Anonymity, Authenticity, and Accountability

Anonymity, authenticity, and accountability (AAA) are three basic security properties of digital information systems. As a systems research lab, our primary interest in this area lies on how to tailor well established crypto models for design of security management architectures, especially for protection of computing resources. Our goal is to provide reasonable protection strengths in response to the threat levels, by harmonizing the strengths of the three properties at the least performance cost.

Many of the Internet based collaborations are fairly autonomous, need based, mission driven. Often organizations have to work together to serve a mission but prefer to remain anonymous, i.e., the notion of anonymous, accountable collaboration. We are investigating security management support for large scale applications that have some of all of these characteristics. Example applications include secure, anonymous workflow management, blinded peer review, policy conformance architecture, and confidential auditing. Many of these applications are involved with the interface and integration of high level system architecture (virtual community, portal, etc.) and document design (e.g., XML). The tradeoff analysis extends from the basic security properties to linkability and traffic patterns dispersing, in order to balance the tradeoff between security properties and performance.

To gain in-depth understanding of these subtle issues we take the electronic cash (E-cash) paradigm as the theoretical bases to investigate system design issues. The off-line transaction model provides an excellent reference architecture for design of distributed systems with high security and privacy requirements. The value of computing resources becomes void with the passing of time, and therefore the classical E-cash algorithms and other similar paradigms need to be tailored for our needs. We are in active development of low-overhead schemes for management of time, transferability, and resource/credit divisibility. In addition to algorithm and protocol designs, a working software prototype is available for both teaching and research purposes.

We propose a timed zero-knowledge proof (TZKP) protocol for session-based access control of computing resources. TZKP is derived from the Eng-Okamoto disposable authentication protocol (DAP) (originally designed for N-divisible E-cash,) by embedding session numbers into its cryptographic constructs. Tokens are accounted for in sessions to reduce the cost of tracking spent tokens and token withdrawals significantly.

Transfer of privileges or responsibility is a basic operation in many distributed applications, e.g., authority delegation, workflow, etc. We propose the notion of multi-source reusability (MSR) to distinguish multiple (legal) transfers of a token from the (illegal) reuse of a token. MSR is constructed by manipulations of the control variables in Eng-Okamoto's DAP. It is proven resistant to the collusion attacks of principals engaged in the transfer process.