v: 3
title: Reference Interaction Models for Remote Attestation Procedures abbrev: REIM docname: draft-ietf-rats-reference-interaction-models-latest wg: RATS Working Group stand_alone: true ipr: trust200902 area: Security kw: Internet-Draft cat: info pi: toc: yes sortrefs: yes symrefs: yes
author:
- ins: H. Birkholz name: Henk Birkholz org: Fraunhofer SIT abbrev: Fraunhofer SIT email: henk.birkholz@sit.fraunhofer.de street: Rheinstrasse 75 code: '64295' city: Darmstadt country: Germany
- ins: M. Eckel name: Michael Eckel org: Fraunhofer SIT abbrev: Fraunhofer SIT email: michael.eckel@sit.fraunhofer.de street: Rheinstrasse 75 code: '64295' city: Darmstadt country: Germany
- ins: W. Pan name: Wei Pan org: Huawei Technologies email: william.panwei@huawei.com
- ins: E. Voit name: Eric Voit org: Cisco Systems abbrev: Cisco email: evoit@cisco.com
normative: RFC3161: TSA RFC5280: X509 RFC7049: CBOR RFC7252: COAP BCP205: RFC8610: CDDL RFC9334: RATS RFC9683: RIV I-D.ietf-rats-epoch-markers: epoch-markers
informative: I-D.birkholz-rats-tuda: TUDA DAA: title: Direct Anonymous Attestation author: - ins: E. Brickell name: Ernie Brickell - ins: J. Camenisch name: Jan Camenisch - ins: L. Chen name: Liqun Chen seriesinfo: ACM: > Proceedings of the 11th ACM conference on Computer and Communications Security page: 132-145 date: 2004 turtles: title: "Turtles All the Way Down: Foundation, Edifice, and Ruin in Faulkner and McCarthy" author: - ins: R. Rudnicki name: Robert Rudnicki seriesinfo: The Faulkner Journal: 25.2 DOI: 10.1353/fau.2010.0002 date: 2010 TNC: title: TCG Trusted Network Communications TNC Architecture for Interoperability author: - ins: TCG name: Trusted Computing Group seriesinfo: Specification: Version 2.0 Revision 13 date: 2017 MQTT: title: Message Queuing Telemetry Transport (MQTT) Version 5.0 Committee Specification 02 author: - ins: OASIS name: Organization for the Advancement of Structured Information Standards seriesinfo: Specification: Version 5.0 date: 2018 DesignPatterns: title: Design Patterns - Elements of Reusable Object-Oriented Software author: - ins: E. Gamma name: Erich Gamma - ins: R. Helm name: Richard Helm - ins: R. Johnson name: Ralph Johnson - ins: J. Vlissides name: John Vlissides seriesinfo: Publisher: Addison-Wesley date: 1994 ISIS: title: Exploiting Virtual Synchrony in Distributed Systems author: - ins: K. Birman name: Ken Paul Birman - ins: T. Joseph name: Thomas A. Joseph seriesinfo: DOI: 10.1145/41457.37515 date: 1987 lampson06: title: Practical Principles for Computer Security author: - ins: B. Lampson name: Butler Lampson date: 2006 ...
--- abstract
This document describes interaction models for remote attestation procedures (RATS) {{-RATS}}. Three conveying mechanisms -- Challenge/Response, Uni-Directional, and Streaming Remote Attestation -- are illustrated and defined. Analogously, a general overview about the information elements typically used by corresponding conveyance protocols are highlighted.
--- middle
Remote ATtestation procedureS (RATS, {{-RATS}}) are workflows composed of roles and interactions, in which Verifiers create Attestation Results about the trustworthiness of an Attester's system component characteristics. The Verifier's assessment in the form of Attestation Results is produced based on Endorsements, Reference Values, Attestation Policies, and Evidence -- trustable and tamper-evident Claims Sets about an Attester's system component characteristics -- generated by an Attester. The roles Attester and Verifier, as well as the Conceptual Messages Evidence and Attestation Results are concepts defined by the RATS Architecture {{-RATS}}. This document defines interaction models that can be used in specific RATS-related solution documents. The primary focus of this document is the conveyance of attestation Evidence. The reference models defined can also be applied to the conveyance of other Conceptual Messages in RATS. This document aims to:
-
prevent inconsistencies in descriptions of interaction models in other documents, which may occur due to text cloning and evolution over time, and to
-
highlight the exact delta/divergence between the core characteristics captured in this document and variants of these interaction models used in other specifications or solutions.
In summary, this document enables the specification and design of trustworthy and privacy-preserving conveyance methods for attestation Evidence from an Attester to a Verifier. While the conveyance of other Conceptual Messages is out of scope, the models described in this document may be adapted to apply to the conveyance of other Conceptual Messages, such as Endorsements or Attestation Results, or supplemental messages, such as Epoch Markers {{-epoch-markers}} or stand-alone event logs.
This document uses the following terms defined in {{Section 4 of -RATS}}: Attester, Verifier, Relying Party, Conceptual Message, Evidence, Endorsement, Attestation Result, Appraisal Policy, Attesting Environment, Target Environment.
A PKIX Certificate is an X.509v3 certificate as specified by {{-X509}}.
{::boilerplate bcp14}
"Remote Attestation" is a common expression often associated or connoted with certain properties. In the context of this document, the term "Remote" does not necessarily refer to a remote entity in the scope of network topologies or the Internet. It rather refers to decoupled systems or entities that exchange the Conceptual Message type called Evidence {{-RATS}}. This conveyance can also be "Local", if the Verifier role is part of the same entity as the Attester role, e.g., separate system components of the same Composite Device (a single RATS entity), or the Verifier and Relying Party roles are hosted by the same entity, for example in a cryptographic key Broker system. If an entity takes on two or more different roles, the functions they provide typically reside in isolated environments that are components of the same entity. Examples of such isolated environments include a Trusted Execution Environment (TEE), Baseboard Management Controllers (BMCs), as well as other physical or logical protected/isolated/shielded Computing Environments (e.g., embedded Secure Elements (eSE) or Trusted Platform Modules (TPM)). It is useful but not necessary for readers of this document to be familiar with the Concept Data/Message flows as described in {{Section 3.1 of -RATS}} and the definition of Attestation in general as described in {{-RIV}}.
