Distributed localization algorithm for wireless sensor networks using range lookup and subregion stitching

Farhan Khan, Sing Kiong Nguang
2021 IET Wireless Sensor Systems  
One of the ways in which localization algorithms in wireless sensor networks (WSNs) have been categorized is whether they are range-based or range-free. Range-based algorithms use expensive hardware to measure one or more physical quantities and, in turn, use them to localize nodes with greater precision. In contrast, range-free algorithms use coarse-grained quantities like connectivity to localize nodes with limited precision. A middle way between these two approaches can be called a partial
more » ... nge-based approach that can utilize the existing received signal strength indicator (RSSI) readings from sensor nodes to improve the already existing coarse-grained localization methods. Another important consideration in WSNs is that a distributed localization algorithm is more computationally feasible as compared to its centralized counterpart. Keeping these two considerations in mind, a distributed localization algorithm is proposed here which falls in the aforementioned partial range-based category. The proposed algorithm called RangeLookup-MDS first creates subregions using connectivity information only. This is followed by the collection of RSSI readings from individual sensor nodes that are used to perform range lookup for inter-node distance estimates in a lookup table. After that, relative localization in every subregion is performed using multidimensional scaling, and then the relative maps are stitched together to create a consistent (but relative) coordinate system. The algorithm also has the capability to compute absolute coordinates in two-dimensional if the stitching step is executed with at least three non-collinear anchor nodes with known locations. Simulation results on uniform as well as irregular networks of various sizes show that the proposed algorithm provides improved localization accuracy and reduces localization error up to 25% in comparison to a previous partial range-based localization algorithm.
doi:10.1049/wss2.12020 fatcat:7tt3fxqxrrbkrpbt5g2kegux24