Wireless Sensor Network Data Gathering Based On Compressed Sensing

To be compared with traditional method, compressed sensing (CS), a novel signal description and processing method, has obvious advantages:sampling data in well below the Shannon-Nyquist sampling rate, achieving signal sampling; compression simultaneously and accurately recovering the original signal with few samples. It has been widely applied in Wireless Sensor Networks (WSN).We carried out the related researches on CS-based data gathering in WSN in this dissertation deeply studies. The main contributions as follows:For problems that data transmission cost and poor quality of signal reconstruction in WSN, caused by raw data variety and poor data compression performance, we present a kind of effective data gathering method based on WSN characteristics with CS. Firstly, getting the clustering routing network model through the analysis of network topology, then processing temporal correlation between raw data with optimized joint sparsity model (JSM), and finally using network cluster model to collect and transmit data via CS. Experimental verification not only significantly improves data compression performance, but also reduces transmission energy consumption effectively.For the core issues that the designs of random measurement matrix and signal reconstruction algorithm, constrained CS application and development, we put forward an effective data gathering method based on optimal measurement matrix. First, elaborating the Semi-QR decomposition, then using it to deal with the matrix, the product of sparse matrix and measurement matrix, to build Gram matrix with mutual coherence, and finally processing coherence threshold and scaling for Gram matrix. It is validated by experiment that the Gram matrix made by the optimized measurement matrix and sparse matrix has better convergence and mutual coherence, while the optimal measurement matrix has a higher accuracy for signal reconstruction.From the above two in-depth study to improve compression performance data gathering based on CS, that is:through the establishment of effective data gathering method based on WSN characteristics with CS to improve the performance of data compression and to reduce energy transmission networks; through the design of effective data gathering method based on optimal measurement matrix to obtain measured values with more information to improve the optimization of the measured value, and to improve the accuracy of the signal reconstruction. Experimental results show that the proposed method can not only improve the data compression performance effectively and reduce network traffic data consumption, but recover the original signal efficiently.