This document:
- outlines common interaction models between RATS roles;
- illustrates interaction models focusing on conveying Evidence about boot-time integrity from Attesters to Verifiers;
- does not exclude the application of those interaction models to runtime integrity or the conveyance of other RATS Conceptual Messages;
- does not cover every detail about Evidence conveyance.
While details regarding Evidence of run-time integrity are not explicitly highlighted, the provided model descriptions serve as a foundation for developing corresponding model extensions. While the interaction models described in this document, including their variants, cover many relevant conveyance models for Conceptual Messages implemented on the Internet, they do not represent an exhaustive list of all possible models.
Procedures, functions, and services needed for a complete semantic binding of the concepts defined in {{-RATS}} are not covered in this document. Examples of such procedures include: identity establishment, key distribution and enrollment, time synchronization, and certificate revocation.
Furthermore, any processes and duties beyond conducting remote attestation procedures are out of scope. For example, utilizing the results of a remote attestation procedure generated by the Verifier, such as triggering remediation actions or recovery processes, as well as the remediation actions and recovery processes themselves, are also out of scope.
The interaction models described in this document are meant to serve as a solid foundation and reference for other solution documents within or outside the IETF. Solution documents of any kind can refer to these interaction models to prevent duplicating text and to avoid the risk of subtle discrepancies. Similarly, deviations from the generic model described in this document can be illustrated in solution documents to highlight distinct contributions.
In order to ensure appropriate conveyance of Evidence, the following requirements MUST be fulfilled:
Integrity: : Information provided by an Attester MUST NOT have been altered since it was created. This may be achieved through a digital signature over Attestation Evidencewhich may be symmetric, like an HMAC, or asymmetric, like ECDSA.
Authentication: : The information provided by the Attester MUST be authentic. To do this, the Attester should authenticate itself to the Verifier. This can be done through implicit authentication using a digital signature over the Attestation Evidence, which does not require additional protocol steps, or by using a confidential channel with encryption.
In order to ensure Evidence is appropriately conveyed through the interaction models described in this document, the following prerequisites MUST be in place to support their implementation:
Authentication Secret:
: An Authentication Secret MUST be exclusively available to an Attesting Environment of the Attester.
: The Attester MUST protect Claims with this Authentication Secret to prove the authenticity of the Claims included in Evidence. The Authentication Secret MUST be established before RATS take place.
Attester Identity:
: A statement made by an Endorser about an Attester that affirms the Attester's distinguishability. (Note that distinguishability does not imply uniqueness.)
: The provenance of Evidence for a distinguishable Attesting Environment MUST be unambiguous.
: An Attester Identity MAY be an Authentication Secret which is available exclusively to one of the Attesting Environments of the Attester. It could be a unique identity, it could be included in a zero-knowledge proof (ZKP), it could be part of a group signature, or it could be a randomized DAA credential {{DAA}}.
Attestation Evidence Authenticity:
: Attestation Evidence MUST be authentic.
: In order to provide a proof of authenticity, Attestation Evidence can be cryptographically associated with an identity document (e.g., a PKIX certificate or trusted key material, or a randomized DAA credential {{DAA}}), or could include a correct, unambiguous, and stable reference to an accessible identity document.
Evidence Freshness:
: Evidence MUST include an indicator about its freshness that can be understood by a Verifier (See also {{Section 10 of -RATS}}). This enables interaction models to support the conveyance of proofs of freshness in a way that is useful to Verifiers and their appraisal procedures.
Evidence Protection:
: Evidence MUST be a set of well-formatted and well-protected Claims that an Attester can create and convey to a Verifier in a tamper-evident manner.
This section describes the essential information elements for the interaction models described in {{interaction-models}}. These generic information elements may be Conceptual Messages included in the protocol messages or may be added as protocol parameters, depending on the specific solution.
Attesting Environment IDs ('attEnvIDs'):
: optional
: A statement representing one or more identifiers that MUST be associated with a corresponding Attestation Environment's keys used to protect Claims in Evidence produced by an Attester. Exemplary identifiers include attestation key material (e.g., the public key associated with the private key that is used to produce Evidence), key identifiers, environment names, or individual hardware-based immutable identifiers.
: While a Verifier does not necessarily have knowledge about an Attesting Environment's identifiers, each distinguishable Attesting Environment typically has access to a protected capability that includes an Attesting Environment ID. If no Attesting Environment IDs are provided, a local default applies based on the Attester. For example, all Attesting Environments will provide Evidence.
Handle (handle):
: mandatory
: An information element provided to the Attester from an external source included in Evidence (or other RATS Conceptual Messages) to determine recentness, freshness, or to protect against replay attacks.
: The term Handle encompasses various data types that can be utilized to determine recentness, freshness, or provide replay protection. Examples include Nonces, which protect against replay attacks, and Epoch Markers that identify distinct periods (Epochs) of freshness {{-epoch-markers}}. Handles can also indicate authenticity or attestation Evidence provenance, as only specific RATS roles (e.g., an Attester and a Verifier in a challenge-response interaction) are meant to know a certain handle.
Claims (claims):
: mandatory
: Claims are assertions that represent characteristics of an Attester's Target Environment.
: Claims are part of a Conceptual Message and are used, for example, to appraise the integrity of Attesters by Verifiers. The other information elements in this section can be presented as Claims in any type of Conceptual Message.
Event Logs (eventLogs):
: optional
: Event Logs accompany Claims by providing event trails of security-critical events in a system. The primary purpose of Event Logs is to ensure Claim reproducibility by providing information on how Claims originated.
