In this paper, the optimization of spectrum allocation in wireless networks based on GA (Genetic Algorithm) is studied. This problem involves how to allocate spectrum intelligently under limited spectrum resources to improve communication system performance while minimizing interference and resource waste. Firstly, we introduce the background and importance of spectrum allocation, and emphasize the limitations of traditional methods, especially in the face of complex communication environment. We designed a series of experiments to evaluate the performance of GA in wireless network spectrum allocation optimization. The experiment covers the efficiency evaluation of the algorithm, the sensitivity analysis of parameters, the complexity study of problem cases and the comparison with previous studies. By comparing the execution time of different algorithms through experiments, we can determine the adaptability of GA to problems with different complexity levels. It may be found that the average processing time of GA in more complex problems is 44 seconds, which is lower than that of greedy algorithm 52 seconds, and GA can also find high-quality solutions. Our research provides a powerful method and theoretical basis for improving the performance of communication system, meeting the needs of users and improving the efficiency of resource utilization.