2.1.Network Endpoint Assessment

Network operators want trustworthy reports that include identityand version information about the hardware and software on themachines attached to their network. Examples of reports include purposes (such as inventory summaries), auditresults, and anomaly notifications (which typically include the maintenance of logrecords or trend reports). The network operator may also want a policy by which full access is onlygranted to devices that meet some definition of hygiene, and so wants to getClaims about such information and verify its validity. Remote attestation isdesired to prevent vulnerable or compromised devices from getting access tothe network and potentially harming others.¶

Typically, a solution starts with a specific component (sometimes referred to as a "root of trust") that oftenprovides a trustworthy device identity and performs a series of operations that enables trustworthiness appraisals for other components.Such components perform operations that help determine the trustworthiness of yet other componentsby collecting, protecting, or signing measurements.Measurements that have been signed by such components are comprised of Evidence that either supports or refutes a claim of trustworthiness when evaluated.Measurements can describe a variety of attributes of system components, such as hardware, firmware, BIOS, software, etc., and how they are hardened.¶

Attester:

A device desiring access to a network.¶

Relying Party:

Network equipment (such as a router, switch, or access point) that isresponsible for admission of the device into the network.¶

2.2.Confidential Machine Learning Model Protection

A device manufacturer wants to protect its intellectual property.The intellectual property's scope primarily encompasses the machine learning (ML) model that is deployed in the devices purchased by its customers.The protection goals include preventing attackers, potentiallythe customer themselves, from seeing the details of the model.¶

Typically, this works by having some protected environmentin the device go through a remote attestation with some manufacturer servicethat can assess its trustworthiness. If remote attestation succeeds,then the manufacturer service releases either the model or a key to decrypta model already deployed on the Attester in encrypted form to the requester.¶

Attester:

A device desiring to run an ML model.¶

Relying Party:

A server or service holding ML models it desires to protect.¶

2.3.Confidential Data Protection

This is a generalization of the ML model use case above wherethe data can be any highly confidential data, such as health dataabout customers, payroll data about employees, future business plans, etc.As part of the attestation procedure, an assessment is made against a setof policies to evaluate the state of the system that is requestingthe confidential data. Attestation is desired to prevent leaking data viacompromised devices.¶

Attester:

An entity desiring to retrieve confidential data.¶

Relying Party:

An entity that holds confidential data for release to authorized entities.¶

2.4.Critical Infrastructure Control

Potentially harmful physical equipment (e.g., power grid, traffic control, hazardous chemical processing, etc.) is connected to a network in support of critical infrastructure. The organization managing such infrastructure needs to ensure that only authorized code and users can control corresponding critical processes, and that these processes are protected from unauthorized manipulation or other threats. When a protocol operation can affect a critical system component of theinfrastructure, devices attached to that critical component require someassurances depending on the security context, including assurances that a requestingdevice or application has not been compromised and the requesters and actorsact on applicable policies. As such, remote attestation can be used to onlyaccept commands from requesters that are within policy.¶

Attester:

A device or application wishing to control physical equipment.¶

Relying Party:

A device or application connected to potentially dangerous physicalequipment (hazardous chemical processing, traffic control, power grid,etc.).¶

2.5.Trusted Execution Environment Provisioning

A Trusted Application Manager (TAM) server is responsiblefor managing the applications running in a Trusted Execution Environment (TEE) of a client device, as described in[TEEP-ARCH].To achieve its purpose, the TAM needs to assess the state of a TEE or applicationsin the TEE of a client device. The TEE conducts remote attestationprocedures with the TAM, which canthen decide whether the TEE is already in compliance with the TAM's latestpolicy. If not, the TAM has to uninstall, update, or install approvedapplications in the TEE to bring it back into compliance with the TAM's policy.¶

Attester:

A device with a TEE capable ofrunning trusted applications that can be updated.¶

Relying Party:

A TAM.¶

2.6.Hardware Watchdog

There is a class of malware that holds a device hostage and doesnot allow it to reboot to prevent updates from being applied.This can be a significant problem because it allows a fleet of devices to be held hostage for ransom.¶

A solution to this problem is a watchdog timer implemented in a protectedenvironment, such as a Trusted Platform Module (TPM),as described in Section 43.3 of[TCGarch].If the watchdog does not receive regular and fresh Attestation Results regarding the system's health, then it forces a reboot.¶

Attester:

The device that should be protected from being held hostage fora long period of time.¶

Relying Party:

A watchdog capable of triggering a procedure that resets a device intoa known, good operational state.¶

2.7.FIDO Biometric Authentication

In the Fast IDentity Online (FIDO) protocol[WebAuthN][CTAP], the device in the user's hand authenticates the human user, whether by biometrics (such as fingerprints) or by PIN and password. FIDO authentication puts a large amount of trust in the device compared to typical password authentication because it is the device that verifies the biometric, PIN, and password inputs from the user, not the server. For the Relying Party to know that the authentication is trustworthy, the Relying Party needs to know that the Authenticator part of the device is trustworthy. The FIDO protocol employs remote attestation for this.¶

The FIDO protocol supports several remote attestation protocols and a mechanism by which new ones can be registered and added; thus, remote attestation defined by the RATS architecture is a candidate for use in the FIDO protocol.¶

Attester:

FIDO Authenticator.¶

Relying Party:

Any website, mobile application backend, or service that relies on authentication data based on biometric information.¶

3.1.Two Types of Environments of an Attester

As shown inFigure 2, an Attester consists of at least one Attesting Environment and at least oneTarget Environment co-located in one entity.In some implementations, the Attesting and Target Environments might be combined into one environment.Other implementations might have multiple Attesting and Target Environments,such as in the examples described in more detail in Sections3.2and3.3. Other examples may exist. All compositions of Attesting and Target Environments discussed in this architecture can be combined into more complex implementations.¶

