基于增强学习的非均匀分簇水声传感器网络能耗研究  

Study on energy-consumption of unequal clustering in Underwater Acoustic Sensor Networks based on reinforcement learning

作　　者：侯睿[1] 何柳婷 HOU Rui;HE Liuting(College of Computer Science, South-Central University for Nationalities, Wuhan 430074, China)

机构地区：[1]中南民族大学计算机科学学院,武汉430074

出　　处：《中南民族大学学报（自然科学版）》2020年第2期205-209,共5页Journal of South-Central University for Nationalities：Natural Science Edition

基　　金：国家自然科学基金资助项目(61972424,60841001);中央高校基本科研业务费专项资金资助项目(CZT19011);中南民族大学研究生学术创新基金资助项目(2019sycxjj116)。

摘　　要：近年来,水声传感器网络越来越成为研究的热点,但由于水下环境复杂多变,导致网络中能量消耗不均的问题.针对此问题提出了一种基于增强学习的非均匀分簇的水声传感器网络路径优化算法.该算法首先根据水声传感器网络中节点的深度和剩余能量把传感器节点分成大小不同的簇;然后根据节点的综合属性值选出最佳簇头;最后在数据传输阶段利用增强学习和ε-greedy策略对簇间的传输路径进行决策和学习,寻找最优路由.实验结果表明:本文方法可以有效均衡能耗,并延长网络寿命.In recent years,Underwater Acoustic Sensor Network has become a hot research topic,but due to the complex and changeable underwater environment,the energy consumption in the network is uneven.In order to solve this problem,a path optimization algorithm of Underwater Acoustic Sensor Network based on reinforcement learning is proposed.The algorithm firstly divides the sensor nodes into clusters of different sizes according to the depth and residual energy of the nodes in the network.Secondly,the best cluster head is selected according to the comprehensive attribute value of nodes.At last,the transmission paths between the cluster-head and the cluster-head are determined and learned by using reinforcement learning andε-greedy strategy in the data transmission stage,and the global optimal path is selected.Experimental results show that the proposed method can effectively balance energy consumption and prolong network life.

关 键 词：水声传感器网络 聚类算法 增强学习 路径优化 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]