Verifier Inputs ('verInputs')
: mandatory
: Appraisal procedures implemented by Verifiers require certain inputs as defined in {{-RATS}}: Reference Values, Endorsements, and Appraisal Policy for Evidence. These Conceptual Messages can take various forms. For example, Reference Values that can be expressed via Reference Integrity Measurements (RIM) or Endorsements that can range from trust anchors to assertions cryptographically bound to the public key associated with an Attesting Environment.
Claim Selection (claimSelection):
: optional
: A (sub-)set of Claims that can be collected and incorporated in Evidence by the Attesting Environments of an Attester.
: Claim Selections act as optional filters to specify the exact set of Claims to be included in Evidence. For example, a Verifier could send a Claim Selection, among other elements, to an Attester. An Attester MAY decide whether or not to provide all requested Claims from a Claim Selection to the Verifier. If there is no way to convey a Claim Selection in a remote attestation protocol, a default Claim Selection (e.g., "all") MUST be defined by the Attester and SHOULD be known by the Verifier. In order to select Claims, Claims that can be potentially included in Evidence by an Attesting Environment have to be known by the Verifier.
Collected Claims (collectedClaims):
: mandatory
: Collected Claims represent a (sub-)set of Claims created by an Attester.
: Collected Claims are gathered based on the Claim Selection. If a Verifier does not provide a Claim Selection, all available Claims on the Attester are part of the Collected Claims.
Evidence (evidence):
: mandatory
: A set of Claims that can include: (a) a list of Attestation Environment IDs, each identifying an Authentication Secret in a single Attesting Environment, (b) the Attester Identity, (c) Claims about the Attester's Target Environment, and (d) a Handle. Attestation Evidence MUST cryptographically bind all of these information elements. Evidence MUST be protected via an Authentication Secret. The Authentication Secret MUST be trusted by the Verifier as authoritatively "speaking for" {{lampson06}} the Attester.
Attestation Result (attestationResult):
: mandatory
: An Attestation Result is produced by the Verifier as the output of the appraisal of Evidence generated by an Attester. Attestation Results include concise assertions about integrity or other characteristics of the appraised Attester that can be processed by Relying Parties.
This document describes three interaction models for Remote Attestation:
- Challenge/Response ({{challenge-response}}),
- Unidirectional ({{unidirectional}}), and
- Streaming ({{streaming}}).
Each section starts with a sequence diagram illustrating the interactions between the involved roles: Attester, Verifier and, optionally, a Relying Party. The presented interaction models focus on the conveyance of Evidence and Attestation Results. The same interaction models may apply to the conveyance of other Conceptual Messages (Endorsements, Reference Values, or Appraisal Policies) with other roles involved. However, that is out of scope for the present document.
All interaction models have a strong focus on the use of a Handle to incorporate a proof of freshness and to prevent replay attacks. The way the Handle is processed is the most prominent difference between the three interaction models.
.----------. .----------.
| Attester | | Verifier |
'----+-----' '-----+----'
| |
=================[Evidence Generation and Conveyance]===================
| |
generateClaims(attestingEnvironment) |
| => claims, ?eventLogs |
| |
|<----- requestEvidence(handle, ?attEnvIDs, ?claimSelection) |
| |
collectClaims(claims, ?claimSelection) |
| => collectedClaims |
| |
generateEvidence(handle, ?attEnvIDs, collectedClaims) |
| => evidence |
| |
| {evidence, ?eventLogs}------------------------------------>|
| |
==========================[Evidence Appraisal]==========================
| |
| appraiseEvidence(evidence, ?eventLogs, verInputs)
| attestationResult <= |
| |
The Attester boots up and thereby produces Claims about its boot state and its operational state. Event Logs may accompany the produced Claims and provide an event trail of security-critical events in the system. Claims are produced by all Attesting Environments of an Attester system.
The Challenge/Response remote attestation procedure is initiated by the Verifier by sending a remote attestation request to the Attester. A request includes a Handle, an optional list of Attestation Key IDs, and an optional Claim Selection.
In the Challenge/Response model, the Handle is composed of qualifying data in the form of a practically infeasible to guess nonce, such as a cryptographically strong random number. The Verifier-generated nonce is intended to guarantee Evidence freshness and to prevent replay attacks.
The list of Attestation Key IDs selects the attestation keys with which the Attester is requested to sign the attestation Evidence. Each selected key is uniquely associated with an Attesting Environment of the Attester. As a result, a single Attestation Key ID identifies a single Attesting Environment. Correspondingly, a particular set of Evidence originating from a particular Attesting Environment in a composite device can be requested via multiple Attestation Key IDs. Methods to acquire Attestation Key IDs or mappings between Attesting Environments to Attestation Key IDs are out of scope of this document.
The Attester selects Claims based on the specified Claim Selection, which is defined by the Verifier. The Claim Selection determines the Collected Claims, which may be a subset of all the available Claims. If the Claim Selection is omitted, then all available Claims on the Attester MUST be used to create corresponding Evidence. For example, when performing a boot integrity evaluation, a Verifier may only request specific claims about the Attester, such as Evidence about the BIOS/UEFI and firmware that the Attester booted up, without including information about all currently running software.
With the Handle, the Attestation Key IDs, and the Collected Claims, the Attester produces signed Evidence. That is, it digitally signs the Handle and the Collected Claims with a cryptographic secret identified by the Attestation Key ID. This is done once per Attesting Environment which is identified by the particular Attestation Key ID. The Attester communicates the signed Evidence as well as all accompanying Event Logs back to the Verifier.
The Claims, the Handle, and the Attester Identity information (i.e., the Authentication Secret) MUST be cryptographically bound to the signature of Evidence. These MAY be presented obfuscated, encrypted, or cryptographically blinded. For further reference see section {{security-and-privacy-considerations}}.