Claims are collected from Target Environments.That is, Attesting Environments collect the values and the information to be represented in Claims by reading system registers and variables, calling into subsystems, and taking measurements on code, memory, or other relevant assets of the Target Environment.Attesting Environments then format the Claims appropriately; typically, theyuse key material andcryptographic functions, such as signing or cipher algorithms, togenerate Evidence.There is no limit or requirement on the types of hardware or software environments that can be used to implement an Attesting Environment. For example, TEEs, embedded Secure Elements(eSEs), TPMs[TCGarch], or BIOS firmware.¶

An arbitrary execution environment may not, by default, be capable of Claims collection for a given Target Environment.Execution environments that are designed specifically to be capable of Claims collection are referred to in this document as "Attesting Environments".For example, a TPM doesn't actively collect Claims itself. Instead, itrequires another component to feed various values to the TPM.Thus, an Attesting Environment in such a case would be the combinationof the TPM together with whatever component is feeding it the measurements.¶

3.2.Layered Attestation Environments

By definition, the Attester role generates Evidence. An Attester may consist of one or more nested environments (layers). The bottom layer of an Attester has an Attesting Environment that is typically designed to be immutable or difficult to modify by malicious code. In order to appraise Evidence generated by an Attester, the Verifier needs to trust various layers, including the bottom Attesting Environment. Trust in the Attester's layers, including the bottom layer, can be established in various ways, as discussed inSection 7.4.¶

In layered attestation, Claims can be collected from or about each layer beginning with an initial layer. The corresponding Claims can be structured in a nested fashion that reflects the nesting of the Attester's layers. Normally, Claims are not self-asserted. Rather, a previous layer acts as the Attesting Environment for the next layer. Claims about an initial layer are typically asserted by an Endorser.¶

The example device illustrated inFigure 3 includes (A) a BIOS stored in read-only memory, (B) a bootloader, and (C) an operating system kernel.¶

The first Attesting Environment (the ROM in this example)has to ensure the integrity of the bootloader (the first Target Environment).There arepotentially multiple kernels to boot; the decision is up to the bootloader.Only a bootloader with intact integrity will make an appropriate decision.Therefore, the Claims relating to the integrity of the bootloader have to be measured securely.At this stage of the boot cycle of thedevice, the Claims collected typically cannot be composed into Evidence.¶

After the boot sequence is started, the BIOS conducts the most important and defining feature of layered attestation: the successfully measured bootloader now becomes (or contains) an Attesting Environment for the next layer. This procedure in layered attestation is sometimes called "staging". It is important that the bootloader not be able to alter any Claims about itself that were collected by the BIOS. This can be ensured having those Claims be either signed by the BIOS or stored in a tamper-proof manner by the BIOS.¶

Continuing with this example, the bootloader's Attesting Environment is now in charge of collecting Claims about the next Target Environment. In this example, it is the kernel to be booted.The final Evidence thus contains two sets of Claims: one set about thebootloader as measured and signed by the BIOS and another set of Claims about thekernel as measured and signed by the bootloader.¶

This example could be extended further by making the kernel become anotherAttesting Environment for an application as another Target Environment.This would result in a third set of Claims in the Evidence pertaining to that application.¶

The essence of this example is a cascade of staged environments. Eachenvironment has the responsibilityof measuring the next environment before the next environment is started.In general, the number of layers may vary by device or implementation,and an Attesting Environment might even have multiple Target Environmentsthat it measures, rather than only one as shown by example inFigure 3.¶

3.3.Composite Device

A composite device is an entity composed of multiple sub-entities such that itstrustworthiness has to be determined by the appraisal of all these sub-entities.¶

Each sub-entity has at least one Attesting Environment collecting the Claimsfrom at least one Target Environment. Then, this sub-entity generates Evidenceabout its trustworthiness; therefore, each sub-entity can be called an "Attester".Among all the Attesters, there may be only some that have the ability to communicatewith the Verifier while others do not.¶

For example, a carrier-grade router consists of a chassis and multiple slots.The trustworthiness of the router depends on all its slots' trustworthiness.Each slot has an Attesting Environment, such as a TEE, collecting theClaims of its boot process, after which it generates Evidence from the Claims.¶

Among these slots, only a "main" slot can communicate with the Verifierwhile other slots cannot. However, other slots can communicate with the mainslot by the links between them inside the router.The main slot collects the Evidence of other slots, produces the final Evidence of the whole router, and conveys the final Evidence to the Verifier.Therefore, the router is a compositedevice, each slot is an Attester, and the main slot is the lead Attester.¶

Another example is a multi-chassis router composed of multiple single carrier-grade routers.Multi-chassis router setups create redundancy groups that provide higher throughput by interconnectingmultiple routers in these groups, which can be treated as one logical router for simpler management.A multi-chassis router setup provides a management point that connects to the Verifier.Typically, one router in the group is designated as the main router.Other routers in the multi-chassis setup are connected to the main router only via physical network links; therefore, they are managed and appraised via the main router's help.Consequently, a multi-chassis router setup is a composite device,each router is an Attester, and the main router is the lead Attester.¶

Figure 4 depicts the conceptual data flow for a composite device.¶

In a composite device, each Attester generates its own Evidence by itsAttesting Environment(s) collecting the Claims from its Target Environment(s).The lead Attester collects Evidence from other Attesters and conveys it to a Verifier.Collection of Evidence from sub-entities may itself be a form of Claims collection that results in Evidence asserted by the lead Attester.The lead Attester generates Evidence about the layout of the whole composite device, while sub-Attesters generate Evidence about their respective (sub-)modules.¶

In this scenario, the trust model described inSection 7 can also be applied to an inside Verifier.¶

3.4.Implementation Considerations

An entity can take on multiple RATS roles (e.g., Attester, Verifier, RelyingParty, etc.) at the same time.Multiple entities can cooperate to implement a single RATS role as well.In essence, the combination of roles and entities can be arbitrary.For example, in the composite device scenario, the entity insidethe lead Attester can also take on the role of a Verifier and theouter entity of Verifier can take on the role of a Relying Party.After collecting the Evidence of other Attesters, this inside Verifier usesEndorsements and appraisal policies (obtained the same way as by any otherVerifier) as part of the appraisal procedures that generate Attestation Results.The inside Verifier then conveys the Attestation Results of other Attesters to the outside Verifier,whether in the same conveyance protocol as part of the Evidence or not.¶

