| CoRE Working Group | M. Tiloca |
| Internet-Draft | RISE SICS AB |
| Intended status: Standards Track | G. Selander |
| Expires: August 16, 2018 | F. Palombini |
| Ericsson AB | |
| J. Park | |
| Universitaet Duisburg-Essen | |
| February 12, 2018 |
Secure group communication for CoAP
draft-ietf-core-oscore-groupcomm-00
This document describes a method for protecting group communication over the Constrained Application Protocol (CoAP). The proposed approach relies on Object Security for Constrained RESTful Environments (OSCORE) and the CBOR Object Signing and Encryption (COSE) format. All security requirements fulfilled by OSCORE are maintained for multicast OSCORE request messages and related OSCORE response messages. Source authentication of all messages exchanged within the group is ensured, by means of digital signatures produced through private keys of sender devices and embedded in the protected CoAP messages.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on August 16, 2018.
Copyright (c) 2018 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.
The Constrained Application Protocol (CoAP) [RFC7252] is a web transfer protocol specifically designed for constrained devices and networks [RFC7228].
Group communication for CoAP [RFC7390] addresses use cases where deployed devices benefit from a group communication model, for example to reduce latencies and improve performance. Use cases include lighting control, integrated building control, software and firmware updates, parameter and configuration updates, commissioning of constrained networks, and emergency multicast (see Appendix A). Furthermore, [RFC7390] recognizes the importance to introduce a secure mode for CoAP group communication. This specification defines such a mode.
Object Security for Constrained RESTful Environments (OSCORE)[I-D.ietf-core-object-security] describes a security protocol based on the exchange of protected CoAP messages. OSCORE builds on CBOR Object Signing and Encryption (COSE) [RFC8152] and provides end-to-end encryption, integrity, and replay protection between a sending endpoint and a receiving endpoint across intermediary nodes. To this end, a CoAP message is protected by including payload (if any), certain options, and header fields in a COSE object, which finally replaces the authenticated and encrypted fields in the protected message.
This document describes multicast OSCORE, providing end-to-end security of CoAP messages exchanged between members of a multicast group. In particular, the described approach defines how OSCORE should be used in a group communication context, while fulfilling the same security requirements. That is, end-to-end security is assured for multicast CoAP requests sent by multicaster nodes to the group and for related CoAP responses sent as reply by multiple listener nodes. Multicast OSCORE provides source authentication of all CoAP messages exchanged within the group, by means of digital signatures produced through private keys of sender devices and embedded in the protected CoAP messages. As in OSCORE, it is still possible to simultaneously rely on DTLS to protect hop-by-hop communication between a multicaster node and a proxy (and vice versa), and between a proxy and a listener node (and vice versa).
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.
Readers are expected to be familiar with the terms and concepts described in CoAP [RFC7252]; group communication for CoAP [RFC7390]; COSE and counter signatures [RFC8152].
Readers are also expected to be familiar with the terms and concepts for protection and processing of CoAP messages through OSCORE, such as "Security Context", "Master Secret" and "Master Salt", defined in [I-D.ietf-core-object-security].
Terminology for constrained environments, such as "constrained device", "constrained-node network", is defined in [RFC7228].
This document refers also to the following terminology.
This section presents a set of assumptions and security objectives for the approach described in this document.
The following assumptions are assumed to be already addressed and are out of the scope of this document.
The approach described in this document aims at fulfilling the following security objectives:
To support multicast communication secured with OSCORE, each endpoint registered as member of a multicast group maintains a Security Context as defined in Section 3 of [I-D.ietf-core-object-security]. In particular, each endpoint in a group stores:
Upon receiving a secure CoAP message, a recipient endpoint relies on the sender endpoint's public key, in order to verify the counter signature conveyed in the COSE Object.
If not already stored in the Recipient Context associated to the sender endpoint, the recipient endpoint retrieves the public key from a trusted key repository. In such a case, the correct binding between the sender endpoint and the retrieved public key MUST be assured, for instance by means of public key certificates.
It is RECOMMENDED that the Group Manager acts as trusted key repository, and hence is configured to store public keys of group members and provide them to other members of the same group upon request. Possible approaches to provision public keys upon joining the group and to retrieve public keys of group members are discussed in Appendix C.2.
