| RFC 8928 | Address Protection ND for LLN | September 2020 |
| Thubert, et al. | Standards Track | [Page] |
This document updates the IPv6 over Low-Power Wireless Personal Area Network (6LoWPAN) Neighbor Discovery (ND) protocol defined in RFCs 6775 and 8505. The new extension is called Address-Protected Neighbor Discovery (AP-ND), and it protects the owner of an address against address theft and impersonation attacks in a Low-Power and Lossy Network (LLN). Nodes supporting this extension compute a cryptographic identifier (Crypto-ID), and use it with one or more of their Registered Addresses. The Crypto-ID identifies the owner of the Registered Address and can be used to provide proof of ownership of the Registered Addresses. Once an address is registered with the Crypto-ID and a proof of ownership is provided, only the owner of that address can modify the registration information, thereby enforcing Source Address Validation.¶
This is an Internet Standards Track document.¶
This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841.¶
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc8928.¶
Copyright (c) 2020 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date 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 in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.¶
Neighbor Discovery optimizations for 6LoWPAN networks (aka 6LoWPAN ND) [RFC6775] adapts the original IPv6 Neighbor Discovery protocols defined in [RFC4861] and [RFC4862] for constrained Low-Power and Lossy Networks (LLNs). In particular, 6LoWPAN ND introduces a unicast host Address Registration mechanism that reduces the use of multicast compared to the Duplicate Address Detection (DAD) mechanism defined in IPv6 ND. 6LoWPAN ND defines a new Address Registration Option (ARO) that is carried in the unicast Neighbor Solicitation (NS) and Neighbor Advertisement (NA) messages exchanged between a 6LoWPAN Node (6LN) and a 6LoWPAN Router (6LR). It also defines the Duplicate Address Request (DAR) and Duplicate Address Confirmation (DAC) messages between the 6LR and the 6LoWPAN Border Router (6LBR). In LLNs, the 6LBR is the central repository of all the registered addresses in its domain.¶
The registration mechanism in "Neighbor Discovery Optimization for IPv6 over Low-Power Wireless Personal Area Networks (6LoWPANs)" [RFC6775] prevents the use of an address if that address is already registered in the subnet (first come, first served). In order to validate address ownership, the registration mechanism enables the 6LR and 6LBR to validate the association between the registered address of a node and its Registration Ownership Verifier (ROVR). The ROVR is defined in "Registration Extensions for IPv6 over Low-Power Wireless Personal Area Network (6LoWPAN) Neighbor Discovery" [RFC8505], and it can be derived from the Medium Access Control (MAC) address of the device (using the 64-bit Extended Unique Identifier (EUI-64) address format specified by IEEE). However, the EUI-64 can be spoofed; therefore, any node connected to the subnet and aware of a registered-address-to-ROVR mapping could effectively fake the ROVR. This would allow an attacker to steal the address and redirect traffic for that address. [RFC8505] defines an Extended Address Registration Option (EARO) that transports alternate forms of ROVRs and is a prerequisite for this specification.¶
In this specification, a 6LN generates a cryptographic identifier (Crypto-ID) and places it in the ROVR field during the registration of one (or more) of its addresses with the 6LR(s). Proof of ownership of the Crypto-ID is passed with the first registration exchange to a new 6LR and enforced at the 6LR. The 6LR validates ownership of the Crypto-ID before it creates any new registration state or changes existing information.¶
The protected address registration protocol proposed in this document provides the same conceptual benefit as Source Address Validation Improvement (SAVI) [RFC7039] in that only the owner of an IPv6 address may source packets with that address. As opposed to [RFC7039], which relies on snooping protocols, the protection is based on a state that is installed and maintained in the network by the owner of the address. With this specification, a 6LN may use a 6LR for forwarding an IPv6 packet if and only if it has registered the address used as the source of the packet with that 6LR.¶
With the 6lo adaptation layer in [RFC4944] and [RFC6282], a 6LN can obtain a better compression for an IPv6 address with an Interface ID (IID) that is derived from a Layer 2 (L2) address. As a side note, this is incompatible with "SEcure Neighbor Discovery (SEND") [RFC3971] and "Cryptographically Generated Addresses (CGAs)" [RFC3972], since they derive the IID from cryptographic keys, whereas this specification separates the IID and the key material.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
The reader may get additional context for this specification from the following references:¶
This document uses the following abbreviations:¶
Section 5.3 of [RFC8505] introduces the ROVR that is used to detect and reject duplicate registrations in the DAD process. The ROVR is a generic object that is designed for both backward compatibility and the capability to introduce new computation methods in the future. Using a Crypto-ID per this specification is the RECOMMENDED method. Section 7.5 discusses collisions when heterogeneous methods to compute the ROVR field coexist inside a same network.¶
This specification introduces a new token called a Crypto-ID that is transported in the ROVR field and used to indirectly prove the ownership of an address that is being registered by means of [RFC8505]. The Crypto-ID is derived from a cryptographic public key and additional parameters.¶