As explained inSection 4, there are a variety of roles in the RATS architecture; they are defined by a unique combination of artifacts they produce and consume.Conversely, artifacts are also defined by the roles that produce or consume them.To produce an artifact means that a given role introduces it into the RATS architecture.To consume an artifact means that a given role has responsibility for processing it in the RATS architecture.Roles also have the ability to perform additional actions, such as caching or forwarding artifacts as opaque data.As depicted inSection 5, these additional actions can be performed by several roles.¶

4.1.Roles

Attester:

A role performed by an entity (typically a device) whose Evidence must be appraised in order to infer the extent to which the Attester is considered trustworthy, such as when deciding whether it is authorized to perform some operation.¶

Produces:

Evidence¶

Relying Party:

A role performed by an entity that depends on the validity of information about an Attester for purposes of reliably applying application-specific actions. Compare: relying party[RFC4949].¶

Consumes:

Attestation Results, Appraisal Policy for Attestation Results¶

Verifier:

A role performed by an entity that appraises the validity of Evidence about an Attester and produces Attestation Results to be used by a Relying Party.¶

Consumes:

Evidence, Reference Values, Endorsements, Appraisal Policy for Evidence¶

Produces:

Attestation Results¶

Relying Party Owner:

A role performed by an entity (typically an administrator) that is authorized to configure an Appraisal Policy for Attestation Results in a Relying Party.¶

Produces:

Appraisal Policy for Attestation Results¶

Verifier Owner:

A role performed by an entity (typically an administrator) that is authorized to configure an Appraisal Policy for Evidence in a Verifier.¶

Produces:

Appraisal Policy for Evidence¶

Endorser:

A role performed by an entity (typically a manufacturer) whose Endorsements may help Verifiers appraise the authenticity of Evidence and infer further capabilities of the Attester.¶

Produces:

Endorsements¶

Reference Value Provider:

A role performed by an entity (typically a manufacturer) whose Reference Values help Verifiers appraise Evidence to determine if acceptable known Claims have been recorded by the Attester.¶

Produces:

Reference Values¶

4.2.Artifacts

Claim:

A piece of asserted information, often in the form of a name/value pair. Claims make up the usual structure of Evidence and other RATS conceptual messages.Compare: claim[RFC7519].¶

Endorsement:

A secure statement that an Endorser vouches for the integrity of an Attester's various capabilities, such as Claims collection and Evidence signing.¶

Consumed By:

Verifier¶

Produced By:

Endorser¶

Evidence:

A set of Claims generated by an Attester to be appraised by a Verifier.Evidence may include configuration data, measurements, telemetry, or inferences.¶

Consumed By:

Verifier¶

Produced By:

Attester¶

Attestation Result:

The output generated by a Verifier, typically including information about an Attester, where the Verifier vouches for the validity of the results.¶

Consumed By:

Relying Party¶

Produced By:

Verifier¶

Appraisal Policy for Evidence:

A set of rules that a Verifier uses to evaluate the validity of information about an Attester. Compare: security policy[RFC4949].¶

Consumed By:

Verifier¶

Produced By:

Verifier Owner¶

Appraisal Policy for Attestation Results:

A set of rules that direct how a Relying Party uses the Attestation Results regarding an Attester generated by the Verifiers. Compare: security policy[RFC4949].¶

Consumed by:

Relying Party¶

Produced by:

Relying Party Owner¶

Reference Values:

A set of values against which values of Claims can be compared as part ofapplying an Appraisal Policy for Evidence. Reference Values are sometimesreferred to in other documents as "known-good values", "golden measurements",or "nominal values". These terms typically assume comparison forequality, whereas here, Reference Values might be more general and be usedin any sort of comparison.¶

Consumed By:

Verifier¶

Produced By:

Reference Value Provider¶

5.1.Passport Model

The Passport Model is so named because of its resemblance to how nations issuepassports to their citizens. The nature of the Evidence that an individual needsto provide to its local authority is specific to the country involved. The citizenretains control of the resulting passport document and presents it to other entitieswhen it needs to assert a citizenship or identity Claim, such as at an airport immigrationdesk. The passport is considered sufficient because it vouches for the citizenship andidentity Claims and it is issued by a trusted authority.¶

Thus, in this immigration desk analogy,the citizen is the Attester,the passport-issuing agency is a Verifier,and the passport application and identifying information (e.g., birth certificate) is the Evidence.The passport is an Attestation Result and the immigration desk is a Relying Party.¶

In this model, an Attester conveys Evidence to a Verifier that comparesthe Evidence against its appraisal policy.The Verifier then gives back an Attestation Result that the Attester treats as opaque data.¶

The Attester does not consume the Attestation Result, but it might cache it.The Attester can then present the Attestation Result (and possibly additional Claims)to a Relying Party, which then compares this information against its ownappraisal policy.The Attester may also present the same Attestation Result to other Relying Parties.¶

There are three ways in which the process may fail:¶

• First, the Verifier may not issue a positive Attestation Result due to the Evidence not passing the Appraisal Policy for Evidence.¶
• The second way in which the process may fail is when the Attestation Resultis examined by the Relying Party, and based upon the Appraisal Policy forAttestation Results, the result does not comply with the policy.¶
• The third way is when the Verifier is unreachable or unavailable.¶

As with any other information needed by the Relying Party to make an authorization decision,an Attestation Result can be carried in a resource access protocol between the Attester and Relying Party.In this model, the details of the resource access protocolconstrain the serialization format of the Attestation Result. On the other hand, theformat of the Evidence is only constrained by theAttester-Verifier remote attestation protocol.This implies that interoperability and standardization is more relevant for Attestation Results than it is for Evidence.¶

5.2.Background-Check Model

The Background-Check Model is so named because of the resemblance of how employers and volunteer organizations perform background checks. When a prospective employee provides Claims about education or previous experience, the employer will contact the respective institutions or former employers to validate the Claim. Volunteer organizations often perform police background checks on volunteers in order to determine the volunteer's trustworthiness. Thus, in this analogy, a prospective volunteer is an Attester, the organization is the Relying Party, and the organization that issues a report is a Verifier.¶

