Self-adaptive random-access protocols for WDM passive star networks

A learning automata based random access protocol for WDM passive star networks is introduced. The proposed protocol makes use of learning automata to achieve a high throughput and a low delay under any load conditions. An array of learning automata that determines the transmission probability of each wavelength is placed at each station. After each slot the transmission probability of each wavelength is modified according to the network feedback information. The asymptotic behaviour of the system which consists of the automata and the network is analysed and it is proved that under any load conditions, the transmission probability asymptotically tends to take its optimum value. Extensive simulation results are presented which indicate that the use of the proposed learning automata based scheme leads to a significant improvement of the network´s performance