The overall mechanism requires the support of Elliptic Curve Cryptography (ECC) and a hash function as detailed in Section 6.2. To enable the verification of the proof, the Registering Node needs to supply certain parameters including a nonce and a signature that will demonstrate that the node possesses the private key corresponding to the public key used to build the Crypto-ID.¶
The elliptic curves and the hash functions listed in Table 1 in Section 8.2 can be used with this specification; more may be added in the future to the corresponding IANA registry. The signature scheme that specifies which combination is used (including the curve and the representation conventions) is signaled by a Crypto-Type in a new IPv6 ND Crypto-ID Parameters Option (CIPO) (see Section 4.3) that contains the parameters that are necessary for the proof, a Nonce Option [RFC3971], and an NDP Signature Option (Section 4.4). The NA(EARO) is modified to enable a challenge and transport a Nonce Option.¶
The Crypto-ID is transported in the ROVR field of the EARO and the Extended Duplicate Address Request (EDAR) message and is associated with the Registered Address at the 6LR and the 6LBR. The ownership of a Crypto-ID can be demonstrated by cryptographic mechanisms, and by association, the ownership of the Registered Address can be ascertained.¶
A node in possession of the necessary cryptographic primitives SHOULD use Crypto-ID by default as ROVR in its registrations. Whether a ROVR is a Crypto-ID is indicated by a new "C" flag in the NS(EARO) message.¶
The Crypto-ID is derived from the public key and a modifier as follows:¶
At the time of this writing, a minimal size for the Crypto-ID of 128 bits is RECOMMENDED unless backward compatibility is needed [RFC8505]. This value is bound to augment in the future.¶
This specification updates the EARO to enable the use of the ROVR field to transport the Crypto-ID. The resulting format is as follows:¶
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 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type | Length | Status | Opaque | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |Rsvd |C| I |R|T| TID | Registration Lifetime | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | ... Registration Ownership Verifier (ROVR) ... | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
This specification uses the Status values "Validation Requested" and "Validation Failed", which are defined in [RFC8505].¶
This specification does not define any new Status values.¶
This specification defines the CIPO. The CIPO carries the parameters used to form a Crypto-ID.¶
In order to provide cryptographic agility [BCP201], this specification supports different elliptic-curve-based signature schemes, indicated by a Crypto-Type field:¶
This specification uses signature schemes that target similar cryptographic strength but rely on different curves, hash functions, signature algorithms, and/or representation conventions. Future specification may extend this to different cryptographic algorithms and key sizes, e.g., to provide better security properties or a simpler implementation.¶
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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Length |Reserved1| Public Key Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Crypto-Type | Modifier | EARO Length | |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ +
| |
. .
. Public Key (variable length) .
. .
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
. Padding .
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The implementation of multiple hash functions in a constrained device may consume excessive amounts of program memory. This specification enables the use of the same hash function SHA-256 [RFC6234] for two of the three supported ECC-based signature schemes. Some code factorization is also possible for the ECC computation itself.¶
[CURVE-REPR] provides information on how to represent Montgomery curves and (twisted) Edwards curves as curves in short-Weierstrass form, and it illustrates how this can be used to implement elliptic curve computations using existing implementations that already provide, e.g., ECDSA and ECDH using NIST [FIPS186-4] prime curves. For more details on representation conventions, refer to Appendix B.¶
This specification defines the NDP Signature Option (NDPSO). The NDPSO carries the signature that proves the ownership of the Crypto-ID. The format of the NDPSO is illustrated in Figure 3.¶
As opposed to the RSA Signature Option (RSAO) defined in Section 5.2 of SEND [RFC3971], the NDPSO does not have a key hash field. Instead, the leftmost 128 bits of the ROVR field in the EARO are used as hash to retrieve the CIPO that contains the key material used for signature verification, left-padded if needed.¶
Another difference is that the NDPSO signs a fixed set of fields as opposed to all options that appear prior to it in the ND message that bears the signature. This allows a CIPO that the 6LR already received to be elided, at the expense of the capability to add arbitrary options that would be signed with an RSAO.¶
An ND message that carries an NDPSO MUST have one and only one EARO. The EARO MUST contain a Crypto-ID in the ROVR field, and the Crypto-ID MUST be associated with the key pair used for the digital signature in the NDPSO.¶
The CIPO may be present in the same message as the NDPSO. If it is not present, it can be found in an abstract table that was created by a previous message and indexed by the hash.¶
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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Length |Reserved1| Signature Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Reserved2 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
. .
