Farooq-I-Azam, Muhammad and Ni, Qiang (2017) Intelligent energy efficient localisation in wireless sensor networks. PhD thesis, Lancaster University.
2017farooq_i_azamphd.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (11MB)
Abstract
A wireless sensor network comprises of tiny sensor nodes which communicate with each other through radio frequency communication links. In many applications of wireless sensor networks, the sensor nodes collect data or detect an event and report it to a central node for further processing. Location information of the node is sent with the data else it may not be useful. Therefore, a sensor node should know its geographic position coordinates. Localization solutions, such as GPS are not feasible due to their energy cost and size. Hence, sensor nodes estimate their positions using an algorithm. This thesis focuses on the localization of sensor nodes and related key issues, such as performance evaluation of localization algorithms, development of analytical model and analysis of localization error. Firstly, this thesis proposes three new novel metrics which can be used for the performance evaluation of three dierent aspects of localization algorithms. Alongside, we also present a comprehensive review of metrics which are used in literature for the measurement and characterization of localization errors. Secondly, we present an intelligent algorithm which we call ripple localization algorithm. The algorithm is distributed, energy efficient and does not require additional hardware for range estimation. The algorithm also provides control over localization granularity which makes it suitable for wide range of wireless sensor network and localization applications. Simulation results show that the algorithm gives good performance and localization accuracy even under irregular radio conditions. Thirdly, we give a new technique for solving multilateration equations and show that the overdetermined system of equations resulting from multilateration can be reduced to a pair of simultaneous equations which can then be solved using conventional techniques, such as Cramer's rule. Fourthly, we develop and present an analytical model of localization error resulting from trilateration. The multilateration solution technique and analytical model are veried using simulation. Finally, we analyze trilateration errors in short range wireless networks, such as wireless sensor networks and internet of things where the distance estimation errors are sometimes comparable to the actual distances. We investigate the minimum and maximum values of localization errors in these networks. We also derive a number of other useful results. For example, we show that the localization error due to positive range estimation errors equal to the actual distances is 3 times the localization error resulting from the same magnitude of negative range estimation errors. All the results are tested and verified using a comprehensive set of simulation experiments.