SINS: A Middleware for Autonomous Agents and Secure Code Mobility

摘要

Building trusted applications is hard, especially in a distributed or mobile setting. Existing methods and tools are inadequate to deal with the multitude of challenges posed by distributed application development. The problem is exacerbated in a hostile environment such as the Internet where, in addition, applications are vulnerable to malicious attacks. It is widely acknowledged that intelligent software agents provide the right paradigm for developing agile, re-configurable, and efficient distributed applications. Distributed processing in general carries with it risks such as denial of service, Trojan horses, information leaks, and malicious code. Agent technology, by introducing autonomy and code mobility, may exacerbate some of these problems. In particular, a malicious agent could do serious damage to an unprotected host, and malicious hosts could damage agents or corrupt agent data. Secure Infrastructure for Networked Systems (SINS) being developed at the Naval Research Laboratory is a middleware for secure agents intended to provide the required degree of trust for mobile agents, in addition to ensuring their compliance with a set of enforceable security policies. An infrastructure such as SINS is central to the successful deployment and transfer of distributed agent technology to Industry because security is a necessary prerequisite for distributed computing.