. Digital Signature (variable length) .
. .
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
. Padding .
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
This specification defines one new capability bit in the 6LoWPAN Capability Indication Option (6CIO), as defined by [RFC7400], for use by the 6LR and 6LBR in IPv6 ND RA messages.¶
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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Length = 1 | Reserved |A|D|L|B|P|E|G|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Reserved |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
New Option Field:¶
The "A" flag is set by the 6LBR that serves the network and is propagated by the 6LRs. It is typically turned on when all 6LRs are migrated to this specification.¶
The scope of the protocol specified here is a 6LoWPAN LLN, typically a stub network connected to a larger IP network via a border router called a 6LBR per [RFC6775]. A 6LBR has sufficient capability to satisfy the needs of DAD.¶
The 6LBR maintains registration state for all devices in its attached LLN. Together with the first-hop router (the 6LR), the 6LBR assures uniqueness and grants ownership of an IPv6 address before it can be used in the LLN. This is in contrast to a traditional network that relies on IPv6 address autoconfiguration [RFC4862], where there is no guarantee of ownership from the network, and each IPv6 Neighbor Discovery packet must be individually secured [RFC3971].¶
---+-------- ............
| External Network
|
+-----+
| | 6LBR
+-----+
o o o
o o o o
o o LLN o o o
o o
o o o(6LR)
^
o o | LLN link
o o v
o(6LN)
o
In a mesh network, the 6LR is directly connected to the host device. This specification mandates that the peer-wise L2 security is deployed so that all the packets from a particular host are securely identifiable by the 6LR. The 6LR may be multiple hops away from the 6LBR. Packets are routed between the 6LR and the 6LBR via other 6LRs.¶
This specification mandates that all the LLN links between the 6LR and the 6LBR are protected so that a packet that was validated by the first 6LR can be safely routed by other on-path 6LRs to the 6LBR.¶
The 6LR/6LBR ensures first come, first served by storing the ROVR associated to the address being registered upon the first registration and rejecting a registration with a different ROVR value. A 6LN can claim any address as long as it is the first to make that claim. After a successful registration, the 6LN becomes the owner of the registered address, and the address is bound to the ROVR value in the 6LR/6LBR registry.¶
This specification protects the ownership of the address at the first hop (the edge). Its use in a network is signaled by the "A" flag in the 6CIO. The flag is set by the 6LBR and propagated unchanged by the 6LRs. The "A" flag enables to migrate a network with the protection off and then turn it on globally.¶
The 6LN places a cryptographic token, the Crypto-ID, in the ROVR that is associated with the address at the first registration, enabling the 6LR to later challenge it to verify that it is the original Registering Node. The challenge may happen at any time at the discretion of the 6LR and the 6LBR. A valid registration in the 6LR or the 6LBR MUST NOT be altered until the challenge is complete.¶
When the "A" flag is on, the 6LR MUST challenge the 6LN when it creates a binding with the "C" flag set in the ROVR and when a new registration attempts to change a parameter of that binding that identifies the 6LN, for instance, its Source Link-Layer Address. The verification protects against a rogue that would steal an address and attract its traffic or use it as source address.¶
The 6LR MUST also challenge the 6LN if the 6LBR directly signals to do so, using an EDAC message with a "Validation Requested" status. The EDAR is echoed by the 6LR in the NA(EARO) back to the Registering Node. The 6LR SHOULD also challenge all its attached 6LNs at the time the 6LBR turns the "A" flag on in the 6CIO in orders to detect an issue immediately.¶
If the 6LR does not support the Crypto-Type, it MUST reply with an EARO Status 10 "Validation Failed" without a challenge. In that case, the 6LN may try another Crypto-Type until it falls back to Crypto-Type 0, which MUST be supported by all 6LRs.¶
A node may use more than one IPv6 address at the same time. The separation of the address and the cryptographic material avoids the need for the constrained device to compute multiple keys for multiple addresses. The 6LN MAY use the same Crypto-ID to prove the ownership of multiple IPv6 addresses. The 6LN MAY also derive multiple Crypto-IDs from a same key.¶
A 6LN registers to a 6LR that is one hop away from it with the "C" flag set in the EARO, indicating that the ROVR field contains a Crypto-ID. The Target Address in the NS message indicates the IPv6 address that the 6LN is trying to register [RFC8505]. The on-link (local) protocol interactions are shown in Figure 6. If the 6LR does not have a state with the 6LN that is consistent with the NS(EARO), then it replies with a challenge NA(EARO, status=Validation Requested) that contains a Nonce Option (shown as NonceLR in Figure 6).¶
6LN 6LR
| |
|<------------------------- RA -------------------------|
| | ^
|---------------- NS with EARO (Crypto-ID) ------------>| |
| | option
|<- NA with EARO(status=Validation Requested), NonceLR | |
| | v
|------- NS with EARO, CIPO, NonceLN and NDPSO -------->|
| |
|<------------------- NA with EARO ---------------------|
| |
...
