【作者】 陈志； 王汝传； 孙力娟；

【Author】 Chen Zhi 1, Wang Ru-Chuan 1,2+, Sun Li-Juan1 1(College of Computer, Nanjing University of Posts and Telecommunications, Nanjing 210003, China) 2(State Key Lab. for Novel Software Technology, Nanjing University, Nanjing 210093, China)

【机构】 南京邮电大学计算机学院；

【摘要】 为实现在普适计算环境下无线传感器网络运行的协调统一和稳定,提出一种基于多Agent增强学习的自组织方法。首先通过汇聚节点训练网络形成初始优化的自组织结构;然后使每个参与信息传送的Agent保持增强学习,优化路径选择;最后以Agent之间的通告,实现主动的网络自组织结构优化。多Agent增强学习方法使在普适计算环境下的无线传感器网络具有相对平滑的自组织运行过程和高效性能。更多还原

【Abstract】 To achieve coherence and stability in operations of Wireless sensor networks (WSNs) in pervasive computing environment, a self-organization method is proposed based on multiagent reinforcement learning. Firstly, the network structure of initial optimization is formed by training through sink node(s) in WSNs. Secondly, the nodes in the transmission of information keep in reinforcement learning to optimize routing. Finally, active optimization of network structure is realized by notice among agents. Analyses show that multiagent reinforcement learning method enable WSNs a comparative smooth self-organization process and effective performance in pervasive computing environment.更多还原