySource PacketFlows"> <!-- ==================================== --> <t> This fully-specified<bcp14>MUST</bcp14> contain an Explicit Source FECScheme defines the Sliding Window Random Linear Codes (RLC) over GF(2^^8). </t> <section anchor="ArbitraryFlows_formatsAndCodes" title="Formats and Codes"> <!-- ==================================== --> <section title="FEC Framework Configuration Information"> <!-- ================ --> <t> FollowingPayload ID that is appended to theguidelinesend of<xref target="RFC6363"/>, section 5.6, this section providesthe packet as illustrated in <xref target="fig_src_pkt_format" format="default" sectionFormat="of" derivedContent="Figure 5"/>. </t> <figure anchor="fig_src_pkt_format" align="left" suppress-title="false" pn="figure-5"> <name slugifiedName="name-structure-of-an-fec-source-">Structure of an FECFramework Configuration Information (or FFCI). This FCCI needs to be shared (e.g., using SDP) betweenSource Packet with theFECFRAME sender and receiver instances in order to synchronize them. It includes aExplicit Source FECEncoding ID, mandatory for anyPayload ID</name> <artwork name="" type="" align="left" alt="" pn="section-4.1.2-2.1"> +--------------------------------+ | IP Header | +--------------------------------+ | Transport Header | +--------------------------------+ | ADU | +--------------------------------+ | Explicit Source FECScheme specification, plus scheme-specific elements. </t> <section title="FEC Encoding ID"> <!-- ================ --> <t> <list style="symbols"> <t>FEC Encoding ID:Payload ID | +--------------------------------+ </artwork> </figure> <t pn="section-4.1.2-3"> More precisely, thevalue assigned to this fully specifiedExplicit Source FECScheme MUST be XXXX, as assigned by IANA (<xref target="iana"/>).</t> </list> </t> <t> When SDPPayload ID isused to communicatecomposed of theFFCI,following field, carried in "big-endian" or "network order" format, that is, most significant byte first (<xref target="fig_src_fpi" format="default" sectionFormat="of" derivedContent="Figure 6"/>): </t> <dl newline="false" spacing="normal" pn="section-4.1.2-4"> <dt pn="section-4.1.2-4.1">Encoding Symbol ID (ESI) (32-bit field):</dt> <dd pn="section-4.1.2-4.2"> this unsigned integer identifies the first source symbol of the ADUI corresponding to this FECEncoding IDSource Packet. The ESI iscarried inincremented for each new source symbol, and after reaching the'encoding-id' parameter. </t>maximum value (2<sup>32</sup>-1), wrapping to zero occurs. </dd> </dl> <figure anchor="fig_src_fpi" align="left" suppress-title="false" pn="figure-6"> <name slugifiedName="name-source-fec-payload-id-encod">Source FEC Payload ID Encoding Format</name> <artwork name="" type="" align="left" alt="" pn="section-4.1.2-5.1"> 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Encoding Symbol ID (ESI) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ </artwork> </figure> </section> <sectionanchor="ArbitraryFlows_fssi" title="FEC Scheme-Specific Information"> <!-- ================ --> <t> The FEC Scheme-Specific Information (FSSI) includes elements that are specific to the presentanchor="ArbitraryFlows_repair_fpi" numbered="true" toc="include" removeInRFC="false" pn="section-4.1.3"> <name slugifiedName="name-repair-fec-payload-id">Repair FECScheme. More precisely: <list style="hanging">Payload ID</name> <thangText="Encoding symbol size (E):"> a non-negative integer that indicates the sizepn="section-4.1.3-1"> A FEC Repair Packet <bcp14>MAY</bcp14> contain one or more repair symbols. When there are several repair symbols, all ofeachthem <bcp14>MUST</bcp14> have been generated from the same encodingsymbol in bytes;</t> <t hangText="Window Size Ratio (WSR) parameter: "> a non-negative integer between 0 and 255 (both inclusive) used to initialize window sizes.window, using Repair_Key values that are managed as explained below. Avaluereceiver can easily deduce the number of0 indicates this parameter is not considered (e.g.,repair symbols within afixed encoding window size may be chosen). A value between 1 and 255 inclusive is requiredFEC Repair Packet bycertain ofcomparing theparameter derivation techniques described in <xref target="possible_param_derivation"/>;</t> </list> This elementreceived FEC Repair Packet size (equal to the UDP payload size when UDP isrequired both bythesender (RLC encoder)underlying transport protocol) and thereceiver(s) (RLC decoder). </t> <t> When SDP is used to communicatesymbol size, E, communicated in theFFCI, thisFFCI. </t> <t pn="section-4.1.3-2"> A FECScheme-specific informationRepair Packet <bcp14>MUST</bcp14> contain a Repair FEC Payload ID that iscarried inprepended to the'fssi' parameter in textual representationrepair symbol asspecifiedillustrated in <xreftarget="RFC6364"/>. For instance: </t> <t> fssi=E:1400,WSR:191 </t> <t> In that case the name values "E" and "WSR" are used to convey the E and WSR parameters respectively.target="fig_repair_pkt_format" format="default" sectionFormat="of" derivedContent="Figure 7"/>. </t><t> If another mechanism requires the FSSI to be carried as<figure anchor="fig_repair_pkt_format" align="left" suppress-title="false" pn="figure-7"> <name slugifiedName="name-structure-of-an-fec-repair-">Structure of anopaque octet string,FEC Repair Packet with the Repair FEC Payload ID</name> <artwork name="" type="" align="left" alt="" pn="section-4.1.3-3.1"> +--------------------------------+ | IP Header | +--------------------------------+ | Transport Header | +--------------------------------+ | Repair FEC Payload ID | +--------------------------------+ | Repair Symbol | +--------------------------------+ </artwork> </figure> <t pn="section-4.1.3-4"> More precisely, theencoding format consistsRepair FEC Payload ID is composed of the followingthree octets,fields wherethe E field isall integer fields are carried in "big-endian" or "network order" format, that is, most significant bytefirst: <list style="hanging"> <t> Encoding symbol length (E): 16-bit field;</t> <t> Window Size Ratio Parameter (WSR): 8-bit field.</t> </list> These three octets can be communicatedfirst (<xref target="fig_repair_fpi" format="default" sectionFormat="of" derivedContent="Figure 8"/>): </t> <dl newline="false" spacing="normal" pn="section-4.1.3-5"> <dt pn="section-4.1.3-5.1">Repair_Key (16-bit field):</dt> <dd pn="section-4.1.3-5.2"> this unsigned integer is used assuch, or for instance,a seed by the coefficient generation function (<xref target="CommonProc_coef_generation_func" format="default" sectionFormat="of" derivedContent="Section 3.6"/>) in order to generate the desired number of coding coefficients. This repair key may besubjecta monotonically increasing integer value that loops back toan additional Base64 encoding. </t> <figure anchor="fig_ArbitraryFlows_fssi_binary" title="FSSI Encoding Format"> <artwork> 0 1 2 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 901 2 3 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Encoding Symbol Length (E) | WSR | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ </artwork> </figure> </section> </section> <section anchor="ArbitraryFlows_src_fpi" title="Explicit Source FEC Payload ID"> <!-- ================ --> <t> Aafter reaching 65535 (see <xref target="ArbitraryFlows_FECCodeSpecification_encoding" format="default" sectionFormat="of" derivedContent="Section 6.1"/>). When a FECSourceRepair PacketMUST contain an Explicit Source FEC Payload ID thatcontains several repair symbols, this repair key value isappended tothat of theendfirst repair symbol. The remaining repair keys can be deduced by incrementing by 1 this value, up to a maximum value of 65535 after which it loops back to 0. </dd> <dt pn="section-4.1.3-5.3">Density Threshold for thepacket as illustrated incoding coefficients, DT (4-bit field):</dt> <dd pn="section-4.1.3-5.4"> this unsigned integer carries the Density Threshold (DT) used by the coding coefficient generation function <xreftarget="fig_src_pkt_format"/>. </t> <figure anchor="fig_src_pkt_format" title="Structuretarget="CommonProc_coef_generation_func" format="default" sectionFormat="of" derivedContent="Section 3.6"/>. More precisely, it controls the probability ofanhaving a nonzero coding coefficient, which equals (DT+1) / 16. When a FECSourceRepair Packetwithcontains several repair symbols, theExplicit Source FEC Payload ID"> <artwork> +--------------------------------+ | IP Header | +--------------------------------+ | Transport Header | +--------------------------------+ | ADU | +--------------------------------+ | ExplicitDT value applies to all of them; </dd> <dt pn="section-4.1.3-5.5">Number of SourceFEC Payload ID | +--------------------------------+ </artwork> </figure> <t> More precisely,Symbols in theExplicit Source FEC Payload ID is composed ofencoding window, NSS (12-bit field):</dt> <dd pn="section-4.1.3-5.6"> this unsigned integer indicates thefollowing field, carriednumber of source symbols in"big-endian" or "network order" format, that is, most significant byte first (<xref target="fig_src_fpi"/>): <list style="hanging"> <t hangText="Encodingthe encoding window when this repair symbol was generated. When a FEC Repair Packet contains several repair symbols, this NSS value applies to all of them; </dd> <dt pn="section-4.1.3-5.7">ESI of First Source SymbolID (ESI)in the encoding window, FSS_ESI (32-bitfield):">field):</dt> <dd pn="section-4.1.3-5.8"> this unsigned integeridentifiesindicates the ESI of the first source symbolofin theADUI corresponding toencoding window when this repair symbol was generated. When a FECSource Packet. The ESI is incremented for each new source symbol, and after reaching the maximumRepair Packet contains several repair symbols, this FSS_ESI value(2^32-1), wrappingapplies tozero occurs. </t> </list></t>all of them; </dd> </dl> <figureanchor="fig_src_fpi" title="Sourceanchor="fig_repair_fpi" align="left" suppress-title="false" pn="figure-8"> <name slugifiedName="name-repair-fec-payload-id-encod">Repair FEC Payload ID EncodingFormat"> <artwork>Format</name> <artwork name="" type="" align="left" alt="" pn="section-4.1.3-6.1"> 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |Encoding Symbol ID (ESI)Repair_Key | DT |NSS (# src symb in ew) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FSS_ESI | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ </artwork> </figure> </section> </section> <sectionanchor="ArbitraryFlows_repair_fpi" title="Repair FEC Payload ID"> <!-- ================ --> <t> A FEC Repair Packet MAY contain one or more repair symbols. When there are several repair symbols, all of them MUST have been generated from the same encoding window, using Repair_Key values that are managed as explained below. A receiver can easily deduceanchor="ArbitraryFlows_Procedures" numbered="true" toc="include" removeInRFC="false" pn="section-4.2"> <name slugifiedName="name-procedures">Procedures</name> <t pn="section-4.2-1"> All thenumberprocedures ofrepair symbols within a<xref target="CommonProcedures" format="default" sectionFormat="of" derivedContent="Section 3"/> apply to this FECRepairscheme. </t> </section> </section> <section anchor="ArbitraryFlows_RLC_GF_2" numbered="true" toc="include" removeInRFC="false" pn="section-5"> <name slugifiedName="name-sliding-window-rlc-fec-schem">Sliding Window RLC FEC Scheme over GF(2) for Arbitrary Packetby comparingFlows</name> <t pn="section-5-1"> This fully-specified FEC scheme defines thereceivedSliding Window Random Linear Codes (RLC) over GF(2) (binary case). </t> <section numbered="true" toc="include" removeInRFC="false" pn="section-5.1"> <name slugifiedName="name-formats-and-codes-2">Formats and Codes</name> <section numbered="true" toc="include" removeInRFC="false" pn="section-5.1.1"> <name slugifiedName="name-fec-framework-configuration-">FEC Framework Configuration Information</name> <section numbered="true" toc="exclude" removeInRFC="false" pn="section-5.1.1.1"> <name slugifiedName="name-fec-encoding-id-2">FEC Encoding ID</name> <dl newline="false" spacing="normal" pn="section-5.1.1.1-1"> <dt pn="section-5.1.1.1-1.1">FEC Encoding ID:</dt> <dd pn="section-5.1.1.1-1.2">the value assigned to this fully specified FECRepair Packet size (equalscheme <bcp14>MUST</bcp14> be 9, as assigned by IANA (<xref target="iana" format="default" sectionFormat="of" derivedContent="Section 9"/>).</dd> </dl> <t pn="section-5.1.1.1-2"> When SDP is used to communicate theUDP payload size when UDPFFCI, this FEC Encoding ID isthe underlying transport protocol) and the symbol size, E, communicatedcarried in theFFCI.'encoding-id' parameter. </t><t> A FEC Repair Packet MUST contain a Repair FEC Payload ID that is prepended to</section> <section numbered="true" toc="exclude" removeInRFC="false" pn="section-5.1.1.2"> <name slugifiedName="name-fec-scheme-specific-informat">FEC Scheme-Specific Information</name> <t pn="section-5.1.1.2-1"> All therepair symbol as illustrated inconsiderations of <xreftarget="fig_repair_pkt_format"/>.target="ArbitraryFlows_fssi" format="default" sectionFormat="of" derivedContent="Section 4.1.1.2"/> apply here. </t><figure anchor="fig_repair_pkt_format" title="Structure of an FEC Repair Packet with the Repair FEC Payload ID"> <artwork> +--------------------------------+ | IP Header | +--------------------------------+ | Transport Header | +--------------------------------+ | Repair</section> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-5.1.2"> <name slugifiedName="name-explicit-source-fec-payload-">Explicit Source FEC PayloadID | +--------------------------------+ | Repair Symbol | +--------------------------------+ </artwork> </figure> <t> More precisely,ID</name> <t pn="section-5.1.2-1"> All theRepairconsiderations of <xref target="ArbitraryFlows_src_fpi" format="default" sectionFormat="of" derivedContent="Section 4.1.2"/> apply here. </t> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-5.1.3"> <name slugifiedName="name-repair-fec-payload-id-2">Repair FEC PayloadID is composedID</name> <t pn="section-5.1.3-1"> All the considerations of <xref target="ArbitraryFlows_repair_fpi" format="default" sectionFormat="of" derivedContent="Section 4.1.3"/> apply here, with thefollowing fields where all integer fields are carried in "big-endian" or "network order" format,only exception thatis, most significant byte first (<xref target="fig_repair_fpi"/>): <list style="hanging"> <t hangText="Repair_Key (16-bit field):"> this unsigned integerthe Repair_Key field isused as a seed byuseless if DT = 15 (indeed, in that case all the coefficients are necessarily equal to 1 and the coefficient generation function(<xref target="CommonProc_coef_generation_func"/>) in orderdoes not use any PRNG). When DT = 15 the FECFRAME sender <bcp14>MUST</bcp14> set the Repair_Key field togeneratezero on transmission and a receiver <bcp14>MUST</bcp14> ignore it on receipt. </t> </section> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-5.2"> <name slugifiedName="name-procedures-2">Procedures</name> <t pn="section-5.2-1"> All thedesired numberprocedures ofcoding coefficients.