All text is normative except for labeled examples and notes.
The key words “MUST”, “MUST
NOT”, “REQUIRED”, “SHALL”, “SHALL NOT”, “SHOULD”, “SHOULD NOT”, “RECOMMENDED”, “MAY”,
and “OPTIONAL” in this document are to be interpreted as described in [RFC2119].
[trust-el-analysis-v1.0]
Analysis of Methods of
Trust Elevation Version 1.0. Edited by Peter Alterman, Shaheen Abdul
Jaabar, Jaap Kuipers, Thomas Hardjono, and Mary Ruddy.Work in progress. https://www.oasis-open.org/committees/download.php/48768
[trust-el-survey-v1.0]
Survey of Methods of Trust
Elevation Version 1.0. Edited by Peter Alterman, Shaheen Abdul Jabbar, Jaap
Kuipers, Thomas Hardjono and Mary Ruddy. Work in progress. https://www.oasis-open.org/committees/download.php/46987
[trust-el-framework-v1.0]
Electronic Identity Credential
Trust Elevation Framework Version 1.0. Edited by Peter Alterman, Shaheen
Abdul Jabbar, Abbie Barbir, Mary Ruddy, and Steve Olshansky. 22 May 2014. OASIS
Committee Specification 01. http://docs.oasis-open.org/trust-el/trust-el-framework/v1.0/cs01/trust-el-framework-v1.0-cs01.html.
Latest version: http://docs.oasis-open.org/trust-el/trust-el-framework/v1.0/trust-el-framework-v1.0.html
[NIST800-63-2] NIST Special Publication 800-63-2,
Electronic Authentication Guideline, August 2013. http://dx.doi.org/10.6028/NIST.SP.800-63-2
[NIST800-162] NIST Special Publication 800-162, Guide
to Attribute Based Access Control (ABAC) Definition and Considerations,
January 2014. http://dx.doi.org/10.6028/NIST.SP.800-162
[RFC2119] Bradner, S.,
"Key words for use in RFCs to Indicate Requirement Levels", BCP 14,
RFC 2119, DOI 10.17487/RFC2119, March 1997, <http://www.rfc-editor.org/info/rfc2119>.
[UMA] Hardjono, T., Maler, E.,
Machulak, M., Catalano, D. User-Managed Access (UMA) Profile of OAuth 2.0,
April 2015. https://docs.kantarainitiative.org/uma/rec-uma-core.html
[X.1252] Recommendation ITU-T X.1252
(2010). Baseline identity management terms and definitions. http://handle.itu.int/11.1002/1000/10440
[XACML3] OASIS Standard, eXtensible Access
Control Markup Language (XACML) Version 3.0, 22 January 2013. http://docs.oasis-open.org/xacml/3.0/xacml-3.0-core-spec-en.doc
[ISO ISMS] ISO/IEC 27000:2014 Information
technology -- Security techniques -- Information security management systems --
Overview and vocabulary, 2014.
[NIST800-37-1] NIST Special Publication 800-37 r1,
Guide for Applying the Risk Management Framework to Federal Information
Systems: A Security Life Cycle Approach, June 2014. http://dx.doi.org/10.6028/NIST.SP.800-37r1
[IDMgmt] Backend Attribute Exchange (BAE)
v2.0 Overview
https://gsageo.force.com/IDM/servlet/fileField?entityId=ka0t0000000TNIkAAO&field=File__Body__s
[OAuth2] Hardt, D., Ed., "The OAuth
2.0 Authorization Framework", RFC 6749, DOI 10.17487/RFC6749, October
2012, <http://www.rfc-editor.org/info/rfc6749>.
[OMB M-04-04] Joshua B. Bolten, U.S.
Government Office of Management and Budget,
E- Authentication Guidance for Federal Agencies, December 2003.
http://www.whitehouse.gov/sites/default/files/omb/memoranda/fy04/m04-04.pdf
[OpenID.Core] Sakimura, N., Bradley, J., Jones, M.,
de Medeiros, B., and C. Mortimore, “OpenID Connect Core 1.0,” August 2015. http://openid.net/specs/openid-connect-core-1_0.html
[SAML2] OASIS Standard, Security Assertion
Markup Language (SAML) Version 2.0, 2 December 2009. https://www.oasis-open.org/committees/download.php/35711/sstc-saml-core-errata-2.0-wd-06-diff.pdf
[SAMLAC] OASIS Standard, Authentication
Context for the OASIS Security Assertion Markup Language (SAML) Version 2.0, 15
March 2005. https://docs.oasis-open.org/security/saml/v2.0/saml-authn-context-2.0-os.pdf
[X.1254] Recommendation ITU-T X.1254
(09/2012). ITU Telecommunication Standardization Sector (ITU-T) Entity
authentication assurance framework. http://handle.itu.int/11.1002/1000/11608
[X.1255] Recommendation ITU-T X.1255
(09/2013). Framework for discovery of identity management information. http://handle.itu.int/11.1002/1000/11951
This document, the fourth deliverable of the OASIS Trust
Elevation Technical Committee, builds on the work of the first three. To recap:
the first deliverable, Survey of Methods of Trust Elevation Version 1.0 [trust-el-survey-v1.0],
consists of a broad overview of current and near-future online trust elevation
techniques used for (or capable of) elevating a relying party’s assurance that
the user requesting access to its resources is actually the person he or she
claims to be. The second deliverable, Analysis of Methods of Trust Elevation
Version 1.0 [trust-el-analysis-v1.0], evaluated how each of the
identified trust elevation mechanisms operated and what threats they mitigated
that added to the relying party’s confidence in the identity asserted. A
discussion of the methodology used to analyze the identified mechanisms has
been included in that deliverable. The third deliverable, Electronic
Identity Credential Trust Elevation Framework Version 1.0 [trust-el-framework-v1.0], is an abstraction intended
to help to develop applications conforming to an accepted way of elevating
trust of a digital identity.
