A multistage self-organizing algorithm combined transiently chaotic neural network for cellular channel assignment

In this paper, a new multistage self-organizing channel assignment algorithm with a transiently chaotic neural network (MSSO-TCNN) is proposed as an optimization algorithm. The algorithm is used for assigning channels in cellular mobile networks to cells in the frequency domain. The MSSO-TCNN consists of a progressively initial channel assignment stage and the TCNN assignment stage. According to the difficulty measure of each cell, the first stage is executed to assign channels cell by cell inspired by the mechanism of bristle. If the optimum assignment solution is not obtained in the first stage, the TCNN stage is then applied to continue the channel assignment until the optimum assignment is made or a maximum number of iterations is reached. A salient feature of the TCNN model is that chaotic neurodynamics are temporarily generated for searching and self-organizing in order to escape local minima. Therefore, the neural network gradually approaches, through transient chaos, a dynamical structure similar to conventional models such as the Hopfield neural network and converges to a stable equilibrium point. A variety of testing problems are used to compare the performance of the MSSO-TCNN against existing heuristic approaches. Simulation results show that the MSSO-TCNN improves performance substantially through solving well-known benchmark problems within comparable numbers of iterations to most existing algorithms.