Resumen de Presenting a learning automata-based spectrum-sharing method in a cognitive radio network
The main objective of this study is to use the learning automata and Time Division Multiplex to reduce interference and improve the throughput in cognitive radio networks. In this method, in addition to the throughput, the cost of operation has been considered for secondary users. The proposed method is a new channel assignment process, which updates the probability of free channels at subsequent intervals, according to the environmental response using the learning automata, and increases throughput by decreasing the interference. For this purpose, three linear models on learning automata including LR-P, LR-I and LReP, were modeled using MATLAB software for access to the spectrum. The results of this simulation in the flexible environment of multiresponse learning automata showed that LR-I model is an optimal method compared to other models. In the event that the primary user traffic model changes, LR-I model cannot adapt itself to the new environment due to the lack of a penalty parameter. This will make it remain in the previous selection and fail to adopt its probability function according to the primary user traffic model. These problems make it possible to use the LReP model in a dynamic environment. This model pays neglible attention to penalty from the learning automata. In fact, LReP model acts like LR-I, with the difference that it can adapt itself to the new environment. LR-P model has severe fluctuations in updating its probability function, due to the equivalence of the reward and penalty rate. This makes it possible for each time slot to give a greater chance of choosing different channels and throughput decreases by increasing the number of switching channels.
