Notarized Federated Identity Management for Web Services
Michael T. Goodrich Roberto Tamassia Danfeng Yao
Outline
Introduction to federated identity management (FIM)
Notarized federated identity management model and protocol
STMS and its application in notarized FIM
Identity theft and proposed countermeasure
Motivation
Digital identity management (DIM)
To protect sensitive personal information in on-line transactions
Users tend to choose weak passwords
As the number of passwords to remember increases
Single sign-on (SSO) and federated identity management
A user logs in only once to a site, then is automatically authenticated
Cookie-based SSO approach (used by Microsoft Passport)
Does not support cross-domain single sign-on
Approach using cryptographic-enabled assertions
Secure Assertion Markup Language (SAML)
SSO and FIM
Provider model in SAML
Specially designed for general cross-domain single sign-on
Identity Provider (IdP)
IdP is the system that asserts information about a subject
Service Provider (SeP)
SeP is the system that relies on the information supplied to it by the identity provider
Relying party
Used in Liberty Alliance Federated Identity Management for single sign-on
Identity Federation
Websites of different admin domains need to trust each other’s access control verdicts
Circle of trust
Issues
How to securely maintain the identity federation when members may leave or join the circle of trust?
How to provide separation of IdP and SeP for the privacy protection of the user?
These questions have not been extensively studied
Existing SSO solutions assume pre-established trust relationship among providers
IdP and SeP communicate to each other during SSO process
Notarized Federated Identity Management
We introduce a trusted third-party, called notary server
The notary information of an assertion provides its trustworthiness
Distributed implementation of the notarized federated identity management framework using STMS
We also present a robust authentication protocol that is resilient against identity theft attacks, using identity-based encryption
Notarization
Notary server
Third party trusted by identity providers and service providers
Notarizes assertions submitted by identity providers
Answers queries on notarized assertions asked by the service providers
Prevents direct communication between the identity provider and the service provider
Notarized assertion
Generated by identity provider
Authenticated by notary server
Trusted by service provider
Security Requirements
Security
A polynomial-time adversary cannot forge a valid notarized assertion
Secrecy
Notarized assertion should not leak sensitive information of a user to unauthorized parties, including the notary server
Accountability
Identity providers should be held accountable for the assertions that they generate; and for any unauthorized information disclosure about the user
Overview of Notarized FIM
Protocol Design Challenges
How to protect the identity of the user from the service provider?
How to blind the content of an assertion from the notary server?
How to unblind by the service provider?
How to hold the identity provider accountable for unauthorized disclosure?
Our solution uses lightweight crypto primitives
hash function
XOR
symmetric encryption
digital signature
Implementation of Notarized FIM Protocol
Two public parameters P1, P2
The user and SeP compute a session_ID N
XOR each party’ s random string
The user requests IdP to generate assertions
Signed request to IdP for accountability
IdP blinds an assertion
Computes the hashed_ID h = Hash(N, P1)
Generates an assertion S using h for index
Computes the blinding factor K = Hash(N, P2)
Encrypts S with K using a symmetric encryption scheme
Blinded assertion is called S’
IdP submits an assertion to the notary server
Sign S’ with its private key
Notary server stores S’ , and stores the signature for accountability
Implementation of the notarized FIM protocol (Cont’ d)
The user queries for an assertion of a hashed_ID
Computes the hashed_ID h = Hash(N, P1)
Queries the notary server for assertions of h
Notary server notarizes an assertion
Retrieves the blinded assertion S’
Signature approach: Signs S’ with its private key
STMS approach: computes the proof for S’
The user unblinds and verifies an assertion
The user verifies the notary information
Computes the blinding factor K = Hash(N, P2)
Decrypts S’ with K and obtains S
Detect unauthorized information disclosure
The service provider unblinds and verifies the assertion
Privacy and Accountability
Notary server realization: STMS
The Secure Transaction Management System [Goodrich, Tamassia et al.] implements an authenticated dictionary
STMS in Notarized FIM
Outline of the talk
Introduction to federated ID
Provider model in SAML
Notarized federated identity management model and protocol
Identity theft and countermeasure
An authentication protocol robust against identity theft
Identity theft causes
Private information insecurely stored and entered
On credit card company’ s computers, in DMV’ s cabinets, in your bank, in your trash can
How to proactively control the release of your private information?
Secure storage
Prevent dumpster diving
Safe disclosure
Prevent shoulder surfing
With minimal changes to current financial and administrative infrastructure
Existing approaches
Centralized processing
Heavy-weight Zero-knowledge proofs
Our approach
To design a lightweight authentication protocol using identity-based encryption
Related work
Anonymous credentials [Camenisch Lysyanskaya 01] [Camenisch Herreweghen 02]
Federated ID management models [Camenisch et al 05] [Bhargav-Spantzel Squicciarini Bertino 05] [Pfitzmann Waidner 03]
Web service framework [Bonatti Samarati 02]
Identity theft detection [van Oorschot Stubblebine 05]
Identity-based encryption [Boneh Franklin 01]
SAML [OASIS], WS-Federation [IBM et al]
Conclusions
Notarized federated identity management is a solution for establishing trust in web services
Notary server provides an anchor of trust
Notarized FIM protocol provides accountability, privacy, and secrecy for participants
IBE-based credentials and exchanges hold promises for identity theft solutions
Acknowledgements
David Croston at IAM Technology, Inc
Generations of the model