As has been the pattern for this TC’s deliverables, this
fourth deliverable builds on the work of the first three and specifies design
considerations, implementation considerations and metadata for the elevation of
trust through increased identification.
Trust Elevation Methods are used to increase assurance in
entity identification using authentication events and related entity
information for the purpose of risk mitigation when making access control
policy decisions.
The goals of this Fourth Deliverable are:
·
To propose simple Trust Elevation
architectural patterns demonstrating the use of Trust Elevation in modern
Access Control architectures.
·
To describe a common metadata set,
mechanisms and protocol elements for Trust Elevation information exchanges.
·
To promote the use of Trust Elevation
elements to facilitate standardization among the many technologies and
approaches currently in use for credential & authentication risk
mitigation.
This section is non-normative.
As described in Electronic
Identity Credential Trust Elevation Framework Version 1.0, the following depicts
the core model for Trust Elevation.
While the flow diagram above is easy to understand, implicit
in the core model are several key components and processes, as shown in Section
4.2. The first of these is a component which functions as a policy engine
capable of consuming the asserted user data and making a determination as to
whether that data satisfies the resource’s policy for authentication risk
mitigation. The resource manager must have previously performed a risk
assessment and adopted a risk mitigation strategy ([NIST RMF] and [ISO
ISMS] are examples of methodologies for these antecedent steps).
The second key component is again an antecedent service
generated during the risk assessment and mitigation process. It is composed of a
capability to recognize which, if any, risks have been adequately mitigated by
the initial transaction, which risks remain to be mitigated and preferred
methods for satisfying the remaining needs.
The third key component is a component for evaluating the
success of the trust elevation transaction. This could be an iteration of the
first component, but it has been broken out in the above graphic to clarify the
decision flow.
While these components are necessary to implement trust
elevation of a presented online identity, they require the resource manager to
have engaged in prior planning and assessments in order to generate the
information necessary to direct the behavior of the components. In addition to
implementing recognized, standards-based risk assessments, the prior work of
this Technical Committee provide the necessary guidance for populating the
decision-making components of the core model as well as most comparable models.
Trust Elevation methods are used to increase confidence in
entity identification using authentication events and related entity
information for the purpose of increased risk mitigation when making access
control policy decisions.
Levels of Assurance models are structured such that
increased risk mitigation results in increased credential or identity assurance
level trust. These models require determination of a given transaction’s
identity and authentication risk, expressed in terms of level of assurance
requirements. Policies are designed such that credential or identity assurance
level must meet or exceed the transaction’s level of assurance requirement.
As described in Electronic Identity Credential Trust
Elevation Framework Version 1.0, entity identification confidence may be
increased by: mitigating an authentication threat not addressed by the original
authentication exchange; improved mitigation of the original authentication
threat, or examination of contextual or environmental factors to corroborate
the existing identification.
The definition of the composition of a particular assurance
level scheme, and related policy evaluation criteria, is the responsibility of
the parties involved in the transactions. The scheme should be tailored to the
risk tolerance and requirements of the relying party. In other words, it is up
to the resource manager to determine when sufficient mitigations of risk have
occurred.
Another way to look at Trust Elevation is as a species of transaction
or access control authorization. From this perspective, evaluation of the
current state versus policy requirements results in decisions to ‘Permit’,
‘Deny’, or ‘Require Elevation’.
The Trust Elevation core model is compatible with other
published authorization models, such as: Attribute Based Access Control (ABAC) [NIST800-162],
User Managed Access ([UMA]), [OAuth2], [XACML3], and SAML
Backend Attribute Exchange.
This section illustrates how Trust Elevation would fit into
an Attribute Based Access Control model.
[NIST SP800-162]
describes the elements of an Attribute Based Access Control Model.
As shown in the figure below,
the primary components of Authorization Services are the Policy Enforcement
Point (PEP) which intercepts resource requests; and, the Policy Decision Point
(PDP) which checks supplied attributes versus access control policy. The PDP can
obtain additional attributes from environmental conditions, Policy Information
Point (PIP) and other sources. Based on the policy evaluation, the PDP
instructs the PEP to permit or deny access to the resource.
In the diagram below, when
the Authorization Services determine that Trust Elevation is required, the
Trust Elevation Services take information from “Authentication Services” and
“Risk-Based Engine” to evaluate what Trust Elevation Method should be used to
achieve the desired result.
The User-Managed Access protocol (UMA)
defines a mechanism for a policy enforcement point – known as the resource
server – to delegate authorization of a requesting party to a policy decision
point – known as the authorization server – using elements of the OAuth 2.0
authorization framework.
