On a guided genetic algorithm for frequency assignment in nonhomogeneous cellular networks

As the demand for bandwidth in cellular communications grows rapidly, fast and effective frequency assignment algorithms are becoming a very critical item in the design process. We consider a realistic radio network with non equal cells and non uniform traffic and propose a stochastic algorithm for the search of a quasi optimal solution for the corresponding frequency assignment problem. The algorithm represents certain modification of the known guided genetic algorithm and is rather fast due to the fact that it escapes local minimums and on the other hand spends more time in the spots with the highest conflict.