<xref target="CommonProcedures" format="default" sectionFormat="of" derivedContent="Section 3"/> apply to this FEC scheme. </t> </section> </section> <section anchor="ArbitraryFlows_FECCodeSpecification" numbered="true" toc="include" removeInRFC="false" pn="section-6"> <name slugifiedName="name-fec-code-specification">FEC Code Specification</name> <section anchor="ArbitraryFlows_FECCodeSpecification_encoding" numbered="true" toc="include" removeInRFC="false" pn="section-6.1"> <name slugifiedName="name-encoding-side">Encoding Side</name> <t pn="section-6.1-1"> Thisrepair key may besection provides amonotonically increasing integer value that loops back to 0 after reaching 65535 (see <xref target="ArbitraryFlows_FECCodeSpecification_encoding"/>). Whenhigh level description of a Sliding Window RLC encoder. </t> <t pn="section-6.1-2"> Whenever a new FEC Repair Packetcontains several repair symbols, this repair key valueisthat ofneeded, the RLC encoder instance first gathers the ew_size source symbols currently in the sliding encoding window. Then it chooses a repairsymbol. The remaining repair keyskey, which can bededuced by incrementing by 1 thisa monotonically increasing integer value, incremented for each repair symbol up to a maximum value of 65535 (as it is carried within a 16-bit field) after which it loops back to 0.</t> <t hangText="Density Threshold for the coding coefficients, DT (4-bit field):"> this unsigned integer carries the Density Threshold (DT) used byThis repair key is communicated to thecodingcoefficient generation function<xref target="CommonProc_coef_generation_func"/>. More precisely, it controls(<xref target="CommonProc_coef_generation_func" format="default" sectionFormat="of" derivedContent="Section 3.6"/>) in order to generate ew_size coding coefficients. Finally, theprobability of havingFECFRAME sender computes the repair symbol as anon zerolinear combination of the ew_size source symbols using the ew_size codingcoefficient, which equals (DT+1) / 16.coefficients (<xref target="CommonProc_gf_specificiation" format="default" sectionFormat="of" derivedContent="Section 3.7"/>). Whena FEC Repair Packet containsE is small and when there is an incentive to pack several repairsymbols, the DT value applies to all of them; </t> <t hangText="Number of Source Symbols insymbols within theencoding window, NSS (12-bit field):"> this unsigned integer indicatessame FEC Repair Packet, the appropriate number of repair symbols are computed. In that case the repair key for each of them <bcp14>MUST</bcp14> be incremented by 1, keeping the same ew_size sourcesymbols insymbols, since only theencoding window when thisfirst repairsymbol was generated. When akey will be carried in the Repair FEC Payload ID. The FEC Repair Packetcontains several repair symbols, this NSS value appliescan then be passed toallthe transport layer for transmission. The source versus repair FEC packet transmission order is out ofthem; </t> <t hangText="ESIscope ofFirst Source Symbol in the encoding window, FSS_ESI (32-bit field):">thisunsigneddocument and several approaches exist that are implementation-specific. </t> <t pn="section-6.1-3"> Other solutions are possible to select a repair key value when a new FEC Repair Packet is needed, for instance, by choosing a random integerindicates the ESI ofbetween 0 and 65535. However, selecting thefirst source symbolsame repair key as before (which may happen in case of a random process) is only meaningful if the encoding windowwhen this repair symbol was generated. When ahas changed, otherwise the same FEC Repair Packetcontains severalwill be generated. In any case, choosing the repairsymbols, this FSS_ESI value applies to all of them; </t> </list> </t> <figure anchor="fig_repair_fpi" title="Repair FEC Payload ID Encoding Format"> <artwork> 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Repair_Key | DT |NSS (# src symb in ew) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FSS_ESI | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ </artwork> </figure> </section> </section> <section anchor="ArbitraryFlows_Procedures" title="Procedures"> <!-- ================ --> <t> Allkey is entirely at theproceduresdiscretion of<xref target="CommonProcedures"/> applythe sender, since it is communicated tothisthe receiver(s) in each Repair FECScheme.Payload ID. A receiver should not make any assumption on the way the repair key is managed. </t> </section></section> <!-- ArbitraryFlows --><sectionanchor="ArbitraryFlows_RLC_GF_2" title="Sliding Window RLC FEC Scheme over GF(2) for Arbitrary Packet Flows"> <!-- ==================================== --> <t>anchor="ArbitraryFlows_FECCodeSpecification_decoding" numbered="true" toc="include" removeInRFC="false" pn="section-6.2"> <name slugifiedName="name-decoding-side">Decoding Side</name> <t pn="section-6.2-1"> Thisfully-specified FEC Scheme defines thesection provides a high level description of a Sliding WindowRandom Linear Codes (RLC) over GF(2) (binary case).RLC decoder. </t><section title="Formats<t pn="section-6.2-2"> A FECFRAME receiver needs to maintain a linear system whose variables are the received andCodes"> <!-- ==================================== --> <section title="FEC Framework Configuration Information"> <!-- ================ --> <section title="FEC Encoding ID"> <!-- ================ --> <t> <list style="symbols"> <t>FEC Encoding ID:lost source symbols. Upon receiving a FEC Repair Packet, a receiver first extracts all thevalue assignedrepair symbols it contains (in case several repair symbols are packed together). For each repair symbol, when at least one of the corresponding source symbols it protects has been lost, the receiver adds an equation to the linear system (or no equation if thisfully specified FEC Scheme MUST be YYYY, as assignedrepair packet does not change the linear system rank). This equation of course re-uses the ew_size coding coefficients that are computed byIANAthe same coefficient generation function (<xreftarget="iana"/>).</t> </list> </t> <t> When SDPtarget="CommonProc_coef_generation_func" format="default" sectionFormat="of" derivedContent="Section 3.6"/>), using the repair key and encoding window descriptions carried in the Repair FEC Payload ID. Whenever possible (i.e., when a sub-system covering one or more lost source symbols isusedof full rank), decoding is performed in order tocommunicate the FFCI,recover lost source symbols. Gaussian elimination is one possible algorithm to solve thisFEC Encoding IDlinear system. Each time an ADUI can be totally recovered, padding is removed (thanks to the Length field, L, of the ADUI) and the ADU iscarried inassigned to the'encoding-id' parameter. </t> </section> <section title="FEC Scheme-Specific Information"> <!-- ================ --> <t> Allcorresponding application flow (thanks to theconsiderationsFlow ID field, F, of<xref target="ArbitraryFlows_fssi"/> apply here. </t> </section> </section> <section title="Explicit Source FEC Payload ID"> <!-- ================ --> <t> Alltheconsiderations of <xref target="ArbitraryFlows_src_fpi"/> apply here. </t> </section> <section title="Repair FEC Payload ID"> <!-- ================ --> <t> AllADUI). This ADU is finally passed to theconsiderations of <xref target="ArbitraryFlows_repair_fpi"/> apply here, withcorresponding upper application. Received FEC Source Packets, containing an ADU, <bcp14>MAY</bcp14> be passed to theonly exception thatapplication either immediately or after some time to guaranty an ordered delivery to theRepair_Key fieldapplication. This document does not mandate any approach as this isuseless if DT = 15 (indeed, inan operational and management decision. </t> <t pn="section-6.2-3"> With real-time flows, a lost ADU thatcase allis decoded after thecoefficients are necessarily equalmaximum latency or an ADU received after this delay has no value to1 andthecoefficient generation function doesapplication. This raises the question of deciding whether or notuse any PRNG). When DT = 15an ADU is late. This decision <bcp14>MAY</bcp14> be taken within the FECFRAMEsender MUST setreceiver (e.g., using theRepair_Key fielddecoding window, see <xref target="CommonProc_rlcParameters" format="default" sectionFormat="of" derivedContent="Section 3.1"/>) or within the application (e.g., using RTP timestamps within the ADU). Deciding which option tozero on transmissionfollow anda receiver MUST ignore itwhether or not to pass all ADUs, including those assumed late, to the application are operational decisions that depend onreceipt. </t> </section> </section> <section title="Procedures"> <!-- ================ --> <t> Alltheproceduresapplication and are therefore out of scope of this document. Additionally, <xreftarget="CommonProcedures"/> applytarget="decodingBeyondMaxLatency" format="default" sectionFormat="of" derivedContent="Appendix D"/> discusses a backward compatible optimization whereby late source symbols <bcp14>MAY</bcp14> still be used within the FECFRAME receiver in order tothis FEC Scheme.improve transmission robustness. </t> </section> </section><!-- ArbitraryFlows --> <section anchor="ArbitraryFlows_FECCodeSpecification" title="FEC Code Specification"> <!-- ================ --><sectionanchor="ArbitraryFlows_FECCodeSpecification_encoding" title="Encoding Side"> <!-- ================ --> <t> This sectionanchor="SecurityConsiderations" numbered="true" toc="include" removeInRFC="false" pn="section-7"> <name slugifiedName="name-security-considerations">Security Considerations</name> <t pn="section-7-1"> The FEC Framework document <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/> provides ahigh level descriptionfairly comprehensive analysis ofasecurity considerations applicable to FEC schemes. Therefore, the present section follows the security considerations section of <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/> and only discusses specific topics. </t> <section numbered="true" toc="include" removeInRFC="false" pn="section-7.1"> <name slugifiedName="name-attacks-against-the-data-fl">Attacks Against the Data Flow</name> <section numbered="true" toc="include" removeInRFC="false" pn="section-7.1.1"> <name slugifiedName="name-access-to-confidential-cont">Access to Confidential Content</name> <t pn="section-7.1.1-1">The Sliding Window RLCencoder. </t> <t> Whenever a newFECRepair Packet is needed, the RLC encoder instance first gathers the ew_size source symbols currentlyscheme specified in this document does not change thesliding encoding window. Then it chooses a repair key, which can be a monotonically increasing integer value, incremented for each repair symbol up to a maximum valuerecommendations of65535 (as it<xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/>. To summarize, if confidentiality iscarried withina16-bit field) after whichconcern, itloops back to 0. This repair keyiscommunicated to<bcp14>RECOMMENDED</bcp14> that one of thecoefficient generation function (<xref target="CommonProc_coef_generation_func"/>)solutions mentioned inorder<xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/> is used with special considerations togenerate ew_size coding coefficients. Finally,theFECFRAME sender computesway this solution is applied (e.g., is encryption applied before or after FEC protection, within therepair symbol asend system or in alinear combination of the ew_size source symbols usingmiddlebox), to theew_size coding coefficients (<xref target="CommonProc_gf_specificiation"/>). When E is smalloperational constraints (e.g., performing FEC decoding in a protected environment may be complicated or even impossible) andwhen there is an incentivetopack several repair symbols withinthesamethreat model. </t> </section> <section anchor="sec_content_corruption" numbered="true" toc="include" removeInRFC="false" pn="section-7.1.2"> <name slugifiedName="name-content-corruption">Content Corruption</name> <t pn="section-7.1.2-1">The Sliding Window RLC FECRepair Packet,scheme specified in this document does not change theappropriate numberrecommendations ofrepair symbols are computed. In<xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/>. To summarize, it is <bcp14>RECOMMENDED</bcp14> thatcase the repair key for eachone ofthem MUST be incremented by 1, keeping the same ew_size source symbols, since onlythefirst repair key will be carriedsolutions mentioned in <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/> is used on both the FEC Source and Repair Packets. </t> </section> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-7.2"> <name slugifiedName="name-attacks-against-the-fec-par">Attacks Against the FECPayload ID.Parameters</name> <t pn="section-7.2-1"> The FECRepair Packetscheme specified in this document defines parameters that canthenbepassed tothetransport layer for transmission. The source versus repair FEC packet transmission order is outbasis ofscopeattacks. More specifically, the following parameters of the FFCI may be modified by an attacker who targets receivers (<xref target="ArbitraryFlows_fssi" format="default" sectionFormat="of" derivedContent="Section 4.1.1.2"/>): </t> <dl newline="false" spacing="normal" pn="section-7.2-2"> <dt pn="section-7.2-2.1">FEC Encoding ID:</dt> <dd pn="section-7.2-2.2">changing thisdocument and several approaches exist that are implementation-specific. </t> <t> Other solutions are possible to selectparameter leads arepair key value whenreceiver to consider anewdifferent FECRepair Packet is needed, for instance, by choosing a random integer between 0 and 65535. However, selectingscheme. The consequences are severe, thesame repair key as before (which may happen in caseformat ofa random process) is only meaningful if the encoding window has changed, otherwisethesameExplicit Source FEC Payload ID and RepairPacket will be generated. In any case, choosing the repair key is entirely at the discretionFEC Payload ID ofthe sender, since it is communicatedreceived packets will probably differ, leading to various malfunctions. Even if thereceiver(s) in each Repairoriginal and modified FECPayload ID. A receiver should not make any assumption onschemes share the same format, FEC decoding will either fail or lead to corrupted decoded symbols. This will happen if an attacker turns value 9 (i.e., RLC over GF(2)) to value 10 (RLC over GF(2<sup>8</sup>)), an additional consequence being a higher processing overhead at thewayreceiver. In any case, therepair key is managed. </t> </section> <section anchor="ArbitraryFlows_FECCodeSpecification_decoding" title="Decoding Side"> <!