| |
|--------------- NS with EARO (Crypto-ID) ------------->|
| |
|<------------------- NA with EARO ---------------------|
| |
...
| |
|--------------- NS with EARO (Crypto-ID) ------------->|
| |
|<------------------- NA with EARO ---------------------|
| |
The Nonce Option contains a nonce value that, to the extent possible for the implementation, was never employed in association with the key pair used to generate the Crypto-ID. This specification inherits the idea from [RFC3971] that the nonce is a random value. Ideally, an implementation uses an unpredictable cryptographically random value [BCP106]. But that may be impractical in some LLN scenarios in which the devices do not have a guaranteed sense of time and for which computing complex hashes is detrimental to the battery lifetime.¶
Alternatively, the device may use an always-incrementing value saved in the same stable storage as the key, so they are lost together, and start at a best-effort random value as either the nonce value or a component to its computation.¶
The 6LN replies to the challenge with an NS(EARO) that includes a new Nonce Option (shown as NonceLN in Figure 6), the CIPO (Section 4.3), and the NDPSO containing the signature. Both nonces are included in the signed material. This provides a "contributory behavior", so either party that knows it generates a good quality nonce and knows that the protocol will be secure.¶
The 6LR MUST store the information associated with a Crypto-ID on the first NS exchange where it appears in a fashion that the CIPO parameters can be retrieved from the Crypto-ID alone.¶
The steps for the registration to the 6LR are as follows:¶
Upon the first exchange with a 6LR, a 6LN will be challenged to prove ownership of the Crypto-ID and the Target Address being registered in the Neighbor Solicitation message. When a 6LR receives an NS(EARO) registration with a new Crypto-ID as a ROVR, and unless the registration is rejected for another reason, it MUST challenge by responding with an NA(EARO) with a status of "Validation Requested".¶
Upon receiving a first NA(EARO) with a status of "Validation Requested" from a 6LR, the Registering Node SHOULD retry its registration with a CIPO (Section 4.3) that contains all the necessary material for building the Crypto-ID, the NonceLN that it generated, and the NDP Signature Option (Section 4.4) that proves its ownership of the Crypto-ID and intent of registering the Target Address. In subsequent revalidation with the same 6LR, the 6LN MAY try to omit the CIPO to save bandwidth, with the expectation that the 6LR saved it. If the validation fails and it gets challenged again, then it SHOULD add the CIPO again.¶
In order to validate the ownership, the 6LR performs the same steps as the 6LN and rebuilds the Crypto-ID based on the parameters in the CIPO. If the rebuilt Crypto-ID matches the ROVR, the 6LN also verifies the signature contained in the NDPSO. At that point, if the signature in the NDPSO can be verified, then the validation succeeds. Otherwise, the validation fails.¶
If the 6LR fails to validate the signed NS(EARO), it responds with a status of "Validation Failed". After receiving an NA(EARO) with a status of "Validation Failed", the Registering Node SHOULD try an alternate Crypto-Type; even if Crypto-Type 0 fails, it may try to register a different address in the NS message.¶
The signature generated by the 6LN to provide proof of ownership of the private key is carried in the NDPSO. It is generated by the 6LN in a fashion that depends on the Crypto-Type (see Table 1 in Section 8.2) chosen by the 6LN as follows:¶
Form the message to be signed, by concatenating the following byte-strings in the order listed:¶
Upon receiving the NDPSO and CIPO options, the 6LR first checks that the EARO Length in the CIPO matches the length of the EARO. If so, it regenerates the Crypto-ID based on the CIPO to make sure that the leftmost bits up to the size of the ROVR match.¶
If, and only if, the check is successful, it tries to verify the signature in the NDPSO using the following steps:¶
Form the message to be verified, by concatenating the following byte-strings in the order listed:¶
A new 6LN that joins the network autoconfigures an address and performs an initial registration to a neighboring 6LR with an NS message that carries an EARO [RFC8505].¶
