Internet-Draft | MQTT-TLS profile of ACE | March 2022 |
Sengul & Kirby | Expires 2 September 2022 | [Page] |
This document specifies a profile for the ACE (Authentication and Authorization for Constrained Environments) framework to enable authorization in a Message Queuing Telemetry Transport (MQTT)-based publish-subscribe messaging system. Proof-of-possession keys, bound to OAuth2.0 access tokens, are used to authenticate and authorize MQTT Clients. The protocol relies on TLS for confidentiality and MQTT server (broker) authentication.¶
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 2 September 2022.¶
Copyright (c) 2022 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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
This document specifies a profile for the ACE framework [I-D.ietf-ace-oauth-authz]. In this profile, Clients and Servers (Brokers) use MQTT to exchange Application Messages. The protocol relies on TLS for communication security between entities. The MQTT protocol interactions are described based on the MQTT v5.0 - the OASIS Standard [MQTT-OASIS-Standard-v5]. Since it is expected that MQTT deployments will continue to support MQTT v3.1.1 clients, this document also describes a reduced set of protocol interactions for MQTT v3.1.1 - the OASIS Standard [MQTT-OASIS-Standard]. However, MQTT v5.0 is the RECOMMENDED version as it works more naturally with ACE-style authentication and authorization.¶
MQTT is a publish-subscribe protocol, and after connecting to the MQTT Server (Broker), a Client can publish and subscribe to multiple topics. The Broker, which acts as the Resource Server (RS), is responsible for distributing messages published by the publishers to their subscribers. In the rest of the document, the terms "RS", "MQTT Server" and "Broker" are used interchangeably.¶
Messages are published under a Topic Name, and subscribers subscribe to the Topic Names to receive the corresponding messages. The Broker uses the Topic Name in a published message to determine which subscribers to relay the messages to. In this document, topics, more specifically, Topic Names, are treated as resources. The Clients are assumed to have identified the publish/subscribe topics of interest out-of-band (topic discovery is not a feature of the MQTT protocol). A Resource Owner can pre-configure policies at the Authorization Server (AS) that give Clients publish or subscribe permissions to different topics.¶
Clients prove their permission to publish and subscribe to topics hosted on an MQTT broker using an access token, bound to a proof-of-possession (PoP) key. This document describes how to authorize the following exchanges between the Clients and the Broker.¶
Clients use the MQTT PUBLISH packet to publish to a topic. This document does not protect the payload of the PUBLISH packet from the Broker. Hence, the payload is not signed or encrypted specifically for the subscribers. This functionality may be implemented using the proposal outlined in the ACE Pub-Sub Profile [I-D.ietf-ace-pubsub-profile].¶
To provide communication confidentiality and Broker authentication to the MQTT clients, TLS is used, and TLS 1.3 [RFC8446] is RECOMMENDED. This document makes the same assumptions as Section 4 of the ACE framework [I-D.ietf-ace-oauth-authz] regarding Client and RS registration with the AS and setting up the keying material. While the Client-Broker exchanges are only over MQTT, the required Client-AS and RS-AS interactions are described for HTTPS-based communication [RFC7230], using 'application/ace+json' content type, and unless otherwise specified, using JSON encoding. The token MAY be a reference or JSON Web Token (JWT) [RFC7519]. For JWTs, this document follows [RFC7800] for PoP semantics for JWTs, and the mechanisms for providing and verifying proof-of-possession are detailed in Section 2.2. The Client-AS and RS-AS MAY also use protocols other than HTTP, e.g. Constrained Application Protocol (CoAP) [RFC7252] or MQTT; it is recommended that TLS is used to secure these communication channels between Client-AS and RS-AS. Implementations MAY also use "application/ace+cbor" content type, and CBOR encoding [RFC8949], and CBOR Web Token (CWT) [RFC8392] and associated PoP semantics to reduce the protocol memory and bandwidth requirements. For more information, see Proof-of-Possession Key Semantics for CBOR Web Tokens (CWTs) [RFC8747]. A JWT token uses JOSE, while a CWT token uses COSE [RFC8152] for security protection.¶
The keywords "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.¶
Certain security-related terms such as "authentication", "authorization", "confidentiality", "(data) integrity", "message authentication code", and "verify" are taken from [RFC4949].¶
The terminology for entities in the architecture is defined in OAuth 2.0 [RFC6749] such as "Client" (C), "Resource Server" (RS) and "Authorization Server" (AS).¶
The term "resource" is used to refer to an MQTT Topic Name, which is defined in Section 1.3. Hence, the "Resource Owner" is any entity that can authoritatively speak for the topic. This document also defines a Client Authorization Server for Clients that are not able to support HTTP.¶
The document describes message exchanges as MQTT protocol interactions. The Clients are MQTT Clients, which connect to the Broker to publish and subscribe to Application Messages, labelled with their topics. For additional information, please refer to the MQTT v5.0 - the OASIS Standard [MQTT-OASIS-Standard-v5] or the MQTT v3.1.1 - the OASIS Standard [MQTT-OASIS-Standard].¶