In this model, an Attester conveys Evidence to a Relying Party, which treats it as opaque and simply forwards it on to a Verifier. The Verifier compares the Evidence against its appraisal policy and returns an Attestation Result to the Relying Party. The Relying Party then compares the Attestation Result against its own appraisal policy.¶

The resource access protocol between the Attester and Relying Party includes Evidence rather than an Attestation Result, but that Evidence is not processed by the Relying Party.¶

Since the Evidence is merely forwarded on to a trusted Verifier, any serialization format can be used for Evidence because the Relying Party does not need a parser for it. The only requirement is that the Evidence can beencapsulated in the formatrequired by the resource access protocol between the Attester and Relying Party.¶

However, as seen in the Passport Model, an Attestation Result is still consumed by theRelying Party. Code footprint and attack surface area can be minimized byusing a serialization format for which the Relying Party already needs aparser to support the protocol between the Attester and Relying Party,which may be an existing standard or widely deployed resource access protocol.Such minimization is especially important if the Relying Party is aconstrained node.¶

5.3.Combinations

One variation of the Background-Check Model is where the Relying Partyand the Verifier are on the same machine, performing both functions together.In this case, there is no need for a protocol between the two.¶

It is also worth pointing out that the choice of model depends on the use case and that different Relying Parties may use different topological patterns.¶

The same device may need to create Evidence for different Relying Parties and/or different use cases.For instance, it would use one model to provide Evidence to a network infrastructure device to gain access to the network andthe other model to provide Evidence to a server holding confidential data to gain access to that data.As such, both models may simultaneously be in use by the same device.¶

Figure 7 shows another example of a combination where Relying Party 1 uses thePassport Model, whereas Relying Party 2 uses an extension of the Background-Check Model.Specifically, in addition to the basic functionality shown inFigure 6, Relying Party 2actually provides the Attestation Result back to the Attester, allowing the Attester touse it with other Relying Parties. This is the model that the TAMplans to support in the TEEP architecture[TEEP-ARCH].¶

7.1.Relying Party

This document covers scenarios for which a Relying Partytrusts a Verifier that can appraise the trustworthiness ofinformation about an Attester. Such trustis expressed by storing one or more "trust anchors" in a secure locationknown as a "trust anchor store".¶

As defined in[RFC6024]:¶

A trust anchor represents an authoritative entity via a publickey and associated data. The public key is used to verify digitalsignatures, and the associated data is used to constrain the typesof information for which the trust anchor is authoritative.¶

The trust anchor may be a certificate or it may be a raw public keyalong with additional data if necessary, such as its public keyalgorithm and parameters.In the context of this document, a trust anchor may also be a symmetric key, asin[TCG-DICE-SIBDA], or the symmetric mode described in[RATS-PSA-TOKEN].¶

Thus, trusting a Verifier might be expressed by having the RelyingParty store the Verifier's key or certificate in its trust anchor store. It might alsobe expressed by storing the public key or certificate of an entity (e.g., a Certificate Authority) that isin the Verifier's certificate path.For example, the Relying Party can verify that the Verifier is an expected one by out-of-band establishment of key material combined with a protocol like TLS to communicate.There is an assumption that the Verifier has not been compromised between the establishment of the trusted key material and the creation of the Evidence.¶

For a stronger level of security, theRelying Party might require that the Verifier first provideinformation about itself that the Relying Party can use to assessthe trustworthiness of the Verifier before accepting its Attestation Results.Such a process would provide a stronger level of confidence in the correctness ofthe information provided, such as a belief that the authentic Verifier hasnot been compromised by malware.¶

For example, one explicit way for a Relying Party "A" to establishsuch confidence in the correctness of a Verifier "B" would be for B to first act as an Attesterwhere A acts as a combined Verifier/Relying Party. If A then accepts B astrustworthy, it can choose to accept B as a Verifier for other Attesters.¶

Similarly, the Relying Party also needs to trust the Relying Party Ownerfor providing its Appraisal Policy for Attestation Results, and,in some scenarios, the Relying Party might even require that theRelying Party Owner go through a remote attestation procedure with it before the Relying Party will acceptan updated policy. This can be done in a manner similar to how a Relying Partycould establish trust in a Verifier as discussed above, i.e., verifying credentials against a trust anchor storeand optionally requiring Attestation Results from the Relying Party Owner.¶

7.2.Attester

In some scenarios, Evidence might contain sensitive information, such asPersonally Identifiable Information (PII) or system identifiable information.Thus, an Attester must trust the entities to which it conveys Evidence to notreveal sensitive data to unauthorized parties.The Verifier might share this information with otherauthorized parties according to a governing policy that addresses thehandling of sensitive information (potentially included in AppraisalPolicies for Evidence).In the Background-Check Model, this Evidence may also be revealed to Relying Parties.¶

When Evidence contains sensitive information, an Attestertypically requires that a Verifier authenticates itself (e.g., at TLS session establishment) and might even request a remote attestation before the Attestersends the sensitive Evidence. This can be done by having theAttester first act as a Verifier/Relying Party and the Verifier act as itsown Attester, as discussed above.¶

7.3.Relying Party Owner

The Relying Party Owner might also require that theRelying Party first act as an Attester by providing Evidence that the Ownercan appraise before the Owner would give the Relying Party an updatedpolicy that might contain sensitive information. In such a case, authentication or attestation in both directions mightbe needed. Typically, one side's Evidence must be considered safe toshare with an untrusted entity in order to bootstrap the sequence.SeeSection 11 for more discussion.¶

7.4.Verifier

The Verifier trusts (or more specifically, the Verifier's securitypolicy is written in a way that configures the Verifier to trust) amanufacturer or the manufacturer's hardware so as to be able toappraise the trustworthiness of that manufacturer's devices. Such trustis expressed by storing one or more trust anchors in the Verifier'strust anchor store.¶

