Study of spring-relaxation technique for cooperative localization in wireless sensor networks

• Main Author: Zhang, Qing
• Other Authors: Foh Chuan Heng
• Format: Theses and Dissertations
• Language: English
• Published: 2013
• Subjects: DRNTU::Engineering::Computer science and engineering::Computer systems organization::Computer-communication networks
• Online Access: http://hdl.handle.net/10356/52663
• Institution: Nanyang Technological University

Self localization of sensor nodes is a important yet difficult problem in large-scale wireless sensor networks. In literature, various localization solutions have been proposed, many of which adopted Received Signal Strength (RSS) as measurements for its advantage of low-cost. However, RSS measurements usually have the drawback of low accuracy, especially in noisy environments, which affects the accuracy of RSS-based localization solutions. Additionally, many RSS-based localization solutions assume an accurate path loss exponent to be known a priori, which is an over-simplified assumption in practice. During the last decade, the concept of cooperative localization has been proposed and applied by simulating the force relaxation process in a mass-spring system, which caught many researchers' attention. Upon these observations, this research is conducted to investigate the suitability of the Spring-Relaxation (SR) based localization using RSS measurements, and extend it to tackle existing problems. At the first stage of this research, an in-depth study is performed on a pure SR algorithm which is in an iterative manner, by directly applying it to sensor localization using RSS measurements. Analysis of convergence and stationary properties is conducted, which serves as the scientific foundation of the SR-based localization. System parameter design is also carried out to ensure the stability of the iterative algorithm and leverage between convergence speed and localization accuracy in a typical setup. The extensive study of the pure SR algorithm confirms the suitability of the SR-based localization, but the localization accuracy is not guaranteed in presence of heavy noise in complex signal propagation environments. Therefore, at the second stage of this research, new localization solutions are proposed to improve localization accuracy in noisy environments from different aspects. From one aspect, Kalman filtering is introduced into the SR-based localization to pre-process noisy distance measurements, which can be translated from RSS measurements if an accurate path loss exponent is assumed available. From another aspect, a neighbor selection scheme designed upon Voronoi diagram is introduced to choose a set of neighboring nodes that are spatially dispersed and distantly close to participate in the cooperative localization. The aim of neighbor selection scheme is to include neighbors with high probability to benefit the localization accuracy, and rule out neighbors that may bring negative impact to the localization accuracy. Finally, aware that the assumption of an accurate path loss exponent is over-simplified thus infeasible, the impact of exponent accuracy is examined on localization accuracy. Based on the study, the relationship between the exponent estimation and localization error is found to be in exponential order, which is named as the error magnification effect. A localization solution named variable-elasticity spring-relaxation is then proposed to suppress the error magnification effect using both passive and active measures. The passive measure enforces SR algorithm to operate at a situation where localization errors are stable, whereas the active measure introduces variable elasticity into SR algorithm to suppress the exponential ranging error. Extensive simulations are conducted using physical layer settings to demonstrate the favorable performance of these localization solutions proposed. Combining all pieces together, a complete study of the spring-relaxation technique is achieved for cooperative localization.