Multipath Error Mitigation of Low-Cost GPS Receiver Using PSO-SVM and GA-SVM Hybrid Methods

One of the major errors that affect GPS accurately is the multi-path effect of each receiver. Multi-paths is receiving an antenna signal from more than one path, multi-path effect is a major source of unknown error in positioning and is not eliminated by differential methods. This effect is largely dependent on the environment specific to each receiver and it is low-frequency effect. The geometry between GPS satellites and the specific location of each receiver is repeated on astronomical days, the multi-path effects tend to behave similarly on consecutive days. In this paper, a method for extracting the multi-path effects behavior was applied to the GPS-code observations, multi-path error mitigation increases the accuracy of positioning. In the proposed method, the residual signal is generated based on the dual difference (DD) and is used as the input of the proposed algorithm. Support Vector Machine (SVM) is used for multi-path approximation. To determine the basic parameters of SVM and its kernel function, particle optimization algorithms (PSO) and genetic algorithm (GA) were used. In order to evaluate the accuracy of the proposed method, simulation and experimental based on two stations (reference and user) and two low-cost receivers were designed. The proposed methods were tested based on practical data. The experiments showed that the multi-path error of the receiver of the user's station decreased by 70% in the static test based on the RMS criterion. Models of this paper have been compared with some recent models presented in the context of multi-path error reduction. The results showed that the proposed model had better performance than other methods. The result is high accuracy and stability in positioning results. Three-dimensional position accuracy improved by about 56% after using the proposed method, reaching 1.60 m.