一种非线性动态自适应的Agent联盟生产算法

代理(Agent)联盟是对无线传感器网络WSN(Wireless Sensor Network)进行管理的重要手段.引入粒子群算法PSO(Particle Swarm Optimization)并对其进行改进,使PSO的参数具有非线性动态自适应性.将改进的PSO用于求解Agent联盟生成问题,并针对WSN的特性设计了一种效益函数用于评价联盟的效益.采用基于非线性动态自适应PSO的Agent联盟生成算法,在联盟生成初期搜索范围较广,搜索后期在局部挖掘上表现出良好的性能.实验证明在解决Agent联盟生成问题中,基于PSO的算法在稳定性上优于其他算法,基于改进PSO的联盟生成算法可以加大搜索空间,更快的收敛到最优解,且该算法可以同时生成多个Agent联盟,支持并行多任务环境下的Agent联盟求解.

Non-linear Dynamic Serf-adaptive Agent Coalition Algorithm

Agent coalition is an important way to manage WSN(Wireless Sensor Network).In the paper,we introduce PSO(Particle Swarm Optimization) and improve the algorithm through making its parameters non-linear self-adaptive.Solve Agent coalition formation using improved PSO and design a utility function for the characteristics of WSN.Algorithm based on non-linear dynamic self-adaptive PSO has larger scope in the early of the search and has a good performance in local mining in the late of the search.It is proved by simulation that algorithm based on PSO is superior to other algorithm in stability.Improved PSO widens the search scope and can convergence to the optimal solution faster.Furthermore,Agent coalition formation algorithm based on non-linear dynamic self-adaptive PSO supports solving Agent coalition formation in parallel multi-tasks environment.