Time Estimation of Superimposed Coherent Multipath Signals using the EM Algorithm for Global Positioning System

Abstract

A novel multipath mitigation technique for Global Positioning System (GPS) receivers using the Expectation-Maximization (EM) algorithm is proposed. It is well-known that conventional propagation delay estimation using parallel sliding correlators is only optimal in additive white Gaussian noise channel. In practical positioning systems, the weak GPS line-of-sight signal is generally embedded in the multipath signals and other source of interference. Although the GPS direct sequence spread spectrum (DS-SS) signal has inherent resistance to interference, the received superimposed multipath signals, which are possibly coherent, are the dominant source of the propagation delay estimation errors. From the parameter estimation point of view, the problem of multipath mitigation is equivalent to estimating the unknown phases, propagation delays and amplitudes of the superimposed multipath signals. The joint maximum likelihood (ML) estimation of all the unknown parameters is optimal and asymptotically efficient. However it involves multi-dimensional search which is computationally expensive. The proposed coarse/acquisition (C/A) code acquisition system using the EM algorithm is an iterative maximum likelihood estimator which decomposes the multi-parameter estimation problem into a number of separate ML optimizations. The performance of the proposed EM algorithm has been tested by simulations. We have observed that the proposed acquisition system is significantly superior to the conventional correlating receiver in a multipath fading channel.