Study On Wireless Sensor Network Node Localization Algorithm Based On Compressive Sensing

Wireless Sensor Network(WSN) is widely applied in many node localization based issue, in field such as military, commercial and scientific research, and has a great market prospects. However, WSN localization algorithm is now revealed to have many defects such as low data processing speed and short endurance, which require less communication cost and lower sampling rate. Compressive sensing(CS) can decrease the sampling rate to a much smaller level and remain accuracy at the same time, this property provide us a new solution to our problem.WSN localization algorithm based on Receive signal strength intensity(RSSI) is wildly used because of it’s easiness of implement and low cost. Fuse the CS theory and localization algorithm based on RSSI together, use the real signal intensity data to build the measurement matrix, divide the network area into grids, and use sparse discrete signals to represent target nodes position information, The localization algorithm based on RSSI and CS theory can effectively reduce the data processing amount, and decrease the inter-nodes traffic when ensure accurate signal reconstruction. By improving greedy CS reconstruction algorithm, Greedy Matching Pursuit(GMP) reconstruction algorithm can lower the localization error further. However, the frame of the former localization algorithm require the position of every grid to localize target nodes and the localization accuracy is limited by the size of the grid. As a result, our goals in this paper are ヾecrease the traffic; ヽhange the frame to realize target nodes positioning without the coordinate information of the grids and improve positioning accuracy at the same time.An improved GMP reconstruction algorithm is presented in this paper according to Greedy Matching Pursuit algorithm, by split K-sparse original signal into K 1-sparse vectors. Experiment shows that the improved GMP has lower communication cost compare to GMP when then-positioning error is in the same level.This paper also proposed a RSSI, trilateration and improved GMP based localization algorithm for WSN, according to many inherent defects of the traditional RSSI and CS based WSN localization algorithms. The proposed algorithm use CS measurement matrix data to calculate distance between the node and the anchor nodes, when distance data is bigger or equal to 3, get the position of the target node by using trilateral positioning method, the positioning procedure do not require the coordinate information of the grids. Experiment shows this algorithm has a higher accuracy compare to other algorithms.