In a multi-hop 6LoWPAN, the registration with Crypto-ID is propagated to 6LBR as shown in Figure 7, which illustrates the registration flow all the way to a 6LowPAN Backbone Router (6BBR) [RFC8929].¶
6LN 6LR 6LBR 6BBR
| | | |
| NS(EARO) | | |
|--------------->| | |
| | Extended DAR | |
| |-------------->| |
| | | proxy NS(EARO) |
| | |--------------->|
| | | | NS(DAD)
| | | | ------>
| | | |
| | | | <wait>
| | | |
| | | proxy NA(EARO) |
| | |<---------------|
| | Extended DAC | |
| |<--------------| |
| NA(EARO) | | |
|<---------------| | |
| | | |
The 6LR and the 6LBR communicate using ICMPv6 EDAR and EDAC messages [RFC8505] as shown in Figure 7. This specification extends EDAR/EDAC messages to carry cryptographically generated ROVR.¶
The assumption is that the 6LR and the 6LBR maintain a security association to authenticate and protect the integrity of the EDAR and EDAC messages, so there is no need to propagate the proof of ownership to the 6LBR. The 6LBR implicitly trusts that the 6LR performs the verification when the 6LBR requires it, and if there is no further exchange from the 6LR to remove the state, the verification succeeded.¶
Only 6LRs that are upgraded to this specification are capable of challenging a registration and repelling an attack. In a brown (mixed) network, an attacker may attach to a legacy 6LR and fool the 6LBR. So even if the "A" flag could be set at any time to test the protocol operation, the security will only be effective when all the 6LRs are upgraded.¶
Observations regarding the following threats to the local network in [RFC3971] also apply to this specification.¶
The threats and mediations discussed in 6LoWPAN ND [RFC6775] [RFC8505] also apply here, in particular, denial-of-service (DoS) attacks against the registry at the 6LR or 6LBR.¶
Secure ND [RFC3971] forces the IPv6 address to be cryptographic since it integrates the CGA as the IID in the IPv6 address. In contrast, this specification saves about 1 KB in every NS/NA message. Also, this specification separates the cryptographic identifier from the registered IPv6 address so that a node can have more than one IPv6 address protected by the same cryptographic identifier.¶
With this specification, the 6LN can freely form its IPv6 address(es) in any fashion, thereby enabling either 6LoWPAN compression for IPv6 addresses that are derived from L2 addresses or temporary addresses, e.g., formed pseudorandomly and released in relatively short cycles for privacy reasons [RFC8064][RFC8065] that cannot be compressed.¶
This specification provides added protection for addresses that are obtained following due procedure [RFC8505] but does not constrain the way the addresses are formed or the number of addresses that are used in parallel by a same entity. A rogue may still perform a DoS attack against the registry at the 6LR or 6LBR or attempt to deplete the pool of available addresses at L2 or L3.¶
This specification distributes the challenge and its validation at the edge of the network, between the 6LN and its 6LR. This protects against DoS attacks targeted at that central 6LBR. This also saves back-and-forth exchanges across a potentially large and constrained network.¶
The downside is that the 6LBR needs to trust the 6LR to perform the checking adequately, and the communication between the 6LR and the 6LBR must be protected to avoid tampering with the result of the test.¶
If a 6LR is compromised, and provided that it knows the ROVR field used by the real owner of the address, the 6LR may pretend that the owner has moved, is now attached to it, and has successfully passed the Crypto-ID validation. The 6LR may then attract and inject traffic at will on behalf of that address, or let a rogue take ownership of the address.¶
A collision of ROVRs (i.e., the Crypto-ID in this specification) is possible, but it is a rare event. Assuming in the calculations/discussion below that the hash used for calculating the Crypto-ID is a well-behaved cryptographic hash, and, thus, random collisions are the only ones possible, the formula (birthday paradox) for calculating the probability of a collision is 1 - e^{-p^2/(2n)}, where n is the maximum population size (2^64 here, 1.84E19), and p is the actual population (number of nodes, assuming one Crypto-ID per node).¶