MQTT sends various control packets across a network connection. The following is not an exhaustive list, and the control packets that are not relevant for authorization are not explained. These include, for instance, the PUBREL and PUBCOMP packets used in the 4-step handshake required for QoS level 2.¶
This section specifies how Client connections are authorized by the AS, and verified by the MQTT Broker. Figure 1 shows the basic protocol flow during connection setup. The token request and response use the token endpoint at the AS, specified for HTTP-based interactions in Section 5.8 of the ACE framework [I-D.ietf-ace-oauth-authz]. Steps (D) and (E) are optional and use the introspection endpoint specified in Section 5.9 of the ACE framework. The discussion in this document assumes that the Client and the Broker use HTTPS to communicate with the AS via these endpoints. The Client and the Broker use MQTT to communicate between them. The C-AS and Broker-AS communication MAY be implemented using protocols other than HTTPS, e.g. CoAP or MQTT. Whatever protocol is used for C-AS and Broker-AS communications must provide mutual authentication, confidentiality protection, and integrity protection.¶
If the Client is resource-constrained or does not support HTTPS, a separate Client Authorization Server may carry out the token request on behalf of the Client, and later, onboard the Client with the token. The interactions between a Client and its Client Authorization Server for token onboarding and support for MQTTS-based token requests at the AS are out of the scope of this document.¶
The first step in the protocol flow (Figure 1 (A)) is the token acquisition by the Client from the AS. The Client and the AS MUST perform mutual authentication. The Client requests an access token from the AS as described in Section 5.8.1 of the ACE framework [I-D.ietf-ace-oauth-authz]. The document follows the procedures defined in Section 3.2.1 of the DTLS profile [I-D.ietf-ace-dtls-authorize] for RPK, and in Section 3.3.1 of the same document for PSK. However, the content type of the request is set to 'application/ace+json', and the AS uses JSON in the payload of its responses to the Client and the RS. As explained earlier, implementations MAY also use "application/ace+cbor" content type.¶
On receipt of the token request, the AS verifies the request. If the AS successfully verifies the access token request and authorizes the Client for the indicated audience (i.e., RS) and scopes (i.e., publish/subscribe permissions over topics as described in Section 2.3), the AS issues an access token (Figure 1 (B)).¶
The response includes the parameters described in Section 5.8.2 of the ACE framework [I-D.ietf-ace-oauth-authz]. For RPK, the parameters are as described in Section 3.2.1 of the DTLS profile [I-D.ietf-ace-dtls-authorize]. For PSK, the document follows Section 3.3.1 of the DTLS profile [I-D.ietf-ace-dtls-authorize]. In both cases, if the response contains an "ace_profile" parameter, this parameter is set to "mqtt_tls". The returned token is a Proof-of-Possession (PoP) token by default.¶
This document follows [RFC7800] for PoP semantics for JWTs (CWTs can also be used). The PoP token includes a 'cnf' parameter with a symmetric or asymmetric PoP key. Note that the contents of the web tokens (including the 'cnf' parameter) are to be consumed by the RS and not the Client (the Client obtains the key information in a different manner). The RPK case is handled as described in in Section 3.2.1 of the DTLS profile [I-D.ietf-ace-dtls-authorize]. For the PSK case, the referenced procedures apply, with the following exceptions to accommodate JWT and JOSE use. In this case, the AS adds a "cnf" parameter to the access information carrying a JWK (JSON Web Key) [RFC7517] object that contains either the symmetric key itself or a key identifier that can be used by the RS to determine the secret key it shares with the Client. The JWT is created as explained in Section 7 of [RFC7519], and the JWT MUST include JWE [RFC7516]. If CWT/COSE is used this information MUST be inside the "COSE_Key" object, and MUST be encrypted using a "COSE_Encrypt0" structure.¶
The AS returns error responses for JSON-based interactions following Section 5.2 of [RFC6749]. When CBOR is used, the interactions MUST implement Section 5.8.3 of the ACE framework [I-D.ietf-ace-oauth-authz].¶
Unless the Client publishes and subscribes to only public topics, the Client and the Broker MUST perform mutual authentication. The Client MUST authenticate to the Broker either over MQTT or TLS before performing any other action. For MQTT, the options are "None" and "ace". For TLS, the options are "Anon" for an anonymous client, and "Known(RPK/PSK)" for Raw Public Keys (RPK) [RFC7250] and Pre-Shared Keys (PSK), respectively. The "None" and "Anon" options do not provide client authentication but can be used either during authentication or in combination with authentication at the other layer. When the Client uses TLS:Anon,MQTT:None, the Client can only publish or subscribe to public topics. Thus, the client authentication procedures involve the following possible combinations:¶
It is RECOMMENDED that the Client implements TLS:Anon,MQTT:ace as the first choice when working with protected topics. However, MQTT v3.1.1 clients that do not prefer to overload username and password fields for ACE (as described in Section 6) MAY implement TLS:Known(RPK/PSK),MQTT:none, and consequently TLS:Anon,MQTT:None to submit their token to "authz-info".¶
