基于时间序列预测模型的簇型数据收集机制

发布时间：2018-05-28 05:46

本文选题：无线传感器网络 + 数据收集　；参考：《计算机应用》2014年10期

【摘要】：由于温度、光照等物理属性的时空连续性,密集部署的传感器网络中节点感知的数据往往具有很高的时空相关性。这种数据相关性产生的数据冗余会带来通信负担,也会缩短网络的生命周期。提出一种基于预测模型的簇型数据收集机制(CDCF),探索数据相关性,减少无线传感器网络的通信量。该机制包括一种基于曲线拟合最小二乘法的时间序列预测模型和简单有效的误差控制方法。在数据收集过程中,簇型结构考虑到了数据间的空间相关性,时间序列预测模型探讨数据的时间相关性。实验仿真表明,在较为稳定的网络环境中,相对于收集原始数据,该机制只需10%~20%的通信量就可完成整个网络的数据收集任务;数据误差控制方法可以确保基站恢复数据的误差控制在用户定义的误差范围之内。
[Abstract]:Due to the spatiotemporal continuity of physical attributes such as temperature and illumination, the data perceived by nodes in the dense deployed sensor networks often have a high spatio-temporal correlation. The data redundancy caused by this data correlation will bring the communication burden and shorten the network life cycle. A cluster data collection mechanism based on predictive model is proposed to explore data correlation and reduce the traffic of wireless sensor networks. The mechanism includes a time series prediction model based on curve fitting least square method and a simple and effective error control method. In the process of data collection, the cluster structure takes into account the spatial correlation between the data, and the time series prediction model discusses the temporal correlation of the data. The experimental results show that in a stable network environment, the mechanism can complete the data collection task of the whole network with only 10% of the traffic compared with the original data collection. The data error control method can ensure that the error control of the base station recovery data is within the user-defined error range.
【作者单位】：南京信息工程大学计算机与软件学院;南京信息工程大学网络信息中心;
【基金】：国家自然科学基金资助项目(61373064,61402236) 公益性行业(气象)科研专项(GYHY201106037) 江苏省普通高校自然科学研究资助项目(12KJB520009)
【分类号】：TP212.9;TN929.5

【共引文献】