The Sender Key/IV stored in the Sender Context and the Recipient Keys/IVs stored in the Recipient Contexts are derived according to the same scheme defined in Section 3.2 of [I-D.ietf-core-object-security].
The approach described in this specification should take into account the risk of compromise of group members. Such a risk is reduced when multicast groups are deployed in physically secured locations, like lighting inside office buildings. Nevertheless, the adoption of key management schemes for secure revocation and renewal of Security Contexts and group keying material should be considered.
Consistently with the security assumptions in Section 2, it is RECOMMENDED to adopt a group key management scheme, and securely distribute a new value for the Master Secret parameter of the group's Security Context, before a new joining endpoint is added to the group or after a currently present endpoint leaves the group. This is necessary in order to preserve backward security and forward security in the multicast group. The Group Manager responsible for the group is entrusted with such a task.
In particular, the Group Manager MUST distribute also a new Group Identifier (Gid) for that group, together with a new value for the Master Secret parameter. An example of how this can be done is provided in Appendix B. Then, each group member re-derives the keying material stored in its own Sender Context and Recipient Contexts as described in Section 3, using the updated Group Identifier.
Especially in dynamic, large-scale, multicast groups where endpoints can join and leave at any time, it is important that the considered group key management scheme is efficient and highly scalable with the group size, in order to limit the impact on performance due to the Security Context and keying material update.
When creating a protected CoAP message, an endpoint in the group computes the COSE object using the untagged COSE_Encrypt0 structure [RFC8152] as defined in Section 5 of [I-D.ietf-core-object-security], with the following modifications.
In particular, "gid" is included as COSE header parameter as defined in Figure 1.
+------+-------+------------+----------------+-------------------+ | name | label | value type | value registry | description | +------+-------+------------+----------------+-------------------+ | gid | TBD | bstr | | Identifies the | | | | | | OSCORE group | | | | | | Security Context | +------+-------+------------+----------------+-------------------+
Figure 1: Additional common header parameter for the COSE object
external_aad = [ version : uint, alg : int, request_kid : bstr, request_piv : bstr, gid : bstr, options : bstr ]
0 1 2 3 4 5 6 7 <----------- n bytes -----------> <-- 1 byte -->
+-+-+-+-+-+-+-+-+---------------------------------+--------------+
|0 0|1|h|1| n | Partial IV (if any) | s (if any) |
+-+-+-+-+-+-+-+-+---------------------------------+--------------+
<------ s bytes ------> <--------- q bytes --------->
-----------------------+-----------------------------+-----------+
Gid (if any) | countersign | kid |
-----------------------+-----------------------------+-----------+
Figure 2: Object-Security Value
Each multicast request message and response message is protected and processed as specified in [I-D.ietf-core-object-security], with the modifications described in the following sections.
Furthermore, endpoints in the multicast group locally perform error handling and processing of invalid messages according to the same principles adopted in [I-D.ietf-core-object-security]. However, a receiver endpoint MUST stop processing and silently reject any message which is malformed and does not follow the format specified in Section 4, without sending back any error message. This prevents listener endpoints from sending multiple error messages to a multicaster endpoint, so avoiding the risk of flooding the multicast group.
A multicaster endpoint transmits a secure multicast request message as described in Section 7.1 of [I-D.ietf-core-object-security], with the following modifications.
Upon receiving a secure multicast request message, a listener endpoint proceeds as described in Section 7.2 of [I-D.ietf-core-object-security], with the following modifications.
A listener endpoint that has received a multicast request message may reply with a secure response message, which is protected as described in Section 7.3 of [I-D.ietf-core-object-security], with the following modifications.
Upon receiving a secure response message, a multicaster endpoint proceeds as described in Section 7.4 of [I-D.ietf-core-object-security], with the following modifications.
The mapping between response messages from listener endpoints and the associated multicast request message from a multicaster endpoint relies on the 3-tuple (Group ID, Sender ID, Partial IV) associated to the secure multicast request message. This is used by listener endpoints as part of the Additional Authenticated Data when protecting their own response message, as described in Section 4.
Upon joining the multicast group, new listeners are not aware of the sequence number values currently used by different multicasters to transmit multicast request messages. This means that, when such listeners receive a secure multicast request from a given multicaster for the first time, they are not able to verify if that request is fresh and has not been replayed. The same applies when a listener endpoint loses synchronization with sequence numbers of multicasters, for instance after a device reboot.
