Research On Node Localization In Wireless Sensor Networks

Wireless Sensor Networks (WSNs) are composed of large numbers of tiny sensor devices with wireless communication capabilities. WSN systems have been developed recently for numerous applications. Because many of them require sensor position information, localization has been an important problem in WSNs and several localization systems and algorithms have been proposed in the past. The key issues are how to improve the efficiency and precision of sensor node localization problem in the particular deployment scenario. Considered the ideal and irregular radio pattern, this thesis discusses some localization approaches in three different scenarios. These three typical application scenarios are range-free localization scenario under mobile-beacon and connectivity assisted, range-free localization scenario under higher dense beacons and connectivity assisted, and range-based localization scenario under Second-Order Cone Programming (SOCP) assisted.In the first scenario, Section 2 proposes a range-free, distributed and probabilistic Mobile Beacon-assisted Localization (MBL) approach, and its update version Adapting MBL (A-MBL), to increase the efficiency and accuracy of MBL. Evaluation results show that the accuracy of MBL and A-MBL outperform both Mobile and Static sensor network Localization (MSL) and Arrival and Departure Overlap (ADO) when both of them use only a single mobile beacon for localization in static WSNs.Section 3 gives the theoretical analysis and experimental evaluations to suggest which probability distribution in the dynamic model should be adopted to improve the efficiency in the prediction stage. Section 3 also gives the condition for whether the unknown node should use the observations from its neighbors to improve the accuracy. Finally, Section 3 proposes a Self-Adapting Mobile Beacon-assisted Localization (SA-MBL) approach to achieve more flexibility and achieve almost the same performance with A-MBL.Section 4 proposes a distributed Mobile beacon-assisted localization scheme based on RSS and Connectivity observations (MRC) with a specific trajectory. Section 4 proposes two improved approaches based on MRC to consider irregular radio scenario in the noisy environment. Compared the performance with three typical range-free localization methods in static WSNs, our lightweight MRC algorithm with limited computation and storage overhead is more suitable for very low-computing power sensor nodes In the second scenario, Section 5 proposes a Range-Free localization via Compressive Sampling (RF-CS). Section 5 shows that the RF-CS scheme performs better than previous classical range-free localization algorithms when locating sparse unknown nodes with higher dense beacons random deployed in sensor networks. In addition, Section 5 proposes an expansion method, called RF-CS* which is workable in locating dense unknown nodes. On the whole, the approaches improve the accuracy of nearly 50%-80% with higher beacons regardless of unknown node density.In the third scenario, Section 6 proposes an efficient SOCP formulation by minimizing the number of auxiliary variable to reduce the size of spare part of Schur complement matrix in solving sensor network localization problem. Compared to the previous SOCP relaxation with larger Schur complement matrix of sparse pattern, the SOCP formulation proposed in Section 6 can be solved faster after the splitting. Also, this SOCP relaxation increases computational efficiency without losing accuracy.