-- ================ --> <t> This section providesattack results in ahigh level descriptionform ofa Sliding Window RLC decoder. </t> <t> A FECFRAME receiver needsDenial of Service (DoS) or corrupted content. </dd> <dt pn="section-7.2-2.3">Encoding symbol length (E):</dt> <dd pn="section-7.2-2.4">setting this E parameter tomaintainalinear system whose variables aredifferent value will confuse a receiver. If thereceived and lost source symbols. Upon receivingsize of a received FEC RepairPacket, a receiver first extracts all the repair symbols it contains (in case several repair symbols are packed together). For each repair symbol, when at least onePacket is no longer multiple of thecorresponding source symbols it protects has been lost, themodified E value, a receiveradds an equation toquickly detects a problem and <bcp14>SHOULD</bcp14> reject thelinear system (or no equation if this repair packet does not changepacket. If thelinear system rank). This equationnew E value is a sub-multiple ofcourse re-uses the ew_size coding coefficients that are computed bythesame coefficient generation function (<xref target="CommonProc_coef_generation_func"/>), usingoriginal E value (e.g., half therepair key and encoding window descriptions carried inoriginal value), then receivers may not detect theRepair FEC Payload ID. Whenever possible (i.e., whenproblem immediately. For instance, asub-system covering one orreceiver may think that a received FEC Repair Packet contains morelost source symbols is of full rank), decoding is performed in order to recover lost source symbols. Gaussian elimination is one possible algorithm to solve this linear system. Each time an ADUI can be totally recovered, paddingrepair symbols (e.g., twice as many if E isremoved (thanksreduced by half), leading to malfunctions whose nature depends on implementation details. Here also, theLength field, L,attack always results in a form ofthe ADUI) and the ADUDoS or corrupted content. </dd> </dl> <t pn="section-7.2-3"> It isassignedtherefore <bcp14>RECOMMENDED</bcp14> that security measures be taken to guarantee thecorresponding application flow (thanksFFCI integrity, as specified in <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/>. How to achieve this depends on theFlow ID field, F, ofway theADUI). This ADUFFCI isfinally passed to the corresponding upper application. Received FEC Source Packets, containing an ADU, MAY be passed tocommunicated from theapplication either immediately or after some time to guaranty an ordered deliverysender to theapplication. This document doesreceiver, which is notmandate any approach asspecified in thisis an operational and management decision.document. </t><t> With real-time flows,<t pn="section-7.2-4"> Similarly, attacks are possible against the Explicit Source FEC Payload ID and Repair FEC Payload ID. More specifically, in case of alost ADU that is decoded afterFEC Source Packet, themaximum latency or an ADU received after this delay has nofollowing valueto the application. This raises the question of deciding whether or not an ADU is late. This decision MAYcan betaken withinmodified by an attacker who targets receivers: </t> <dl newline="false" spacing="normal" pn="section-7.2-5"> <dt pn="section-7.2-5.1">Encoding Symbol ID (ESI):</dt> <dd pn="section-7.2-5.2">changing theFECFRAMEESI leads a receiver(e.g., using the decoding window, see <xref target="CommonProc_rlcParameters"/>) or withinto consider a wrong ADU, resulting in severe consequences, including corrupted content passed to theapplication (e.g.,receiving application; </dd> </dl> <t pn="section-7.2-6"> And in case of a FEC Repair Packet: </t> <dl newline="false" spacing="normal" pn="section-7.2-7"> <dt pn="section-7.2-7.1">Repair Key:</dt> <dd pn="section-7.2-7.2">changing this value leads a receiver to generate a wrong coding coefficient sequence, and therefore any source symbol decoded usingRTP timestamps withintheADU). Deciding which optionrepair symbols contained in this packet will be corrupted; </dd> <dt pn="section-7.2-7.3">DT:</dt> <dd pn="section-7.2-7.4">changing this value also leads a receiver tofollowgenerate a wrong coding coefficient sequence, andwhether or not to pass all ADUs, including those assumed late, totherefore any source symbol decoded using theapplication are operational decisions that depend onrepair symbols contained in this packet will be corrupted. In addition, if theapplication and are therefore out of scopeDT value is significantly increased, it will generate a higher processing overhead at a receiver. In case of very large encoding windows, thisdocument. Additionally, <xref target="decodingBeyondMaxLatency"/> discussesmay impact the terminal performance; </dd> <dt pn="section-7.2-7.5">NSS:</dt> <dd pn="section-7.2-7.6">changing this value leads abackward compatible optimization whereby latereceiver to consider a different set of sourcesymbols MAY still be used withinsymbols, and therefore any source symbol decoded using theFECFRAME receiverrepair symbols contained inorder to improve transmission robustness. </t> </section> </section> <section anchor="implementationStatus" title="Implementation Status"> <!-- ====================== --> <t> Editor's notes: RFC Editor, please removethissection motivated by RFC 6982 before publishing the RFC. Thanks. </t> <t>An implementation ofpacket will be corrupted. In addition, if theSliding Window RLC FEC Scheme for FECFRAME exists: <list style="symbols"> <t>Organisation: Inria</t> <t>Description: ThisNSS value isan implementation ofsignificantly increased, it will generate a higher processing overhead at a receiver, which may impact theSliding Window RLC FEC Scheme limitedterminal performance; </dd> <dt pn="section-7.2-7.7">FSS_ESI:</dt> <dd pn="section-7.2-7.8">changing this value also leads a receiver toGF(2^^8). It relies onconsider amodified versiondifferent set ofour OpenFEC (http://openfec.org) FEC code library.source symbols and therefore any source symbol decoded using the repair symbols contained in this packet will be corrupted. </dd> </dl> <t pn="section-7.2-8"> It isintegratedtherefore <bcp14>RECOMMENDED</bcp14> that security measures are taken to guarantee the FEC Source and Repair Packets as stated inour FECFRAME software (see<xreftarget="fecframe-ext"/>).</t> <t>Maturity: prototype.</t> <t>Coverage: this software complies with thetarget="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/>. </t> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-7.3"> <name slugifiedName="name-when-several-source-flows-a">When Several Source Flows are to be Protected Together</name> <t pn="section-7.3-1">The Sliding Window RLC FECScheme.</t> <t>Licensing: proprietary.</t> <t>Contact: vincent.roca@inria.fr</t> </list></t> </section> <!-- =========================================================================================== --> <section anchor="SecurityConsiderations" title="Security Considerations"> <!-- ====================== --> <t> The FEC Frameworkscheme specified in this document<xref target="RFC6363"/> provides a fairly comprehensive analysis of security considerations applicable to FEC Schemes. Therefore, the present section followsdoes not change thesecurity considerations sectionrecommendations of <xreftarget="RFC6363"/> and only discusses specific topics. </t> <section title="Attacks Against the Data Flow"> <!-- ====================== -->target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/>.</t> </section> <sectiontitle="Access to Confidential Content"> <!-- ====================== --> <t>Thenumbered="true" toc="include" removeInRFC="false" pn="section-7.4"> <name slugifiedName="name-baseline-secure-fec-framewo">Baseline Secure FEC Framework Operation</name> <t pn="section-7.4-1">The Sliding Window RLC FECSchemescheme specified in this document does not change the recommendations of <xreftarget="RFC6363"/>. To summarize, if confidentiality istarget="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/> concerning the use of the IPsec/Encapsulating Security Payload (ESP) security protocol as aconcern, itmandatory-to-implement (but not mandatory-to-use) security scheme. This isRECOMMENDED thatwell suited to situations where the only insecure domain is the oneofover which thesolutions mentionedFEC Framework operates. </t> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-7.5"> <name slugifiedName="name-additional-security-conside">Additional Security Considerations for Numerical Computations</name> <t pn="section-7.5-1"> In addition to the above security considerations, inherited from <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/>, the present document introduces several formulae, in particular in <xreftarget="RFC6363"/>target="param_derivation_cbr_realtime" format="default" sectionFormat="of" derivedContent="Appendix C.1"/>. It isused with special considerations<bcp14>RECOMMENDED</bcp14> to check that theway this solutioncomputed values stay within reasonable bounds since numerical overflows, caused by an erroneous implementation or an erroneous input value, may lead to hazardous behaviors. However, what "reasonable bounds" means isapplied (e.g.,use-case and implementation dependent and isencryption applied before or after FEC protection, withinnot detailed in this document. </t> <t pn="section-7.5-2"><xref target="param_derivation_other_realtime_flows" format="default" sectionFormat="of" derivedContent="Appendix C.2"/> also mentions theend-system orpossibility of "using the timestamp field of an RTP packet header" when applicable. A malicious attacker may deliberately corrupt this header field in order to trigger hazardous behaviors at amiddlebox),FECFRAME receiver. Protection against this type of content corruption can be addressed with the above recommendations on a baseline secure operation. In addition, it is also <bcp14>RECOMMENDED</bcp14> to check that theoperational constraints (e.g., performingtimestamp value be within reasonable bounds. </t> </section> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-8"> <name slugifiedName="name-operations-and-management-c">Operations and Management Considerations</name> <t pn="section-8-1"> The FECdecoding inFramework document <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/> provides aprotected environment may be complicated or even impossible)fairly comprehensive analysis of operations and management considerations applicable to FEC schemes. Therefore, thethreat model.present section only discusses specific topics. </t></section><sectiontitle="Content Corruption" anchor="sec_content_corruption"> <!-- ====================== --> <t>The Sliding Window RLC FEC Scheme specified in thisanchor="oprecom_ff_considerations" numbered="true" toc="include" removeInRFC="false" pn="section-8.1"> <name slugifiedName="name-operational-recommendations">Operational Recommendations: Finite Field GF(2) Versus GF(2<sup>8</sup>)</name> <t pn="section-8.1-1"> The present documentdoes not changespecifies two FEC schemes that differ on therecommendations of <xref target="RFC6363"/>. To summarize, itFinite Field used for the coding coefficients. It isRECOMMENDEDexpected thatonethe RLC over GF(2<sup>8</sup>) FEC scheme will be mostly used since it warrants a higher packet loss protection. In case of small encoding windows, thesolutions mentionedassociated processing overhead is not an issue (e.g., we measured decoding speeds between 745 Mbps and 2.8 Gbps on an ARM Cortex-A15 embedded board in <xreftarget="RFC6363"/> is usedtarget="Roca17" format="default" sectionFormat="of" derivedContent="Roca17"/> depending onboththeFEC Sourcecode rate andRepair Packets. </t> </section> </section> <section title="Attacks AgainsttheFEC Parameters"> <!