The exact way to address this issue depends on the specific use case and its synchronization requirements. The Group Manager should define also how to handle synchronization of sequence numbers, as part of the policies enforced in the multicast group. In particular, the Group Manager can suggest to single specific listener endpoints how they can exceptionally behave in order to synchronize with sequence numbers of multicasters. Appendix D describes three possible approaches that can be considered.
The same security considerations from OSCORE (Section 11 of [I-D.ietf-core-object-security]) apply to this specification. Additional security aspects to be taken into account are discussed below.
The approach described in this document relies on commonly shared group keying material to protect communication within a multicast group. This means that messages are encrypted at a group level (group-level data confidentiality), i.e. they can be decrypted by any member of the multicast group, but not by an external adversary or other external entities.
In addition, it is required that all group members are trusted, i.e. they do not forward the content of group messages to unauthorized entities. However, in many use cases, the devices in the multicast group belong to a common authority and are configured by a commissioner. For instance, in a professional lighting scenario, the roles of multicaster and listener are configured by the lighting commissioner, and devices strictly follow those roles.
TBD. Header parameter 'gid'.
The authors sincerely thank Stefan Beck, Rolf Blom, Carsten Bormann, Klaus Hartke, Richard Kelsey, John Mattsson, Jim Schaad and Ludwig Seitz for their feedback and comments.
| [I-D.ietf-core-object-security] | Selander, G., Mattsson, J., Palombini, F. and L. Seitz, "Object Security for Constrained RESTful Environments (OSCORE)", Internet-Draft draft-ietf-core-object-security-08, January 2018. |
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997. |
| [RFC7252] | Shelby, Z., Hartke, K. and C. Bormann, "The Constrained Application Protocol (CoAP)", RFC 7252, DOI 10.17487/RFC7252, June 2014. |
| [RFC8032] | Josefsson, S. and I. Liusvaara, "Edwards-Curve Digital Signature Algorithm (EdDSA)", RFC 8032, DOI 10.17487/RFC8032, January 2017. |
| [RFC8152] | Schaad, J., "CBOR Object Signing and Encryption (COSE)", RFC 8152, DOI 10.17487/RFC8152, July 2017. |
| [RFC8174] | Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, May 2017. |
Group Communication for CoAP [RFC7390] provides the necessary background for multicast-based CoAP communication, with particular reference to low-power and lossy networks (LLNs) and resource constrained environments. The interested reader is encouraged to first read [RFC7390] to understand the non-security related details. This section discusses a number of use cases that benefit from secure group communication. Specific security requirements for these use cases are discussed in Section 2.
This section provides an example of how the Group Identifier (Gid) can be specifically formatted. That is, the Gid can be composed of two parts, namely a Group Prefix and a Group Epoch.
The Group Prefix is uniquely defined in the set of all the multicast groups associated to the same Group Manager. The choice of the Group Prefix for a given group's Security Context is application specific. Group Prefixes are random as well as long enough, in order to achieve a negligible probability of collisions between Group Identifiers from different Group Managers.
The Group Epoch is set to 0 upon the group's initialization, and is incremented by 1 upon completing each renewal of the Security Context and keying material in the group (see Section 3.1). In particular, once a new Master Secret has been distributed to the group, all the group members increment by 1 the Group Epoch in the Group Identifier of that group (see Section 3).
An endpoint joins a multicast group by explicitly interacting with the responsible Group Manager. All communications between a joining endpoint and the Group Manager rely on the CoAP protocol and MUST be secured. Specific details on how to secure communications between joining endpoints and a Group Manager are out of the scope of this specification.
In order to receive multicast messages sent to the group, a joining endpoint has to register with a network router device [RFC3376][RFC3810], signaling its intent to receive packets sent to the multicast IP address of that group. As a particular case, the Group Manager can also act as such a network router device. Upon joining the group, endpoints are not required to know how many and what endpoints are active in the same group.
Furthermore, in order to participate in the secure group communication, an endpoint needs to maintain a number of information elements stored in its own Security Context (see Section 3). The following Appendix C.1 describes which of this information is provided to an endpoint upon joining a multicast group through the responsible Group Manager.
An endpoint requests to join a multicast group by sending a confirmable CoAP POST request to the Group Manager responsible for that group. The join request is addressed to a CoAP resource associated to that group and carries the following information.