Upon receiving the Evidence and Event Logs, the Verifier validates the signature, Attester Identity, and Handle, and then appraises the Claims. Claim appraisal is driven by Policy and takes Reference Values and Endorsements as input. The Verifier outputs Attestation Results. Attestation Results create new Claim Sets about the properties and characteristics of an Attester, which enable Relying Parties to assess an Attester's trustworthiness.
According to the RATS Architecture, two reference models for Challenge/Response Attestation have been proposed. This section highlights the information flows between the Attester, Verifier, and Relying Party undergoing Remote Attestation Procedure, using these models.
The passport model is so named because of its resemblance to how nations issue passports to their citizens. In this model, the attestation sequence is a two-step procedure. In the first step, an Attester conveys Evidence to a Verifier, which appraises the Evidence according to its Appraisal Policy. The Verifier then gives back an Attestation Result to the Attester, which caches it. In the second step, the Attester presents the Attestation Result (and possibly additional Claims/Evidence) to a Relying Party, which appraises this information according to its own Appraisal Policy to establish the trustworthiness of the Attester.
.----------. .----------. .---------------.
| Attester | | Verifier | | Relying Party |
'----+-----' '-----+----' '-------+-------'
| | |
=================[Evidence Generation and Conveyance]===================
| | |
generateClaims(attestingEnvironment) | |
| => claims, ?eventLogs | |
| | |
|<------------------------ requestEvidence(handle, |
| ?attEnvIDs, ?claimSelection) |
| | |
collectClaims(claims, ?claimSelection) | |
| => collectedClaims | |
| | |
generateEvidence(handle, | |
?attEnvIDs, collectedClaims) | |
| => evidence | |
| | |
| {evidence, ?eventLogs} ------------->| |
| | |
==========================[Evidence Appraisal]==========================
| | |
| appraiseEvidence(evidence, |
| ?eventLogs, verInputs) |
| attestationResult <= | |
| | |
|<---------------- {attestationResult} | |
| | |
| {evidence, attestationResult} ------------------------>|
| | |
====================[Attestation Result Generation]=====================
| | |
| | appraiseResult(policy,
| | attestationResult)
| | |
The background-check model is so named because of the resemblance of how employers and volunteer organizations perform background checks. In this model, the attestation sequence is initiated by a Relying Party. The Attester conveys Evidence to the Relying Party, which does not process its payload, but relays the message and optionally checks its signature against a policed trust anchor store. Upon receiving the Evidence, the Relying Party initiates a session with the Verifier. Once the session is established, it forwards the received Evidence to the Verifier. The Verifier appraises the received Evidence according to its appraisal policy for Evidence and returns a corresponding Attestation Result to the Relying Party. The Relying Party then checks the Attestation Result against its own appraisal policy to conclude attestation.
.----------. .---------------. .----------.
| Attester | | Relying Party | | Verifier |
'----+-----' '-------+-------' '-----+----'
| | |
=================[Evidence Generation and Conveyance]===================
| | |
|<------------------------ requestEvidence(handle, |
| ?attEnvIDs, ?claimSelection) |
| | |
generateClaims(attestingEnvironment) | |
| => {claims, ?eventLogs} | |
| | |
collectClaims(claims, | |
?claimSelection) | |
| => collectedClaims | |
| | |
generateEvidence(handle, | |
?attEnvIDs, collectedClaims) | |
| => evidence | |
| | |
| {evidence, ?eventLogs} ---------->| |
| | |
==========================[Evidence Appraisal]==========================
| | |
| | {handle, evidence, |
| | ?eventLogs} -------->|
| | |
| | appraiseEvidence(evidence,
| | ?eventLogs, verInputs)
| | attestationResult <= |
| | |
| |<---------- {evidence, |
| | attestationResult} |
| | |
====================[Attestation Result Generation]=====================
| | |
| appraiseResult(policy, |
| attestationResult) |
| | |
.----------. .--------------------. .----------.
| Attester | | Handle Distributor | | Verifier |
'----+-----' '---------+----------' '-----+----'
| | |
.--------[loop]------------------------------------------------------.
| | | |
| ========================[Handle Generation]========================= |
| | generateHandle() | |
| | | => handle | |
| | | | |
| |<---------------------------- {handle} | {handle} --------->| |
| | | | |
| | x | |
| | | |
| ===============[Evidence Generation and Conveyance]================= |
| | | |
| generateClaims(attestingEnvironment) | |
| | => claims, eventLogs | |
| | | |
| collectClaims(claims, ?claimSelection) | |
| | => collectedClaims | |
| | | |
| generateEvidence(handle, attEnvIDs, collectedClaims) | |
| | => evidence | |
| | | |
| | {evidence, eventLogs} ------------------------------------>| |
| | | |
| ========================[Evidence Appraisal]======================== |
| | | |
| | appraiseEvidence(evidence, |
| | eventLogs, |
| | verInputs) |
| | attestationResult <= | |
| ~ ~ |
| | | |
| .-------[loop]-----------------------------------------------------. |
|| | | ||
|| ============[Delta Evidence Generation and Conveyance]============ ||
|| | | ||
|| generateClaims(attestingEnvironment) | ||
|| | => claimsDelta, eventLogsDelta | ||
|| | | ||
|| collectClaims(claimsDelta, ?claimSelection) | ||
|| | => collectedClaimsDelta | ||
|| | | ||
|| generateEvidence(handle, attEnvIDs, collectedClaimsDelta) | ||
|| | => evidence | ||
|| | | ||
|| | {evidence, eventLogsDelta} ------------------------------->| ||
|| | | ||
|| ====================[Delta Evidence Appraisal]==================== ||
|| | | ||
|| | appraiseEvidence(evidence, ||
|| | eventLogsDelta, ||
|| | verInputs) ||
|| | attestationResult <= | ||
|| | | ||
| '------------------------------------------------------------------' |
'--------------------------------------------------------------------'
| |
Uni-Directional Remote Attestation procedures can be initiated both by the Attester and by the Verifier. Initiation by the Attester can result in unsolicited pushes of Evidence to the Verifier. Initiation by the Verifier always results in solicited pushes to the Verifier.