If the Crypto-ID is 64 bits (the least possible size allowed), the chance of a collision is 0.01% for a network of 66 million nodes. Moreover, the collision is only relevant when this happens within one stub network (6LBR). In the case of such a collision, a third-party node would be able to claim the registered address of an another legitimate node, provided that it wishes to use the same address. To prevent address disclosure and avoid the chances of collision on both the ROVR and the address, it is RECOMMENDED that nodes do not derive the address being registered from the ROVR.¶
The signature schemes referenced in this specification comply with NIST [FIPS186-4] or Crypto Forum Research Group (CFRG) standards [RFC8032] and offer strong algorithmic security at roughly a 128-bit security level. These signature schemes use elliptic curves that either were specifically designed with exception-free and constant-time arithmetic in mind [RFC7748] or have extensive implementation experience of resistance to timing attacks [FIPS186-4].¶
However, careless implementations of the signing operations could nevertheless leak information on private keys. For example, there are micro-architectural side channel attacks that implementors should be aware of [breaking-ed25519]. Implementors should be particularly aware that a secure implementation of Ed25519 requires a protected implementation of the hash function SHA-512, whereas this is not required with implementations of the hash function SHA-256 used with ECDSA256 and ECDSA25519.¶
The key pair used in this specification can be self-generated, and the public key does not need to be exchanged, e.g., through certificates, with a third party before it is used.¶
New key pairs can be formed for new registration as the node desires. On the other hand, it is safer to allocate a key pair that is used only for the address protection and only for one instantiation of the signature scheme (which includes the choice of elliptic curve domain parameters, a used hash function, and applicable representation conventions).¶
The same private key MUST NOT be reused with more than one instantiation of the signature scheme in this specification. The same private key MUST NOT be used for anything other than computing NDPSO signatures per this specification.¶
ECDSA shall be used strictly as specified in [FIPS186-4]. In particular, each signing operation of ECDSA MUST use randomly generated ephemeral private keys and MUST NOT reuse these ephemeral private keys k across signing operations. This precludes the use of deterministic ECDSA without a random input for the determination of k, which is deemed dangerous for the intended applications this document aims to serve.¶
Public keys contained in the CIPO field (which are used for signature verification) shall be verified to be correctly formed, by checking that this public key is indeed a point of the elliptic curve indicated by the Crypto-Type and that this point does have the proper order.¶
For points used with the signature scheme Ed25519, one MUST check that this point is not in the small subgroup (see Appendix B.1 of [CURVE-REPR]); for points used with the signature scheme ECDSA (i.e., both ECDSA256 and ECDSA25519), one MUST check that the point has the same order as the base point of the curve in question. This is commonly called "full public key validation" (again, see Appendix B.1 of [CURVE-REPR]).¶
The ROVR field in the EARO introduced in [RFC8505] extends the EUI-64 field of the ARO defined in [RFC6775]. One of the drawbacks of using an EUI-64 as ROVR is that an attacker that is aware of the registrations can correlate traffic for a same 6LN across multiple addresses. Section 3 of [RFC8505] indicates that the ROVR and the address being registered are decoupled. A 6LN may use the same ROVR for multiple registrations or a different ROVR per registration, and the IID must not be derived from the ROVR. In theory, different 6LNs could use the same ROVR as long as they do not attempt to register the same address.¶
The modifier used in the computation of the Crypto-ID enables a 6LN to build different Crypto-IDs for different addresses with a same key pair. Using that facility improves the privacy of the 6LN at the expense of storage in the 6LR, which will need to store multiple CIPOs that contain the same public key. Note that if the attacker is the 6LR, then the modifier alone does not provide a protection, and the 6LN would need to use different keys and MAC addresses in an attempt to obfuscate its multiple ownership.¶