The Group Manager MUST be able to verify that the joining enpoint is authorized to become a member of the multicast group. To this end, the Group Manager can directly authorize the joining endpoint, or expect it to provide authorization evidence previously obtained from a trusted entity. Appendix C.3 describes how this can be achieved by leveraging the ACE framework for Authentication and Authorization in constrained environments [I-D.ietf-ace-oauth-authz].
In case of successful authorization check, the Group Manager generates an Endpoint ID assigned to the joining node, before proceeding with the rest of the join process. Instead, in case the authorization check fails, the Group Manager MUST abort the join process. Further details about the authorization of joining endpoint are out of the scope of this specification.
As discussed in Section 3.1, it is then RECOMMENDED that the Security Context is renewed before the joining endpoint becomes a new active member of the multicast group. This is achieved by securely distributing a new Master Secret and a new Group Identifier to the endpoints currently present in the same group.
Once renewed the Security Context in the multicast group, the Group Manager replies to the joining endpoint with a CoAP response carrying the following information.
As mentioned in Section 3, it is RECOMMENDED that the Group Manager acts as trusted key repository, stores public keys of group members and provide them to other members of the same group upon request. In such a case, a joining endpoint provides its own public key to the Group Manager, as "Identity credentials" of the join request, when joining the multicast group (see Appendix C.1).
After that, the Group Manager MUST verify that the joining endpoint actually owns the associated private key, for instance by performing a proof-of-possession challenge-response. In case of success, the Group Manager stores the received public key as associated to the joining endpoint and its Endpoint ID, before sending the join response and continuing with the rest of the join process. From then on, that public key will be available for secure and trusted delivery to other endpoints in the multicast group.
The joining node does not have to provide its own public key if that already occurred upon previously joining the same or a different multicast group under the same Group Manager. However, separately for each multicast group under its control, the Group Manager maintains an updated list of active Endpoint IDs associated to a same endpoint's public key.
Instead, in case the Group Manager does not act as trusted key repository, the following information is exchanged with the Group Manager during the join process.
When a group member X wants to retrieve the public key of another group member Y in the same multicast group, the endpoint X proceeds as follows.
The join process to register an endpoint as a new member of a multicast group can be based on the ACE framework for Authentication and Authorization in constrained environments [I-D.ietf-ace-oauth-authz], built on re-use of OAuth 2.0 [RFC6749].
In particular, the approach described in [I-D.tiloca-ace-oscoap-joining] uses the ACE framework to delegate the authentication and authorization of joining endpoints to an Authorization Server in a trust relation with the Group Manager. At the same time, it allows a joining endpoint to establish a secure channel with the Group Manager, by leveraging protocol-specific profiles of ACE [I-D.ietf-ace-oscore-profile] [I-D.ietf-ace-dtls-authorize] [I-D.aragon-ace-ipsec-profile] to achieve communication security, proof-of-possession and server authentication.
More specifically and with reference to the terminology defined in OAuth 2.0:
Both the joining endpoint and the Group Manager MUST adopt secure communication also for any message exchange with the Authorization Server. To this end, different alternatives are possible, such as OSCORE, DTLS [RFC6347] or IPsec [RFC4301].
This section describes three possible approaches that can be considered by listener endpoints to synchronize with sequence numbers of multicasters.
Upon receiving a multicast request from a multicaster, a listener endpoint does not take any action to synchonize with the sequence number of that multicaster. This provides no assurance at all as to message freshness, which can be acceptable in non-critical use cases.
Upon receiving a multicast request from a given multicaster for the first time, a listener endpoint initializes its last-seen sequence number in its Recipient Context associated to that multicaster. However, the listener drops the multicast request without delivering it to the application layer. This provides a reference point to identify if future multicast requests from the same multicaster are fresher than the last one received.
A replay time interval exists, between when a possibly replayed message is originally transmitted by a given multicaster and the first authentic fresh message from that same multicaster is received. This can be acceptable for use cases where listener endpoints admit such a trade-off between performance and assurance of message freshness.
A listener endpoint performs a challenge-response exchange with a multicaster, by using the Repeat Option for CoAP described in Section 2 of [I-D.amsuess-core-repeat-request-tag].
