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

Wireless Sensor and Actuator Network (WSAN) is a kind of new network derived from the traditional Wireless Sensor Network (WSN) and based on the introduction of actuators. In WSAN, the common sensors sense the surrounding environment information, while actuators which are relatively powerful process information, make decision and then act on the environment. Due to the introduction of actuators, WSAN expands the sensor networks in functionality and application areas tremendously, but it also faces many challenges, such as the coordination, communication and scheduling problem arising from heterogeneous nodes. In addition, to most of applications, it makes no sense for WSAN to gather sensor data without sensor location information. Thus, node localization is also the supporting technology to WSAN. Consequently, research on the localization techniques in WSAN is conducted in the thesis on the basis of investigation in WSN. Recent research shows that semi-definite programming (SDP) localization techniques can provide high accuracy with less beacon points. Thus, the WSAN localization techniques based on SDP are mainly studied in the thesis.First of all, the SDP relaxation model of node localization named fullSDP is presented. In fullSDP, the localization problem is modeled as an optimization problem which takes the minimum distance error as the objective function, and the distance between nodes as the constraints, and then relaxed to an SDP problem. Meanwhile, four related range-free SDP localization methods are investigated.Secondly, a new SDP algorithm for WSAN node localization in Gaussian noise environment is proposed aiming at the problem that the localization result of fullSDP trends to crowd towards the center of the convex hull of beacon points. Analysis and simulation results show that the algorithm can effectively solve the problem confronted by fullSDP, resulting in obvious improvement to the localization accuracy with almost the same computational complexity. Additionally, taking the results of the proposed algorithm as the initial point of gradient search method, the localization accuracy will be further improved.Then, several methods of solving the problem that the computational complexity of SDP localization techniques is high are presented, and the distributed localization technique is mainly investigated. In order to fully take advantage of the actuator’s powerful functionality of power control and mobility, a range-free SDP distributed localization algorithm based on the multi-power levels of mobile actuators is investigated. Analysis and simulation results show that it can immensely improve the localization accuracy when actuators transmit beacons with multi-power levels.Finally, aiming at the problem that the computational complexity of the cluster based SDP distributed localization algorithm in some clusters is still high due to the non-uniformly clustering, a sensor based distributed localization algorithm is proposed. It distributes the computation to every sensor, which can effectively solve the high computational complexity problem in large scale WSAN. Meanwhile, this algorithm calculates the distances between tow-hop neighbors and adds it to the constraint conditions, which ensures the localization accuracy. Analysis and simulation results show that the proposed distributed algorithm is better than the cluster based algorithm in both complexity and accuracy.