The Broker MUST support TLS:Anon,MQTT:ace. To support Clients with different capabilities, the Broker MAY provide multiple client authentication options, e.g. support TLS:Known(RPK),MQTT:none and TLS:Anon,MQTT:None, to enable RPK-based client authentication.¶
The Client MUST authenticate the Broker during the TLS handshake. If the Client authentication uses TLS:Known(RPK/PSK), then the Broker is authenticated using the respective method. Otherwise, to authenticate the Broker, the client MUST validate a public key from a X.509 certificate or an RPK from the Broker against the 'rs_cnf' parameter in the token response. The AS MAY include the thumbprint of the RS's X.509 certificate in the 'rs_cnf' (thumbprint as defined in [I-D.ietf-cose-x509]). In this case, the client MUST validate the RS certificate against this thumbprint.¶
In the cases when the Client must transport the token to the Broker first, the Client connects to the Broker to publish its token to the "authz-info" topic. The "authz-info" topic MUST be publish-only (i.e., the Clients are not allowed to subscribe to it). "authz-info" is not protected, and hence, the Client uses the TLS:Anon,MQTT:None option over a TLS connection. After publishing the token, the Client disconnects from the Broker and is expected to reconnect using client authentication over TLS (i.e., TLS:Known(RPK/PSK),MQTT:none).¶
The Broker stores and indexes all tokens received to the "authz-info" topic in its key store (similar to the DTLS profile for ACE [I-D.ietf-ace-dtls-authorize]). This profile follows the recommendation of Section 5.10.1 of the ACE framework [I-D.ietf-ace-oauth-authz], and expects that the Broker stores only one token per proof-of-possession key, and any other token linked to the same key overwrites an existing token.¶
The Broker MUST verify the validity of the token (i.e., through local validation or introspection, if the token is a reference) as described in Section 2.2.5. If the token is not valid, the Broker MUST discard the token. Depending on the QoS level of the PUBLISH packet, the Broker returns the error response as a PUBACK, PUBREC, or DISCONNECT packet as explained below.¶
If the QoS level is equal to 0, and the token is invalid or the claims cannot be obtained in the case of an introspected token, the Broker MUST send a DISCONNECT packet with the reason code '0x87 (Not authorized)'. If the PUBLISH payload does not parse to a token, the RS MUST send a DISCONNECT with the reason code '0x99 (Payload format invalid)'.¶
If the QoS level of the PUBLISH packet is greater than or equal to 1, and the token is invalid, or the claims cannot be obtained in the case of an introspected token, the Broker MUST send the reason code 0x87 ('Not authorized') in the PUBACK or PUBREC. If the PUBLISH payload does not parse to a token, the PUBACK/PUBREC reason code is '0x99 (Payload format invalid)'.¶
It must be noted that when the RS sends the 'Not authorized' response, this corresponds to the token being invalid, and not that the actual PUBLISH packet was not authorized. Given that the "authz-info" is a public topic, this response is not expected to cause confusion.¶
This document supports TLS with Raw Public Keys (RPK) [RFC7250] and with Pre-Shared Keys (PSK). The TLS session setup follows the DTLS profile for ACE [I-D.ietf-ace-dtls-authorize], as the profile applies to TLS equally well [I-D.ietf-ace-extend-dtls-authorize]. When there are exceptions to the DTLS profile, these are explicitly stated in the document. If TLS 1.2 is used, [RFC7925] describes how TLS can be used for constrained devices, alongside recommended cipher suites. Additionally, TLS 1.2 implementations MUST use the "Extended Master Secret" extension [RFC7627] to incorporate the handshake transcript into the master secret. TLS implementations SHOULD use the SNI (Server Name Indication) [RFC6066] and APLN (Application-Layer Protocol Negotiation) [RFC7301] extensions so the TLS handshake authenticates as much of the protocol context as possible.¶
This document follows the procedures defined in Section 3.2.2 of the DTLS profile for ACE [I-D.ietf-ace-dtls-authorize] with the following exceptions. The client MUST upload the access token to the Broker using the method specified in Section 2.2.2 before initiating the handshake.¶
This document follows the procedures defined in Section 3.3.2 of DTLS profile for ACE [I-D.ietf-ace-dtls-authorize] with the following exceptions.¶
To use TLS 1.3 with pre-shared keys, the Client utilizes the PSK key extension specified in [RFC8446] using the key conveyed in the "cnf" parameter of the AS response. The same key is bound to the access token in the "cnf" claim. The Client can upload the token as specified in Section 2.2.2 before initiating the handshake. When using a previously uploaded token, the Client MUST indicate during the handshake which previously uploaded access token it intends to use. To do so, it MUST create a "COSE_Key" or "JWK" structure with the "kid" that was conveyed in the "rs_cnf" claim in the token response from the AS and the key type "symmetric". This structure is then included as the only element in the "cnf" structure and the encoded value of that "cnf" structure used as a PSK identity in TLS. As an alternative to the access token upload, the client can provide the most recent access token, JWT or CWT, as a PSK identity.¶
In contrast to DTLS profile for ACE [I-D.ietf-ace-dtls-authorize], a client MAY omit support for the cipher suites TLS_PSK_WITH_AES_128_CCM_8 and TLS_ECDHE_ECDSA_WITH_AES_128_CCM_8, but for TLS 1.2, MUST support TLS_ECDHE_PSK_WITH_AES_128_GCM_SHA256 for PSK and TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256 for RPK, as recommended in [RFC7525] (and adjusted to be a PSK cipher suite as appropriate).¶
