The research considers agent-based systems, and develops techniques to support the development of reliable agent-based applications in large and open distributed systems. Such applications are important for the growth of electronic commerce, and Internet computing.
The main goals of the research is to provide a framework for secure, fault-tolerant, and heterogeneous agents and to study fundamental limitations on such a framework. The desired agent characteristics are achieved automatically with little or no modification to the source code. The research uses compiler techniques to provide the ability for applications to save and restore their execution states in a heterogeneous environment. Code hiding techniques are developed to provide secure agent execution in hostile environments without compromising source code or execution results. Compiler techniques are also used to provide fault tolerance, and new techniques are developed for tolerating failures in heterogeneous environments.
The research is complemented by an educational component, the main goal of which is to attract undergraduate students to do research in computer science.