Uncertainty in target self-localization
Published:
Abstract
This work considers the problem of self-localization of a target node using quantized power measurements from BLE beacon nodes transmitting from known locations. We propose two approaches to studying the performance of target localization. Firstly, the average area uncertainty in localization is derived in terms of expected area coverage of a beacon. Using the derived expression, the optimal beacon radius which minimizes the average area uncertainty is determined. Secondly, the geographical area is overlaid with a virtual grid, and the problem is treated as one of testing overlapping subsets of grid cells for the presence of the target node.
The benefit of considering the problem in this framework is that it then becomes one of group testing, where, in each test, subsets of individuals (grid locations) are tested for the presence of defective individuals (targets). From the literature on group testing, a column matching algorithm is considered for devising the target localization algorithm. The minimum node density required to localize the target within a region of accuracy, with high probability, is determined using the tools from Poisson point processes and order statistics.
The derived theoretical expressions, proposed algorithm, and design procedure are validated both by Monte Carlo simulations as well as using experimental data collected from commercially-off-the-shelf bluetooth low energy (BLE) beacon nodes. We also extend the problem to an outdoor setting, where energy-harvesting beacons are used. We performed Monte Carlo experiments on MATLAB to verify our claims with empirical results.
The benefit of considering the problem in this framework is that it then becomes one of group testing, where, in each test, subsets of individuals (grid locations) are tested for the presence of defective individuals (targets). From the literature on group testing, a column matching algorithm is considered for devising the target localization algorithm. The minimum node density required to localize the target within a region of accuracy, with high probability, is determined using the tools from Poisson point processes and order statistics.
The derived theoretical expressions, proposed algorithm, and design procedure are validated both by Monte Carlo simulations as well as using experimental data collected from commercially-off-the-shelf bluetooth low energy (BLE) beacon nodes. We also extend the problem to an outdoor setting, where energy-harvesting beacons are used. We performed Monte Carlo experiments on MATLAB to verify our claims with empirical results.