Adhoc-like routing in wired networks with genetic algorithms

Routing of packets in networks requires that a path be selected either dynamically while the packets are being forwarded, or statically (in advance) as in source routing from a source node to a destination. Quality of service (QoS) driven routing has been proposed using a protocol called the “Cognitive Packet Network” (CPN) which dynamically selects paths through a store and forward packet network so as to provide best effort QoS to route peer-to-peer connections. CPN operates very much as an adhoc protocol within a wired setting, and uses smart packets to select routes based on QoS requirements. We extend the path discovery process in CPN to include a genetic algorithm which can help discover new paths that may not have been discovered by smart packets. We describe how possible routes can “evolve” from prior knowledge, and then be selected based on “fitness” with respect to QoS. We detail the design of the algorithm and of its implementation, and report on resulting QoS measurements.