The Research Of Indoor Localization Algorithm Based On Compressive Sensing

In the wireless sensor network,the location information of the sensor nodes is very significant for practical application, for example,in the event monitoring,obtaining the content of events from the wireless sensor network only is not enough,at this moment you need to know the information about the location of the sensor nodes,then you can make the monitoring meaningful.So the positioning technology of the sensor nodes is one of the supporting technology of wireless sensor network. The technology is applied in an important occasion,which called indoor positionong,besides,it is also widely used in the airports,exhibition halls,supermarkets and other places.At present,there are many solutions of indoor positioning technology,the existing positioning schemes can be divided into two kinds,one based on the fingerprint map algorithm,and the other based on the signal propagation model algorithm.The algorithm based on fingerprint map has a better accuracy in positioning,but we need to update the fingerprints when the environment changes,which means a large workload.And it is not practical in large area localization.Conversely,it doesn’t need a lot of offline training missions based on the signal propagation model algorithm,and thus the workload can be reduced greatly.In the processing of localization,however, there is a mount of data collection and processing,in addition,there are weaknesses of the sensors in calculation capability,storage and limited battery capacity,leading to some limitations of the existing schemes.In recent years,The newly developed compressive sensing theory can use the sparsity with respect to the location area of nodes,and the node coordinates can be achieved by signal measurement of a few anchor nodes towards to the unknown nodes.In view of the existing positioning algorithm based on compressive sensing, it is difficult to meet real time requirements,because of the redundancy of measurement and the meshing area are oversize, thus we put forward the improved algorithm.According to the idea of Bounding-Box,the meshing area is reduced by estimating the unknown nodes to a smaller rectangular area of possibility.The size of potential areas is changed with the number of anchor nodes and communication radius,so in order to make the positioning problem be converted to the problem of compressive sensing signal reconstruction, we designed the dynamic measurement matrix.Besides,in order to ruduce the redundancy of measurement matrix,only the anchor nodes that having a communication relationship to the unknown nodes can be used as the measuring nodes,at the same time, setting the maximun number of measurements can ensure the accuracy of the reconstruction,and it reduced the redundancy of measurement matix further.Besides,the RSSI difference comparison algorithm is proposed,which is similar to the fingerprint matching,in order to improve the positioning accuracy and efficiency further.At last,using the MATLAB simulates and analyses the performance of the improved algorithm,and compared with another two kinds of wireless sensor network localization algorithm that based on compressive sensing,experiments show that the improved algorithm has certain advantages on the localization of average effective rate and the average time.