The Uni-Directional model uses the same information elements as the Challenge/Response model. In the sequence diagram above, the Attester initiates the conveyance of Evidence (comparable with a RESTful POST operation). While a request of Evidence from the Verifier would result in a sequence diagram more similar to the Challenge/Response model (comparable with a RESTful GET operation). The specific manner how Handles are generated is not in scope of this document. One example of a specific handle representation is {{-epoch-markers}}.
In the Uni-Directional model, handles are composed of cryptographically signed trusted timestamps as shown in {{-TUDA}}, potentially including other qualifying data. The Handles are created by an external trusted third party (TTP) -- the Handle Distributor -- which includes a trustworthy source of time, and takes on the role of a Time Stamping Authority (TSA, as initially defined in {{-TSA}}). Timestamps created from local clocks (absolute clocks using a global timescale, as well as relative clocks, such as tick-counters) of Attesters and Verifiers MUST be cryptographically bound to fresh Handles received from the Handle Distributor. This binding provides a proof of synchronization that MUST be included in all produced Evidence. This model provides proof that Evidence generation happened after the Handle generation phase. The Verifier can always determine whether the received Evidence includes a fresh Handle, i.e., one corresponding to the current Epoch.
The lifecycle of a handle is a critical aspect of ensuring the freshness and validity of attestation Evidence. When a new handle is generated by the Handle Distributor, it effectively supersedes the previous handle. However, due to network latencies and propagation delays, there may be a period during which both the old and new handles are in circulation. This "grey zone" can potentially lead to situations where Evidence may be associated with an outdated handle yet still appear to be valid.
To manage this complexity, it is essential to define a clear policy for handle validity and expiration:
-
Handle Expiry: Each handle should have a well-defined expiration time, after which it is considered invalid. This expiry must account for expected propagation delays and be clearly communicated to all entities in the attestation process.
-
Synchronization Checks: Implement periodic synchronization checks between the Handle Distributor and both Attesters and Verifiers to ensure that handles are updated consistently across all participants.
-
Grace Periods: Define grace periods during which a newly issued handle starts being accepted, and the old handle stops being valid. This period should be long enough to account for the maximum expected propagation delay across the network.
Implementing these measures will help mitigate the risks associated with the handle lifecycle, particularly in environments where propagation delays are significant. This careful management ensures that the integrity and trustworthiness of the attestation process are maintained.
While periodically pushing Evidence to the Verifier, the Attester only needs to generate and convey evidence generated from Claim values that have changed and new Event Log entries since the previous conveyance. These updates reflecting the differences are called "delta" in the sequence diagram above.
Effectively, the Uni-Directional model allows for a series of Evidence to be pushed to multiple Verifiers simultaneously. Methods to detect excessive time drift that would mandate a fresh Handle to be received by the Handle Distributor as well as timing of Handle distribution are out-of-scope of this document.
Streaming Remote Attestation serves as the foundational concept for both the observer pattern {{ISIS}} and the publish-subscribe pattern {{DesignPatterns}}. It entails establishing subscription states to enable continuous remote attestation. In the observer pattern, observers are directly connected to target resources without a Broker, while the publish-subscribe pattern involves a central Broker for message distribution.
.----------. .----------.
| Attester | | Verifier |
'----+-----' '-----+----'
| |
.--------[loop]------------------------------------------------------.
| | | |
| ========================[Handle Generation]========================= |
| | | |
| | generateHandle() |
| | handle<= | |
| | | |
| |<------------ subscribe(handle, attEnvIDs, ?claimSelection) | |
| | {handle} ------------------------------------------------->| |
| | | |
| ===============[Evidence Generation and Conveyance]================= |
| | | |
| generateClaims(attestingEnvironment) | |
| | => claims, eventLogs | |
| | | |
| collectClaims(claims, ?claimSelection) | |
| | => collectedClaims | |
| | | |
| generateEvidence(handle, attEnvIDs, collectedClaims) | |
| | => evidence | |
| | | |
| | {handle, evidence, eventLogs} ---------------------------->| |
| | | |
| ========================[Evidence Appraisal]======================== |
| | | |
| | appraiseEvidence(evidence, |
| | eventLogs, |
| | verInputs) |
| | attestationResult <= | |
| ~ ~ |
| | | |
| .--------[loop]----------------------------------------------------. |
|| | | ||
|| ============[Delta Evidence Generation and Conveyance]============ ||
|| | | ||
|| generateClaims(attestingEnvironment) | ||
|| | => claimsDelta, eventLogsDelta | ||
|| | | ||
|| collectClaims(claimsDelta, ?claimSelection) | ||
|| | => collectedClaimsDelta | ||
|| | | ||
|| generateEvidence(handle, attEnvIDs, collectedClaimsDelta) | ||
|| | => evidence | ||
|| | | ||
|| | {evidence, eventLogsDelta} ------------------------------->| ||
|| | | ||
|| ====================[Delta Evidence Appraisal]==================== ||
|| | | ||
|| | appraiseEvidence(evidence, ||
|| | eventLogsDelta, ||
|| | verInputs) ||
|| | attestationResult <= | ||
|| | | ||
| '------------------------------------------------------------------' |
'--------------------------------------------------------------------'
| |
In the observer pattern, an observer establishes a direct connection to the observed resources through a subscription mechanism, which is designed specifically for conveying conceptual messages for remote attestation purposes. This mechanism not only facilitates the initial subscription request but also actively maintains the state of the subscription, ensuring that any changes in the observed resources are consistently communicated to the observer. It handles the complexities of managing these connections, including the maintenance of pertinent information about the observer's preferences and security requirements, ensuring that the transmission of attestation data remains both secure and relevant to the observer’s specific context.
