A Hybrid TDOA-Fingerprinting-Based Localization System for LTE Network

To the best of our knowledge, effective indoor positioning for the long term evolution (LTE) has not been addressed in the literature. In this paper, the uplink sounding reference signal (SRS), which contains timing and received signal strength (RSS) information, is exploited to achieve this task. Accordingly, an LTE based localization system using time-difference-of-arrival (TDOA) and fingerprint of RSS that operates in two steps is devised to meet the requirement of high-precision indoor positioning. In the first step, peak value detection is applied to the SRSs from spatially separated sensors to estimate the TDOAs, from which a coarse target location is computed with the use of least squares. Based on deep neural network, fingerprinting in a subarea containing the coarse solution is then performed to obtain the final position estimate. Due to the instability in the RSS values caused by non-line-of-sight and multipath propagation, we propose a novel extraction method to select reliable RSSs as input to the network. Experimental results show that our two-step localization system can provide an accuracy of sub-meter level.