The Research Of Missing Data Recovery Method Based On Feature Analysis In Wireless Sensor Networks

With the rapid development of micro electro mechanical system, chip system, wireless communication technology and low power consumption embedded technology, wireless sensor network(WSN) is widely used in industrial process control, health monitoring and environmental sensing and other fields. The sensor data as the research focus of wireless sensor network is becoming more and more important. However, the sensor data loss is inevitably due to the limitation of nodes energy, link instability and other reasons which will influence the integrity and accuracy of sensor dataset. The recovery of missing data will help to improve the quality of sensor data in the process of data acquisition and transmission. At present, the data recovery methods based on retransmission and redundancy mechanism have high energy consumption, and another kind of methods based on sensor data feature and mathematical models is mostly aimed at the single sensor attribute, which has many limitations. Based on multi-sensor integration and fusion technology, it combines with the analysis of characteristics and correlation of multi sensor attribute data in this thesis to discuss the data recovery method to meet the requirement of data accuracy.Firstly, we introduce data interleaving technology in the communication system by considering the effects of data missing caused by data transmission, and dispersing the errors maximum which may occur in the process of transmission, thereby put forward the sensing data transmission scheme based on data interleaving and build a data recovery overall framework. In data transmission phase, the original sensor data is interleaved based on matrix in in-network which can be regrouped and dispersion transmission; in the data processing phase, de-interleaving the received data and recording a corresponding information by judging the missing data; in the data recovery phase, based on the analysis of sensor data features and sensor attribute correlation to recover the missing data.Secondly, considering that most of the physical attributes of nature have a certain correlation, which can be used to improve the accuracy of data recovery. A sensor data recovery algorithm based on temporal stability and attribute correlation(TS-AC) is proposed by analyzing the characteristics of the multi attribute data. On the one hand, the regression model is established based on attribute correlation; on the other hand linear interpolation model is built based on the temporal stability. In order to improve the accuracy of missing data recovery, the estimation results of the two methods are combined by calculating the average.Finally, the performance of the proposed data recovery framework and the sensor data recovery algorithm based on attribute correlation and time stability are analyzed based on the theoretical and experimental analysis, the performance of the proposed data recovery framework and the sensor data recovery algorithm based on attribute correlation and time stability are evaluated. Compared with KNN and other classical algorithms, the proposed method can improve the accuracy of the data recovery effectively.