Setting up subscription state between a Verifier and an Attester is conducted via a subscribe operation. The subscribe operation is used to convey Handles required for Evidence generation. Effectively, this allows for a series of Evidence to be pushed to a Verifier, similar to the Uni-Directional model. In the observer pattern, the Handle Distributor role is optional. While the model is typically used for direct, bi-lateral subscription relationships where the Verifier generates and provides Handles directly, it is also possible to include the trusted third party that is a Handle Distributor. A Handle Distributor independently manages the generation and distribution of Handles to other RATS roles. As a result, scenarios involving more than bi-lateral interactions are enabled. However, in its basic form, the model assumes direct interaction between an Attester and a Verifier, where the Handle generation is a responsibility taken on by the Verifier roles.
Handles provided by a specific subscribing Verifier MUST be used in Evidence generation for that specific Verifier. The streaming model without a Broker uses the same information elements as the Challenge/Response and the Uni-Directional model. Methods to detect excessive time drift that would render Handles stale and mandate a fresh Handles to be conveyed via another subscribe operation are out-of-scope of this document.
This model includes a Broker to facilitate the distribution of messages between RATS roles, such as Attesters and Verifiers. The Broker is a trusted third party and acts as an intermediary that ensures messages are securely and reliably conveyed between involved RATS roles. The publish-subscribe messaging pattern is widely used for communication in different areas. Unlike the Streaming Remote Attestation without a Broker interaction model, Attesters are not required to be aware of corresponding Verifiers. In scenarios with large numbers of Attesters and Verifiers, the publish-subscribe pattern may reduce interdependencies and improve scalability.
With publish-subscribe, clients typically connect to (or register with) a publish-subscribe server (PubSub server or Broker). Clients may publish data in the form of a message under a certain topic. Subscribers to that topic get notified whenever a message arrives under a topic, and the appropriate message is forwarded to them. Depending on particular publish-subscribe models and implementations, involved roles can be publishers, subscribers or both.
The Broker and Handle Distributor are considered to be trusted third parties (TTPs) for all other participating roles, including Attesters and Verifiers (see also {{security-and-privacy-considerations}}). All roles must establish a trust relationship with the Broker and Handle Distributor, as those are responsible for the secure and reliable dissemination of critical protocol information, such as Handles and Attestation Results.
Ensuring the security of these trusted third parties is vital, as any compromise could undermine the entire remote attestation procedure. Therefore, the deployment of Brokers and Handle Distributors requires stringent security measures to protect against unauthorized access and to ensure that they operate as trustworthy facilitators within the remote attestation framework.
In the following sections, the interaction models Challenge/Response Remote Attestation over Publish-Subscribe and Uni-Directional Remote Attestation over Publish-Subscribe are described. There are different phases that both models go through:
- Handle Generation
- Evidence Generation and Conveyance
- Evidence Appraisal
- Attestation Result Generation
The models only differ in the handle generation phase. From a remote attestations procedure's point of view Evidence Generation, Conveyance, and Appraisal, as well as Attestation Result Generation are identical in both models.
.----------. .---------------. .----------.
| Attester | | PubSub Server | | Verifier |
'----+-----' '-------+-------' '-----+----'
| | |
==========================[Handle Generation]===========================
| | |
| | generateHandle()
| | handle <= |
sub(topic=AttReq) ------------------->| |
| |<--------- pub(topic=AttReq,
| | handle)
|<---------------------- notify(topic=AttReq, handle) |
| | |
~ ~ ~
The Challenge/Response Remote Attestation over Publish-Subscribe interaction model uses the same information elements as the Challenge/Response Remote Attestation interaction model. Handles are generated by the Verifier on a per-request basis. In the sequence diagram above, the Verifier initiates an attestation by generating a new handle and publishing it to the "AttReq" (= Attestation Request) topic on the PubSub server. The PubSub server then forwards this handle to the Attester by notifying it. This mechanism ensures that each handle is uniquely associated with a specific attestation request, thereby enhancing security by preventing replay attacks.
.----------. .-------------. .---------------. .----------.
| Attester | | Handle | | PubSub Server | | Verifier |
'----+-----' | Distributor | '-------+-------' '-----+----'
| '------+------' | |
| | | |
==========================[Handle Generation]===========================
| | | |
| | | |
sub(topic=Handle) ------------------->| |
| | | |
| | |<--------- sub(topic=Handle)
| | | |
| | | |
| generateHandle() | |
| | => handle | |
| | | |
| pub(topic=Handle, | |
| | handle) --------->| |
| x | |
| | |
|<--------------------- notify(topic=Handle, handle) |
| | |
| notify(topic=Handle, handle) ------->|
| | |
~ ~ ~
Handles are created by a trusted third party, the Handle Distributor (see {{security-and-privacy-considerations}}). In the sequence diagram above, both an Attester and a Verifier subscribe to the topic "Handle" on the PubSub server. When the Handle Distributor generates and publishes a Handle to the "Handle" topic on the PubSub server, the PubSub server notifies the subscribers, Attester and Verifier, and forwards ("notify") the Handle to them during Handle Generation.
