Agent-based communication security

This paper presents an application of agent technology to the field of communication security. The proposed method, Split and Merge, is presented. The generic approach is on the denying of access to the message, instead of ciphering the message at the source and then sending it over a communication channel. This is achieved by splitting the message in parts and sending them to the destination through different routes, in an ad-hoc network of cooperating agents. The implementation of the method using multi-agent systems is detailed. The solutions found to deal with error detection and correction (loss of message fragments, message integrity and node misbehavior) are then discussed. An emphasis is put on the use reputation and trust mechanisms, with a fuzzy approach, to detect and cope with nodes with different levels of reliability.