异构无线传感器网络中基于POMDP的实时调度算法  

POMDP-based Real-time Scheduling Algorithm in Heterogeneous Wireless Sensor Networks

作　　者：冯延蓬[1] 仵博[1,2] 郑红燕[1] 

机构地区：[1]深圳职业技术学院教育技术与信息中心,广东深圳518055 [2]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《仪表技术与传感器》2012年第8期101-104,共4页Instrument Technique and Sensor

基　　金：国家自然科学基金项目(61074058);广东省自然科学基金项目(S2001040004769)

摘　　要：针对异构无线传感器网络节点性能存在差异和易受环境影响的特点,提出一种基于部分可观察Markov决策过程(partially observable markov decision process,POMDP)的实时休眠调度算法,使用状态转移函数和观察函数表示系统完成用户请求任务中存在的环境噪声和传输冲突等不确定性,使用回报函数表示采用不同调度策略对异构网络感知准确度和能量消耗的影响,采用基于当前信念点的在线求解算法求取最优策略。仿真结果表明:该算法能够平衡数据准确性与能量消耗,延长网络生存时间。To handle the issues of node performance heterogeneity and environmental sensitivity with heterogeneous wireless sensor networks,a partially observable markov decision process（POMDP） based real-time sleeping scheduling algorithm was proposed.The algorithm used state transition function and observation function to represent the uncertainty in the network performance caused by environmental noise and transmission conflicts,and used reward function to represent the impacts of different scheduling strategies on the sensing accuracy and energy consumption.It solved for the optimal scheduling policy using belief point based online algorithm.Simulation results show that the algorithm can achieve a good tradeoff between data accuracy and energy consumption,and can prolong the network lifetime.

关 键 词：异构无线传感器网络 部分可观察Markov决策过程 调度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]