This section describes how the Client transports the token to the Broker inside the CONNECT packet. If this method is used, the Client TLS connection is expected to be anonymous, and the Broker is authenticated during the TLS connection setup. The approach described in this section is similar to an earlier proposal by Fremantle et al [fremantle14].¶
After sending the CONNECT, the client MUST wait to receive the CONNACK from the Broker. The only packets it is allowed to send are DISCONNECT or AUTH that is in response to the Broker AUTH. Similarly, except for a DISCONNECT and AUTH response from the Client, the Broker MUST NOT process any packets before sending a CONNACK.¶
Figure 2 shows the structure of the MQTT CONNECT packet used in MQTT v5.0. A CONNECT packet is composed of a fixed header, a variable header and a payload. The fixed header contains the Control Packet Type (CPT), Reserved, and Remaining Length fields. Remaining Length is a Variable Byte Integer that represents the number of bytes remaining within the current Control Packet, including data in the Variable Header and the Payload. The Variable Header contains the Protocol Name, Protocol Level, Connect Flags, Keep Alive, and Properties fields. The Connect Flags in the variable header specify the properties of the MQTT session. It also indicates the presence or absence of some fields in the Payload. The payload contains one or more encoded fields, namely a unique Client Identifier for the Client, a Will Topic, Will Payload, User Name and Password. All but the Client Identifier can be omitted depending on the flags in the Variable Header. The Client Identifier identifies the Client to the Broker, and therefore, is unique for each Client. It must be noted that the Client Identifier is an unauthenticated identifier used within the MQTT protocol and so is not bound to the access token.¶
The CONNECT flags are Username, Password, Will retain, Will QoS, Will Flag, Clean Start, and Reserved. Figure 3 shows how the flags MUST be set to use AUTH packets for authentication and authorization, i.e., the username and password flags MUST be set to 0. An MQTT v5.0 RS MAY also support token transport using Username and Password to provide a security option for MQTT v3.1.1 clients, as described in Section 6.¶
The Will Flag indicates that a Will message needs to be sent if the network connection is not closed normally. The situations in which the Will message is published include disconnections due to I/O or network failures and the server closing the network connection due to a protocol error. The Client MAY set the Will Flag as desired (marked as 'X' in Figure 3). If the Will Flag is set to 1, the Broker MUST check that the token allows the publication of the Will message (i.e., the Will Topic filter is in the scope array). The check is performed against the token scope described in Section 2.3. If the Will authorization fails, the connection is refused as described in Section 2.4.1. If the Broker accepts the connection request, the Broker stores the Will message and publishes it when the network connection is closed according to Will QoS, and Will retain parameters and MQTT Will management rules. To avoid publishing the Will Messages in the case of temporary network disconnections, the Client specifies a Will Delay Interval in the Will Properties. Section 5 explains how the Broker deals with the retained messages in further detail.¶
In MQTT v5.0, the Client signals a clean session (i.e., that the session does not continue an existing session) by setting the Clean Start Flag to 1 in the CONNECT packet. In this profile, the Client SHOULD always start with a clean session. The Broker MAY also signal that it does not support session continuation by setting Session Expiry Interval to 0 in the CONNACK. If the Broker starts a clean session, the Broker MUST set the Session Present flag to 0 in the CONNACK packet to signal this to the Client.¶
The Broker MAY support session continuation, e.g. if the Broker requires it for QoS reasons. In this case, if a CONNECT packet is received with Clean Start set to 0 and there is a Session associated with the Client Identifier, the Broker MUST resume communications with the Client based on the state from the existing Session. In its response, the Broker MUST set the Session Present flag to 1 in the CONNACK packet to signal session continuation to the Client. The session state stored by the Client and the Broker is described in Section 5.¶
When reconnecting to a Broker that supports session continuation, the Client MUST still provide a token, in addition to using the same Client Identifier, and setting the Clean Start to 0. The Broker MUST still perform proof-of-possession validation on the provided token. If the token matches the stored state, the Broker MAY skip introspecting a token-by-reference and use the stored introspection result. The Broker MUST also verify the Client is authorized to receive or send MQTT packets that are pending transmission. When a Client connects with a long Session Expiry Interval, the Broker may need to maintain the Client's MQTT session state after it disconnects for an extended period. Brokers SHOULD implement administrative policies to limit misuse.¶
