Research On K-Cover And Compressed Sensing In Wireless Sensor Networks

Coverage and in-network data processing problems are the key points in wirelesssensor networks(WSNs). Coverage problem means that both the coverage and the life-time of the WSNs have been maximized by making sensors’ configuration algorithms.In-network data processing means that the WSNs can finish the data gathering and re-construction task with a minimized energy consumption by investigating the networktopology, route protocol and data aggregation methods. K Cover and in-network dataprocessing are the key points in this dissertation. By fully taking the advantage of theredundancy in WSNs, the sensors are partitioned into K SETS and their correspondingalgorithm is proposed for solving the K Cover problem, which is based on game theory.The measurement matrix construction and the compressed data aggregation algorithm-s are presented based on compressed data gathering theory. The main contents of thisdissertation are the follows:An N Person Card Game Algorithm(NPCGA) is proposed to solve the K Coverproblem in homogeneous WSNs. NPCGA is a pure distributed algorithm, only pure localinformation is needed for all the sensors to complete the whole game process withoutcentre control unit and global communications. In NPCGA, the strategy profile of all thesensors is converged to pure nash equilibrium and the average coverage has reached therelative optimum at the same time. By NPCGA, lifetime of the WSNs is prolonged to Ktimes and the monitoring performance achieved the relative optimal case. NPCGA showssuperiority in coverage, convergence speed, robustness and network design.Greedy Game Algorithms(GGAs) are presented to solve the K Cover problem inheterogeneous WSNs. A typical heterogeneous network model is proposed first in thescheme. The neighbors can be divided into four parts, that are the equal neighbors, blindneighbors, potential neighbors and explicit neighbors respectively. GGAs is one categoryof the pure distributed algorithms. By GGAs, the K Cover problem has been well solvedin heterogeneous WSNs. The strategy profile of the whole sensors constitutes the purenash equilibrium finally, and the average coverage has been maximized under the lifetimeguarantee K. As the heterogeneous ratio increases, GGAs can decrease the efect of theunbalanced information to the network average coverage ratio at the most extent. GGAs can still maintain the better performance in coverage, convergence speed, robustness andnetwork design in heterogeneous WSNs.The Measurement Matrix Construction Algorithm(MMCA) is introduced for gath-ering the in-network compressed data. Shrinkage and alternating projection theory havebeen adopted to optimize the measurement matrix Φ iteratively under the prefixed sparsematrix Ψ. Finally, the optimized matrix Φ is obtained, which shows the relative minimumcoherence with Ψ. Diferent kinds of measurement matrices can be optimized and thesame optimal result can be achieved by MMCA. According to the analysis of the mutualcoherence, z average coherence and cumulative coherence, the superiority of the pro-posed MMCA has been proved in nature. Under the guarantee of recovery performancefor the sink, both the simulations and the real experiments present the efectively reduc-tions of the data transmission based on the measurement matrix Φ, which is optimized byMMCA.Integrative solution of in-network data processing is put forward based on K Coverin WSNs. Under the guarantee of prolonged lifetime of the WSNs, the efcient in-networkdata processing has been achieved by the proposed scheme. An Optimal Data AggregationTree(ODAT) algorithm has been proposed based on the minimum energy consumptioncriteria. ODAT is rooted at the sink node, and as a whole integrity is constituted by theMinimum Spanning Tree(MST) and the Shortest Path Forest(SPF). Based on schedulingthe K Cover sensors and adopting the ODAT algorithm, the in-network data transmissionhas been minimized and the lifetime of the whole WSNs has been prolonged to K times.Finally, the sink node finishes the efcient source data recovery based on MMCA anddifusion wavelets sparse matrix construction algorithm.