Method and Analysis of Spectrally Compressed Radio Images for Mobile-Centric Indoor Localization

Abstract

Large databases with Received Signal Strength (RSS) measurements are essential for various use cases in mobile wireless communications and navigation, including radio resource management algorithms and network-based localization. Because of the constantly increasing number of radio transmitters with various wireless technologies and with the advent of 5G cloud computing and Internet of Things (IoT), the required size of the RSS databases are becoming unmanageably large. Thus, the requirements for the bandwidth and data rates for accessing the memory might become too costly. Therefore, in order to reduce the size of the RSS database, while maintaining the data quality, we have previously proposed the method of spectrally compressed RSS images, which are able to achieve considerable data compression of up to 70%. In this paper, we deeply analyze the process of spectral compression and introduce error sources, which affect the compression performance. Based on the analysis, we propose a novel theoretical framework and methods to optimize the performance of the spectral compression. In addition, we derive the Cramér-Rao Lower Bound (CRLB) for the RSS-based localization error and compare the CRLB between separate baseline localization approaches. The theoretical analysis is justified and compared with experimental RSS measurements taken from several multi-storey buildings.