engleski

Adaptive Routing Strategies Using Learning Automata

Modern end-to-end switching networks have forced us to reconsider the routing schemes that are used in traditional networks. The reexamination is necessary since the choice on the call's route is predefined in these networks and does not dynamically adapt itself to the environmental changes. In these paper we present and evaluate adaptive routing schemes based on the concept of the learning automata. Our concern is to get efficient routing with a better utilization of network resources as compared to traditional fixed routing schemes i.e. to provide a higher network throughput. To show the feasibility of this routing as well to provide a base for simulating and implementing a variety of routing schemes, we have designed LAG, a tool which should help us in the network design process

Automata theory ; Reinforcement learning ; learning automaton

nije evidentirano