Note that, according to the MQTT standard, the Broker uses the Client Identifier to identify the session state. In the case of a Client Identifier collision, a client may take over another client's session. Given that clients provide a token at each connection, clients will only send or receive messages to their authorized topics. Therefore, while this issue is not expected to affect security, it may affect QoS (i.e., PUBLISH or QoS messages saved for Client A may be delivered to a Client B). In addition, if this Client Identifier represents a Client already connected to the Broker, the Broker sends a DISCONNECT packet to the existing Client with Reason Code of '0x8E (Session taken over)', and closes the connection to the Client.¶
To use AUTH, the Client MUST set the Authentication Method as a property of a CONNECT packet by using the property identifier 21 (0x15). This is followed by a UTF-8 Encoded String containing the name of the Authentication Method, which MUST be set to "ace". If the RS does not support this profile, it sends a CONNACK with a Reason Code of '0x8C (Bad authentication method)'.¶
The Authentication Method is followed by the Authentication Data, which has a property identifier 22 (0x16) and is Binary Data. Based on the Authentication Data, RS MUST support both options below:¶
For this option, the Authentication Data inside the client's CONNECT MUST contain the two-byte integer token length, the token, and the keyed message digest (MAC) or the Client signature (as shown in Figure 4). The Proof-of-Possession key in the token is used to calculate the keyed message digest (MAC) or the Client signature based on the content obtained from the TLS exporter ([RFC5705] for TLS 1.2, and Section 7.5 of [RFC8446]) for TLS 1.3. This content is exported from the TLS session using the exporter label 'EXPORTER-ACE-MQTT-Sign-Challenge', an empty context, and length of 32 bytes. The token is also validated as described in Section 2.2.5, and the Broker responds with a CONNACK with the appropriate response code. The Client cannot reauthenticate using this method during the same TLS session (see Section 4).¶
For this option, the Broker follows a Broker-generated challenge/response protocol. If the Authentication Data inside the Client's CONNECT contains only the two-byte integer token length and the token (as shown in Figure 5), the RS MUST respond with an AUTH packet, with the Authenticate Reason Code set to "0x18 (Continue Authentication)". The Broker also uses this method if the Authentication Data does not contain a token, but the Broker has a token stored for the connecting client.¶
The AUTH packet to continue authentication includes the Authentication Method, which MUST be set to "ace" and Authentication Data. The Authentication Data MUST NOT be empty and contains an 8-byte RS nonce as a challenge for the Client (Figure 6).¶
The Client responds to this with an AUTH packet with a reason code "0x18 (Continue Authentication)". Similarly, the Client packet sets the Authentication Method to "ace". The Authentication Data in the Client's response is formatted as shown in Figure 7 and includes the the 8-byte Client nonce, and the signature or MAC computed over the RS nonce concatenated with the Client nonce using PoP key in the token.¶
Next, the token is validated as described in Section 2.2.5. The success case is illustrated in Figure 8. The client MAY also re-authenticate using this challenge-response flow, as described in Section 4.¶
The Broker MUST verify the validity of the token either locally (e.g. in the case of a self-contained token) or MAY send a request to the introspection endpoint of the AS (as described for HTTP-based interactions in Section 5.9 of the ACE framework [I-D.ietf-ace-oauth-authz]). The Broker MUST verify the claims in the access token according to the rules set in Section 5.10.1.1 of the ACE framework [I-D.ietf-ace-oauth-authz].¶
To authenticate the Client, the Broker validates the signature or the MAC, depending on how the PoP protocol is implemented. For self-contained tokens, the Broker MUST process the security protection of the token first, as specified by the respective token format, i.e. a CWT token uses COSE, while a JWT token uses JOSE. For a token-by-reference, the Broker uses the "cnf" structure returned as a result of token introspection as specified in [RFC7519]. HS256 (HMAC-SHA-256) [RFC6234] and Ed25519 [RFC8032] are mandatory to implement for the Broker. The Client MUST implement at least one of them depending on the choice of symmetric or asymmetric validation. Validation of the signature or MAC MUST fail if the signature algorithm is set to "none", when the key used for the signature algorithm cannot be determined, or the computed and received signature/MAC do not match.¶
The Broker MUST check if the access token is still valid, if it is the intended destination (i.e., the audience) of the token, and if the token was issued by an authorized authorization server. If the Client is using TLS RPK mode to authenticate to the Broker, the AS constructs the access token so that the Broker can associate the access token with the Client's public key. The "cnf" claim MUST contain either the client's RPK or, if the key is already known by the RS (e.g., from previous communication), a reference to it.¶
The scope field contains the publish and subscribe permissions for the Client. Therefore, the token or its introspection result MUST be cached to allow a Client's future PUBLISH and SUBSCRIBE packets. During the CONNECT, if the Will Flag is set to 1, the Broker MUST also authorize the publication of the Will Topic and message using the token's scope field. The Broker uses the scope to match against the Topic Name in a PUBLISH packet (including Will Topic in the CONNECT) or a Topic Filter in a SUBSCRIBE packet.¶
