Research On Distributed Localization Techniques Based On Semidefinite Programming In Wireless Sensor And Actuator Network

As the augment of actuators, wireless sensor and actuator network (WSAN) becomes a more powerful network compared with wireless sensor network (WSN). Through the coordination between the actuators, actuators and sensor nodes in these networks, the sensor nodes can detect the physical information in the around environment and send it to the right actuator nodes, then the actuator nodes will do corresponding action to the environment. In some sense, WSAN can not only perceive the physical world, but also change the physical world at the same time. As the development of microelectronics, embedded technology and wireless communication technology, the cost and power consumption of nodes have been greatly reduced, the perceiving and transmission capacity of nodes have been greatly improved, so the network size is becoming larger and larger. Meanwhile, as one of the key supporting technologies in WSAN, more challenges are faced for the positioning technology because of the new network features. Considering the localization requirements in WSAN and based on the study of the traditional location algorithm, this paper focuses on the research of distributed localization algorithms based on semi-definite programming (SDP) in the large-scaled static WSAN.First of all, the network architecture and the characteristics of WSAN are briefly described, the challenges of WSAN localization techniques are introduced, an overview of several basic ranging technologies and positioning algorithms is provided. Secondly the detailed study of the centralized and distributed positioning classification criteria is performed. Six classical positioning algorithms under these two classification criteria are discussed and simulated to evaluate the performance of the two kinds of algorithms. Thirdly the SDP positioning models are established both in the noise and noiseless environment, a kind of classic SDP refinement algorithm called the gradient search method is analyzed, the high rank problem and the congregations toward the center problem of the SDP solution are discussed. Then three kinds of SDP based location methods are summarized for large-scaled network, three typical algorithms are discussed and tested to evaluate their performance under different impact factors such as different radio ranges or noise factors, corresponding application environments for these algorithms are discussed based on the simulation and analysis.At last, a new edge sparsification method is studied, the performance evaluation of FSDP and SSDP algorithms based on this method is executed from different aspects. Aiming at the large-scaled net localization problem that the computational complexity of the cluster based SDP distributed localization algorithm in some clusters is high due to the non-uniformly clustering, a new distributed localization algorithm named EES-Cluster (Equivalent Edge Sparsification Cluster) is proposed. Based on the sparsification processing to each cluster diagram of the whole network, the number of edges is reduced. When the number of cluster is limited, this algorithm can effectively reduce the computation complexity in the localization process, at the same time it can keep the location accuracy, and decrease the power consumption in cluster header nodes. Simulation results and analysis show that EES-Cluster can effectively decrease the computation complexity, and improve the location efficiency of large-scaled WSAN.