To gain access to a protected resource, an
UMA client (web or mobile application operating on behalf of a requesting
party) must present a valid access token, called a requesting party token
(RPT), to the resource server. The RPT must be valid and associated with
sufficient authorization data, issued through a trust elevation process, before
the resource server can grant access.
The authorization server, guided by
policies set by the owner of the protected resource, elevates trust by testing
whether the requesting party meets the policies. As part of this process, it could
demand that the requesting party (or the client on their behalf) provide
claims, such as identity information or even promises to adhere to constraints
set by the resource owner, such as an embargo on information release until a
certain date.
One policy the authorization server can
consider is what mechanism was used to authenticate the person. UMA doesn’t
require use of any particular authentication protocol, but works especially
well with OpenID Connect.
The OpenID Connect Core specification
defines two claims in the ID Token format called acr and amr, which
provide details about what type of authentication was performed. Their values
can be defined by a domain, a federation, a global registry, or some other
trust framework. An UMA authorization server can test a requesting party
against policies to evaluate the sufficiency of the authentication mechanism as
provided in values of these claims.
In the event that the mechanism was not
sufficient, the authorization server can indicate the reason for the
authorization failure and what type of credentials would satisfy the policy. At
this point, the client can request re-authentication from the OpenID Provider
and ultimately re-request the RPT token. This flow would constitute trust
elevation by step-up authentication.
The
eXtensible Access Control Markup Language (XACML) standard defines a reference
architecture for Attribute-Based Access Control
(ABAC), a language for expressing access control rules and policies, and a
protocol for generating and processing access control requests and returning
responses.
Access to resources is mediated by a Policy Enforcement Point (PEP), which relies on decisions from a Policy
Decision Point (PDP). When a user attempts to access a protected resource, the
PEP assembles a request, which provides attributes about the user, the
resource, the environment, and the action requested.
The PEP communicates the request to the PDP, which evaluates it according to
pre-defined policies.
To perform Trust Elevation, the access control policy can
specify how users must be authenticated, including parameters such as
authentication method, credentials accepted, and
levels of assurance. Trust elevation in this context means enhancing
authentication and/or authorization by means of requiring additional
attributes.
Consider the following example: a user requests access to
a protected resource. The access control policy
governing the resource requires multi-factor authentication using a strongly
vetted identity credential by means of setting the MustBePresent attribute to
TRUE. The PEP controlling access to the resource has only hitherto validated the user identity by means of a lower assurance
username/password combination. When the PEP initially formulates the request,
it bases the user identity attribute on the previous username/password
authentication event. When the PDP receives the request,
it evaluates the request according to the appropriate policy, based on the
resource. Since MustBePresent = TRUE, the PDP renders an “Indeterminate” decision, with
a status code of “urn:oasis:names:tc:xacml:1.0:status:missing-attribute”. Upon
receiving this “Indeterminate” with
MissingAttribute status decision from the PDP, the PEP may resubmit a request
after acquiring the proper attributes. In this case, the proper attributes
could only be gathered through a step-up authentication event. This sequence constitutes a sample Trust Elevation event.
Alternatively, security administrators and resource owners
may devise a series of Boolean attributes to test for authentication methods
used, i.e.:
subject-id-authenticated-by-password
subject-id-authenticated-by-smart-card
subject-id-authenticated-by-biometric-iris-scan
subject-id-authenticated-by-biometric-fingerprint
subject-id-authenticated-by-two-factors
subject-id-authenticated-by-three-factors
This would allow policy authors to specify which methods
are acceptable by testing for a TRUE result among the
list they define as meeting security requirements.
Lastly, the Obligation element of XACML could be used to
perform Trust Elevation. Any rule that permits access and specifies the
authentication level required would add an obligation
stating the minimum required authentication level. e.g.,
if “User authorized” then Permit.
FulfillOn=Permit -> authenticated-by-two-factors-obligation.
In this case, the PEP does not need any special
attributes. It makes a normal authorization request.
If the response is Deny or NotApplicable, then the authentication level is
irrelevant because the user is not allowed access. If the response is Permit
without any authentication level obligations, then access is allowed even at the lowest authentication level. If the response is
Permit with specific authentication level obligations, then the PEP
must perform step-up authentication to the authentication level of the highest
level of the obligations it received. If the highest
level is satisfied, then any lower levels are satisfied. If that step-up fails
or cannot be attempted, then access is denied. If step-up succeeds then access
is allowed without needing an additional authorization request.
The Security Assertion Markup Language (SAML) standard
defines a means for representing authentication events between different
trusting security domains. A SAML assertion may contain a variety of
attributes about the requesting subject and the
conditions of the authentication event. Subject and Issuer attributes
generally relate the name of the subject and the name of the organization with
which the subject is associated in the AuthenticationStatement element. The
AuthenticationStatement also contains an
AuthenticationContext attribute, which details how the subject was
authenticated in the context of the current assertion.
SAML-aware relying party applications can request
additional attributes via the AttributeQuery element. Moreover, SAML authorities can request full attribute evaluations via the
AuthzDecisionQuery element. Relying parties may specify acceptable
authentication methods and credentials by using the RequestedAuthnContext
element, and can force a fresh authentication event by setting ForceAuthn to true.
