Nonstationary models of learning automata routing in data communication networks

摘要: In a data communication network the message traffic has peak and slack periods topology may change. When learning approach is applied to routing, automation situation at each node in network. Each selects routing choices its modifies strategy according conditions. A model of nonstationary automaton environment, with response characteristics dynamically related probabilities actions performed on it, proposed. The limiting behavior identical that earlier models. Simulation studies automata operating simple queuing networks reinforce analytical results show parameters proposed can be chosen predict transient behavior.