-- ====================== --> <t>channel conditions, using an encoding window of size 18 or 23 symbols; see the above article for the details). Of course the CPU overhead will increase with the encoding window size, because more operations in the GF(2<sup>8</sup>) finite field will be needed. </t> <t pn="section-8.1-2"> The RLC over GF(2) FECScheme specified in this document defines parametersscheme offers an alternative. In that case operations symbols can bethe basis of attacks. More specifically, the following parameters of the FFCI maydirectly XOR-ed together which warrants high bitrate encoding and decoding operations, and can bemodifiedan advantage with large encoding windows. However, packet loss protection is significantly reduced byan attacker who targets receivers (<xref target="ArbitraryFlows_fssi"/>): <list style="symbols"> <t>FEC Encoding ID: changingusing thisparameter leads a receiver to consider a differentFECScheme. The consequences are severe,scheme. </t> </section> <section numbered="true" toc="include" removeInRFC="false" pn="section-8.2"> <name slugifiedName="name-operational-recommendations-">Operational Recommendations: Coding Coefficients Density Threshold</name> <t pn="section-8.2-1"> In addition to theformatchoice of theExplicit Source FEC Payload ID and RepairFinite Field, the two FECPayload ID of received packets will probably differ, leadingschemes define a coding coefficient density threshold (DT) parameter. This parameter enables a sender tovarious malfunctions. Even ifcontrol theoriginal and modified FEC Schemes sharecode density, i.e., thesame format, FEC decoding will either fail or lead to corrupted decoded symbols. This will happen if an attacker turns value YYYY (i.e.,proportion of coefficients that are nonzero on average. With RLC overGF(2))GF(2<sup>8</sup>), it is usually appropriate that small encoding windows be associated tovalue XXXX (RLC over GF(2^^8)), an additional consequence beingahigher processing overhead at the receiver. In any case,density threshold equal to 15, theattack resultsmaximum value, ina form of Denial of Service (DoS) or corrupted content. </t> <t>Encoding symbol length (E): setting this E parameterorder to warrant adifferent value will confuse a receiver. Ifhigh loss protection. </t> <t pn="section-8.2-2"> On thesize of a received FEC Repair Packetopposite, with larger encoding windows, it isno longer multiple ofusually appropriate that themodified E value, a receiver quickly detects a problemdensity threshold be reduced. With large encoding windows, an alternative can be to use RLC over GF(2) andSHOULD reject the packet. If the new E value isasub-multiple ofdensity threshold equal to 7 (i.e., an average density equal to 1/2) or smaller. </t> <t pn="section-8.2-3"> Note that using a density threshold equal to 15 with RLC over GF(2) is equivalent to using an XOR code that computes theoriginal E value (e.g., halfXOR sum of all theoriginal value), then receivers may not detectsource symbols in theproblem immediately. For instance, a receiver may thinkencoding window. In that case: (1) only areceived FEC Repair Packet contains moresingle repairsymbols (e.g., twice as many if E is reduced by half), leading to malfunctions whose nature dependssymbol can be produced for any encoding window, and (2) the repair_key parameter becomes useless (the coding coefficients generation function does not rely onimplementation details. Here also,theattack always results in a form of DoS or corrupted content.PRNG). </t></list></section> </section> <section anchor="iana" numbered="true" toc="include" removeInRFC="false" pn="section-9"> <name slugifiedName="name-iana-considerations">IANA Considerations</name> <t pn="section-9-1"> This document registers two values in the "FEC Framework (FECFRAME) FEC Encoding IDs" registry <xref target="RFC6363" format="default" sectionFormat="of" derivedContent="RFC6363"/> as follows: </t><t> It is therefore RECOMMENDED that security measures be taken<ul spacing="normal" bare="false" empty="false" pn="section-9-2"> <li pn="section-9-2.1">9 refers toguaranteetheFFCI integrity,Sliding Window Random Linear Codes (RLC) over GF(2) FEC Scheme for Arbitrary Packet Flows, asspecifieddefined in <xreftarget="RFC6363"/>. How to achievetarget="ArbitraryFlows_RLC_GF_2" format="default" sectionFormat="of" derivedContent="Section 5"/> of thisdepends on the way the FFCI is communicated from the senderdocument.</li> <li pn="section-9-2.2">10 refers to thereceiver, which is not specifiedSliding Window Random Linear Codes (RLC) over GF(2<sup>8</sup>) FEC Scheme for Arbitrary Packet Flows, as defined in <xref target="ArbitraryFlows_RLC_GF_28" format="default" sectionFormat="of" derivedContent="Section 4"/> of thisdocument. </t> <t> Similarly, attacks are possible against the Explicit Source FEC Payload ID and Repair FEC Payload ID. More specifically,document.</li> </ul> </section> </middle> <back> <references pn="section-10"> <name slugifiedName="name-references">References</name> <references pn="section-10.1"> <name slugifiedName="name-normative-references">Normative References</name> <reference anchor="C99" quoteTitle="true" derivedAnchor="C99"> <front> <title>Programming languages - C: C99, correction 3:2007</title> <seriesInfo name="ISO/IEC" value="9899:1999/Cor 3:2007"/> <author> <organization showOnFrontPage="true">International Organization for Standardization</organization> </author> <date month="November" year="2007"/> </front> </reference> <reference anchor="RFC2119" target="https://www.rfc-editor.org/info/rfc2119" quoteTitle="true" derivedAnchor="RFC2119"> <front> <title>Key words for use in RFCs to Indicate Requirement Levels</title> <author initials="S." surname="Bradner" fullname="S. Bradner"> <organization showOnFrontPage="true"/> </author> <date year="1997" month="March"/> <abstract> <t>In many standards track documents several words are used to signify the requirements incase of a FEC Source Packet,thefollowing value canspecification. These words are often capitalized. This document defines these words as they should bemodified byinterpreted in IETF documents. This document specifies anattacker who targets receivers: <list style="symbols"> <t>Encoding Symbol ID (ESI): changingInternet Best Current Practices for theESI leads a receiver to considerInternet Community, and requests discussion and suggestions for improvements.</t> </abstract> </front> <seriesInfo name="BCP" value="14"/> <seriesInfo name="RFC" value="2119"/> <seriesInfo name="DOI" value="10.17487/RFC2119"/> </reference> <reference anchor="RFC6363" target="https://www.rfc-editor.org/info/rfc6363" quoteTitle="true" derivedAnchor="RFC6363"> <front> <title>Forward Error Correction (FEC) Framework</title> <author initials="M." surname="Watson" fullname="M. Watson"> <organization showOnFrontPage="true"/> </author> <author initials="A." surname="Begen" fullname="A. Begen"> <organization showOnFrontPage="true"/> </author> <author initials="V." surname="Roca" fullname="V. Roca"> <organization showOnFrontPage="true"/> </author> <date year="2011" month="October"/> <abstract> <t>This document describes awrong ADU, resultingframework for using Forward Error Correction (FEC) codes with applications insevere consequences, including corrupted content passedpublic and private IP networks tothe receiving application; </t> </list> And in case of aprovide protection against packet loss. The framework supports applying FECRepair Packet: <list style="symbols"> <t>Repair Key: changing this value leads a receivertogenerate a wrong coding coefficient sequence,arbitrary packet flows over unreliable transport andtherefore any source symbol decodedis primarily intended for real-time, or streaming, media. This framework can be used to define Content Delivery Protocols that provide FEC for streaming media delivery or other packet flows. Content Delivery Protocols defined usingthe repair symbols containedthis framework can support any FEC scheme (and associated FEC codes) that is compliant with various requirements defined in thispacket willdocument. Thus, Content Delivery Protocols can becorrupted; </t> <t>DT: changing this value also leads a receiverdefined that are not specific togenerateawrong coding coefficient sequence,particular FEC scheme, andtherefore any source symbol decoded using the repair symbols contained in this packet willFEC schemes can becorrupted. In addition, if the DT value is significantly increased, it will generate a higher processing overhead at a receiver. In case of very large encoding windows, this may impact the terminal performance; </t> <t>NSS: changing this value leads a receiverdefined that are not specific toconsideradifferent setparticular Content Delivery Protocol. [STANDARDS-TRACK]</t> </abstract> </front> <seriesInfo name="RFC" value="6363"/> <seriesInfo name="DOI" value="10.17487/RFC6363"/> </reference> <reference anchor="RFC6364" target="https://www.rfc-editor.org/info/rfc6364" quoteTitle="true" derivedAnchor="RFC6364"> <front> <title>Session Description Protocol Elements for the Forward Error Correction (FEC) Framework</title> <author initials="A." surname="Begen" fullname="A. Begen"> <organization showOnFrontPage="true"/> </author> <date year="2011" month="October"/> <abstract> <t>This document specifies the use ofsource symbols,the Session Description Protocol (SDP) to describe the parameters required to signal the Forward Error Correction (FEC) Framework Configuration Information between the sender(s) andtherefore any source symbol decoded usingreceiver(s). This document also provides examples that show the semantics for grouping multiple source and repairsymbols containedflows together for the applications that simultaneously use multiple instances of the FEC Framework. [STANDARDS-TRACK]</t> </abstract> </front> <seriesInfo name="RFC" value="6364"/> <seriesInfo name="DOI" value="10.17487/RFC6364"/> </reference> <reference anchor="RFC8174" target="https://www.rfc-editor.org/info/rfc8174" quoteTitle="true" derivedAnchor="RFC8174"> <front> <title>Ambiguity of Uppercase vs Lowercase inthis packet willRFC 2119 Key Words</title> <author initials="B." surname="Leiba" fullname="B. Leiba"> <organization showOnFrontPage="true"/> </author> <date year="2017" month="May"/> <abstract> <t>RFC 2119 specifies common key words that may becorrupted. In addition, ifused in protocol specifications. This document aims to reduce theNSS value is significantly increased, it will generate a higher processing overhead at a receiver, which may impactambiguity by clarifying that only UPPERCASE usage of theterminal performance; </t> <t>FSS_ESI: changing this value also leads a receiverkey words have the defined special meanings.</t> </abstract> </front> <seriesInfo name="BCP" value="14"/> <seriesInfo name="RFC" value="8174"/> <seriesInfo name="DOI" value="10.17487/RFC8174"/> </reference> <reference anchor="RFC8680" target="https://www.rfc-editor.org/info/rfc8680" quoteTitle="true" derivedAnchor="RFC8680"> <front> <title>Forward Error Correction (FEC) Framework Extension toconsider a different setSliding Window Codes</title> <seriesInfo name="RFC" value="8680"/> <seriesInfo name="DOI" value="10.17487/RFC8680"/> <author initials="V" surname="Roca" fullname="Vincent Roca"> <organization showOnFrontPage="true"/> </author> <author initials="A" surname="Begen" fullname="Ali Begen"> <organization showOnFrontPage="true"/> </author> <date month="January" year="2020"/> </front> </reference> <reference anchor="RFC8682" target="https://www.rfc-editor.org/info/rfc8682" quoteTitle="true" derivedAnchor="RFC8682"> <front> <title>TinyMT32 Pseudorandom Number Generator (PRNG)</title> <seriesInfo name="RFC" value="8682"/> <seriesInfo name="DOI" value="10.17487/RFC8682"/> <author initials="M" surname="Saito" fullname="Mutsuo Saito"> <organization showOnFrontPage="true"/> </author> <author initials="M" surname="Matsumoto" fullname="Makoto Matsumoto"> <organization showOnFrontPage="true"/> </author> <author initials="V" surname="Roca" fullname="Vincent Roca" role="editor"> <organization showOnFrontPage="true"/> </author> <author initials="E" surname="Baccelli" fullname="Emmanuel Baccelli"> <organization showOnFrontPage="true"/> </author> <date month="January" year="2020"/> </front> </reference> </references> <references pn="section-10.2"> <name slugifiedName="name-informative-references">Informative References</name> <reference anchor="PGM13" target="http://web.eecs.utk.edu/~plank/plank/papers/UT-CS-13-717.html" quoteTitle="true" derivedAnchor="PGM13"> <front> <title>A Complete Treatment ofsource symbolsSoftware Implementations of Finite Field Arithmetic for Erasure Coding Applications</title> <seriesInfo name="University of Tennessee Technical Report" value="UT-CS-13-717"/> <author initials="J." surname="Plank"> <organization showOnFrontPage="true"/> </author> <author initials="K." surname="Greenan"> <organization showOnFrontPage="true"/> </author> <author initials="E." surname="Miller"> <organization showOnFrontPage="true"/> </author> <date month="October" year="2013"/> </front> </reference> <reference anchor="RFC5170" target="https://www.rfc-editor.org/info/rfc5170" quoteTitle="true" derivedAnchor="RFC5170"> <front> <title>Low Density Parity Check (LDPC) Staircase andtherefore any source symbol decoded using the repair symbols contained in this packet will be corrupted. </t> </list> It is therefore RECOMMENDED that security measures are taken to guarantee the FEC SourceTriangle Forward Error Correction (FEC) Schemes</title> <author initials="V." surname="Roca" fullname="V. Roca"> <organization showOnFrontPage="true"/> </author> <author initials="C." surname="Neumann" fullname="C. Neumann"> <organization showOnFrontPage="true"/> </author> <author initials="D." surname="Furodet" fullname="D. Furodet"> <organization showOnFrontPage="true"/> </author> <date year="2008" month="June"/> <abstract> <t>This document describes two Fully-Specified Forward Error Correction (FEC) Schemes, Low Density Parity Check (LDPC) Staircase andRepair Packets as stated in <xref target="RFC6363"/>. </t> </section> <section title="When Several Source Flows areLDPC Triangle, and their application tobe Protected Together"> <!-- ====================== --> <t>The Sliding Window RLC FEC Scheme specified in this document does not change the recommendations of <xref target="RFC6363"/>.</t> </section> <section title="Baseline Secure FEC Framework Operation"> <!-- ====================== --> <t>The Sliding Window RLC FEC Scheme specified in this document does not changetherecommendations of <xref target="RFC6363"/> concerning the usereliable delivery of data objects on theIPsec/ESP security protocol aspacket erasure channel (i.e., amandatory to implement (but not mandatory to use) security scheme. This is well suited to situationscommunication path wherethe only insecure domain is the one over which thepackets are either received without any corruption or discarded during transmission). These systematic FECFramework operates. </t> </section> <section title="Additional Security Considerations for Numerical Computations"> <!