Trust Elevation can be exemplified in the following
scenario using SAML: a user attempts to access content protected by a
SAML-aware relying party (RP) application. The user posts a SAML assertion
containing Subject/Issuer attributes and indicates a
low level assurance authentication event to the RP. The RP’s access
control policy requires additional attributes and a higher strength credential
and authentication event. The RP initiates a SAML authentication request to
the user’s home domain. This forces a step-up authentication event
and retrieval of additional attributes, as required by the attribute contract.
As with the XACML model, trust elevation means enhancing authentication and/or
authorization by means of requiring additional
attributes.
This section is normative.
The participants, authentication methods, communication
protocols and authorization methods of the Trust Elevation system MUST be
agreed upon among the participants.
If new participants and/or methods are introduced to the
Trust Elevation system, appropriate onboarding processes MUST be used.
The lack of generally agreed-upon criteria and evaluations
of an authentication method’s efficacy to counter threats, mitigate impacts or
reduce negative occurrence frequency, as well as local extrinsic concerns makes
dynamic addition of new authentication methods problematic. One Trust Elevation
system may consider a password-based authenticator to be sufficient for
identification whereas another Trust Elevation system may require additional
fraud detection infrastructure to realize the same degree of sufficiency.
The Trust Elevation system MUST use business rules and
technologies related to authentication and authorization for performing trusted
transactions that are shared among participants. A Trust Elevation system could
refer to: federated systems; systems controlled by a single governing entity;
or a single system.
This section is normative.
There are several assumptions that help set the context for
this work:
·
The resource manager MUST have a defined set of requirements for
authentication and/or authorization control. The requirements MAY include
combinations of static rules and dynamic risk evaluations.
·
In the case of federated services, the federation agreement MUST
define the available identification and authentication methods and their
relationship to discrete ‘levels’ of assurance that map to risk mitigation or
compensating controls.
·
Authentication methods MUST be described sufficiently to allow
creation of sets of compatible methods that cover identifiable risks or threats
to allow implementers to choose independent authentication factors.
This section is normative.
There are many potential factors that influence the design
specific Trust Elevation architectures. The nature and impact of the factors is
determined by local requirements.
The semantics of combining authentication methods to
increase risk mitigation MUST be dependent on local definition of
authentication method characteristics within a Trust Elevation system.
The risk models of the resource manager and/or federation
that comprise the Trust Elevation system MUST be considered when defining how
combinations of methods modify risk mitigation.
For example, in one federation repetition of a password
authentication to re-confirm the authenticator may change the risk mitigation
from ‘Low’ to ‘Medium’. In a different federation, the same risk mitigation
change might require a second authentication method which is different from the
first one used.
The full range of permitted combinations and their effect on
risk mitigation SHOULD be defined for the local entities.
As with authentication method combinations, the
specification of each permitted authentication method MUST be shared within a
Trust Elevation system.
Note: if a fingerprint template biometric is to be used,
common specification of sampling mechanics, template calculation and comparison
algorithms is essential. Variance in specification within a Trust Elevation
system will result in different semantic meaning when combining authentication
methods.
Authentication state per Subject MAY need to be kept.
The Trust Elevation system MAY need to know which
authentication methods have been attempted in prior transaction attempts in
order to select the a different authentication method or factor to be attempted
next.
Tracking state per Subject and transaction attempt may prove
to be a complex undertaking unless care is taken when designing elevation
policy.
The architecture and design SHOULD be able to accommodate local,
remote and distributed policy evaluation. Policy evaluation for trust elevation
purposes may occur within a single system, or may occur in several different
systems then combined.
A mechanism for calculating the combined result of the
policy evaluation MUST be designed.
When designing the state model for the authorization system,
time-related degradation of information quality or authenticator validity
SHOULD be considered. The degradation COULD be defined as nil, or according to
a specified time function.
The effect of changes in the threat environment might cause
changes of calculated assurance levels. Designers SHOULD determine if and how
to respond to changes to the threat environment.
If a system component is observed to be under active attack,
the authorization system SHOULD require increased assurance levels through use
of additional authentication methods.
4.4
Trust Elevation Architecture Components
This section is normative.
The following architecture diagram shows Trust Elevation
system components and other components related to Trust Elevation systems and
their core functions. The dashed line boxes represent the boundary for each
major component. The solid line boxes represent the functions within the major
components. In other authorization model representations, the functions may
have different names and may possibly appear within different major component
boundaries.
The Trust Elevation Services component is comprised of the
Trust Elevation Method Determiner and the Trust Elevation Method Repository.
When the Authorization Services component determines that
the Subject is not permitted to access the resources due to insufficient
identification and authentication assurance, the Trust Elevation Services
component is used to select an additional authentication method or methods
which would allow the Subject to access the resources.
The Trust Elevation Services Component enables the
Authorization Services to ask the Subject to retry access using different or
additional authenticators.
The Trust Elevation Services are aware of the methods and
authenticators previously used by the Subject to attempt access. This enables
mitigation of identification threats different from the initial authentication
methods and authenticators, without having to hard code all combinations of
authenticators that could be used.