That is, upon receiving a multicast request from a particular multicaster for the first time, the listener processes the message as described in Section 5.2 of this specification, but, even if valid, does not deliver it to the application. Instead, the listener replies to the multicaster with a 4.03 Forbidden response message including a Repeat Option, and stores the option value included therein.
Upon receiving a 4.03 Forbidden response that includes a Repeat Option and originates from a verified group member, a multicaster MUST send a group request as a unicast message addressed to the same listener, echoing the Repeat Option value. In particular, the multicaster does not necessarily resend the same group request, but can instead send a more recent one, if the application permits it. This makes it possible for the multicaster to not retain previously sent group requests for full retransmission, unless the application explicitly requires otherwise. In either case, the multicaster uses the sequence number value currently stored in its own Sender Context. If the multicaster stores group requests for possible retransmission with the Repeat Option, it should not store a given request for longer than a pre-configured time interval. Note that the unicast request echoing the Repeat Option is correctly treated and processed as a group message, since the "gid" field including the Group Identifier of the OSCORE group is still present in the Object-Security Option as part of the COSE object (see Section 4).
Upon receiving the unicast group request including the Repeat Option, the listener verifies that the option value equals the stored and previously sent value; otherwise, the request is silently discarded. Then, the listener verifies that the unicast group request has been received within a pre-configured time interval, as described in [I-D.amsuess-core-repeat-request-tag]. In such a case, the request is further processed and verified; otherwise, it is silently discarded. Finally, the listener updates the Recipient Context associated to that multicaster, by setting the Replay Window according to the Sequence Number from the unicast group request conveying the Repeat Option. The listener either delivers the request to the application if it is an actual retransmission of the original one, or discard it otherwise. Mechanisms to signal whether the resent request is a full retransmission of the original one are out of the scope of this specification.
In case it does not receive a valid group request including the Repeat Option within the configured time interval, the listener node SHOULD perform the same challenge-response upon receiving the next multicast request from that same multicaster.
A listener SHOULD NOT deliver group request messages from a given multicaster to the application until one valid group request from that same multicaster has been verified as fresh, as conveying an echoed Repeat Option [I-D.amsuess-core-repeat-request-tag]. Also, a listener MAY perform the challenge-response described above at any time, if synchronization with sequence numbers of multicasters is (believed to be) lost, for instance after a device reboot. It is the role of the application to define under what circumstances sequence numbers lose synchronization. This can include a minimum gap between the sequence number of the latest accepted group request from a multicaster and the sequence number of a group request just received from the same multicaster. A multicaster MUST always be ready to perform the challenge-response based on the Repeat Option in case a listener starts it.
Note that endpoints configured as pure listeners are not able to perform the challenge-response described above, as they do not store a Sender Context to secure the 4.03 Forbidden response to the multicaster. Therefore, pure listeners should adopt alternative approaches to achieve and maintain synchronization with sequence numbers of multicasters.
This approach provides an assurance of absolute message freshness. However, it can result in an impact on performance which is undesirable or unbearable, especially in large multicast groups where many nodes at the same time might join as new members or lose synchronization.
There are some application scenarios using group communications that have particularly strict requirements. One example of this is the requirement of low message latency in non-emergency lighting applications [I-D.somaraju-ace-multicast]. For those applications which have tight performance constraints and relaxed security requirements, it can be inconvenient for some endpoints to verify digital signatures in order to assert source authenticity of received group messages. In other cases, the signature verification can be deferred or only checked for specific actions. For instance, a command to turn a bulb on where the bulb is already on does not need the signature to be checked. In such situations, the counter signature needs to be included anyway as part of the group message, so that an endpoint that needs to validate the signature for any reason has the ability to do so.
In this specification, it is NOT RECOMMENDED that endpoints do not verify the counter signature of received group messages. However, it is recognized that there may be situations where it is not always required. The consequence of not doing the signature validation is that security in the group is based only on the group-authenticity of the shared keying material used for encryption. That is, endpoints in the multicast group have evidence that a received message has been originated by a group member, although not specifically identifiable in a secure way. This can violate a number of security requirements, as the compromise of any element in the group means that the attacker has the ability to control the entire group. Even worse, the group may not be limited in scope, and hence the same keying material might be used not only for light bulbs but for locks as well. Therefore, extreme care must be taken in situations where the security requirements are relaxed, so that deployment of the system will always be done safely.