This document defines a new 128-bit value of a Message Type tag under the CGA Message Type [RFC3972] namespace: 0x8701 55c8 0cca dd32 6ab7 e415 f148 84d0.¶
IANA has created the "Crypto-Types" subregistry in the "Internet Control Message Protocol version 6 (ICMPv6) Parameters" registry. The registry is indexed by an integer in the interval 0..255 and contains an elliptic curve, a hash function, a signature algorithm, representation conventions, public key size, and signature size, as shown in Table 1, which together specify a signature scheme (and are fully specified in Appendix B).¶
The following Crypto-Type values are defined in this document:¶
| Crypto-Type Value | 0 (ECDSA256) | 1 (Ed25519) | 2 (ECDSA25519) |
|---|---|---|---|
| Elliptic Curve | NIST P-256 [FIPS186-4] | Curve25519 [RFC7748] | Curve25519 [RFC7748] |
| Hash Function | SHA-256 [RFC6234] | SHA-512 [RFC6234] | SHA-256 [RFC6234] |
| Signature Algorithm | ECDSA [FIPS186-4] | Ed25519 [RFC8032] | ECDSA [FIPS186-4] |
| Representation Conventions | Weierstrass, (un)compressed, MSB/msb first, [RFC7518] | Edwards, compressed, LSB/lsb first, [RFC8037] | Weierstrass, (un)compressed, MSB/msb first, [CURVE-REPR] |
| Public Key Size | 33/65 bytes (compressed/uncompressed) | 32 bytes (compressed) | 33/65 bytes (compressed/uncompressed) |
| Signature Size | 64 bytes | 64 bytes | 64 bytes |
| Reference | RFC 8928 | RFC 8928 | RFC 8928 |
New Crypto-Type values providing similar or better security may be defined in the future.¶
Assignment of new values for new Crypto-Type MUST be done through IANA with either "Specification Required" or "IESG Approval" as defined in BCP 26 [RFC8126].¶
This document registers two new ND option types under the subregistry "IPv6 Neighbor Discovery Option Formats":¶
| Description | Type | Reference |
|---|---|---|
| Crypto-ID Parameters Option (CIPO) | 39 | RFC 8928 |
| NDP Signature Option (NDPSO) | 40 | RFC 8928 |
IANA has made an addition to the subregistry for "6LoWPAN Capability Bits" created for [RFC7400] as follows:¶
| Bit | Description | Reference |
|---|---|---|
| 9 | AP-ND Enabled (1 bit) | RFC 8928 |
In this section, the requirements of a secure Neighbor Discovery protocol for LLNs are stated.¶
The signature scheme ECDSA256 corresponding to Crypto-Type 0 is ECDSA, as specified in [FIPS186-4], instantiated with the NIST prime curve P-256, as specified in Appendix B of [FIPS186-4], and the hash function SHA-256, as specified in [RFC6234], where points of this NIST curve are represented as points of a short-Weierstrass curve (see [FIPS186-4]) and are encoded as octet strings in most-significant-bit (msb) first and most-significant-byte (MSB) first order. The signature itself consists of two integers (r and s), which are each encoded as fixed-size octet strings in msb first and MSB first order. For further details, see [FIPS186-4] for ECDSA, see Appendix B.3 for the encoding of public keys, and see Appendix B.2 for signature encoding.¶
The signature scheme Ed25519 corresponding to Crypto-Type 1 is EdDSA, as specified in [RFC8032], instantiated with the Montgomery curve Curve25519, as specified in [RFC7748], and the hash function SHA-512, as specified in [RFC6234], where points of this Montgomery curve are represented as points of the corresponding twisted Edwards curve Edwards25519 (see Appendix B.4) and are encoded as octet strings in least-significant-bit (lsb) first and least-significant-byte (LSB) first order. The signature itself consists of a bit string that encodes a point of this twisted Edwards curve, in compressed format, and an integer encoded in lsb first and LSB first order. For details on EdDSA and the encoding of public keys and signatures, see the specification of pure Ed25519 in [RFC8032].¶
The signature scheme ECDSA25519 corresponding to Crypto-Type 2 is ECDSA, as specified in [FIPS186-4], instantiated with the Montgomery curve Curve25519, as specified in [RFC7748], and the hash function SHA-256, as specified in [RFC6234], where points of this Montgomery curve are represented as points of the corresponding short-Weierstrass curve Wei25519 (see Appendix B.4) and are encoded as octet strings in msb first and MSB first order. The signature itself consists of a bit string that encodes two integers, each encoded as fixed-size octet strings in msb first and MSB first order. For further details, see [FIPS186-4] for ECDSA, see Appendix B.3 for the encoding of public keys, and see Appendix B.2 for signature encoding.¶
