Research On Data Gathering Based On Compressive Sensing In Wireless Sensor Networks

Wireless sensor networks (WSNs), as a kind of function networks, is widely used in environment monitoring, medical0020observing, military target-tracking and other areas. However, in large-scale WSNs, the network lifetime is limited by the sensor nodes energy. Due to the dense layout of the sensor nodes, the sensed data has a strong correlation and is compressible. So it is an effective method to reduce the energy consumption and prolong the network lifetime by adopting the compression algorithm to reduce the number of data transmission. In compressive sensing theory (CS), data is compressed and transmitted at the same time, which is different from the traditional compression methods. What’s more, the application of CS in WSN data gathering has the characteristics of simple compression, easy to distributed processing and accurate reconstruction.However, there are still some problems to be solved in CS-based data gathering in WSNs. In this dissertation, we focus on the problem that the measurement matrix does not match the routing. To improve the performance of the network on the condition of high reconstruction accuracy, efficient routing strategy is designed based on the existing measurement matrix. The main research contents are as summarized follows1. For the dense layout of the nodes, there are too much relay nodes involved in data gathering, which leads to excessive energy consumption. To solve the problem, a compressive sensing data gathering algorithm based on minimum energy tree is proposed. Sparse random matrix is used as the measurement matrix to reduce the number of source nodes. A minimum energy consumption relay node selection scheme is given by analyzing the energy model to achieve the tradeoff between the transmission distance and relay node number. The spanning tree routing is created based on the idea of centralized greedy increasing tree to match the projection matrix. The simulation shows that compared with the existing data gathering algorithm, the proposed algorithm can reduce the participated node numb and save the energy consumption of the network.2. Aiming at the problem of too much observation and the unreliable communication links in large-scale WSN, a sparse block diagonal matrix based clustering data gathering algorithm is presented. Data is collected through distributed cluster routing to reduce the relay number. The sparse block diagonal matrix is used as the measurement matrix to reduce the number of the observation number of each cluster. The optimal number of cluster head is obtained by analyzing the energy consumption model of the data gathering to minimize the energy consumption. On the basis of the above analysis, a distributed clustering data gathering strategy based on the optimal cluster head number is proposed. The proposed algorithm could effectively balance the energy load and prolong 20%lifetime of the network compared with the existing cluster data gathering methods.3. Concerning the poor data compressibility caused by too many influence factors in complex WSNs, a spatial-correlation routing based CS data gathering algorithm is proposed. First, the joint spare model of distributed compressive sensing is used to analyze the data sparsity. Second, the relationship between spatial correlation and node distance is gained by analyzing the existing spatial correlation model. Third, a distributed cluster scheme based on the spatial correlation is presented to make sure low data sparsity in each cluster. At last, the data is gathered in cluster independently. The simulation showing that the proposed algorithm could reduce the amount of communication data and improve the network performance.4. To apply the CS in actual WSN data gathering, we design a simple tree type routing scheme based on the uC/OS-II real-time operating system. Three function module, network setup, route update and data forward module, are designed to realize the route setup and maintenance according to the requirements. Data frames are designed according to the networks interface. The route development is finished based on the μC/OS-Ⅱ system and the function module is tested on the STM32F407-Discovery platform.