The scope in the token is a single value. For a JWT, the single scope is base64url encoded string with any padding characters removed, which has an internal structure of a JSON array. For a CWT, this information is represented in CBOR. The internal structure follows the Authorization Information Format (AIF) for ACE [I-D.ietf-ace-aif]. Using the Concise Data Definition Language (CDDL) [RFC8610], the specific data model for MQTT is:¶
Topic filters are implemented according to Section 4.7 of MQTT v5.0 - the OASIS Standard [MQTT-OASIS-Standard-v5]. By default, Wildcard Subscriptions are supported, and so, the topic filter may include special wildcard characters. The multi-level wildcard, '#', matches any number of levels within a topic, and the single-level wildcard, '+', matches one topic level. The Broker MAY signal in the CONNACK explicitly whether wildcard subscriptions are supported by returning a CONNACK property "Wildcard Subscription Available". A value of 0 means that Wildcard Subscriptions are not supported. A value of 1 means Wildcard Subscriptions are supported.¶
Following this model, an example scope may contain:¶
This access token gives publish ("pub") and subscribe ("sub") permissions to the "topic1", publish permission to all the subtopics of "topic2", and subscribe permission to all "topic3", skipping one level.¶
If the scope is empty, the Broker records no permissions for the client for any topic. In this case, the Client is not able to publish or subscribe to any protected topics. The non-empty scope is used to authorize the Will Topic, if provided, in the CONNECT packet, during connection setup, and if the connection request succeeds, the Topic Names or Topic Filters requested in the future PUBLISH and SUBSCRIBE packets. For the authorization to succeed, the Broker MUST verify that the topic name or filter in question is either an an exact match to or a subset of at least one 'topic_filter' in the scope.¶
Based on the validation result (obtained either via local inspection or using the introspection interface of the AS), the Broker MUST send a CONNACK packet to the Client.¶
If the Client does not provide a valid token or omits the Authentication Data field and the Broker has no token stored for the Client or the token or Authentication data are malformed, or if the Will flag is set, the authorization checks for the Will topic fails, authentication fails. The Broker responds with the CONNACK reason code "0x87 (Not Authorized)" or any other applicable reason code.¶
The Broker MAY also trigger AS discovery and include a User Property (identified as property type 38 (0x26)) in the CONNACK for the AS Request Creation Hints. The User Property is a UTF-8 string pair, composed of a name and a value. The name of the User Property MUST be set to "ace_as_hint". The value of the user property is a UTF-8 encoded JSON object containing the mandatory "AS" parameter, and the optional parameters "audience", "kid", "cnonce", and "scope" as defined in Section 5.3 of the ACE framework [I-D.ietf-ace-oauth-authz].¶
On success, the reason code of the CONNACK is "0x00 (Success)". If the Broker starts a new session, it MUST also set Session Present to 0 in the CONNACK packet to signal a clean session to the Client. Otherwise, it MUST set Session Present to 1.¶
Having accepted the connection, the Broker MUST be prepared to store the token during the connection and after disconnection for future use. If the token is not self-contained and the Broker uses token introspection, it MAY cache the validation result to authorize the subsequent PUBLISH and SUBSCRIBE packets. PUBLISH and SUBSCRIBE packets, which are sent after a connection setup, do not contain access tokens. If the introspection result is not cached, the Broker needs to introspect the saved token for each request. The Broker SHOULD also use a cache timeout to introspect tokens regularly. The timeout value is application-specific and SHOULD be chosen to reduce the risk of using stale introspection responses.¶
The Broker MUST check for token expiration whenever a CONNECT, PUBLISH or SUBSCRIBE is received or sent. The Broker SHOULD check for token expiration on receiving a PINGREQUEST. The Broker MAY also check for token expiration periodically, e.g. every hour. This may allow for early detection of a token expiry.¶
The token expiration is checked by checking the 'exp' claim of a JWT or introspection response or via performing an introspection request with the AS as described in Section 5.9 of the ACE framework [I-D.ietf-ace-oauth-authz]. Token expirations may trigger the Broker to send PUBACK, SUBACK and DISCONNECT packets with return code set to "Not authorized". After sending a DISCONNECT, the network connection is closed, and no more messages can be sent.¶
The Client MAY reauthenticate as a response to the PUBACK and SUBACK that signal loss of authorization. The Clients MAY also proactively update their tokens, i.e., before they receive a packet with a 'Not authorized' return code. To start reauthentication, the Client MUST send an AUTH packet with the reason code '0x19 (Re-authentication)'. The Client MUST set the Authentication Method as "ace" and transport the new token in the Authentication Data. To re-authenticate, the Client MUST NOT use Proof-of-Possession using a challenge from the TLS session during the same TLS session to avoid re-using the same challenge value from the TLS-Exporter. Note that this means that servers will either need to record in the session ticket or database entry whether the TLS-Exporter-derived challenge was used, or always deny use of the TLS-Exporter-derived challenge for resumed sessions. In TLS 1.3, the resumed connection would have a new exporter value, but the requirement is phrased this way for simplicity. For re-authentications in the same TLS-session, the Client MUST use the challenge-response PoP as defined in Section 2.2.4.2.2. The Broker accepts reauthentication requests if the Client has already submitted a token (may be expired), for which it performed proof-of-possession. Otherwise, the Broker MUST deny the request. If the reauthentication fails, the Broker MUST send a DISCONNECT with the reason code '0x87 (Not Authorized)'.¶