~ ~ ~
| | |
.----+-----. .-------+-------. .-----+----.
| Attester | | PubSub Server | | Verifier |
'----+-----' '-------+-------' '-----+----'
| | |
| |<---------- sub(topic=AttEv)
| | |
.--------[loop]------------------------------------------------------.
| | | | |
| ===============[Evidence Generation and Conveyance]================= |
| | | | |
| generateClaims(attestingEnvironment) | | |
| | => claims, eventLogs | | |
| | | | |
| collectClaims(claims, ?claimSelection) | | |
| | => collectedClaims | | |
| | | | |
| generateEvidence(handle, attEnvIDs, | | |
| | collectedClaims) | | |
| | => evidence | | |
| | | | |
| pub(topic=AttEv, | | |
| | evidence, eventLogs) ------------>| | |
| | | | |
| | notify(topic=AttEv, | |
| | | evidence, | |
| | | eventLogs) ---------->| |
| | | | |
| ========================[Evidence Appraisal]======================== |
| | | | |
| | | appraiseEvidence( |
| | | evidence, |
| | | eventLogs, |
| | | verInputs) |
| | | attestationResult <= | |
| | | | |
'--------------------------------------------------------------------'
| | |
~ ~ ~
Exactly as in the Challenge/Response and Uni-Directional interaction models, there is an Evidence Generation-Appraisal loop, in which the Attester generates Evidence and the Verifier appraises it. In the Publish-Subscribe model above, the Attester publishes Evidence to the topic "AttEv" (= Attestation Evidence) on the PubSub server, to which a Verifier subscribed before. The PubSub server notifies Verifiers, accordingly, by forwarding the attestation Evidence. Although the above diagram depicts only full attestation Evidence and Event Logs, later attestations may use "deltas' for Evidence and Event Logs. Verifiers appraise the Evidence and publish the Attestation Result to topic "AttRes" (= Attestation Result) on the PubSub server.
~ ~ ~ ~
| | | |
.----+-----. .--+------------. .-------+-------. .-----+----.
| Attester | | Relying Party | | PubSub Server | | Verifier |
'----+-----' '--+------------' '-------+-------' '-----+----'
| | | |
====================[Attestation Result Generation]=====================
| | | |
| sub(topic=AttRes, | |
| handle) ----------------->| |
| | | |
.--------[loop]------------------------------------------------------.
| | | | | |
| | | |<--------- pub(topic=AttRes, |
| | | | attestationResult) |
| | | | | |
| | |<----------------- notify(topic=AttRes, | |
| | | | attestationResult) | |
| | | | | |
'--------------------------------------------------------------------'
| | | |
~ ~ ~ ~
Attestation Result Generation is the same for both publish-subscribe models,Challenge/Response Remote Attestation over Publish-Subscribe and Uni-Directional Remote Attestation over Publish-Subscribe.
Relying Parties subscribe to topic AttRes (= Attestation Result) on the PubSub server.
The PubSub server forwards Attestation Results to the Relying Parties as soon as they are published to topic AttRes.
Note to RFC Editor: Please remove this section as well as references to {{BCP205}} before AUTH48.
This section records the status of known implementations of the protocol defined by this specification at the time of posting of this Internet-Draft, and is based on a proposal described in {{BCP205}}. The description of implementations in this section is intended to assist the IETF in its decision processes in progressing drafts to RFCs. Please note that the listing of any individual implementation here does not imply endorsement by the IETF. Furthermore, no effort has been spent to verify the information presented here that was supplied by IETF contributors. This is not intended as, and must not be construed to be, a catalog of available implementations or their features. Readers are advised to note that other implementations may exist.
According to {{BCP205}}, "this will allow reviewers and working groups to assign due consideration to documents that have the benefit of running code, which may serve as evidence of valuable experimentation and feedback that have made the implemented protocols more mature. It is up to the individual working groups to use this information as they see fit".
The open-source implementation was initiated and is maintained by the Fraunhofer Institute for Secure Information Technology SIT.
The open-source implementation is named "CHAllenge-Response based Remote Attestation" or in short: CHARRA.
The open-source implementation project resource can be located via: https://github.com/fraunhofer-sit/charra
The code's level of maturity is considered to be "prototype".
The current version ('6194b3b') implements a challenge/response interaction model and is aligned with the exemplary specification of the CoAP FETCH bodies defined in Section {{coap-fetch-bodies}} of this document.
The CHARRA project and all corresponding code and data maintained on GitHub are provided under the BSD 3-Clause "New" or "Revised" license.
The implementation requires the use of the Trusted Computing Group (TCG) Trusted Software Stack (TSS), and an HSM interoperable with the Trusted Platform Module Library specifications, e.g., a Trusted Platform Module (TPM) 2.0 or equivalent implementation. The corresponding project resources (code and data) for Linux-based operating systems are maintained on GitHub at https://github.com/tpm2-software/tpm2-tss/.
The implementation uses the Constrained Application Protocol {{-COAP}} (http://coap.technology/) and the Concise Binary Object Representation {{-CBOR}} (https://cbor.io/).
Michael Eckel (michael.eckel@sit.fraunhofer.de)
{: #security-and-privacy-considerations}
This document outlines three interaction models for remote attestation procedures (RATS) {{-RATS}}. While the subsequent sections address additional security and privacy considerations, the security considerations from {{Section 12 of -RATS}} must also be adhered to. Additionally, for TPM-based remote attestation, the security considerations outlined in {{Section 5 of -RIV}} should be taken into account.
In a remote attestation procedure, both the Verifier and the Attester may choose to cryptographically blind certain attributes to enhance privacy. For example, specific information can be included in the signature after being processed through a one-way function, such as a hash function.
Additionally, there is an option to scramble the Nonce or Attester Identity using other information that is known to both parties. A common example is the IP address of the Attester, which is typically accessible to both the Attester and the Verifier. This additional information can be utilized to scramble the Nonce, thereby mitigating certain types of relay attacks.
The handle distributor, as a third party in remote attestation scenarios, holds a critical position similar to that of a Trusted Third Party (TTP). Given its role in generating handles, it has the potential to influence the attestation process significantly. The integrity and reliability of the handles it produces are pivotal for ensuring that the attestation evidence remains trustworthy and that the attestation process is not susceptible to manipulation or interference.
-
Integrity and Authenticity: It is assumed that the handle distributor operates with high levels of integrity and authenticity. Handles generated by the distributor must be secure, unique, and verifiable. This ensures that they cannot be forged or reused maliciously.
