A Framework for Improving the Performance of Direct Position Determination by Normal Hedge Algorithm for Emitter Localization in Indoor Applications

Indoor localization is a way to determine the location of transmitter(s) by using wireless network sensors without using GPS. In this paper, it is assumed that a random signal is transmitted by an unknown-position emitter through a noisy and lossy channel and then is received by known-position wireless sensors. Unlike the classical algorithms such as TDOA or AOA, the direct position determination (DPD) approach estimates the position of emitter in one step using the observation signal from all the collector receivers. In this paper a framework of methods is proposed to develop the DPD formulations. Originally, a preprocessing is applied on signals and later, a tree search algorithm and a fine localization algorithm will be proposed in order to achieve a higher resolution and with a less volume of computing. Furthermore, an approach is proposed to increase the performance of this algorithm. Using Normal Hedge algorithm with a proposed loss function, fuses the estimation results of several runs of the fine DPD algorithm. The simulation results are shown that the proposed framework would eventually improve the performance of the DPD algorithm.