System with Database for Evaluating Routing Algorithms in Computer Networks

This paper concerns a complex experimentation system for evaluating routing algorithms. The optimisation problem consists of choosing the route in a computer network, in order to find a path with the lowest cost and to balance traffic load, but cost changes affected by transmissions and other factors are taken into consideration. Three algorithms are presented: a fixed routing method with additional paths (DRMAP), an adaptive distance-vector routing (DVR) and the proposed by authors, an adaptive method (MQR) based on reinforcement learning approach. The properties of the algorithms and opportunities of the designed experimentation system with database are shown using examples. It is demonstrated that the proposed simulation model retains the relation between the applied routing algorithms and network environment, as in real networks.