-- ====================== --> <t> In additioncodes belong to theabove security considerations, inherited from <xref target="RFC6363"/>, the present document introduces several formulae, in particularwell- known class of "Low Density Parity Check" codes, and are large block FEC codes in<xref target="param_derivation_cbr_realtime"/>. It is RECOMMENDED to check thatthecomputed values stay within reasonable bounds since numerical overflows, caused by an erroneous implementation or an erroneous input value, may lead to hazardous behaviours. However, what "reasonable bounds" means is use-case and implementation dependent andsense of RFC 3453. [STANDARDS-TRACK]</t> </abstract> </front> <seriesInfo name="RFC" value="5170"/> <seriesInfo name="DOI" value="10.17487/RFC5170"/> </reference> <reference anchor="RFC5510" target="https://www.rfc-editor.org/info/rfc5510" quoteTitle="true" derivedAnchor="RFC5510"> <front> <title>Reed-Solomon Forward Error Correction (FEC) Schemes</title> <author initials="J." surname="Lacan" fullname="J. Lacan"> <organization showOnFrontPage="true"/> </author> <author initials="V." surname="Roca" fullname="V. Roca"> <organization showOnFrontPage="true"/> </author> <author initials="J." surname="Peltotalo" fullname="J. Peltotalo"> <organization showOnFrontPage="true"/> </author> <author initials="S." surname="Peltotalo" fullname="S. Peltotalo"> <organization showOnFrontPage="true"/> </author> <date year="2009" month="April"/> <abstract> <t>This document describes a Fully-Specified Forward Error Correction (FEC) Scheme for the Reed-Solomon FEC codes over GF(2^^m), where m isnot detailedinthis document. </t> <t> <xref target="param_derivation_other_realtime_flows"/> also mentions{2..16}, and its application to thepossibilityreliable delivery of"usingdata objects on thetimestamp field of an RTPpacketheader" when applicable. A malicious attacker may deliberately corrupt this header field in order to trigger hazardous behaviours aterasure channel (i.e., aFECFRAME receiver. Protection against this type of contentcommunication path where packets are either received without any corruptioncan be addressed with the above recommendations onor discarded during transmission). This document also describes abaseline secure operation. In addition, itFully-Specified FEC Scheme for the special case of Reed-Solomon codes over GF(2^^8) when there isalso RECOMMENDED to check thatno encoding symbol group. Finally, in thetimestamp value be within reasonable bounds. </t> </section> </section> <section title="Operations and Management Considerations"> <!-- ====================== --> <t> Thecontext of the Under-Specified Small Block Systematic FECFrameworkScheme (FEC Encoding ID 129), this document<xref target="RFC6363"/> provides a fairly comprehensive analysisassigns an FEC Instance ID to the special case ofoperations and management considerations applicableReed-Solomon codes over GF(2^^8).</t> <t>Reed-Solomon codes belong toFEC Schemes. Therefore,thepresent section only discusses specific topics. </t> <section anchor="oprecom_ff_considerations" title="Operational Recommendations: Finite Field GF(2) Versus GF(2^^8)"> <!-- ================ --> <t>class of Maximum Distance Separable (MDS) codes, i.e., they enable a receiver to recover the k source symbols from any set of k received symbols. Thepresentschemes described here are compatible with the implementation from Luigi Rizzo. [STANDARDS-TRACK]</t> </abstract> </front> <seriesInfo name="RFC" value="5510"/> <seriesInfo name="DOI" value="10.17487/RFC5510"/> </reference> <reference anchor="RFC6681" target="https://www.rfc-editor.org/info/rfc6681" quoteTitle="true" derivedAnchor="RFC6681"> <front> <title>Raptor Forward Error Correction (FEC) Schemes for FECFRAME</title> <author initials="M." surname="Watson" fullname="M. Watson"> <organization showOnFrontPage="true"/> </author> <author initials="T." surname="Stockhammer" fullname="T. Stockhammer"> <organization showOnFrontPage="true"/> </author> <author initials="M." surname="Luby" fullname="M. Luby"> <organization showOnFrontPage="true"/> </author> <date year="2012" month="August"/> <abstract> <t>This documentspecifies two FECdescribes Fully-Specified Forward Error Correction (FEC) Schemesthat differ on the Finite Field usedfor thecoding coefficients. It is expected thatRaptor and RaptorQ codes and their application to reliable delivery of media streams in the context of theRLC over GF(2^^8)FECScheme will be mostly used since it warrantsFramework. The Raptor and RaptorQ codes are systematic codes, where ahigher packet loss protection. In casenumber ofsmall encoding windows, the associated processing overhead is not an issue (e.g., we measured decoding speeds between 745 Mbpsrepair symbols are generated from a set of source symbols and2.8 Gbps on an ARM Cortex-A15 embedded boardsent in<xref target="Roca17"/> depending on the code rate and the channel conditions, using an encoding window of size 18one or23 symbols; see the above article for the details). Of course the CPU overhead will increase with the encoding window size, becausemoreoperationsrepair flows in addition to theGF(2^^8) finite field will be needed. </t> <t> The RLC over GF(2) FEC Scheme offers an alternative. In that case operationssource symbolscan be directly XOR-ed together which warrants high bitrate encoding and decoding operations,that are sent to the receiver(s) within a source flow. The Raptor andcan be an advantage with large encoding windows. However, packet lossRaptorQ codes offer close to optimal protectionis significantly reduced by using thisagainst arbitrary packet losses at a low computational complexity. Six FECScheme. </t> </section> <section title="Operational Recommendations: Coding Coefficients Density Threshold"> <!-- ================ --> <t> In addition toSchemes are defined: two for thechoiceprotection ofthe Finite Field, thearbitrary packet flows, twoFEC Schemes definethat are optimized for small source blocks, and two for the protection of a single flow that already contains acoding coefficient density threshold (DT) parameter. This parameter enablessequence number. Repair data may be sent over arbitrary datagram transport (e.g., UDP) or using RTP. [STANDARDS-TRACK]</t> </abstract> </front> <seriesInfo name="RFC" value="6681"/> <seriesInfo name="DOI" value="10.17487/RFC6681"/> </reference> <reference anchor="RFC6726" target="https://www.rfc-editor.org/info/rfc6726" quoteTitle="true" derivedAnchor="RFC6726"> <front> <title>FLUTE - File Delivery over Unidirectional Transport</title> <author initials="T." surname="Paila" fullname="T. Paila"> <organization showOnFrontPage="true"/> </author> <author initials="R." surname="Walsh" fullname="R. Walsh"> <organization showOnFrontPage="true"/> </author> <author initials="M." surname="Luby" fullname="M. Luby"> <organization showOnFrontPage="true"/> </author> <author initials="V." surname="Roca" fullname="V. Roca"> <organization showOnFrontPage="true"/> </author> <author initials="R." surname="Lehtonen" fullname="R. Lehtonen"> <organization showOnFrontPage="true"/> </author> <date year="2012" month="November"/> <abstract> <t>This document defines File Delivery over Unidirectional Transport (FLUTE), asender to control the code density, i.e.,protocol for theproportionunidirectional delivery ofcoefficients that are non zero on average. With RLCfiles overGF(2^^8), itthe Internet, which isusually appropriate that small encoding windows be associated to a density threshold equalparticularly suited to15,multicast networks. The specification builds on Asynchronous Layered Coding, themaximum value, in order to warrantbase protocol designed for massively scalable multicast distribution. This document obsoletes RFC 3926. [STANDARDS-TRACK]</t> </abstract> </front> <seriesInfo name="RFC" value="6726"/> <seriesInfo name="DOI" value="10.17487/RFC6726"/> </reference> <reference anchor="RFC6816" target="https://www.rfc-editor.org/info/rfc6816" quoteTitle="true" derivedAnchor="RFC6816"> <front> <title>Simple Low-Density Parity Check (LDPC) Staircase Forward Error Correction (FEC) Scheme for FECFRAME</title> <author initials="V." surname="Roca" fullname="V. Roca"> <organization showOnFrontPage="true"/> </author> <author initials="M." surname="Cunche" fullname="M. Cunche"> <organization showOnFrontPage="true"/> </author> <author initials="J." surname="Lacan" fullname="J. Lacan"> <organization showOnFrontPage="true"/> </author> <date year="2012" month="December"/> <abstract> <t>This document describes ahigh loss protection. </t> <t> On the opposite, with larger encoding windows, it is usually appropriatefully specified simple Forward Error Correction (FEC) scheme for Low-Density Parity Check (LDPC) Staircase codes thatthe density threshold be reduced. With large encoding windows, an alternativecan be used touse RLC over GF(2)protect media streams along the lines defined by FECFRAME. These codes have many interesting properties: they are systematic codes, they perform close to ideal codes in many use-cases, and they also feature very high encoding and decoding throughputs. LDPC-Staircase codes are therefore adensity threshold equal to 7 (i.e., an average density equalgood solution to1/2) or smaller. </t> <t> Note that usingprotect adensity threshold equal to 15 with RLC over GF(2) is equivalent to using an XOR code that computes the XOR sum of all thesingle high bitrate sourcesymbols in the encoding window. In that case: (1) onlyflow or to protect globally several mid-rate flows within a singlerepair symbol canFECFRAME instance. They are also a good solution whenever the processing load of a software encoder or decoder must beproducedkept to a minimum.</t> </abstract> </front> <seriesInfo name="RFC" value="6816"/> <seriesInfo name="DOI" value="10.17487/RFC6816"/> </reference> <reference anchor="RFC6865" target="https://www.rfc-editor.org/info/rfc6865" quoteTitle="true" derivedAnchor="RFC6865"> <front> <title>Simple Reed-Solomon Forward Error Correction (FEC) Scheme forany encoding window, and (2) the repair_key parameter becomes useless (the coding coefficients generation function does not rely on the PRNG). </t> </section> </section> <section anchor="iana" title="IANA Considerations"> <!-- ====================== --> <t> ThisFECFRAME</title> <author initials="V." surname="Roca" fullname="V. Roca"> <organization showOnFrontPage="true"/> </author> <author initials="M." surname="Cunche" fullname="M. Cunche"> <organization showOnFrontPage="true"/> </author> <author initials="J." surname="Lacan" fullname="J. Lacan"> <organization showOnFrontPage="true"/> </author> <author initials="A." surname="Bouabdallah" fullname="A. Bouabdallah"> <organization showOnFrontPage="true"/> </author> <author initials="K." surname="Matsuzono" fullname="K. Matsuzono"> <organization showOnFrontPage="true"/> </author> <date year="2013" month="February"/> <abstract> <t>This documentregisters two values indescribes a fully-specified simple Forward Error Correction (FEC) scheme for Reed-Solomon codes over the"FEC Framework (FECFRAME) FEC Encoding IDs" registry <xref target="RFC6363"/>finite field (also known asfollows: <list style="symbols"> <t>YYYY refers totheSliding Window Random Linear Codes (RLC) over GF(2) FEC Scheme for Arbitrary Packet Flows, as defined in <xref target="ArbitraryFlows_RLC_GF_2"/> of this document.</t> <t>XXXX refersGalois Field) GF(2^^m), with 2 <= m <= 16, that can be used to protect arbitrary media streams along theSliding Window Random Linear Codes (RLC) over GF(2^^8) FEC Scheme for Arbitrary Packet Flows, aslines definedin <xref target="ArbitraryFlows_RLC_GF_28"/> of this document.</t> </list> </t> </section> <section title="Acknowledgments"> <!-- ====================== --> <t>by FECFRAME. Theauthors would like to thank the three TSVWG chairs, Wesley Eddy, our shepherd, David Black and Gorry Fairhurst, as well as Spencer Dawkins, our responsible AD, and all those who provided comments, namely (alphabetical order) Alan DeKok, Jonathan Detchart, Russ Housley, Emmanuel Lochin, Marie-Jose Montpetit,Reed-Solomon codes considered have attractive properties, since they offer optimal protection against packet erasures andGreg Skinner. Last but not least,theauthors are really gratefulsource symbols are part of the encoding symbols, which can greatly simplify decoding. However, the price to pay is a limit on theIESG members, in particular Benjamin Kaduk, Mirja Kuhlewind, Eric Rescorla, Adam Roach,maximum source block size, on the maximum number of encoding symbols, andRoman Danyliw for their highly valuable feedbacksa computational complexity higher than thatgreatly contributed to improve this specification. </t> </section> </middle> <back> <references title="Normative References"> <!-- ====================== --> &rfc2119; &rfc8174; &rfc6363; &rfc6364;of the Low-Density Parity Check (LDPC) codes, for instance.