Comprehensive Analysis of Multipath Estimation Algorithms in the Framework of Information Theoretic Learning

Abstract

Multipath interference is considered as one of the dominant error sources for high-precision positioning systems. Multipath estimation plays a significant role in eliminating multipath error to improve the positioning precision. In this paper, a category of information theoretic learning (ITL)-based multipath estimation algorithms are studied in non-Gaussian noise. Apart from the previously proposed ITL-based multipath estimation algorithms, the minimum error entropy criterion-based algorithm and the centered error entropy criterion-based algorithm, a new multipath estimation algorithm using survival information potential (SIP) is also proposed. For the SIP-based multipath estimation algorithm, the SIP criterion is adopted as an index to measure the performance of estimation results. The goal of applying SIP in multipath estimation is to solve the problem that the previously proposed algorithms based on ITL are time consuming due to the calculation of multi-dimensional Gaussian kernel function. The comparisons of the three ITL-based algorithms are comprehensively analyzed in theory at first. Then, a case study of multipath environment in non-Gaussian noise is given to further demonstrate the effectiveness of the proposed algorithm in terms of estimation accuracy, randomness, and the calculation complexity for multipath estimation.