A Pre-processing Algorithm Utilizing a Paired CRLB for TDoA Based IoT Positioning

Abstract

Time Difference of Arrival (TDoA) is currently viewed as an important technique for the positioning capabilities in the Internet of Things (IoT). However, in the case of practical measurement, not all the TDoA values between the gateways have the same impact on the localization accuracy. In this paper, a novel TDoA pre-processing methodology for dropping out the outlier TDoA values is presented, after instrumentalizing a paired Cramér-Rao lower bound (CRLB). Thus, the proposed approach is detecting the best TDoA values, which have the lowest paired CRLB values specifically, in the vicinity of the guessed node location, based on a robust thresholding method. A comparison is performed investigating the attainable accuracies for localizing based on this pre-processing algorithm, on a well-defined simulation environment. This simulator is based on a Poisson distribution approach for defining the gateways and node positions, as well as a noise model for emulating the timestamp imperfections. In the given results, the feasibility of the proposed technique is asserted by a drastic improvement over a wide range of the number of gateways as well as measurement noise variances. This manifests the robustness of the contributed method to the outlier TDoA values and its valuable rendering.