</t> </abstract> </front> <seriesInfo name="RFC" value="6865"/> <seriesInfo name="DOI" value="10.17487/RFC6865"/> </reference> <referenceanchor="fecframe-ext" target="https://tools.ietf.org/html/draft-ietf-tsvwg-fecframe-ext">anchor="RFC8406" target="https://www.rfc-editor.org/info/rfc8406" quoteTitle="true" derivedAnchor="RFC8406"> <front><title>Forward Error Correction (FEC) Framework Extension to Sliding Window Codes</title><title>Taxonomy of Coding Techniques for Efficient Network Communications</title> <authorinitials='V.' surname='Roca'>initials="B." surname="Adamson" fullname="B. Adamson"> <organization/>showOnFrontPage="true"/> </author> <authorinitials='A.' surname='Begen'>initials="C." surname="Adjih" fullname="C. Adjih"> <organization/>showOnFrontPage="true"/> </author><date month="January" year="2019" /> </front> <seriesInfo name='Transport Area Working Group (TSVWG)' value='draft-ietf-tsvwg-fecframe-ext (Work in Progress)' /> </reference> <reference anchor="tinymt32" target="https://tools.ietf.org/html/draft-roca-tsvwg-tinymt32"> <front> <title>TinyMT32 Pseudo Random Number Generator (PRNG)</title><authorinitials="M" surname="Saito">initials="J." surname="Bilbao" fullname="J. Bilbao"> <organization/>showOnFrontPage="true"/> </author> <authorinitials="M" surname="Matsumoto">initials="V." surname="Firoiu" fullname="V. Firoiu"> <organization showOnFrontPage="true"/> </author> <author initials="F." surname="Fitzek" fullname="F. Fitzek"> <organization showOnFrontPage="true"/> </author> <author initials="S." surname="Ghanem" fullname="S. Ghanem"> <organization showOnFrontPage="true"/> </author> <author initials="E." surname="Lochin" fullname="E. Lochin"> <organization showOnFrontPage="true"/> </author> <author initials="A." surname="Masucci" fullname="A. Masucci"> <organization showOnFrontPage="true"/> </author> <author initials="M-J." surname="Montpetit" fullname="M-J. Montpetit"> <organization showOnFrontPage="true"/> </author> <author initials="M." surname="Pedersen" fullname="M. Pedersen"> <organization/>showOnFrontPage="true"/> </author> <author initials="G." surname="Peralta" fullname="G. Peralta"> <organization showOnFrontPage="true"/> </author> <author initials="V."surname="Roca">surname="Roca" fullname="V. Roca" role="editor"> <organization/>showOnFrontPage="true"/> </author> <authorinitials="E" surname="Baccelli">initials="P." surname="Saxena" fullname="P. Saxena"> <organization showOnFrontPage="true"/> </author> <author initials="S." surname="Sivakumar" fullname="S. Sivakumar"> <organization/>showOnFrontPage="true"/> </author> <datemonth="February" year="2019" /> </front> <seriesInfo name='Transport Area Workingyear="2018" month="June"/> <abstract> <t>This document summarizes recommended terminology for Network Coding concepts and constructs. It provides a comprehensive set of terms in order to avoid ambiguities in future IRTF and IETF documents on Network Coding. This document is the product of the Coding for Efficient Network Communications Research Group(TSVWG)' value='draft-roca-tsvwg-tinymt32 (Work(NWCRG), and it is inProgress)' /> </reference> <reference anchor="C99"> <front> <title>Programming languages - C: C99, correction 3:2007</title> <author /> <date month="November" year="2007" />line with the terminology used by the RFCs produced by the Reliable Multicast Transport (RMT) and FEC Framework (FECFRAME) IETF working groups.</t> </abstract> </front> <seriesInfoname="International Organization for Standardization," value="ISO/IEC 9899:1999/Cor 3:2007" />name="RFC" value="8406"/> <seriesInfo name="DOI" value="10.17487/RFC8406"/> </reference></references> <references title="Informative References"> <!-- ====================== --> &rfc5170; &rfc5510; &rfc6726; &rfc6681; <!-- raptor(Q) for ff --> &rfc6816; <!-- ldpc-staircase for ff --> &rfc6865; <!-- r-s for ff --> &rfc8406;<reference anchor="Roca16"target="https://hal.inria.fr/hal-01395937/en/">target="https://hal.inria.fr/hal-01395937/en/" quoteTitle="true" derivedAnchor="Roca16"> <front> <title>Block or Convolutional AL-FEC Codes? A Performance Comparison for Robust Low-Latency Communications</title> <seriesInfo name="HAL ID" value="hal-01395937v2"/> <authorinitials='V.' surname='Roca'>initials="V." surname="Roca"> <organization/>showOnFrontPage="true"/> </author> <authorinitials='B.' surname='Teibi'>initials="B." surname="Teibi"> <organization/>showOnFrontPage="true"/> </author> <authorinitials='C.' surname='Burdinat'>initials="C." surname="Burdinat"> <organization/>showOnFrontPage="true"/> </author> <authorinitials='T.' surname='Tran'>initials="T." surname="Tran-Thai"> <organization/>showOnFrontPage="true"/> </author> <authorinitials='C.' surname='Thienot'>initials="C." surname="Thienot"> <organization/>showOnFrontPage="true"/> </author> <datemonth="November" year="2016" />month="February" year="2017"/> </front><seriesInfo name='HAL open-archive document,hal-01395937' value='https://hal.inria.fr/hal-01395937/en/' /></reference> <reference anchor="Roca17"target="https://hal.inria.fr/hal-01571609v1/en/">target="https://hal.inria.fr/hal-01571609v1/en/" quoteTitle="true" derivedAnchor="Roca17"> <front> <title>Less Latency and Better Protection with AL-FEC Sliding Window Codes: a Robust Multimedia CBR Broadcast Case Study</title><author initials='V.' surname='Roca'> <organization /> </author> <author initials='B.' surname='Teibi'> <organization /> </author> <author initials='C.' surname='Burdinat'> <organization /> </author> <author initials='T.' surname='Tran'> <organization /> </author> <author initials='C.' surname='Thienot'> <organization /> </author> <date month="October" year="2017" /> </front><seriesInfoname='13th IEEE International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob17), October 2017' value='https://hal.inria.fr/hal-01571609v1/en/' /> </reference> <reference anchor="PGM13" target="http://web.eecs.utk.edu/~plank/plank/papers/UT-CS-13-717.html"> <front> <title>A Complete Treatment of Software Implementations of Finite Field Arithmetic for Erasure Coding Applications</title>name="HAL ID" value="hal-01571609"/> <author initials="V." surname="Roca"> <organization showOnFrontPage="true"/> </author> <authorinitials="J." surname="Plank">initials="B." surname="Teibi"> <organization/>showOnFrontPage="true"/> </author> <authorinitials="K." surname="Greenan">initials="C." surname="Burdinat"> <organization/>showOnFrontPage="true"/> </author> <authorinitials="E." surname="Miller">initials="T." surname="Tran"> <organization showOnFrontPage="true"/> </author> <author initials="C." surname="Thienot"> <organization/>showOnFrontPage="true"/> </author> <date month="October"year="2013" />year="2017"/> </front><seriesInfo name="University of Tennessee Technical Report UT-CS-13-717," value="http://web.eecs.utk.edu/~plank/plank/papers/UT-CS-13-717.html" /><refcontent>13th IEEE International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob17)</refcontent> </reference> </references><!-- ====================== --></references> <section anchor="annex_tinymt32_validation"title="TinyMT32numbered="true" toc="include" removeInRFC="false" pn="section-appendix.a"> <name slugifiedName="name-tinymt32-validation-criteri">TinyMT32 Validation Criteria(Normative)"> <!-- ====================== --> <t>(Normative)</name> <t pn="section-appendix.a-1"> PRNG determinism, for a given seed, is a requirement. Consequently, in order to validate an implementation of the TinyMT32 PRNG, the following criteriaMUST<bcp14>MUST</bcp14> be met. </t><t><t pn="section-appendix.a-2"> The first criterionfocussesfocuses on the tinymt32_rand256(), where the 32-bit integer of the core TinyMT32 PRNG is scaled down to an 8-bit integer. Using a seed value of 1, the first 50 values returned by: tinymt32_rand256() as 8-bit unsigned integersMUST<bcp14>MUST</bcp14> be equal to values provided in <xreftarget="fig_tinymt32_out_truncated_256"/>,target="fig_tinymt32_out_truncated_256" format="default" sectionFormat="of" derivedContent="Figure 9"/>, to be read line by line. </t> <figure anchor="fig_tinymt32_out_truncated_256"title="Firstalign="left" suppress-title="false" pn="figure-9"> <name slugifiedName="name-first-50-decimal-values-to-">First 50 decimal values (to be read per line) returned by tinymt32_rand256() as 8-bit unsigned integers, with a seed value of1."> <artwork><![CDATA[1</name> <artwork name="" type="" align="left" alt="" pn="section-appendix.a-3.1"> 37 225 177 176 21 246 54 139 168 237 211 187 62 190 104 135 210 99 176 11 207 35 40 113 179 214 254 101 212 211 226 41 234 232 203 29 194 211 112 107 217 104 197 135 23 89 210 252 109 166]]></artwork></artwork> </figure><t><t pn="section-appendix.a-4"> The second criterionfocussesfocuses on the tinymt32_rand16(), where the 32-bit integer of the core TinyMT32 PRNG is scaled down to a 4-bit integer. Using a seed value of 1, the first 50 values returned by: tinymt32_rand16() as 4-bit unsigned integersMUST<bcp14>MUST</bcp14> be equal to values provided in <xreftarget="fig_tinymt32_out_truncated_16"/>,target="fig_tinymt32_out_truncated_16" format="default" sectionFormat="of" derivedContent="Figure 10"/>, to be read line by line. </t> <figure anchor="fig_tinymt32_out_truncated_16"title="Firstalign="left" suppress-title="false" pn="figure-10"> <name slugifiedName="name-first-50-decimal-values-to-b">First 50 decimal values (to be read per line) returned by tinymt32_rand16() as 4-bit unsigned integers, with a seed value of1."> <artwork><![CDATA[1</name> <artwork name="" type="" align="left" alt="" pn="section-appendix.a-5.1"> 5 1 1 0 5 6 6 11 8 13 3 11 14 14 8 7 2 3 0 11 15 3 8 1 3 6 14 5 4 3 2 9 10 8 11 13 2 3 0 11 9 8 5 7 7 9 2 12 13 6]]></artwork></artwork> </figure> </section> <section anchor="annex_assessing_prng"title="Assessingnumbered="true" toc="include" removeInRFC="false" pn="section-appendix.b"> <name slugifiedName="name-assessing-the-prng-adequacy">Assessing the PRNG Adequacy(Informational)"> <!-- ====================== --> <t>(Informational)</name> <t pn="section-appendix.b-1"> This annex discusses the adequacy of the TinyMT32 PRNG and the tinymt32_rand16() and tinymt32_rand256() functions, to the RLC FECSchemes.schemes. The goal is to assess the adequacy of these two functions in producing coding coefficients that are sufficiently different from one another, across various repair symbols with repair key values in sequence (we can expect this approach to be commonly used by implementers, see <xreftarget="ArbitraryFlows_FECCodeSpecification_encoding"/>).target="ArbitraryFlows_FECCodeSpecification_encoding" format="default" sectionFormat="of" derivedContent="Section 6.1"/>). This section is purely informational and does not claim to be a solid evaluation. </t><t><t pn="section-appendix.b-2"> The two RLC FECSchemesschemes use the PRNG to producepseudo-randompseudorandom coding coefficients (<xreftarget="CommonProc_coef_generation_func"/>),target="CommonProc_coef_generation_func" format="default" sectionFormat="of" derivedContent="Section 3.6"/>), each time a new repair symbol is needed. A different repair key is used for each repair symbol, usually by incrementing the repair key value (<xreftarget="ArbitraryFlows_FECCodeSpecification_encoding"/>).target="ArbitraryFlows_FECCodeSpecification_encoding" format="default" sectionFormat="of" derivedContent="Section 6.1"/>). For each repair symbol, a limited number ofpseudo-randompseudorandom numbers is needed, depending on the DT and encoding window size (<xreftarget="CommonProc_coef_generation_func"/>),target="CommonProc_coef_generation_func" format="default" sectionFormat="of" derivedContent="Section 3.6"/>), using either tinymt32_rand16() or tinymt32_rand256().ThereforeTherefore, we are more interested in the randomness of small sequences of random numbers mapped to 4-bit or 8-bit integers, than in the randomness of a very large sequence of random numbers which is not representative of the usage of the PRNG. </t><t><t pn="section-appendix.b-3"> Evaluation of tinymt32_rand16(): We first generate a huge number (1,000,000,000) of small sequences (20pseudo-randompseudorandom numbers per sequence), increasing the seed value for each sequence, and perform statistics on the number of occurrences of each of the 16 possible values across all sequences. In this first test we consider 32-bit seed values in order to assess the PRNG quality after output truncation to 4 bits.<figure anchor="fig_tinymt32_out_truncated_16_huge_nb_small_seq" title="tinymt32_rand16(): occurrence statistics across</t> <table anchor="table_tinymt32_out_truncated_16_huge_nb_small_seq" align="center" pn="table-1"> <name slugifiedName="name-tinymt32_rand16-occurrence-">tinymt32_rand16() Occurrence Statistics</name> <thead> <tr> <th align="left" colspan="1" rowspan="1">Value</th> <th align="left" colspan="1" rowspan="1"> Occurrences</th> <th align="left" colspan="1" rowspan="1">Percentage (%)</th> </tr> </thead> <tbody> <tr> <td align="left" colspan="1" rowspan="1">0</td> <td align="left" colspan="1" rowspan="1">1250036799</td> <td align="left" colspan="1" rowspan="1">6.2502</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">1</td> <td align="left" colspan="1" rowspan="1">1249995831</td> <td align="left" colspan="1" rowspan="1">6.2500</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">2</td> <td align="left" colspan="1" rowspan="1">1250038674</td> <td align="left" colspan="1" rowspan="1">6.2502</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">3</td> <td align="left" colspan="1" rowspan="1">1250000881</td> <td align="left" colspan="1" rowspan="1">6.2500</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">1250023929</td> <td align="left" colspan="1" rowspan="1">6.2501</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">1249986320</td> <td align="left" colspan="1" rowspan="1">6.2499</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">6</td> <td align="left" colspan="1" rowspan="1">1249995587</td> <td align="left" colspan="1" rowspan="1">6.2500</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">7</td> <td align="left" colspan="1" rowspan="1">1250020363</td> <td align="left" colspan="1" rowspan="1">6.2501</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">8</td> <td align="left" colspan="1" rowspan="1">1249995276</td> <td align="left" colspan="1" rowspan="1">6.2500</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">9</td> <td align="left" colspan="1" rowspan="1">1249982856</td> <td align="left" colspan="1" rowspan="1">6.2499</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">10</td> <td align="left" colspan="1" rowspan="1">1249984111</td> <td align="left" colspan="1" rowspan="1">6.2499</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">11</td> <td align="left" colspan="1" rowspan="1">1250009551</td> <td align="left" colspan="1" rowspan="1">6.2500</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">12</td> <td align="left" colspan="1" rowspan="1">1249955768</td> <td align="left" colspan="1" rowspan="1">6.2498</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">13</td> <td align="left" colspan="1" rowspan="1">1249994654</td> <td align="left" colspan="1" rowspan="1">6.2500</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">14</td> <td align="left" colspan="1" rowspan="1">1250000569</td> <td align="left" colspan="1" rowspan="1">6.2500</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">15</td> <td align="left" colspan="1" rowspan="1">1249978831</td> <td align="left" colspan="1" rowspan="1">6.2499</td> </tr> </tbody> </table> <t pn="section-appendix.b-5"> Evaluation of tinymt32_rand16(): We first generate a huge number (1,000,000,000) of small sequences (20pseudo-randompseudorandom numbers per sequence),with 0 asincreasing thefirst PRNG seed."