An Effective Method for Multipath Mitigationunder Severe Multipath Environments

摘要

The support systems to supply highly accurate positioning,such as DGPS or RTK-GPS, must be effective even in thecase of severe multipath environments. Therefore, someresearches related to the code multipath mitigation underrelatively severe multipath environments have beenpresented in recent years. Generally speaking, multipathphenomenon is caused not only by reflection but also bydiffraction and obstruction. Sometimes we happen toreceive only signals reflected by tall buildings. Althoughsome correlator equipped with multipath mitigationtechniques can reduce the multipath error considerably inthe case of direct and multipath signals only, thesetechniques can't cope with the cases that the multipathsignal is dominant, diffraction loss is large and the delayof multipath signal is shorter than about 20 m. In fact,these types of signal can be received frequently when wedrive in dense urban environments with slow speed ofseveral km/hour. We propose two techniques in this paperto reduce the multipath error in the above cases.Signal strength is used as a criterion to remove themultipath dominant signals. Furthermore, in order toreduce the effect of the short delay multipath, thetechnique aided by maximum likelihood criteria, whichhas already been proposed by some researchers, is appliedto estimate the multipath signal. The feature of ourmultipath estimation technique is to infer multipathparameters rapidly by making use of multi-correlatoroutputs. Reducing search space by deducing accurateinitial multipath parameters enables us to reducecomputational time. The analyses of many field data,obtained by ourselves, have indicated that our twotechniques were effective to reduce errors by themultipath dominant signal and the short delay multipathsignal. Simulation was also conducted to evaluate ourmultipath estimation technique, which was tested in theboth static and semi-static cases. The method to removemultipath dominant signals was tried in all conditions.