For example, if the initial authenticator used
username/password (a ‘know’ factor), the Trust Elevation Services would not
recommend that authenticator if asked for another single factor authenticator:
it might return a ‘have’ or ‘are’ factor authentication method, or a ‘know’
factor authentication method that is not username/password.
The Trust Elevation Method Determiner makes Trust Elevation
policy decisions.
It receives requests from the Authorization Services
component that MUST include current authentication state information of the
Subject and the desired Level of Assurance.
The Trust Elevation Method Determiner uses policies stored
in the Trust Elevation Method Repository to determine which, if any,
authentication methods could be used to achieve the desired Level of Assurance.
The Trust Elevation Policy MUST map the combinations of
authenticators to the desired assurance levels.
Given the desired assurance level, the Trust Elevation
Method Determiner MUST be able to evaluate Trust Elevation Policy to identify
the list of authentication methods that could be used to achieve the desired
assurance level.
The current authentication state information MAY include
data about: authenticators presented to the Authorization Services component;
authentication methods that were used by the Subject to achieve the current
authentication state; and, the current Level of Assurance of the Subject.
If the authentication capabilities of Subjects (user, device
or client) are dynamic or dependent on device, user or software abilities and
features, the Method Determiner MAY need information about the specific
capabilities of the specific Subject in order to avoid unnecessary round trips
to the Subject.
The Trust Elevation Method Repository contains information necessary
to the functions of the Trust Elevation Method Determiner.
The Trust Elevation Method Repository MUST contain
information about the implemented authentication methods and their
characteristics. These characteristics are used in the Trust Elevation Policy
when the concepts of ‘stronger’ authenticators or ‘more’ assurance are
represented.
If the Trust Elevation system uses authentication factors to
determine authenticator strength, it COULD treat a single factor authenticator
as weaker than a two-factor authenticator. In this case the characteristics
SHOULD include details of which authentication factors are used.
4.5 Other Architecture Components
This section is non-normative.
These components interact with Trust Elevation systems but
are not part of the Trust elevation systems.
The Authorization Services component must be capable of
requesting and processing Trust Elevation information. Trust Elevation Services
may be treated as an information source or a remote policy engine.
The Authorization Services component may need additional
functionality to handle and track multiple access attempts by the Subject as
the Subject responds to elevation requests.
If a Risk-Based Engine Component exists, it represents
systems that may be used by the resource manager to detect, measure and respond
to threats in the operational environment. Detection of increased online
attacks could cause the resource manager to require a greater degree of
identification or authentication for access to resources.
This section is normative.
Orchestration of Trust Elevation systems interaction with
access control system components is required.
The access control components MUST be capable of requesting
additional authentication or information from the Subject.
Since the Trust Elevation services component determines
which authentication methods are required after the first round of policy
evaluation, all components in the access control service MUST be able to handle
the extra requests.
The implemented authentication methods MUST be enumerated
and details stored in the Trust Elevation Repository.
The details that SHOULD be captured are identified in
Deliverable 2, comprised of threats eliminated and risks mitigated. The
detailed information will enable analysts to design Trust Elevation sequences
that use complementary authentication methods to strengthen risk mitigation.
Authentication methods recorded in the Trust Elevation
Method Repository MAY involve any combination of User, Device and Client.
Because the Subject might interact with the Authorization
Services at different points in time with different User, Device or Client
elements, authentication methods MUST NOT make assumptions about the
relationships between the Subject, User, Device or Client.
Note: the same User attempting access from a different
device that has an identical device model has lower assurance than use of the
originally registered device. Authentication methods involving the device need
to be able to differentiate between those devices.
Devices may have different authentication method
capabilities at enrolment versus at the time of the transaction. Device
hardware used for authentication SHOULD NOT be assumed to be available or
functioning.
Enrolment is a key phase to support execution of Trust
Elevation. At enrollment time, the Trust Elevation system MUST identify, record
and possibly provision authentication methods. These authentication methods
COULD include user, device, geo-location, network location and environmental
elements.
This section is non-normative.
The specific structure and content of the Policy Table and
Methods Table are defined within the Trust Elevation system, driven by the Relying
Party’s authentication policies.
In this simple example, a static mapping of Relying Party
defined Transaction Risk Levels to pre-defined authentication strengths encoded
as “Authentication Levels” (AL) is shown. The Relying Party defines which
Authentication Level transitions are required for each Transaction Risk Level.
The policies are based on the ‘authentication factors’
approach to risk mitigation. The Relying Party policy sets out the permitted
combinations of authentication factors required to move from one Authentication
Level to another Authentication Level.
Note that all transitions for all risk levels are not
necessarily defined. The Policy Table only shows valid policies for this
Relying Party within this trust system. If a particular transition is not
defined, it is deemed to be invalid.
A bank customer (Subject) initially logs on to the bank site
(through a browser or mobile app) to view their account balance. Then, they
decide to perform a higher risk transaction that requires a higher level of
authentication: a funds transfer of $X.