> <artwork><![CDATA[seed value for each sequence, and perform statistics on the number of occurrencespercentage (%) (totalof20000000000) 0 1250036799 6.2502 1 1249995831 6.2500 2 1250038674 6.2502 3 1250000881 6.2500each of the 16 possible values across the 20,000,000,000 numbers of all sequences. In this first test, we consider 32-bit seed values in order to assess the PRNG quality after output truncation to 41250023929 6.2501 5 1249986320 6.2499 6 1249995587 6.2500 7 1250020363 6.2501 8 1249995276 6.2500 9 1249982856 6.2499 10 1249984111 6.2499 11 1250009551 6.2500 12 1249955768 6.2498 13 1249994654 6.2500 14 1250000569 6.2500 15 1249978831 6.2499 ]]></artwork> </figure>bits. </t> <t pn="section-appendix.b-6"> The results (<xreftarget="fig_tinymt32_out_truncated_16_huge_nb_small_seq"/>)target="table_tinymt32_out_truncated_16_huge_nb_small_seq" format="default" sectionFormat="of" derivedContent="Table 1"/>) show that all possible values are almost equally represented, or said differently, that the tinymt32_rand16() output converges to a uniform distribution where each of the 16 possible values would appear exactly 1 / 16 * 100 = 6.25% of times. </t><t><t pn="section-appendix.b-7"> Since the RLC FECSchemesschemes use of this PRNG will be limited to 16-bit seed values, we carried out the same test for the first2^^162<sup>16</sup> seed values only. The distribution (not shown) is of course less uniform, with valueoccurencesoccurrences ranging between 6.2121% (i.e., 81,423occurencesoccurrences out of a total of 65536*20=1,310,720) and 6.2948% (i.e., 82,507occurences).occurrences). However, we donot believe it significantly impacts the RLC FEC Scheme behavior. </t> <t> Other types of biases may exist that may be visible with smaller tests, for instance to evaluate the convergence speed to a uniform distribution. We therefore perform 200 tests, each of them consisting in producing 200 sequences, keeping only the first value of each sequence. We use non overlapping repair keys for each sequence, starting with value 0 and increasing it after each use. <!-- <figure anchor="fig_tinymt32_out_truncated_16_small_nb_small_seq" title="tinymt32_rand16(): occurrence statistics across a small number (100) of sequences, keeping only the first value of each sequence, with 0 as the first PRNG seed."> <artwork><![CDATA[ value occurrences percentage (total of 200) 0 13 6.5000 1 11 5.5000 2 15 7.5000 3 10 5.0000 4 15 7.5000 5 17 8.5000 6 11 5.5000 7 14 7.0000 8 10 5.0000 9 11 5.5000 10 12 6.0000 11 11 5.5000 12 12 6.0000 13 17 8.5000 14 13 6.5000 15 8 4.0000 ]]></artwork> </figure> --> <figure anchor="fig_tinymt32_out_truncated_16_small_nb_small_seq" title="tinymt32_rand16(): occurrence statistics acrossnot believe it significantly impacts the RLC FEC scheme behavior. </t> <t pn="section-appendix.b-8"> Other types of biases may exist that may be visible with smaller tests, for instance to evaluate the convergence speed to a uniform distribution. We therefore perform 200 tests, each of themconsisting inproducing 200sequencessequences, keeping only the first value of1 pseudo-random number each, with non overlapping PRNG seeds in sequenceeach sequence. We use non-overlapping repair keys for each sequence, startingfrom 0."> <artwork><![CDATA[with valuemin occurrences max occurrences average occurrences04 21 6.3675 1 4 22 6.0200 2 4 20 6.3125 3 5 23 6.1775 4 5 24 6.1000 5 4 21 6.5925 6 5 30 6.3075 7 6 22 6.2225 8 5 26 6.1750 9 3 21 5.9425 10 5 24 6.3175 11 4 22 6.4300 12 5 21 6.1600 13 5 22 6.3100 14 4 26 6.3950 15 4 21 6.1700 ]]></artwork> </figure> <xref target="fig_tinymt32_out_truncated_16_small_nb_small_seq"/>and increasing it after each use. </t> <table anchor="table_tinymt32_out_truncated_16_small_nb_small_seq" align="center" pn="table-2"> <name slugifiedName="name-tinymt32_rand16-occurrence-s">tinymt32_rand16() Occurrence Statistics</name> <thead> <tr> <th align="left" colspan="1" rowspan="1">Value</th> <th align="left" colspan="1" rowspan="1">Min Occurrences</th> <th align="left" colspan="1" rowspan="1">Max Occurrences</th> <th align="left" colspan="1" rowspan="1">Average Occurrences</th> </tr> </thead> <tbody> <tr> <td align="left" colspan="1" rowspan="1">0</td> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">21</td> <td align="left" colspan="1" rowspan="1">6.3675</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">1</td> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">22</td> <td align="left" colspan="1" rowspan="1">6.0200</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">2</td> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">20</td> <td align="left" colspan="1" rowspan="1">6.3125</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">3</td> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">23</td> <td align="left" colspan="1" rowspan="1">6.1775</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">24</td> <td align="left" colspan="1" rowspan="1">6.1000</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">21</td> <td align="left" colspan="1" rowspan="1">6.5925</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">6</td> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">30</td> <td align="left" colspan="1" rowspan="1">6.3075</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">7</td> <td align="left" colspan="1" rowspan="1">6</td> <td align="left" colspan="1" rowspan="1">22</td> <td align="left" colspan="1" rowspan="1">6.2225</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">8</td> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">26</td> <td align="left" colspan="1" rowspan="1">6.1750</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">9</td> <td align="left" colspan="1" rowspan="1">3</td> <td align="left" colspan="1" rowspan="1">21</td> <td align="left" colspan="1" rowspan="1">5.9425</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">10 </td> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">24</td> <td align="left" colspan="1" rowspan="1">6.3175</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">11 </td> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">22</td> <td align="left" colspan="1" rowspan="1">6.4300</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">12 </td> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">21</td> <td align="left" colspan="1" rowspan="1">6.1600</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">13 </td> <td align="left" colspan="1" rowspan="1">5</td> <td align="left" colspan="1" rowspan="1">22</td> <td align="left" colspan="1" rowspan="1">6.3100</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">14 </td> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">26</td> <td align="left" colspan="1" rowspan="1">6.3950</td> </tr> <tr> <td align="left" colspan="1" rowspan="1">15 </td> <td align="left" colspan="1" rowspan="1">4</td> <td align="left" colspan="1" rowspan="1">21</td> <td align="left" colspan="1" rowspan="1">6.1700</td> </tr> </tbody> </table> <t pn="section-appendix.b-10"><xref target="table_tinymt32_out_truncated_16_small_nb_small_seq" format="default" sectionFormat="of" derivedContent="Table 2"/> shows across all 200 tests, for each of the 16 possiblepseudo-randompseudorandom number values, the minimum (resp. maximum) number of times it appeared in a test, as well as the average number of occurrences across the 200 tests. Although the distribution is not perfect, there is no major bias. On theopposite,contrary, in the same conditions, the Park-Miller linear congruential PRNG of <xreftarget="RFC5170"/>target="RFC5170" format="default" sectionFormat="of" derivedContent="RFC5170"/> with a result scaled down to 4-bit values, using seeds in sequence starting from 1,returnssystematically returns 0 as the first value during sometime, thentime. Then, after a certain repair key value threshold, it systematically returns 1, etc. </t><t><t pn="section-appendix.b-11"> Evaluation of tinymt32_rand256(): The same approach is used here. Results (not shown) are similar: occurrences vary between 7,810,3368 (i.e., 0.3905%) and 7,814,7952 (i.e., 0.3907%). Here also we see a convergence to the theoretical uniform distribution where each of the 256 possible values would appear exactly 1 / 256 * 100 = 0.390625% of times. </t> </section> <section anchor="possible_param_derivation"title="Possiblenumbered="true" toc="include" removeInRFC="false" pn="section-appendix.c"> <name slugifiedName="name-possible-parameter-derivati">Possible Parameter Derivation(Informational)"> <!-- ====================== --> <t> <xref target="CommonProc_rlcParameters"/>(Informational)</name> <t pn="section-appendix.c-1"><xref target="CommonProc_rlcParameters" format="default" sectionFormat="of" derivedContent="Section 3.1"/> defines several parameters to control the encoder or decoder. This annex proposes techniques to derive these parameters according to the target use-case. This annex is informational, in the sense that using a different derivation technique will not prevent the encoder and decoder to interoperate: a decoder can still recover an erased source symbol without any error. However, in case of a real-time flow, an inappropriate parameter derivation may lead to the decoding of erased source packets after their validity period, making them useless to the target application. This annex proposes an approach to reduce this risk, among other things. </t><t><t pn="section-appendix.c-2"> The FECSchemesschemes defined in this document can be used in various manners, depending on the target use-case:<list style="symbols"> <t></t> <ul spacing="normal" bare="false" empty="false" pn="section-appendix.c-3"> <li pn="section-appendix.c-3.1"> the source ADU flow they protect may or may not have real-timeconstraints;</t> <t>constraints;</li> <li pn="section-appendix.c-3.2"> the source ADU flow may be a Constant Bitrate (CBR) or VariableBitRateBitrate (VBR)flow;</t> <t>flow;</li> <li pn="section-appendix.c-3.3"> with a VBR source ADU flow, the flow's minimum and maximum bitrates may or may not beknown;</t> <t>known;</li> <li pn="section-appendix.c-3.4"> and the communication path between encoder and decoder may be a CBR communication path (e.g., as with certain LTE-based broadcast channels) or not (general case, e.g., withInternet).</t> </list>Internet).</li> </ul> <t pn="section-appendix.c-4"> The parameter derivation technique should be suited to the use-case, as described in the following sections. </t> <section anchor="param_derivation_cbr_realtime"title="Casenumbered="true" toc="include" removeInRFC="false" pn="section-c.1"> <name slugifiedName="name-case-of-a-cbr-real-time-flo">Case of a CBR Real-TimeFlow"> <!-- ====================== --> <t>Flow</name> <t pn="section-c.1-1"> In the following, we consider a real-time flow with max_lat latency budget. The encoding symbol size, E, is constant. The code rate, cr, is also constant, its value depending on the expected communication loss model (this choice is out of scope of this document). </t><t><t pn="section-c.1-2"> In a first configuration, the source ADU flow bitrate at the input of the FECFRAME sender is fixed and equal to br_in (in bits/s), and this value is known by the FECFRAME sender. It follows that the transmission bitrate at the output of the FECFRAME sender will be higher, depending on the added repair flow overhead. In order to comply with the maximum FEC-related latency budget, we have:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-c.1-3"> <li pn="section-c.1-3.1"> dw_max_size = (max_lat * br_in) / (8 * E)</t> </list></li> </ul> <t pn="section-c.1-4"> assuming that the encoding and decoding times are negligible with respect to the target max_lat. This is a reasonable assumption in many situations (e.g., see <xreftarget="oprecom_ff_considerations"/>target="oprecom_ff_considerations" format="default" sectionFormat="of" derivedContent="Section 8.1"/> in case of small window sizes). Otherwise the max_lat parameter should be adjusted in order to avoid the problem. In any case, interoperability will never becompromizedcompromised by choosing a too large value. </t><t><t pn="section-c.1-5"> In a second configuration, the FECFRAME sender generates a fixed bitrate flow, equal to the CBR communication path bitrate equal to br_out (in bits/s), and this value is known by the FECFRAME sender, as in <xreftarget="Roca17"/>.target="Roca17" format="default" sectionFormat="of" derivedContent="Roca17"/>. The maximum source flow bitrate needs to be such that, with the added repair flow overhead, the total transmission bitrate remains inferior or equal to br_out. We have:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-c.1-6"> <li pn="section-c.1-6.1"> dw_max_size = (max_lat * br_out * cr) / (8 * E)</t> </list></li> </ul> <t pn="section-c.1-7"> assuming here also that the encoding and decoding times are negligible with respect to the target max_lat. </t><t><t pn="section-c.1-8"> For decoding to be possible within the latency budget, it is required that the encoding window maximum size be smaller than or at most equal to the decoding window maximum size. The ew_max_size is the main parameter at a FECFRAME sender, but its exact value has no impact on thetheFEC-related latency budget. The ew_max_size parameter is computed as follows:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-c.1-9"> <li pn="section-c.1-9.1"> ew_max_size = dw_max_size * WSR /255</t> <!-- <t> ew_max_size = dw_max_size * 0.75 </t> --> </list>255</li> </ul> <t pn="section-c.1-10"> In line with <xreftarget="Roca17"/>,target="Roca17" format="default" sectionFormat="of" derivedContent="Roca17"/>, WSR = 191 is considered as areasonable value (the resulting encoding to decoding window size ratio is then close to 0.75), but other values between 1 and 255 inclusive are possible, depending on the use-case. <!-- It is always RECOMMENDED to check that the ew_max_size value stays within reasonable bounds in order to avoid hazardous behaviours. --> <!