Research On Distributed Localization And Data Collection Technology In Wireless Sensor Networks

With the characteristics of easy deployment, self-organizing, low-cost, strong self-healing capability, wireless sensor networks have broad application prospects in military reconnaissance, environment monitoring, medical treatment, space exploration, disaster relief, and so on. Node localization is the basis for numerous applications of wireless sensor networks, and data collection is a key support technology. It is very important to design node localization and data collection mechanisms fitting for wireless sensor networks’characteristics.Focusing on the problem of limited resources of sensor nodes, with the sum of neighbor distance error as utility function, a distributed localization model based on game theory is presented. It is formally proved as a type of potential game, and the existence of Nash Equilibrium is testified theoretically. A distributed localization algorithm based on potential game is proposed, where each sensor node exchanges information only with its neighbors. To avoid converging in local optimum and accelerate the convergence speed, the strategy space determination and unknown nodes elevation are developed. Further, by judging the relationship of measuring distance and estimated distance, the node localization problem is established as virtual force based distributed localization model. The concept of confidence level is introduced to measure localization accuracy of nodes, and virtual force based distributed localization algorithm is proposed. Exploiting push-pull force of neighboring nodes who own higher confidence levels, all of the unknown nodes in the network can be localized correctly and quickly. A series of simulation results demonstrate the effectiveness of our proposed algorithms.In data collection process, multi-hop communication and many-to-one traffic characteristics can easily cause data loss and node failure, which impairs reliability of data collection seriously. Therefore, an energy consumption model of nodes based on compressive sensing is presented. Data traffic in the network is reduced and the performance of data collection is improved through data compression. Taking advantage of wireless power transfer technology, an energy supplement mechanism of nodes is designed. Respectively, the shortest-distance based wireless charging vehicle stop-point selection algorithm and genetic algorithm based wireless charging vehicle stop-point selection algorithm are proposed to improve wireless charging efficiency. Further, for the problem of data recovery, a weight factor based regularized adaptive matching pursuit algorithm is proposed, which can reconstruct the signal accurately without knowing signal sparsity. The effectiveness of our proposed algorithms is verified by extensive simulation results.