Super-Resolution-Aided Positioning Fingerprinting Based on Channel Impulse Response Measurement

Abstract

Position fingerprinting (FP), in which the signature of a position is captured from the radio frequency signal, is one of the most efficient indoor positioning algorithms. Besides the received signal strength (RSS), the channel impulse response (CIR) is regarded as a linear temporal filter, which characterizes the multipath channel of the operating environment. Since the CIR requires a larger system bandwidth to distinguishing individual paths along which the signal waves travel, a smaller bandwidth may limit the performance of CIR- based fingerprinting. In this paper, we utilize a super-resolution method, i.e. the multiple signal classification (MUSIC) algorithm, to obtain the pseudo-spectrum for the enhanced resolution of the arriving paths. We create an offline database by the implementation of an OFDM-based channel sounder and obtain the cumulative pseudo-spectrum based on the alignment of the maximal power path. Based on the online#x002F;offline measurements with enhancing resolutions, we propose a super-resolution-aided fingerprinting (S- FP) to estimate the position by finding the reference points (RPs) with the highest similarity of the cumulative pseudo-spectrum. The experimental results show that S-FP reduces the localization error compared with the conventional CIR FP.