The Research Of Compressive Sensing And Its Application In UWB And Wireless Sensor Networks

Compressive sensing theory is the latest research in the information processing field. It has good application prospects in ultra-wideband (UWB) communication system and wireless sensor networks (WSN). Topic of this dissertation comes from the project such as National Natural Science Foundation of China. They have important significance on both theory and application.Based on the deep research about distributed Bayesian compressive sensing theory and the basic principles of its reconstruction algorithm, this dissertation focuses on its application on the multiuser UWB communication system and wireless sensor networks. The main innovation ideas are as followed.In order to solve the limitation of the traditional Bayesian compressive sensing algorithm, we propose the Laplace prior probability based distributed Bayesian compressive sensing algorithm. Distributed Bayesian compressive sensing algorithm utilizes the property that the different signals satisfy the same priori probability distribution, to jointly estimate the probability distribution parameters of the original signals. Therefore, the distributed Bayesian compressive sensing improves the original signal reconstruction performance of some statistically related signals through the joint reconstruction algorithm. At the same time, the Laplace prior probability distribution is better than the Gaussian prior probability distribution, so the signal reconstruction performance of distributed Bayesian compressive sensing algorithm is better than the Gaussian prior probability distribution based multi-task Bayesian compressive sensing algorithm.Also, this paper considers the multiuser UWB system. The channels of different user signals received at one time are statistical relation. Therefore, we set the distributed compressive sensing signal model corresponding to multiuser UWB channel model, and build the multiuser signal processing frame corresponding to the distributed Bayesian compressive sensing reconstruction algorithm. The receiver jointly reconstructs the received user signals and gets the parameters of the channel models which are utilized to restore the received signals. The simulation results show that the proposed method reduces the necessary samples for channel estimation of multiuser UWB system and improves the BER performance.For the source detection of wireless sensor networks, an algorithm combining LEACH algorithm and Bayesian compressive sensing is proposed. LEACH algorithm divides the sensors into some clusters and chooses the clusterheads. The information of sensors in the cluster is collected by the clusterhead, and only the clusterheads are allowed transmitted information to the fusion center. It reduces the number of sensors which send the information to the fusion center. The fusion center utilizes Bayesian compressive sensing to recover the sources from a few measurements transmitted by clusterheads, and detects the sources accurately. At the same time, a threshold is used in this paper to optimize the performance of reconstruction and improve the accuracy of source detection.In order to solve the energy limitation of sensor node, we propose an energy balance based adaptive compressive sensing. Through making some operations on the measure vector of measure matrix, the algorithm makes that the accurate reconstruction needs less measurements. The path choice of the proposed algorithm not only considers the reconstruction performance, but also takes into account the balance of energy consumption. It prevents the network from dying because of energy depletion of some node. The adaptive compressive sensing algorithm is combined with the energy balance based compressive sensing algorithm to improve the convergence of algorithm. The flexible configuration purpose is achieved by choosing the threshold. Compared with the adaptive algorithms, the proposed algorithm can extend the survival time of the network and achieve flexible configuration.Finally the content of dissertation is summarized and the future research works are discussed.