无线传感器网络中基于马尔科夫链的数据预测算法  

Data Prediction Algorithm Based on Markov Chain in Wireless Sensor Networks

作　　者：吏济新 张雅云[1,2] 黄福川[1,2] 程红举[1,2] 

机构地区：[1]福州大学数学与计算机科学学院,福州350108 [2]福州大学福建省网络计算与智能信息处理重点实验室,福州350108

出　　处：《小型微型计算机系统》2016年第11期2498-2503,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61370210)资助;福建省教育厅A类科技项目(2013JA12027)资助;福州大学科技发展基金项目(2013-XQ-35)资助

摘　　要：无线传感器节点持续感知的数据具有高度的时间关联性,所产生的冗余数据会给传感器节点带来不必要的数据传输和能量消耗.数据预测算法通过预测节点的感知数据序列可以有效避免上述问题.提出一种基于马尔科夫链的数据预测算法(MC-DP),该算法首先对传感器节点感知的数据进行弱化处理来提高数据序列的光滑性,然后采用离散灰色预测模型对节点需要向sink传输的数据序列进行预测,当节点预测的数据精度不够时,进一步使用基于马尔科夫链的数据修正过程来对其进行改进.实验结果表明,MC-DP算法预测准确率更高,预测序列的数据误差率更低,使用该预测算法使得传感器节点可以节约更多能量.The data periodically sensed by wireless sensor nodes usually is of high temporal redundancy, and the final redundant data leads to unnecessary data transmission and energy consumption for sensor nodes. Data prediction algorithm is an effective way to avoid these problems by predicting the sensed data sequence. In this paper, we propose a Data Prediction algorithm based on Markov Chain （ MC-DP ）. The proposed algorithm firstly improves the smoothness of sensed data sequence by the weakening operation, then predicts the data sequence which shall be sent to the sink using discrete grey model. In case that the predictive data does not satisfy the data accuracy requirement,it is improved with the amending process based on Markov chain. The experimental results show that MC-DP algorithm has higher prediction accuracy rate and lower data error rate for the predictive data sequence,and thus the proposed algorithm can save more energy for nodes in the network.

关 键 词：无线传感器网络 离散灰色模型 马尔科夫链 数据预测 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]