Cognitive radio is an emerging technique to improve the efficiency and flexibility of spectrum utilization.In cognitive radio system

dynamic spectrum allocation is essential to make use of free frequency without interfering licensed users. This paper analyzes a spectrum allocation system based on graph coloring model and proposes a improved artificial bee colony algorithm. The basic artificial bee colony algorithm suffers from the problems of slower convergence and low exploration precision. The improved artificial bee colony algorithm introduces a search strategy based on differential evolution and Gaussian mutation detection strategy. In addition

a novel probability choice model is proposed to increase population diversity. The simulations are conducted to apply the improved ABC algorithm into spectrum allocation.The simulation results show that the improved ABC algorithm can achieve better system benefit and accelerate the convergence speed compared with the basic ABC algorithm.