In the case of a Client DISCONNECT, if the Session Expiry Interval is set to 0, the Broker doesn't maintain session state but MUST keep the retained messages. If the Broker maintains session state, the state MAY include token and its introspection result (for reference tokens) in addition to the MQTT session state. The MQTT session state is identified by the Client Identifier and includes state on client subscriptions; messages with QoS levels 1 and 2, and which have not been completely acknowledged or are pending transmission to the Client; and if the Session is currently not connected, the time at which the Session will end and Session State will be discarded. The token/introspection state is not part of the MQTT session state, and proof-of-possession is required for each new connection, regardless of whether MQTT session continuation is used.¶
The messages to be retained are indicated to the Broker by setting a RETAIN flag in a PUBLISH packet. This way, the publisher signals to the Broker to store the most recent message for the associated topic. Hence, the new subscribers can receive the last sent message from the publisher for that particular topic without waiting for the next PUBLISH packet. The Broker MUST continue publishing the retained messages as long as the associated tokens are valid. In the MQTT standard, if QoS is 0 for the PUBLISH packet, the Broker may discard the retained message any time. For QoS>1, the message expiry interval dictates how long the retained message is kept. However, it is important that the Broker avoids sending messages indefinitely for the Clients that never update their tokens (i.e., the Client connects briefly with a valid token, sends a PUBLISH packet with RETAIN flag set to 1 and QoS>1, disconnects and never connects again). Therefore, the Broker MUST use the minimum of token expiry and message expiry interval to discard a retained message.¶
In case of disconnections due to network errors or server disconnection due to a protocol error (which includes authorization errors), the Will message is sent if the Client supplied a Will in the CONNECT packet. The Client's token scope array MUST include the Will Topic. The Will message MUST be published to the Will Topic regardless of whether the corresponding token has expired (as it has been validated and accepted during CONNECT).¶
This section describes a reduced set of protocol interactions for the MQTT v3.1.1 Clients. An MQTT v5.0 Broker MAY implement these interactions for the MQTT v3.1.1 clients; MQTT v5.0 clients are NOT RECOMMENDED to use the flows described in this section. Brokers that do not support MQTT v3.1.1 clients return a CONNACK packet with Reason Code '0x84 (Unsupported Protocol Version)' in response to the connection requests.¶
As in MQTT v5.0, the token MAY either be transported before, by publishing to the "authz-info" topic, or inside the CONNECT packet. If the Client provided the token via the "authz-info" topic and will not update the token in the CONNECT packet, it MUST authenticate over TLS. The Broker SHOULD still be prepared to store the Client access token for future use (regardless of the method of transport).¶
In MQTT v3.1.1, after the Client published to the "authz-info" topic, the Broker cannot communicate the result of the token validation as PUBACK reason codes or server-side DISCONNECT packets are not supported. In any case, an invalid token would fail the subsequent TLS handshake, which can prompt the Client to obtain a valid token.¶
To transport the token to the Broker inside the CONNECT packet, the Client uses the username and password fields. Figure 11 shows the structure of the MQTT CONNECT packet.¶
Figure 12 shows how the MQTT connect flags MUST be set to initiate a connection with the Broker.¶
The Client SHOULD set the Clean flag to 1 to always start a new session. If the Clean flag is set to 0, the Broker MUST resume communications with the Client based on the state from the current Session (as identified by the Client Identifier). If there is no Session associated with the Client Identifier the Broker MUST create a new session. The Broker MUST set the Session Present flag in the CONNACK packet accordingly, i.e. 0 to indicate a clean session to the Client and 1 to indicate session continuation. The Broker MUST still perform proof-of-possession validation on the provided Client token. MQTT v3.1.1 does not use a Session Expiry Interval, and the Client expects that the Broker maintains the session state after it disconnects. However, stored Session state can be discarded as a result of administrator policies, and Brokers SHOULD implement the necessary policies to limit misuse.¶
The Client MAY set the Will Flag as desired (marked as 'X' in Figure 12). Username and Password flags MUST be set to 1 to ensure that the Payload of the CONNECT packet includes both Username and Password fields. The MQTT Username is a UTF-8 encoded string, and the MQTT Password is Binary Data.¶