In a typical solution, a Verifier comes to trust an Attester indirectly by having an Endorser (such as a manufacturer) vouch for the Attester's ability to securely generate Evidence through Endorsements (seeSection 8.2). Endorsements might describe the ways in which the Attester resists attacks, protects secrets, and measures Target Environments. Consequently, the Endorser's key material is stored in the Verifier's trust anchor store so that Endorsements can be authenticated and used in the Verifier's appraisal process.¶

In some solutions, a Verifier might be configured to directly trust an Attester by having the Verifier possess the Attester's key material (rather than the Endorser's) in its trust anchor store.¶

Such direct trust must first be established at the time of trust anchorstore configuration either by checking with an Endorser at thattime or by conducting a security analysis of the specific device.Having the Attester directly in the trust anchor store narrowsthe Verifier's trust to only specific devices rather than all devicesthe Endorser might vouch for, such as all devices manufactured by thesame manufacturer in the case that the Endorser is a manufacturer.¶

Such narrowing is often important since physical possession of a devicecan also be used to conduct a number of attacks, and so a device ina physically secure environment (such as one's own premises) may beconsidered trusted, whereas devices owned by others would not be.This often results in a desire either to have the owner run theirown Endorser that would only endorse devices one owns or to useAttesters directly in the trust anchor store. When there are manyAttesters owned, the use of an Endorser enables better scalability.¶

That is, a Verifier might appraise the trustworthiness of an application component, operatingsystem component, or service under the assumption that informationprovided about it by the lower-layer firmware or software is true.A stronger level of assurance of security comes when information can be vouchedfor by hardware or by ROM code, especially if such hardware isphysically resistant to hardware tampering.In most cases, components that have to be vouched for via Endorsements (because no Evidence is generated about them) are referred to as "roots of trust".¶

The manufacturer having arranged for an Attesting Environment to be provisioned with key material with which to sign Evidence, the Verifier is then provided withsome way of verifying the signature on the Evidence. This may be in the form of an appropriate trust anchor or the Verifier may be provided with a database of public keys (rather than certificates) or even carefully curated and secured lists of symmetric keys.¶

The nature of how the Verifier manages to validate the signatures produced by the Attester is critical to the secure operation of a remote attestation system but is not the subject of standardization within this architecture.¶

A conveyance protocol that provides authentication and integrity protection can be usedto convey Evidence that is otherwise unprotected (e.g., not signed). Appropriate conveyance of unprotected Evidence (e.g.,[RATS-UCCS]) relies on the following conveyance protocol's protection capabilities:¶

1. The key material used to authenticate and integrity protect the conveyance channel is trusted by the Verifier to speak for the Attesting Environment(s) that collected Claims about the Target Environment(s).¶
2. All unprotected Evidence that is conveyed is supplied exclusively by the Attesting Environment that has the key material that protects the conveyance channel.¶
3. A trusted environment protects the conveyance channel's key material, which may depend on other Attesting Environments with equivalent strength protections.¶

As illustrated in[RATS-UCCS], an entity that receives unprotected Evidence via a trusted conveyance channel always takes on the responsibility of vouching for the Evidence's authenticity and freshness.If protected Evidence is generated, the Attester's Attesting Environments take on that responsibility.In cases where unprotected Evidence is processed by a Verifier, Relying Parties have to trust that the Verifier is capable of handling Evidence in a manner that preserves the Evidence's authenticity and freshness.Generating and conveying unprotected Evidence always creates significant risk and the benefits of that approach have to be carefully weighed against potential drawbacks.¶

SeeSection 12 for discussion on security strength.¶

7.5.Endorser, Reference Value Provider, and Verifier Owner

In some scenarios, the Endorser, Reference Value Provider, and Verifier Owner may need to trust the Verifierbefore giving the Endorsement, Reference Values, or appraisal policy to it. This can be done in a similar manner to how a Relying Party might establish trust in a Verifier.¶

As discussed inSection 7.3, authentication or attestation in both directions might beneeded. Typically, one side's identity orEvidence in this case must be considered safe to share with an untrusted entityin order to bootstrap the sequence.SeeSection 11 for more discussion.¶

8.1.Evidence

Evidence is a set of Claims about the Target Environment that reveal operationalstatus, health, configuration, or construction that have security relevance.Evidence is appraised by a Verifier to establish its relevance, compliance, and timeliness.Claims need to be collected in a manner that is reliable such that a Target Environment cannot lie to the Attesting Environment about its trustworthiness properties.Evidence needs to be securely associated with the Target Environmentso that the Verifier cannot be tricked into accepting Claims originatingfrom a different environment (that may be more trustworthy).Evidence also must be protected from an active on-path attacker who may observe,change, or misdirect Evidence as it travels from the Attester to the Verifier.The timeliness of Evidence can be captured using Claims that pinpoint the timeor interval when changes in operational status, health, and so forth occur.¶

8.2.Endorsements

An Endorsement is a secure statement that some entity (e.g., a manufacturer) vouches for the integrity of the device's various capabilities, such as Claims collection, signing, launching code, transitioning to other environments, storing secrets, and more. For example, if the device's signing capability is in hardware, then an Endorsement might be a manufacturer certificate that signs a public key whose corresponding private key is only known inside the device's hardware. Thus, when Evidence and such an Endorsement are used together, an appraisal procedure can be conducted based on appraisal policies that may not be specific to the device instance but are merely specific to the manufacturer providing the Endorsement. For example, an appraisal policy might simply check that devices from a given manufacturer have information matching a set of Reference Values. An appraisal policy might also have a set of more complex logic on how to appraise the validity of information.¶

However, while an appraisal policy that treats all devices from a given manufacturer the same may be appropriate for some use cases, it would be inappropriate to use such an appraisal policy as the sole means of authorization for use cases that wish to constrainwhich compliant devices are considered authorized for some purpose. For example, an enterprise using remote attestation for Network Endpoint Assessment (NEA)[RFC5209] may not wish to let every healthy laptop from the same manufacturer onto the network. Instead, it may only want to let devices that it legally owns onto the network. Thus, an Endorsement may be helpful information in authenticating information about a device, but is not necessarily sufficient to authorize access to resources that may need device-specific information, such as a public key for the device or component or user on the device.¶