- Subject has an existing relationship with the bank (i.e.,
is an account holder)
- Subject has previously registered their authentication
methods (e.g., password, device, biometric)
- There are three Authentication Levels defined by the bank
(the Relying Party)
|
Transaction Designation
|
Transaction Name
|
Transaction Risk Level
|
|
T1
|
Check Account Balance
|
Low
|
|
T2
|
Transfer Funds Out
|
Med
|
The Policy Table is defined
during system design by the Relying Party.
|
Transaction
Risk Level
|
Initial Strength
|
Desired Strength
|
Authentication needed*
|
Policy designation
|
|
Low
|
AL0
|
AL1
|
One factor, either what you
know or have
|
P1
|
|
Med
|
AL0
|
AL2
|
Two factors, any class
|
P2
|
|
|
AL1
|
AL2
|
One factor, different than
used for AL1 authentication
|
P3
|
|
High
|
AL0
|
AL3
|
Three factors
|
P4
|
|
|
AL1
|
AL3
|
Two factors, any class,
different than used for AL1 authentication
|
P5
|
|
|
AL2
|
AL3
|
One factor, different than
used for AL1 OR AL2 authentication
|
P6
|
Where AL0 represents a "user not logged in" state.
*Authentication policies are set by the relying party.
The Methods Table enumerates
the authentication methods available in the trust system.
|
Method designation
|
Method description
|
Class(es)
|
SF strength
|
Threats addressed*
|
|
M1
|
PIN (>=4 char)
|
Know
|
1
|
|
|
M2
|
Password (>=8char)
|
Know
|
1
|
|
|
M3
|
Device ID
|
Have
|
1
|
|
|
M4
|
Crypto key (TLS protocol)
|
Have
|
2
|
|
|
M5
|
Biometric – face
|
Are
|
NA
|
|
|
M6
|
Biometric – fingerprint
|
Are
|
NA
|
|
|
M7
|
PIN + Device ID
|
K+H
|
2
|
|
|
M8
|
Crypto key + face
|
H+A
|
3
|
|
*For the benefit of relying
party operators setting up policies.
Note: In the process flow the
PEP is not shown and is assumed to be part of the Resource.
Title
Transaction 1: Check Account Balance
Note
over Subject: Initial State \nLoA0 Not-logged-in
Subject->Bank
Site\nResource: CheckAccountBalance(T1)
Bank
Site\nResource->Authorization\nService: Evaluate Subject authorization for
T1
Authorization\nService->
Authorization\nService: Look up T1 Access Policy
Note
over Authorization\nService: Policy P1 selected
Authorization\nService->
Authorization\nService: Evaluate P1
Note
over Authorization\nService: 'Subject Not Authorized'\n(Not Authenticated)
Authorization\nService->
Authorization\nService: Select Method from \nlist of Methods for P1
note
over Authorization\nService: Method M2 selected
Authorization\nService-->Bank
Site\nResource: 'Not Authorized', Try M2 Method
Bank
Site\nResource-> Authentication\nService: Authenticate Subject with M2
Authentication\nService->Subject:
Prompt for M2 \n (UserID + Password)
Subject-->
Authentication\nService: M2 Authenticator
Authentication\nService->
Authentication\nService: Verify Authenticator
Authentication\nService-->
Bank Site\nResource: 'Subject is Authenticated with M2'
Note
over Subject: Subject State \nLoA1 using M2
Bank
Site\nResource-> Authorization\nService: Evaluate Subject authorization for
T1
Authorization\nService->
Authorization\nService: Look up T1 Access Policy
note
over Authorization\nService: Policy P1 selected
Authorization\nService->
Authorization\nService: Evaluate P1
note
over Authorization\nService: 'Subject is Authorized' \n(M2 satisfies P1)
Authorization\nService-->Bank
Site\nResource: 'Subject is Authorized'
Bank
Site\nResource-->Subject: Access Granted, Transaction Proceeds
Note: In the process flow the
PEP is not shown and is assumed to be part of the Resource.
Note: In the process flow the
PEP is not shown and is assumed to be part of the Resource.
Title
Transaction 2: Transfer Funds Out
Note
over Subject: Subject State \nLoA1 Using M2
Subject->Bank
Site\nResource: TransferFundsOut(T2)
Bank
Site\nResource->Authorization\nService: Evaluate Subject authorization for
T2
Authorization\nService->
Authorization\nService: Look up T2 Access Policy\nPolicy P3 selected\nEvaluate
P3
Note
over Authorization\nService: 'Subject Not Authorized'\n(M2 Insufficient for P3)
Authorization\nService->TE
Method\nDeterminer: Determine List of Methods for P3\n{CurrentLoA, TargetLoA,
CurrentLoAContext}
TE
Method\nDeterminer->TE Method\nRepository: Look Up List of Methods to try
TE
Method\nRepository-->TE Method\nDeterminer: List of Methods\nto go from
CurrentLoA to TargetLoA\nwithin CurrentLoAContext
TE
Method\nDeterminer-->Authorization\nService: List of Methods
Authorization\nService-->Bank
Site\nResource: 'Not Authorized',\nTry one from List of Methods
Bank
Site\nResource-> Authentication\nService: Authenticate Subject using a
selection from List of Methods
Authentication\nService->
Authentication\nService: Select Method to try from \nlist of Methods
note
over Authentication\nService: Method M6 selected\n(Biometric-Fingerprint)
Authentication\nService->Subject:
Prompt for M6
Subject-->
Authentication\nService: M6 Authenticator
Authentication\nService->
Authentication\nService: Verify Authenticator
Authentication\nService-->
Bank Site\nResource: 'Subject is Authenticated with M6'
Note
over Subject: Subject State \nLoA2 using M2+M6
Bank
Site\nResource->Authorization\nService: Evaluate Subject authorization for
T2
Authorization\nService->
Authorization\nService: Look up T2 Access Policy\nPolicy P3 selected\nEvaluate
P3
Note
over Authorization\nService: 'Subject Is Authorized'\n(M2+M6 Sufficient for P3)
Authorization\nService-->Bank
Site\nResource: 'Subject is Authorized'
Bank
Site\nResource-->Subject: Access Granted, Transaction Proceeds
Note: In the process flow the
PEP is not shown and is assumed to be part of the Resource.
