Neural network-based dynamic channel assignment for cellular mobilecommunication systems

摘要

Conventional dynamic channel assignment schemes are bothntime-consuming and algorithmically complex. An alternative approach,nbased on cascaded multilayered feedforward neural networks, is proposednand examined on two cellular systems with different configurations.nSimulation results showed that the blocking performance of ournmultistage neural network approach can match that of an examplenconventional scheme with less complexity and higher computationalnefficiency. The example scheme considered here is the ordered channelnsearch, which can achieve a reasonably high spectral efficiency asncompared to that of an ideal dynamic channel allocation algorithm. Wenconclude that our neural network approach is well-suited to the dynamicnchannel allocation problem of future cellular or microcellular systemsnwith decentralized control