8.3.Reference Values

Reference Values used in appraisal procedures come from a Reference Value Provider and are then used by the Verifier to compare to Evidence. Reference Values with matching Evidence produce acceptable Claims. Additionally, an appraisal policy may play a role in determining the acceptance of Claims.¶

8.4.Attestation Results

Attestation Results are the input used by the Relying Party to decide the extent to which it will trust a particular Attester and allow it to access some data or perform some operation.¶

Attestation Results may carry a boolean value indicating compliance or non-compliance with a Verifier's appraisal policy or may carry a richer set of Claims about the Attester, against which the Relying Party applies its Appraisal Policy for Attestation Results.¶

The quality of the Attestation Results depends upon the ability of the Verifier to evaluate the Attester.Different Attesters have a differentStrength of Function[strengthoffunction], which results in the Attestation Results being qualitatively different in strength.¶

An Attestation Result that indicates non-compliance can be used by an Attester (in the Passport Model) ora Relying Party (in the Background-Check Model) to indicate that the Attestershould not be treated as authorized and may be in need of remediation. In some cases,it may even indicate that the Evidence itself cannot be authenticated as being correct.¶

By default, the Relying Party does not believe the Attester to be compliant.Upon receipt of an authentic Attestation Result and given the Appraisal Policy for Attestation Results is satisfied,the Attester is allowed to perform the prescribed actions or access.The simplest such appraisal policy might authorize granting the Attester full access or control over the resources guarded by the Relying Party.A more complex appraisal policy might involve using the informationprovided in the Attestation Result to compare against expected values or to apply complex analysisof other information contained in the Attestation Result.¶

Thus, Attestation Results can contain detailed information about an Attester, which can include privacy sensitive information as discussed inSection 11. Unlike Evidence, which is often very device- and vendor-specific, Attestation Results can be vendor-neutral, if the Verifier has a way to generate vendor-agnostic information based on the appraisal of vendor-specific information in Evidence. This allows a Relying Party's appraisal policy to be simpler, potentially based on standard ways of expressing the information, while still allowing interoperability with heterogeneous devices.¶

Finally, whereas Evidence is signed by the device (or indirectly by a manufacturer if Endorsements are used), Attestation Results are signed by a Verifier, allowing a Relying Party to only need a trust relationship with one entity rather than a larger set of entities for purposes of its appraisal policy.¶

8.5.Appraisal Policies

The Verifier (when appraising Evidence) or the Relying Party (whenappraising Attestation Results) checks the values of matched Claimsagainst constraints specified in its appraisal policy.Examples of such constraints checking include the following:¶

• Comparison for equality against a Reference Value.¶
• A check for being in a range bounded by Reference Values.¶
• Membership in a set of Reference Values.¶
• A check against values in other Claims.¶

Upon completing all appraisal policy constraints, the remaining Claims are acceptedas input toward determining Attestation Results (when appraising Evidence)or as input to a Relying Party (when appraising Attestation Results).¶

10.1.Explicit Timekeeping Using Synchronized Clocks

The first approach is to rely on synchronized and trustworthy clocks andinclude a signed timestamp (see[RATS-TUDA]) along with theClaims in the Evidence or Attestation Result. Timestamps can also be added on aper-Claim basis to distinguish the time of generation of Evidence or AttestationResult from the time that a specific Claim was generated. The clock'strustworthiness can generally be established via Endorsements and typically requires additional Claims about the signer's timesynchronization mechanism.¶

However, a trustworthy clock might not be available in some use cases. Forexample, in many TEEs today, a clock is onlyavailable outside the TEE; thus, it cannot be trusted by the TEE.¶

10.2.Implicit Timekeeping Using Nonces

A second approach places the onus of timekeeping solely on the Verifier (for Evidence) or the Relying Party (for Attestation Results). For example, this approach might be suitable in case the Attester does not have a trustworthy clock or time synchronization is otherwise impaired. In this approach, an unpredictable nonce is sent by the appraising entity and the nonce is then signed and included along with the Claims in the Evidence or Attestation Result. After checking that the sent and received nonces are the same, the appraising entity knows that the Claims were signed after the nonce was generated. This allows associating a "rough" epoch to the Evidence or Attestation Result. In this case, the epoch is said to be rough because:¶

• The epoch applies to the entire Claim set instead of a more granularassociation, and¶
• The time between the creation of Claims and the collection of Claims isindistinguishable.¶

10.3.Implicit Timekeeping Using Epoch IDs

A third approach relies on having epoch identifiers (IDs)periodically sent to both the sender and receiver of Evidence orAttestation Results by some "epoch ID distributor".¶

Epoch IDs are different from nonces as they can be used more than once andcan even be used by more than one entity at the same time.Epoch IDs are different from timestamps as they do not have to convey information about a point in time, i.e., they are not necessarily monotonically increasing integers.¶

Like the nonce approach, this allows associating a "rough" epoch withoutrequiring a trustworthy clock or time synchronization in order to generate orappraise the freshness of Evidence or Attestation Results. Only theepoch ID distributor requires access to a clock so it can periodically sendnew epoch IDs.¶

The most recent epoch ID is included in the produced Evidence or AttestationResults, and the appraising entity can compare the epoch ID in receivedEvidence or Attestation Results against the latest epoch ID it received fromthe epoch ID distributor to determine if it is within the current epoch.An actual solution also needs to take into account race conditionswhen transitioning to a new epoch, such as by using a counter signedby the epoch ID distributor as the epoch ID, by including both the current andprevious epoch IDs in messages and/or checks by requiring retriesin case of mismatching epoch IDs, or by buffering incoming messagesthat might be associated with an epoch ID that the receiver has not yetobtained.¶

More generally, in order to prevent an appraising entity from generating falsenegatives (e.g., discarding Evidence that is deemed stale even if it isnot), the appraising entity should keep an "epoch window" consisting of themost recently received epoch IDs. The depth of such epoch window is directlyproportional to the maximum network propagation delay between the first to receive the epoch ID and the last to receive the epoch ID and it is inversely proportional to the epoch duration.The appraising entity shall compare theepoch ID carried in the received Evidence or Attestation Result with the epoch IDsin its epoch window to find a suitable match.¶

