Adaptive Genetic Algorithms for Dynamic Channel Assignment in Mobile Cellular Communication Systems

Abstract

Two adaptive genetic algorithms (GAs), namely GA for locking channel (GALC) and GA for switching channel (GASC), are proposed for a dynamic channel assignment in mobile cellular communication systems. The algorithms aim to minimize the blocking probability of new calls and the dropping probability of handoff calls in channelized systems, simultaneously considering three types of electromagnetic compatibility (EMC) constraints: 1) the cochannel; 2) the adjacent channel; and 3) the cosite. The proposed algorithms add a number of mechanisms to the canonical GA in order to increase their efficiency and velocity of convergence. Such mechanisms are adaptive parameters, random immigrants, a greedy policy, a reservoir to assist the initial population, a truncation selection scheme, and a three-point crossover. The GASC allows call switching between channels during the call holding time, whereas the GALC does not allow it. Computer simulations evaluated the performance of the proposed models considering a benchmark cellular environment formed by 49 cells with 70 channels and nonuniform traffic load characteristics. The impact of EMC constraints on the blocking probability of new calls and on the dropping probability of handoff calls was assessed, and the proposed models reached suitable performance. Equipment failure tests showed robust performance of the two adaptive GA schemes during the fault occurrence and recovery capability after the fault ends. The results suggest that the GASC has lower overall blocking probability of new calls than the GALC; however, the GALC may do better than the GASC in a number of combinations of handoff requests and EMC restrictions.