The CONNECT in MQTT v3.1.1 does not have a field to indicate the authentication method. To signal that the Username field contains an ACE token, this field MUST be prefixed with 'ace' keyword, i.e., the Username field is a concatenation of 'a', 'c', 'e' and the access token represented as:¶
To this end, the access token MUST be base64url encoded, omitting the '=' padding characters [RFC4648].¶
The password field MUST be set to the keyed message digest (MAC) or signature associated with the access token for proof-of-possession. The Client MUST apply the PoP key on the challenge derived from the TLS session as described in Section 2.2.4.2.1.¶
Error handling is more primitive in MQTT v3.1.1 due to not having appropriate error fields, error codes, and server-side DISCONNECTs. Therefore, the broker will disconnect on almost any error and may not keep the session state, necessitating that clients make a greater effort to ensure that tokens remain valid and not attempt to publish to topics that they do not have permissions for. The following lists how the broker responds to specific errors.¶
Note to RFC Editor: Please replace all occurrences of "[this document]" with the RFC number of this specification and delete this paragraph.¶
This document registers 'EXPORTER-ACE-MQTT-Sign-Challenge' (introduced in Section 2.2.4.2.1 in this document) in the TLS Exporter Label Registry [RFC8447].¶
This document registers the 'application/ace+json' media type for messages of the protocols defined in this document carrying parameters encoded in JSON.¶
Additional information:¶
The following registrations are done for the ACE OAuth Profile Registry following the procedure specified in [I-D.ietf-ace-oauth-authz].¶
This document specifies a profile for the Authentication and Authorization for Constrained Environments (ACE) framework [I-D.ietf-ace-oauth-authz]. Therefore, the security considerations outlined in [I-D.ietf-ace-oauth-authz] apply to this work.¶
In addition, the security considerations outlined in MQTT v5.0 - the OASIS Standard [MQTT-OASIS-Standard-v5] and MQTT v3.1.1 - the OASIS Standard [MQTT-OASIS-Standard] apply. Mainly, this document provides an authorization solution for MQTT, the responsibility of which is left to the specific implementation in the MQTT standards. In the following, we comment on a few relevant issues based on the current MQTT specifications.¶
After the Broker validates an access token and accepts a connection from a client, it caches the token to authorize a Client's publish and subscribe requests in an ongoing session. Broker does not cache any invalid tokens. If a client's permissions get revoked, but the access token has not expired, the Broker may still grant publish/subscribe to revoked topics. If the Broker caches the token introspection responses, then the Broker SHOULD use a reasonable cache timeout to introspect tokens regularly. The timeout value is application-specific and SHOULD be chosen to reduce the risk of using stale introspection responses. When permissions change dynamically, it is expected that AS also follows a reasonable expiration strategy for the access tokens.¶
The Broker may monitor Client behaviour to detect potential security problems, especially those affecting availability. These include repeated token transfer attempts to the public "authz-info" topic, repeated connection attempts, abnormal terminations, and Clients that connect but do not send any data. If the Broker supports the public "authz-info" topic, described in Section 2.2.2, then this may be vulnerable to a DDoS attack, where many Clients use the "authz-info" public topic to transport tokens that are not meant to be used, and which the Broker may need to store until the tokens expire.¶
For MQTT v5.0, when a Client connects with a long Session Expiry Interval, the Broker may need to maintain the Client's MQTT session state after it disconnects for an extended period. For MQTT v3.1.1, the session state may need to be stored indefinitely, as it does not have a Session Expiry Interval feature. The Broker SHOULD implement administrative policies to limit misuse of the session continuation by the Client.¶
The privacy considerations outlined in [I-D.ietf-ace-oauth-authz] apply to this work.¶
In MQTT, the RS is a central trusted party and may forward potentially sensitive information between Clients. The mechanisms defined in this document do not protect the contents of the PUBLISH packet from the Broker, and hence, the content of the PUBLISH packet is not signed or encrypted separately for the subscribers. This functionality may be implemented using the proposal outlined in the ACE Pub-Sub Profile [I-D.ietf-ace-pubsub-profile]. However, this solution would still not provide privacy for other fields of the packet, such as Topic Name.¶
Version 11 to 14: Addressing AD review comments.¶
Version 10 to 11: Clarified the TLS use between RS-AS and Client-AS.¶
Version 09 to 10: Fixed version issues for references.¶
Version 08 to 09: Fixed spacing issues and references.¶
Version 07 to 08:¶
Version 06 to 07:¶
Version 05 to 06:¶
Version 04 to 05:¶
Version 03 to 04:¶
Version 02 to 03:¶
Version 01 to 02:¶
Version 00 to 01:¶
The authors would like to thank Ludwig Seitz for his review and his input on the authorization information endpoint, and Benjamin Kaduk for his review, insightful comments, and contributions to resolving issues. The authors would like to thank Paul Fremantle for the initial discussions on MQTT v5.0 support.¶