Performance analysis of indoor localization based on channel state information ranging model

Due to robustness against multi-path effect, channel state information (CSI) of Orthogonal Frequency Division Multiplexing (OFDM) systems is supposed to provide accurate distance measurement for indoor localization. However, we find that the original CSI ranging model is biased, so the model cannot be used to directly derive Cramer-Rao lower bound (CRLB) of positioning error for CSI-ranging based localization scheme. In this paper we first analyze the estimation bias of the original CSI ranging model according to indoor wireless channel model. Then we propose a negative power summation ranging model which can be used as an unbiased ranging model for both Line-Of-Sight (LOS) and Non-LOS scenarios. Subsequently, based on the proposed model, we derive both the CRLB of ranging error and the CRLB of positioning error for CSI-ranging localization scheme. Through simulation we validate the bias of the original ranging model and the approximately zero bias of our proposed ranging model. Through comprehensive experiments in different indoor scenarios, localization errors by different ranging models are compared to the CRLB, meanwhile our proposed ranging model is demonstrated to have better ranging and localization accuracy than the original ranging model.