With ECDSA, each signature is an ordered pair (r, s) of integers [FIPS186-4], where each integer is represented as a 32-octet string according to the Field-Element-to-Octet-String conversion rules in [SEC1] and where the ordered pair of integers is represented as the right concatenation of these representation values (thereby resulting in a 64-octet string). The inverse operation checks that the signature is a 64-octet string and represents the left-side and right-side halves of this string (each a 32-octet string) as the integers r and s, respectively, using the Octet-String-to-Field-Element conversion rules in [SEC1].¶
ECDSA is specified to be used with elliptic curves in short-Weierstrass form. Each point of such a curve is represented as an octet string using the Elliptic-Curve-Point-to-Octet-String conversion rules in [SEC1], where point compression may be enabled (which is indicated by the leftmost octet of this representation). The inverse operation converts an octet string to a point of this curve using the Octet-String-to-Elliptic-Curve-Point conversion rules in [SEC1], whereby the point is rejected if this is the so-called point at infinity. (This is the case if the input to this inverse operation is an octet string of length 1.)¶
The elliptic curve Curve25519, as specified in [RFC7748], is a so-called Montgomery curve. Each point of this curve can also be represented as a point of a twisted Edwards curve or as a point of an elliptic curve in short-Weierstrass form, via a coordinate transformation (a so-called isomorphic mapping). The parameters of the Montgomery curve and the corresponding isomorphic curves in twisted Edwards curve and short-Weierstrass form are as indicated below. Here, the domain parameters of the Montgomery curve Curve25519 and of the twisted Edwards curve Edwards25519 are as specified in [RFC7748]; the domain parameters of the elliptic curve Wei25519 in short-Weierstrass form comply with Section 6.1.1 of [FIPS186-4]. For further details on these curves and on the coordinate transformations referenced above, see [CURVE-REPR].¶
******** OPTION A: dl¶
General parameters (for all curve models):¶
******** OPTION B: sourcecode¶
Montgomery curve-specific parameters (for Curve25519):¶
A 486662 B 1 Gu 9 (=0x9) Gv 147816194475895447910205935684099868872646061346164752889648818377 55586237401 (=0x20ae19a1 b8a086b4 e01edd2c 7748d14c 923d4d7e 6d7c61b2 29e9c5a2 7eced3d9)¶
******** OPTION B (again): sourcecode¶
Twisted Edwards curve-specific parameters (for Edwards25519):¶
a -1 (-0x01) d -121665/121666 (=3709570593466943934313808350875456518954211387984321901638878553 3085940283555) (=0x52036cee 2b6ffe73 8cc74079 7779e898 00700a4d 4141d8ab 75eb4dca 135978a3) Gx 151122213495354007725011514095885315114540126930418572060461132839 49847762202 (=0x216936d3 cd6e53fe c0a4e231 fdd6dc5c 692cc760 9525a7b2 c9562d60 8f25d51a) Gy 4/5 (=4631683569492647816942839400347516314130799386625622561578303360 3165251855960) (=0x66666666 66666666 66666666 66666666 66666666 66666666 66666666 66666658)¶
******** OPTION C: dictionary list (using <tt> for fixed-width font)¶
Weierstrass curve-specific parameters (for Wei25519):¶
a 19298681539552699237261830834781317975544997444273427339909597334573241639236
(=0x2aaaaaaa aaaaaaaa aaaaaaaa aaaaaaaa aaaaaaaa aaaaaaaa aaaaaa98 4914a144)¶
b 55751746669818908907645289078257140818241103727901012315294400837956729358436
(=0x7b425ed0 97b425ed 097b425e d097b425 ed097b42 5ed097b4 260b5e9c 7710c864) ¶
GX 19298681539552699237261830834781317975544997444273427339909597334652188435546
(=0x2aaaaaaa aaaaaaaa aaaaaaaa aaaaaaaa aaaaaaaa aaaaaaaa aaaaaaaa aaad245a) ¶
GY 14781619447589544791020593568409986887264606134616475288964881837755586237401
(=0x20ae19a1 b8a086b4 e01edd2c 7748d14c 923d4d7e 6d7c61b2 29e9c5a2 7eced3d9) ¶
Many thanks to Charlie Perkins for his in-depth review and constructive suggestions. The authors are also especially grateful to Robert Moskowitz and Benjamin Kaduk for their comments and discussions that led to many improvements. The authors wish to also thank Shwetha Bhandari for actively shepherding this document and Roman Danyliw, Alissa Cooper, Mirja Kuehlewind, Eric Vyncke, Vijay Gurbani, Al Morton, and Adam Montville for their constructive reviews during the IESG process. Finally, many thanks to our INT area ADs, Suresh Krishnan and Erik Kline, who supported us along the whole process.¶