-
Isolation and Protection: The handle distributor must be isolated and protected from unauthorized access and attacks. This includes physical security measures for hardware that might house the handle distributor's operations, as well as cybersecurity measures to protect against online threats.
-
Auditability: The operations of the handle distributor should be auditable. This allows for the verification of its compliance with security policies and the integrity of its operations. Regular audits help to ensure that the handle distributor is functioning as expected and has not been compromised.
-
Transparency: While maintaining security and confidentiality, the processes by which handles are generated and distributed should be as transparent as possible to authorized parties. This helps in building trust and verifying that handles are distributed in a fair and unbiased manner.
To mitigate risks associated with the handle distributor being a central point of potential failure or attack, several measures should be implemented:
-
Redundancy: Deploying multiple, geographically dispersed handle distributors can ensure continuity of service even if one distributor is compromised or fails.
-
Cryptographic Security: Using strong cryptographic techniques to protect the generation and transmission of handles ensures that they cannot be tampered with during distribution.
-
Certification and Compliance: The handle distributor should comply with relevant security standards and undergo regular security certifications. This ensures that they meet industry-wide security benchmarks and maintain high levels of trust.
By defining and adhering to these security assumptions, the role of the handle distributor in remote attestation procedures can be securely managed, minimizing risks and enhancing the overall trust in the attestation process.
The role of the Broker in the "Streaming Remote Attestation with a Broker model" introduces potential security vulnerabilities, including the ability to perform cross-application attacks by manipulating handles and topics. To mitigate these risks, it is essential to implement robust security measures:
-
End-to-End Authentication: Establishing end-to-end authenticated channels between Attesters and Verifiers ensures that data integrity and authenticity are preserved across the communication process. This measure prevents the Broker from altering the content of the messages, including Handles and other sensitive data.
-
Strong Isolation of Topics: Implementing strong isolation mechanisms for topics can help prevent the Broker from inadvertently or maliciously routing notifications to unauthorized parties. This includes using secure naming conventions and access controls that restrict the Broker's ability to manipulate topic subscriptions.
-
Trusted Association Verification: To further safeguard against confusion attacks where the Broker might misroute notifications, mechanisms should be in place to verify the trust association between senders and receivers continuously. This can be facilitated by cryptographic assurances, such as digital signatures and trusted certificates that validate the sender's identity and the integrity of the message content.
-
Audit and Monitoring: Regular audits and real-time monitoring of Broker activities can detect and respond to anomalous behavior that might indicate security breaches or manipulation attempts. Logging all actions performed by the Broker provides an audit trail that can be critical for forensic analysis.
-
Broker as a Trusted Third Party (TTP): Recognizing the Broker as a TTP necessitates stringent security certifications and compliance with security standards to ensure that they operate under strict governance and security protocols. This includes regular security assessments and certifications that validate the Broker's security practices.
By addressing these vulnerabilities proactively, the integrity and confidentiality of the attestation process can be maintained, reducing the risks associated with Broker-mediated communication in remote attestation scenarios. It is crucial for solution architects to incorporate these security measures during the design and deployment phases to ensure that the attestation process remains secure and trustworthy.
The security and privacy requirements for remote attestation can vary significantly based on the deployment environment, the nature of the attestation mechanisms used, and the specific threats each scenario faces. This section details additional security considerations that are pertinent to the interaction models discussed in this document.
The need for confidentiality in the transmission of attestation information is critical, particularly when exchanges occur over public or untrusted networks, such as the public Internet. For instance, in the Streaming Remote Attestation with a Broker model (cf.{{streaming-with-broker}}), where data might traverse multiple nodes, employing TLS can provide necessary confidentiality protections. Similarly, for scenarios involving sensitive environments like carrier management networks, evaluating the confidentiality of the transport medium is crucial to ensure that attestation data remains secure against interception or eavesdropping.
Mutual authentication is particularly relevant in models such as the Challenge/Response Remote Attestation (cf. {{challenge-response}}) where both the Attester and the Verifier engage in bidirectional exchanges of sensitive information. Ensuring that both parties can authenticate each other prevents impersonation attacks, enhancing the trustworthiness of the attestation results.
In environments where the integrity and security of attestation evidence are paramount, hardware-based security features play a critical role. Technologies like Hardware Security Modules (HSMs), Physically Unclonable Functions (PUFs), and Trusted Execution Environments (TEEs) provide robust protections against tampering and unauthorized access. These are especially important in high-security settings such as financial services or military applications, where attestation processes must rely on physically secure and tamper-resistant components to meet stringent regulatory and security standards.
By addressing these application-specific security requirements within the context of defined interaction models, security strategies can be tailored to fit the unique challenges and operational contexts of different attestation scenarios.
Olaf Bergmann, Michael Richardson, and Ned Smith
--- back
{: #coap-fetch-bodies}
The following CDDL specification is an exemplary proof-of-concept to illustrate a potential implementation of the Challenge/Response Interaction Model. The communication protocol used is CoAP. Both the request message and the response message are exchanged via the FETCH operation and corresponding FETCH request and FETCH response body.
In this example, Evidence is created via the root-of-trust for reporting primitive operation "quote" that is provided by a TPM 2.0.
charra-bodies = charra-attestation-request / charra-attestation-response
charra-attestation-request = [
hello: bool, ; if true, the TPM 2.0 AK Cert shall be conveyed
key-id: bytes, ; the key ID to use for signing
nonce: bytes, ; a (random) nonce, providing freshness and/or recentness
pcr-selections: [ * pcr-selection ]
]
pcr-selection = [
tcg-hash-alg-id: uint .size 2, ; TPM2_ALG_ID
pcrs: [
pcr: uint .size 2
]
]
charra-attestation-response = [
attestation-data: bytes, ; TPMS_ATTEST.quoted
tpm2-signature: bytes,
? ak-cert: bytes, ; TPM2 attestation key certificate (AK Cert)
]