摘要
针对无线传感器网络节点能量有限、数据采集易受环境影响的问题,提出一种基于可分解部分可观察Markov决策过程FPOMDP(Factored Partially Observable Markov Decision Process)的节点休眠调度算法。通过节点空时相关模型求取休眠节点数据,利用网络数据准确性和节点能量间的条件独立关系,构造状态转移函数、观察函数和奖赏函数,采用值迭代求解算法求取最优策略,实现节点动态调度。仿真结果表明,该算法能够在保证数据准确性的前提下,有效降低节点能量消耗,延长网络生存时间。
An FPOMDP-based node sleep scheduling algorithm is proposed to address the problems of limited node energy and environ- ment-prone data acquisition in wireless sensor networks. The algorithm estimates the data of sleeping nodes based on spatiotemporal correla- tions model. By exploiting conditional independence between the networks data accuracy and the node energy, the algorithm then constructs transfer function, observation function and reward function of the state, uses value iteration to find the solution of the algorithm to obtain opti- mal node scheduling policy, and implements dynamic node scheduling. Simulation results show that this algorithm can effectively reduce node energy consumption and prolong network lifetime without compromising data accuracy.
出处
《计算机应用与软件》
CSCD
北大核心
2012年第8期55-58,77,共5页
Computer Applications and Software
基金
国家自然科学基金项目(61074058)
广东省自然科学基金项目(S2011040004769)
关键词
无线传感器网络
可分解部分可观察Markov决策过程
空时相关模型
Wireless sensor networks Factored partially observable Markov decision process (FPOMDP) Spatiotemporalcorrelation model