This section is non-normative.
7.1 Component-Component
Communications
Content of Authorization Service (PDP) to Trust Elevation
Method Determiner request:
- Current Authentication Level
- Method(s) that were used to achieve current Authentication
Level
- Target Authentication Level
Content of Trust Elevation Method Determiner to
Authorization Service (PDP) response:
- List of methods that could be used to achieve target
Authentication Level
Content of Authorization Service (PDP)-Authentication
Service request:
- Subject ID
- List of methods to choose from
The fragments below are examples showing the kinds of
information to exchange between components.
<trustel:MethodTypeRequest>
<trustel:CurrentLoA>....</trustel:CurrentLoA> //current
Authentication Level in numerical value
<trustel:TargetLoA>...</trustel:TargetLoA> //Target
Authentication Level in numerical value
<trustel:CurrentLoAContext>
<trustel:Method>...</trustel:Method>
//could be "|" delimited array of methods
<trustel:AuthnDeviceSig>..</trustel:AuthnDeviceSig> //Device
Fingerprint
<trustel:AuthnLocation>...</trustel:AuthnLocation> //Device
location
<trustel:AuthnIP>...</trustel:AuthnIP> //IP
of the device
<trustel:AuthnTime>...</trustel:AuthnTime> //time
of request
</trustel:CurrentLoAContext>
</trustel:MethodTypeRequest>
<trustel:MethodTypeResponse>
<trustel:Method>...</trustel:Method> //could
be "|" delimited array of methods
</trustel:MethodTypeResponse>
In order to conform with this specification, the Trust
Elevation system under consideration:
[1] MUST be designed and use an architecture that conforms
to the normative statements in section 4
[2] MUST be implemented in conformance with the normative
statements in section 5
This section is non-normative.
The following individuals have participated in the creation
of this specification and are gratefully acknowledged:
Chairs:
Abbie Barbir, Aetna
Don Thibeau, OIX
Editors:
Andrew Hughes, Individual
Shaheen Abdul Jabbar, JPMorgan Chase Bank, N.A.
Peter Alterman, SAFE-BioPharma Association
Abbie Barbir, Aetna
Mary Ruddy, Identity Commons
Document Contributors:
Peter Alterman, SAFE-BioPharma Association
Abbie Barbir, Aetna
Andrew Hughes, Individual
Shaheen Abdul Jabbar, JPMorgan Chase Bank, N.A.
Eve Maler, Forgerock
Steven Legg, Individual
Diana Proud-Madruga, Individual
Mary Ruddy, Identity Commons
Michael Schwartz, Gluu
Martin Smith, Individual
Don Thibeau, OIX
Cathy Tilton, Daon
John Tolbert, Queralt, Inc
David Turner, Voltage Gate
Colin Wallis, New Zealand Government
Technical Committee Member Participants:
Shaheen Abdul Jabbar, JPMorgan Chase Bank, N.A.
Orlando Adams, U.S. Bank
Peter Alterman, SAFE-BioPharma Assn
Abbie Barbir, Aetna
John Bradley, Open Identity Exchange
David Brossard, Axiomatics
Doron Cohen, SafeNet, Inc.
Ed Coyne, Veterans Health Administration
John Davis, Veterans Health Administration
Suzanne Gonzales-Webb, Veterans Health
Administration
Dazza Greenwood, M.I.T.
Richard Grow, Veterans Health Administration
Thomas Hardjono, M.I.T.
Rainer Hoerbe, Individual
Andrew Hughes, Individual
Mohammad Jafari, Veterans Health Administration
Gershon Janssen, Individual
Kevin Mangold, NIST
Carl Mattocks, Individual
Steve Olshansky, Internet Society (ISOC)
Brendan Peter, CA Technologies
Diana Proud-Madruga, Veterans Health Administration
Mary Ruddy, Identity Commons
Anthony Rutkowski, Yaana Technologies, LLC
Marty Schleiff, The Boeing Company
Michael Schwartz, Individual
Shahrokh Shahidzadeh, Intel Corporation
Jeffrey Shultz, NIST
Don Thibeau, Open Identity Exchange
Cathy Tilton, Daon
John Tolbert, Queralt, Inc.
David Turner, Voltage Gate
Colin Wallis, New Zealand Government
This section is non-normative.
One of the core assumptions of Trust Elevation is that a
subject attempting a transaction is unable to meet the policy requirements for
identification certainty unless an Elevation event occurs.
