一种基于深度学习的改进萤火虫频谱分配算法

An improved firefly spectrum allocation algorithm based on deep learning

针对使用萤火虫算法(FA)解决认知无线电中频谱供需矛盾分配不均时,FA全局寻优能力欠佳以及过早的收敛使得频谱分配效率并不理想的问题,基于深度学习思想提出了一种改进萤火虫算法(IFA),IFA将频谱分配变量映射为萤火虫的位置信息,再将平均最大化网络效益转化为萤火虫的亮度函数,然后在萤火虫寻优中采用了移动变化规则模式,利用深度学习思想寻找最优中心粒子,提高了中心粒子的搜索精度。学习后的粒子引导种群进化,从而提升寻优性能。与其他智能优化算法相比,IFA频谱分配算法寻优精度及收敛速度更佳,所对应通信状态下的频谱分配更优。

In light of the lack of spectrum resources and uneven distribution in cognitive radio,the Firefly Algorithm(FA)is used to solve the problem.However,FA’s global optimization ability and premature convergence are not ideal during simulation.This paper proposes an improved Firefly Algorithm(IFA)to address the spectrum allocation problem.Firstly,the inverse function of the maximizing the total benefit of the network is used as the fitness function to initialize the luciferin value.Secondly,the random search mode is adopted in the firefly evolution process,and the optimal central particle is found by using deep learning in the search process of the central particle.and the accuracy of the central particle is enhanced.The learned particle guides the population evolution and thus the optimization is improved.The experimental results indicate with the comparison with other intelligent optimization algorithms,and the IFA spectrum allocation algorithm has better searching accuracy and convergence speed,and the corresponding spectrum allocation in the communication state is better.