Application of Superresolution Concepts to the GPS Multipath Mitigation Problem

Lawrence R. Weill

Abstract:	Although it is now known how to achieve code pseudoranging accuracy at or near theoretical limits for single-path propagation, it is recognized that multipath remains as a dominant source of ranging error in differential positioning. It is a critical issue in the development of high perfomance applications such as high-productivity survey equipment, high-precision aircraft landing systems, and the emerging WAAS system. Examples of recently developed high-speed in-receiver multipath mitigation methods include the narrow correlator, Multipath Elimination Technology (MET), the Multipath Estimating Delay Lock Loop (MEDLL), and the strobe correlator. The best of these methods rely on estimation-theoretic principles and are driven by the desire to approach theoretical performance limits (for example, MEDLL). However, they are complex and expensive to implement because of the need to measure the shape of the received signal cross-correlation function with multiple correlators and to process these measurements with complex algorithms. In this paper, we take a different approach to the multipath mitigation problem which is both effective and very low in cost: Instead of generating a correlation function in the usual way and then attempting to measure its shape, pre-process the received signal so that it produces a correlation function with much higher range resoution (superresolution). This approach has several important advantages: 1. The pre-processing requires only one linear filter common to all satellite channels. 2. The improved range resolution greatly reduces correlation function distortion, thus permitting the use of standard code tracking loops without the computational load inherent in shape-based methods. 3. Just one punctual-only, code-tracking correlator is required per satellite channel, together with an additional correlator for data recovery and removal of data influence on the tracking correlator. 4. Performance is coparable to the best of published methods. Code and phase multipath bias errors are completely eliminated for path separations greater than 15-20 meters, with substantial mitigation otherwise, using the full GPS signal bandwidth. The paper first reviews theoretical aspects of the multipath problem, followed by a description of the superresolution cocept. A definition of resolution is given and an analytical tool for assessing error due to random noise is developed. Two equivalent implementations of superresolution processing are presented, followed by examples of performance, including tradeoffs which are possible between the magnitudes of bias and random error components of the range estimates. Although the superresolution approach to multipath mitigation does not reach theoretical accuracy limits for multipath-induced error previously established by the author, it offers performance comparable to the best of published methods at a much lower cost.
Published in:	Proceedings of the 1998 National Technical Meeting of The Institute of Navigation January 21 - 23, 1998 Westin Long Beach Hotel Long Beach, CA
Pages:	673 - 682
Cite this article:	Weill, Lawrence R., "Application of Superresolution Concepts to the GPS Multipath Mitigation Problem," Proceedings of the 1998 National Technical Meeting of The Institute of Navigation, Long Beach, CA, January 1998, pp. 673-682.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In