THE PERFORMANCE EVALUATION OF A NEW NEURAL-NETWORK-BASED TRAFFIC MANAGEMENT SCHEME FOR A SATELLITE COMMUNICATION-NETWORK

This paper presents the first attempt to incorporate the map configura tion task with the routing task in the traffic management scheme of a satellite communication network. The map configuration task, which is automated by a self-organization scheme modified from Kohonen's model, consists of three stages. The first stage is the pattern recognition task, which selects an exemplar map that best meets the current networ k requirement. The second stage consists of analyzing the discrepancy between the chosen exemplar map and the state of the network, and adap tively modifying the chosen exemplar map to conform closely to the net work requirement (input data pattern) by self-organization. Based on c ertain performance criteria, whether a new map is generated to replace the original chosen map is decided in the third stage. The routing ta sk is implemented by a state-dependent routing algorithm which arrange s the incoming call to some proper path to make the network more effic ient, and lowers the call block rate. The new scheme inheriting merits of both the map configuration and routing tasks provides the best com promised performance in terms of block rates at various load condition s as compared to status quo approaches schemes with either only the ma p configuration task or only the routing task.