Whereas the nonce approach typically requires the appraising entityto keep state for each nonce generated, the epoch ID approach minimizesthe state kept to be independent of the number of Attesters or Verifiersfrom which it expects to receive Evidence or Attestation Results as longas all use the same epoch ID distributor.¶

10.4.Discussion

Implicit and explicit timekeeping can be combined into hybrid mechanisms.For example, if clocks exist within the Attesting Environment and are considered trustworthy (tamper-proof) but are not synchronized, a nonce-based exchange may be used to determine the (relative) time offset between the involved peers followed by any number of timestampbased exchanges.¶

It is important to note that the actual values in Claims might have beengenerated long before the Claims are signed. If so, it is the signer'sresponsibility to ensure that the values are still fresh when they aresigned. For example, values generated at boot time might have been saved tosecure storage until network connectivity is established to the remote Verifierand a nonce is obtained.¶

A more detailed discussion with examples appears inAppendix A.¶

For a discussion on the security of epoch IDs seeSection 12.3.¶

12.1.Attester and Attestation Key Protection

Implementers need to pay close attention to the protection of the Attester and the manufacturing processes for provisioning attestation key material. If either of these are compromised, intended levels of assurance for remote attestation procedures are compromised because attackers can forge Evidence or manipulate the Attesting Environment.For example, a Target Environment should not be able to tamper with theAttesting Environment that measures it by isolating the two environmentsfrom each other in some way.¶

Remote attestation applies to use cases with a range of security requirements. The protections discussed here range from low to high security: low security may be limited to application or process isolation by the device's operating system and high security may involve specialized hardware to defend against physical attacks on a chip.¶

12.2.Conceptual Message Protection

Any solution that conveys information in any conceptual message (seeSection 8)must support end-to-end integrity protectionand replay attack prevention. It often also needs to support additionalsecurity properties, including:¶

• end-to-end encryption,¶
• denial-of-service protection,¶
• authentication,¶
• auditing,¶
• fine-grained access controls, and¶
• logging.¶

Section 10 discusses ways in which freshness can be used in thisarchitecture to protect against replay attacks.¶

To assess the security provided by a particular appraisal policy, itis important to understand the strength of the root of trust, e.g.,whether it is mutable software or firmware that is read-only afterboot or immutable hardware/ROM.¶

It is also important that the appraisal policy was obtained securely itself.If an attacker can configure or modify appraisal policies and Endorsements or Reference Values for a Relying Party or a Verifier, then integrity of the process is compromised.¶

Security protections in the RATS architecture may be applied at different layers, whether by a conveyance protocol or an information encoding format.This architecture expects conceptual messages to be end-to-end protected based on the role interaction context.For example, if an Attester produces Evidence that is relayed through some other entity that doesn't implement the Attester or the intended Verifier roles, then the relaying entity should not expect to have access to the Evidence.¶

The RATS architecture allows for an entity to function in multiple roles(Section 6) and for composite devices (Section 3.3).Implementers need to evaluate their designs to ensure that the assumed security properties of the individual components and roles still hold despite the lack of separation and that emergent risk is not introduced. The specifics of this evaluation will depend on the implementation and the use case; hence, they are out of scope for this document.Isolation mechanisms in software or hardware that separate Attesting Environments and Target Environments (Section 3.1) can support an implementer's evaluation and resulting design decisions.¶

12.3.Attestation Based on Epoch ID

Epoch IDs, described inSection 10.3, can be tampered with, replayed, dropped, delayed, andreordered by an attacker.¶

An attacker could either be external or belong to the distribution group (forexample, if one of the Attester entities have been compromised).¶

An attacker who is able to tamper with epoch IDs can potentially lock all theparticipants in a certain epoch of choice forever, effectively freezing time.This is problematic since it destroys the ability to ascertain freshness ofEvidence and Attestation Results.¶

To mitigate this threat, the transport should be at least integrity protectedand provide origin authentication.¶

Selective dropping of epoch IDs is equivalent to pinning the victim node to a past epoch.An attacker could drop epoch IDs to only some entities and not others, which will typically result in a denial of service due to the permanent staleness of the Attestation Result or Evidence.¶

Delaying or reordering epoch IDs is equivalent to manipulating the victim'stimeline at will. This ability could be used by a malicious actor (e.g., acompromised router) to mount a confusion attack. For example, a Verifiercan be tricked into accepting Evidence coming from a past epoch as fresh, while, inthe meantime, the Attester has been compromised.¶

Reordering and dropping attacks are mitigated if the transport provides the ability to detect reordering and drop.However, the delay attack described above can't be thwarted in this manner.¶

12.4.Trust Anchor Protection

As noted inSection 7, Verifiers and Relying Parties have trust anchor storesthat must be secured.[RFC6024] contains more discussion of trust anchor store requirementsfor protecting public keys.Section 6 of[NIST-800-57-p1] contains a comprehensive treatment of thetopic, including the protection of symmetric key material.Specifically, a trust anchor store must resist modification against unauthorized insertion, deletion, and modification.Additionally, if the trust anchor is a symmetric key, the trust anchor store mustnot allow unauthorized read.¶

If certificates are used as trust anchors, Verifiers and Relying Parties are alsoresponsible for validating the entire certificate path up to the trust anchor,which includes checking for certificate revocation.For an example of such a procedure, seeSection 6 of [RFC5280].¶

A.1.Example 1: Timestamp-Based Passport Model

Figure 10illustrates a hypothetical Passport Modelsolution that uses timestamps and requires roughly synchronizedclocks between the Attester, Verifier, and Relying Party, whichdepends on using a secure clock synchronization mechanism.As a result, the receiver of a conceptual message containing atimestamp can directly compare it to its own clock and timestamps.¶

The Verifier can check whether the Evidence is fresh when appraisingit at time(RG_v) by checkingtime(RG_v) - time(EG_a) < Threshold, where theVerifier's threshold is large enough to account for the maximumpermitted clock skew between the Verifier and the Attester.¶