An important concept is that measured assurance levels
change over time due to many factors. At the instant of authorization policy
evaluation, the current state of identity attribute assurance level and
authenticator assurance level are compared to the Transaction’s Assurance Level
Requirement. If the measured assurance levels are greater or equal to the
requirement, the transaction proceeds.
The graphics show that the assurance level of the Identity
Information Attributes established via the Identity Proofing and Verification
processes are separate and unlinked to the assurance level of the Authentication
Event (which includes Credential and Authenticator details). This approach is
consistent with the NIST SP800-63 LOA calculation method.
This graphic illustrates a
scenario where the levels of identity attribute assurance and authenticator
assurance are determined in advance and do not degrade over time.
The vertical dashed lines represent the potential points in
time of the transaction event. The identity attribute assurance and authenticator
assurance levels are compared to the transaction assurance level requirement.
If both values are greater than the requirement, the transaction can proceed
(check mark). If one or both are lower, the transaction cannot proceed (X mark)
and is either rejected or directed to a trust elevation event.
Trust Elevation in this scenario combines authentication
factors to step up combined authenticator assurance to meet or exceed the
transaction requirement.
Notes:
·
The ‘Assurance Score’ is a simple numerical representation of the
degree of certainty for illustrative purposes. ‘Assurance Level 3’ has been
arbitrarily defined as ‘30’ on the scale
·
The Grey line represents the assurance level resulting from the
Identity Proofing and Verification process; established at Subject Registration
time by the Registration Agent.
·
The Black line represents the authenticator assurance level
resulting from the Authentication event. It takes credential, authentication
secrets and authenticator generation factors into account.
·
The Green line represents the Resource Owner defined assurance
score/level required for the transaction. It is based on the Resource Owner’s
risk determination methods. In this example, the Transaction Requirement is
’30’ or ‘LOA3’
·
The Black line initially shows the effect of a single
authenticator, then two authenticators, then three authenticators.
The assurance level of the
Authenticator is important. This graphic illustrates a scenario where the
authenticator assurance level changes over time due to time-based degradation
of the credential, secrets and authenticator generation processes.
The vertical dashed lines represent the potential points in
time of the transaction event. The identity attribute assurance and
authenticator assurance levels are compared to the transaction assurance level
requirement. If both values are greater than the requirement, the transaction
can proceed (check mark). If one or both are lower, the transaction cannot proceed
(X mark) and is either rejected or directed to a trust elevation event.
This scenario shows that due to rapid degradation of
authenticator assurance for most time periods, Trust Elevation to three
authenticators is needed for the transaction policy.
Notes:
·
The ‘Assurance Score’ is a simple numerical representation of the
degree of certainty for illustrative purposes. ‘Assurance Level 3’ has been
arbitrarily defined as ‘30’ on the scale
·
The Grey line represents the assurance level resulting from the
Identity Proofing and Verification process; established at Subject Registration
time by the Registration Agent.
·
The Black line represents the authenticator assurance level
resulting from the Authentication event. It takes credential, authentication
secrets and authenticator generation factors into account.
·
The Green line represents the Resource Owner defined assurance
score/level required for the transaction. It is based on the Resource Owner’s
risk determination methods. In this example, the Transaction Requirement is
’30’ or ‘LOA3’
·
The Black line initially shows the effect of a single
authenticator, then two authenticators, then three authenticators.
·
The downward slopes represent the time-based degradation of
certainty of the authenticator
·
Although not shown explicitly, refresh to original values could
be achieved by re-issuance of credentials, or generation of new keys.
The last graphic illustrates a more complex example in which
the overall threat level affects the Authenticator assurance level. A
simplistic calculation is used where increasing threat environment, increasing
detected fraud and decreased system security subtract directly from the
authenticator assurance score.
This mimics the effect that a risk-based authentication
system or risk engine might have on transaction assurance requirement
evaluation.
As in the previous illustrations, the vertical dashed lines
represent the potential points in time of the transaction event.
Where the increased threat level causes the effective
authenticator assurance level to dip below the green transaction requirement
line, Trust Elevation could be used to achieve the minimums necessary. Note
that in the ‘Two Authenticators’ region, the transaction could proceed or fail
depending on the magnitude of the threat levels. If the transaction fails, the
Relying Party could choose to retry at a later time, or request additional
Authenticators.
This section is non-normative.
|
Revision
|
Date
|
Editor
|
Changes Made
|
|
Draft 04b
|
2015-09-04
|
A. Hughes
|
Minor updates to base draft.
|
|
Draft 04c
|
2015-09-26
|
A. Hughes
|
Added contributions to draft.
|
|
Draft 05
|
2015-10-15
|
A. Hughes
|
Final draft version for TC review and approval.
|
|
Draft 06
|
2015-11-07
|
A. Hughes
|
Minor updates for Final draft version
|
|
Draft 07
|
2015-11-07
|
A. Hughes
|
Minor updates for Final draft version
|
|
Draft 08
|
2016-03-15
|
A. Hughes
|
Disposition of comments received
|
|
Draft 09
|
2016-08-18
|
A. Hughes
|
Updates to Conformance Clause and related items.
|
|
Draft 091
|
2016-09-12
|
A. Hughes
|
Finalization of Conformance Clause
|
|
Darft 092
|
2016-09-14
|
A. Hughes
|
Corrected acknowledgements items
|