-- However, any value ew_max_size < dw_max_size can be used without impact on the FEC-related latency budget. Another value could be determined depending on the use-case details, which is out of scope of this document. Whenever the FEC protection (i.e., cr value) is sufficient in front of the experienced packet losses, the ew_max_size guaranties that the recovery of lost ADUs will happen at a FECFRAME receiverreasonable value (the resulting encoding to decoding window size ratio is then close to 0.75), but other values between 1 and 255 inclusive are possible, depending ontime. -->the use-case. </t><t><t pn="section-c.1-11"> The dw_max_size is computed by a FECFRAME sender but not explicitly communicated to a FECFRAME receiver. However, a FECFRAME receiver can easily evaluate the ew_max_size by observing the maximum Number of Source Symbols (NSS) value contained in the Repair FEC Payload ID of received FEC Repair Packets (<xreftarget="ArbitraryFlows_repair_fpi"/>).target="ArbitraryFlows_repair_fpi" format="default" sectionFormat="of" derivedContent="Section 4.1.3"/>). A receiver can then easily compute dw_max_size:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-c.1-12"> <li pn="section-c.1-12.1"> dw_max_size = max_NSS_observed * 255 / WSR</t> <!-- <t> dw_max_size = max_NSS_observed / 0.75 </t> --> </list></li> </ul> <t pn="section-c.1-13"> A receiver can thenchosechoose an appropriate linear system maximum size:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-c.1-14"> <li pn="section-c.1-14.1"> ls_max_size≥>= dw_max_size</t> </list></li> </ul> <t pn="section-c.1-15"> It is good practice to use a larger value for ls_max_size as explained in <xreftarget="decodingBeyondMaxLatency"/>,target="decodingBeyondMaxLatency" format="default" sectionFormat="of" derivedContent="Appendix D"/>, which does not impact maximum latency nor interoperability. </t><t><t pn="section-c.1-16"> In any case, for a given use-case (i.e., for target encoding and decoding devices and desired protection levels in front of communication impairments) and for the computed ew_max_size, dw_max_size and ls_max_size values, it isRECOMMENDED<bcp14>RECOMMENDED</bcp14> to check that the maximum encoding time and maximum memory requirements at a FECFRAME sender, and maximum decoding time and maximum memory requirements at a FECFRAME receiver, stay within reasonable bounds. When assuming that the encoding and decoding times are negligible with respect to the target max_lat, this should be verified as well, otherwise the max_latSHOULD<bcp14>SHOULD</bcp14> be adjusted accordingly. </t><t><t pn="section-c.1-17"> The particular case of session start needs to be managed appropriately since the ew_size, starting at zero, increases each time a new source ADU is received by the FECFRAME sender, until it reaches the ew_max_size value.ThereforeTherefore, a FECFRAME receiverSHOULD<bcp14>SHOULD</bcp14> continuously observe the received FEC Repair Packets, since the NSS value carried in the Repair FEC Payload ID will increase too, and adjust its ls_max_size accordingly if need be. With a CBR flow, session start is expected to be the only moment when the encoding window size will increase. Similarly, with a CBR real-time flow, the session end is expected to be the only moment when the encoding window size will progressively decrease. No adjustment of the ls_max_size is required at the FECFRAME receiver in that case. </t> </section> <section anchor="param_derivation_other_realtime_flows"title="Othernumbered="true" toc="include" removeInRFC="false" pn="section-c.2"> <name slugifiedName="name-other-types-of-real-time-fl">Other Types of Real-TimeFlow"> <!-- ====================== --> <t>Flow</name> <t pn="section-c.2-1"> In the following, we consider a real-time source ADU flow with a max_lat latency budget and a variable bitrate (VBR) measured at the entry of the FECFRAME sender. A first approach consists in considering the smallest instantaneous bitrate of the source ADU flow, when this parameter is known, and to reuse the derivation of <xreftarget="param_derivation_cbr_realtime"/>.target="param_derivation_cbr_realtime" format="default" sectionFormat="of" derivedContent="Appendix C.1"/>. Considering the smallest bitrate means that the encoding and decoding window maximum size estimations are pessimistic: these windows have the smallest size required to enable on-time decoding at a FECFRAME receiver. If the instantaneous bitrate is higher than this smallest bitrate, this approach leads to an encoding window that is unnecessarily small, which reduces robustness in front of long erasure bursts. </t><t><t pn="section-c.2-2"> Another approach consists in using ADU timing information (e.g., using the timestamp field of an RTP packet header, or registering the time upon receiving a new ADU). From the global FEC-related latency budget, the FECFRAME sender can derive a practical maximum latency budget for encoding operations, max_lat_for_encoding. For the FECSchemesschemes specified in this document, this latency budgetSHOULD<bcp14>SHOULD</bcp14> be computed with:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-c.2-3"> <li pn="section-c.2-3.1"> max_lat_for_encoding = max_lat * WSR / 255</t> <!-- <t> max_lat_for_encoding = max_lat * 0.75 </t> --> </list></li> </ul> <t pn="section-c.2-4"> It follows that any source symbols associated to an ADU that has timed-out with respect to max_lat_for_encodingSHOULD<bcp14>SHOULD</bcp14> be removed from the encoding window. With this approach there is no pre-determined ew_size value: this value fluctuates over the time according to the instantaneous source ADU flow bitrate. For practical reasons, a FECFRAME sender may still require that ew_size does not increase beyond a maximum value (<xreftarget="param_derivation_non_realtime"/>).target="param_derivation_non_realtime" format="default" sectionFormat="of" derivedContent="Appendix C.3"/>). </t><t><t pn="section-c.2-5"> With both approaches, and no matter the choice of the FECFRAME sender, a FECFRAME receiver can still easily evaluate the ew_max_size by observing the maximum Number of Source Symbols (NSS) value contained in the Repair FEC Payload ID of received FEC Repair Packets. A receiver can then compute dw_max_size and derive an appropriate ls_max_size as explained in <xreftarget="param_derivation_cbr_realtime"/>.target="param_derivation_cbr_realtime" format="default" sectionFormat="of" derivedContent="Appendix C.1"/>. </t><t><t pn="section-c.2-6"> When the observed NSS fluctuates significantly, a FECFRAME receiver may want to adapt its ls_max_size accordingly. In particular when the NSS is significantly reduced, a FECFRAME receiver may want to reduce the ls_max_size too in order to limit computation complexity. A balance must be found between using an ls_max_size "too large" (which increases computation complexity and memory requirements) and the opposite (which reduces recovery performance). </t><!-- <t> Beyond these general guidelines, the details of how to manage these situations at a FECFRAME sender and receiver can depend on additional considerations that are out of scope of this document. </t> --></section> <section anchor="param_derivation_non_realtime"title="Casenumbered="true" toc="include" removeInRFC="false" pn="section-c.3"> <name slugifiedName="name-case-of-a-non-real-time-flo">Case of aNon Real-Time Flow"> <!-- ====================== --> <t>Non-Real-Time Flow</name> <t pn="section-c.3-1"> Finally there are configurations where a source ADU flow has no real-time constraints. FECFRAME and the FECSchemesschemes defined in this document can still be used. The choice of appropriate parameter values can be directed by practical considerations. For instance, it can derive from an estimation of the maximum memory amount that could be dedicated to the linear system at a FECFRAME receiver, or the maximum computation complexity at a FECFRAME receiver, both of them depending on the ls_max_size parameter. The same considerations also apply to the FECFRAME sender, where the maximum memory amount and computation complexity depend on the ew_max_size parameter. </t><t><t pn="section-c.3-2"> Here also, the NSS value contained in FEC Repair Packets is used by a FECFRAME receiver to determine the current coding window size and ew_max_size by observing its maximum value over the time. </t><!-- <t> Beyond these general guidelines, the details of how to manage these situations at a FECFRAME sender and receiver can depend on additional considerations that are out of scope of this document. </t> --></section> </section> <section anchor="decodingBeyondMaxLatency"title="Decodingnumbered="true" toc="include" removeInRFC="false" pn="section-appendix.d"> <name slugifiedName="name-decoding-beyond-maximum-lat">Decoding Beyond Maximum Latency Optimization(Informational)"> <!-- ====================== --> <t>(Informational)</name> <t pn="section-appendix.d-1"> This annex introducesnon normativenon-normative considerations. It is provided as suggestions, without any impact on interoperability. For more information see <xreftarget="Roca16"/>.target="Roca16" format="default" sectionFormat="of" derivedContent="Roca16"/>. </t><t><t pn="section-appendix.d-2"> With a real-time source ADU flow, it is possible to improve the decoding performance ofsliding window codesSliding Window Codes without impacting maximum latency, at the cost of extra memory and CPU overhead. The optimization consists, for a FECFRAME receiver, to extend the linear system beyond the decoding window maximum size, by keeping a certain number of old source symbols whereas their associated ADUs timed-out:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-appendix.d-3"> <li pn="section-appendix.d-3.1"> ls_max_size>> dw_max_size</t> </list></li> </ul> <t pn="section-appendix.d-4"> Usually the following choice is a good trade-off between decoding performance and extra CPU overhead:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-appendix.d-5"> <li pn="section-appendix.d-5.1"> ls_max_size = 2 * dw_max_size</t> </list> </t> <t></li> </ul> <t pn="section-appendix.d-6"> When the dw_max_size is very small, it may be preferable to keep a minimum ls_max_size value (e.g., LS_MIN_SIZE_DEFAULT = 40 symbols). Going below this threshold will not save a significant amount of memory nor CPU cycles. Therefore:<list style="none"> <t></t> <ul empty="true" spacing="normal" bare="false" pn="section-appendix.d-7"> <li pn="section-appendix.d-7.1"> ls_max_size = max(2 * dw_max_size, LS_MIN_SIZE_DEFAULT)</t> </list> </t> <t></li> </ul> <t pn="section-appendix.d-8"> Finally, it is worth noting that a receiver that benefits from an FEC protection significantly higher than what is required to recover from packet losses, can choose to reduce the ls_max_size. In that case lost ADUs will be recovered without relying on this optimization. </t> <figureanchor="fig_decoding_beyond_max_laetency" title="Relationshipanchor="fig_decoding_beyond_max_latency" align="left" suppress-title="false" pn="figure-11"> <name slugifiedName="name-relationship-between-parame">Relationship betweenparametersParameters todecodeDecode beyondmaximum latency."> <artwork>Maximum Latency</name> <artwork name="" type="" align="left" alt="" pn="section-appendix.d-9.1"> ls_max_size /---------------------------------^-------------------------------\ late source symbols (pot. decoded but not delivered) dw_max_size /--------------^-----------------\ /--------------^---------------\ src0 src1 src2 src3 src4 src5 src6 src7 src8 src9 src10 src11 src12 </artwork> </figure><t><t pn="section-appendix.d-10"> It means that source symbols, and therefore ADUs, may be decoded even if the added latency exceeds the maximum value permitted by the application (the "late source symbols" of <xreftarget="fig_decoding_beyond_max_laetency"/>).target="fig_decoding_beyond_max_latency" format="default" sectionFormat="of" derivedContent="Figure 11"/>). It follows that the corresponding ADUs will not be useful to the application. However, decoding these "late symbols" significantly improves the global robustness in bad reception conditions and is therefore recommended for receivers experiencing bad communication conditions <xreftarget="Roca16"/>.target="Roca16" format="default" sectionFormat="of" derivedContent="Roca16"/>. In any case whether or not to use this optimization and what exact value to use for the ls_max_size parameter are local decisions made by each receiver independently, without any impact on the other receivers nor on the source. </t> </section> <section numbered="false" toc="include" removeInRFC="false" pn="section-appendix.e"> <name slugifiedName="name-acknowledgments">Acknowledgments</name> <t pn="section-appendix.e-1"> The authors would like to thank the three TSVWG chairs, Wesley Eddy (our shepherd), David Black, and Gorry Fairhurst; as well as Spencer Dawkins, our responsible AD; and all those who provided comments -- namely (in alphabetical order), Alan DeKok, Jonathan Detchart, Russ Housley, Emmanuel Lochin, Marie-Jose Montpetit, and Greg Skinner. Last but not least, the authors are really grateful to the IESG members, in particular Benjamin Kaduk, Mirja Kuehlewind, Eric Rescorla, Adam Roach, and Roman Danyliw for their highly valuable feedback that greatly contributed to improving this specification. </t> </section> <section anchor="authors-addresses" numbered="false" removeInRFC="false" toc="include" pn="section-appendix.f"> <name slugifiedName="name-authors-addresses">Authors' Addresses</name> <author fullname="Vincent Roca" initials="V" surname="Roca"> <organization showOnFrontPage="true">INRIA</organization> <address> <postal> <street/> <city/> <code/> <extaddr>Univ. Grenoble Alpes</extaddr> <country>France</country> </postal> <email>vincent.roca@inria.fr</email> </address> </author> <author fullname="Belkacem Teibi" initials="B" surname="Teibi"> <organization showOnFrontPage="true">INRIA</organization> <address> <postal> <street/> <city/> <code/> <extaddr>Univ. Grenoble Alpes</extaddr> <country>France</country> </postal> <email>belkacem.teibi@gmail.com</email> </address> </author> </section> </back> </rfc>