Temporal compression in wireless sensor networks using compressive sensing and ARMA modeling

In this paper a new method is proposed which reduce the redundancy of data transmission in Wireless Sensor Network (WSN). In a dense environment of sensors, energy efficiency is key factor for prolonged battery life. Here in this paper temporal redundancy of sensor node is reduced using hybrid of two methods, one is compressive sensing using Discrete Cosine Transform (DCT) and other one is Linear Predictive Coding (LPC). Given method is not suitable for real time data monitoring rather it is better for delay tolerant services as it require processing time before transmission. Here it is assumed that data monitored by sensors is piecewise linear. The relevancy in temporal data exploited here for data compression. First data monitored from each node for a long enough span of time which is further segmented according to their linearity. As compressive sensing compress the data and ARMA also use only few of the transmission bits, combination of both provide satisfactory compression of stored data. The segmented data used as input for compression and further processed to sink node. Result shows that this scheme provide satisfactory compression rate of monitored data.