The design of a secure agent platform

This paper contains a description of the enhancement of TEEMA, an extensible, general-purpose mobile agent execution environment, to support secure communications. Software agents are proactive, autonomous programs that are given the option to migrate among supporting platforms. Agents can mimic the behavior of a number of systems, from simple cellular automata to complex handoff sequences in cellular systems. They can also be adopted to represent users and act on their behalf in such environments as economic commerce and online communities. To be used in this capacity, agents must be trusted by users. Unfortunately, agents must often rely on the good faith of the execution environments and must be trusted within any environment they are allowed to reside. This poses a serious security threat when agents, for example, need to establish a transaction that involves the exchange of sensitive information, like, for example, credit card numbers, social insurance numbers, but also telephones, medical information, and other sensitive data. The obvious solution to this problem is to make the channels transmitting such sensitive information safe from tampering and interception. The construction of a secure system starts by building a secure environment for the execution environments. TEEMA is written completely in Java, and this makes it possible to adopt existing security protocols.