If time(VG_a) is included in the Evidence along with the Claim valuegenerated at that time, and the Verifier decides that it can trust thetime(VG_a) value, the Verifier can also determine whether the Claim value isrecent by checkingtime(RG_v) - time(VG_a) < Threshold.The threshold is decided by the Appraisal Policy for Evidence and, again, needs to takeinto account the maximum permitted clock skew betweenthe Verifier and the Attester.¶

The Attester does not consume the Attestation Result but might cache it.¶

The Relying Party can check whether the Attestation Result is freshwhen appraising it at time(RA_r) by checking thetime(RA_r) - time(RG_v) < Threshold,where the Relying Party's threshold is large enough to account for themaximum permitted clock skew between the Relying Party and the Verifier.The result might then be used for some time (e.g., throughout the lifetimeof a connection established at time(RA_r)). However, the Relying Party must becareful not to allow continued use beyond the period for whichit deems the Attestation Result to remain fresh enough. Thus,it might allow use (at time(OP_r)) as long astime(OP_r) - time(RG_v) < Threshold.However, if the Attestation Result contains an expiry time time(RX_v), thenit could explicitly checktime(OP_r) < time(RX_v).¶

A.2.Example 2: Nonce-Based Passport Model

Figure 11 illustrates a hypothetical Passport Modelsolution that uses nonces instead of timestamps. Compared to thetimestamp-based example, it requires an extra round tripto retrieve a nonce and requires that the Verifier and Relying Partytrack state to remember the nonce for some period of time.¶

The advantage is that it does not require that any clocksare synchronized.As a result, the receiver of a conceptual message containing atimestamp cannot directly compare it to its own clock or timestamps.Thus, we use a suffix ("a" for Attester, "v" for Verifier, and "r" for Relying Party) on the IDs below indicating which clock generated them since times from different clocks cannot be compared.Only the delta between two events from the sender can be used by the receiver.¶

In this example solution, the Verifier can check whether the Evidence isfresh attime(RG_v) by verifying thattime(RG_v)-time(NS_v) < Threshold.¶

However, the Verifier cannot simply rely on a Nonce todetermine whether the value of a Claim is recent since the Claim valuemight have been generated long before the nonce was sent by the Verifier.Nevertheless, if the Verifier decides that the Attester can be trusted tocorrectly provide the deltatime(EG_a)-time(VG_a), then it can determine recencyby checkingtime(RG_v)-time(NS_v) + time(EG_a)-time(VG_a) < Threshold.¶

Similarly if, based on an Attestation Result from a Verifier it trusts, theRelying Party decides that the Attester can be trusted to correctly providetime deltas, then it can determine whether the Attestation Result is fresh bycheckingtime(OP_r)-time(NS_r) + time(RR_a)-time(EG_a) <Threshold. Although the Nonce2 andtime(RR_a)-time(EG_a) valuescannot be inside the Attestation Result, they might be signed by the Attestersuch that the Attestation Result vouches for the Attester's signingcapability.¶

However, the Relying Party must still be careful not to allow continueduse beyond the period for which it deems the Attestation Result to remainvalid. Thus, if the Attestation Result sends a validity lifetimein terms oftime(RX_v)-time(RG_v), then the Relying Party can checktime(OP_r)-time(NS_r) < time(RX_v)-time(RG_v).¶

A.3.Example 3: Passport Model Based on Epoch ID

The example inFigure 12 illustrates a hypothetical Passport Modelsolution that uses epoch IDs instead of nonces or timestamps.¶

The epoch ID distributor broadcasts epoch IDI, which starts a newepochE for a protocol participant upon reception attime(IR).¶

The Attester generates Evidence incorporating epoch IDI and conveys it to the Verifier.¶

The Verifier appraises that the received epoch IDI is "fresh" according to thedefinition provided inSection 10.3 whereby retries are required in the case of mismatching epoch IDs; then the Verifier generates an Attestation Result. TheAttestation Result is conveyed to the Attester.¶

After the transmission of epoch IDI' a new epochE' isestablished whenI' is received by each protocol participant. The Attesterrelays the Attestation Result obtained during epochE (associated with epoch IDI) to the Relying Party using the epoch ID for the current epochI'.If the Relying Party had not yet receivedI', then the Attestation Result would be rejected. The Attestation Result is received in this example.¶

InFigure 12, the epoch ID for relaying an Attestation Result to the Relying Party is current while a previous epoch ID was used to generate Verifier evaluated Evidence.This indicates that at least one epoch transition has occurred and the Attestation Results may only be as fresh as the previous epoch.If the Relying Party remembers the previous epoch IDI during an epoch windowas discussed inSection 10.3, and the message is received duringthat window, the Attestation Result is accepted as fresh; otherwise, it isrejected as stale.¶

A.4.Example 4: Timestamp-Based Background-Check Model

Figure 13 illustrates a hypothetical Background-Check Modelsolution that uses timestamps and requires roughly synchronizedclocks between the Attester, Verifier, and Relying Party.The Attester conveys Evidence to the Relying Party, which treats it as opaque and simply forwards it on to the Verifier.¶

The time considerations in this example are equivalent to thosediscussed underExample 1.¶

A.5.Example 5: Nonce-Based Background-Check Model

Figure 14 illustrates a hypothetical Background-Check Modelsolution that uses nonces; thus, it does not require that any clocksbe synchronized.In this example solution, a nonce is generated by a Verifier at the request of a Relying Party when the Relying Party needs to send one to an Attester.¶

The Verifier can check whether the Evidence is fresh and a Claimvalue is recent, which is the same asExample 2.¶

However, unlike inExample 2, the Relying Party can use the Nonce to determine whether the Attestation Result is fresh by verifying thattime(OP_r)-time(NR_r) < Threshold.¶

However, the Relying Party must still be careful not to allow continueduse beyond the period for which it deems the Attestation Result to remainvalid. Thus, if the Attestation Result sends a validity lifetimein terms oftime(RX_v)-time(RG_v), then the Relying Party can checktime(OP_r)-time(ER_r) < time(RX_v)-time(RG_v).¶