Multipath Estimation Based On Differential Evolution

Multipath interference is one of the main sources of error in high precision navigation systems.The occurrence of multipath is uncertain in respect of time and irrelevant at different positions,which makes it impossible to eliminate by the differential techniques.Thus,multipath elimination has become the main obstacle affecting the positioning precision of navigation systems.To solve this problem,this paper focuses on the multipath error elimination method based on data processing.This method is widely used in software receivers because of its low cost,good generality,and good popularity.The key of multipath error elimination method based on data processing is parameter estimation.The commonly used multipath estimation methods are Extended Kalman Filter(EKF)multipath estimation methods and Particle Filter(PF)based multipath estimation methods.However,these two kind of methods have the following disadvantages:(1)The EKF multipath estimation method is only suitable for Gaussian noise.Linearization errors are inevitably generated when nonlinear systems are linearized.Apart from that,the algorithm is also sensitive to the initial state;(2)The PF multipath estimation method is suitable for non-Gaussian noise,but it cannot avoid the problem of particle impoverishment in multipath estimation,which may lead to the convergence of the estimation result to the local optimal solution,and the estimation result fluctuates greatly;(3)Under multipath condition,there is a problem of degradation of estimation performance;(4)A multipath estimation algorithm that can be applied to both Gaussian noise and non-Gaussian noise has not been proposed yet.To solve problem(1)and(2),a new PF algorithm based on improved differential evolution(DEPF)is proposed.In order to overcome the problem of particle exhaustion caused by standard resampling,the DE algorithm instead of the re-sampling process is used to generate new particles in PF.The simulation results shows that the DEPF algorithm has better estimation performance than PF and EKF in non-Gaussian noise.To solve problem(2)and(3),a multipath estimation algorithm based on improved intelligent optimization algorithm is proposed in this paper,namely?RDE algorithm.By using the second moment of the estimation error as the objective function of the ?RDE algorithm,the prior information of multipath parameters and the instantaneous error as constraints of the ?RDE algorithm,the multipath estimation problem is transferred into an optimization problem with constrained conditions.The simulation results show that the ?RDE algorithm has good steady-state performance,and has solved the multipath estimation problem of single-path and two-path multipath at the same time.As for problem(4),this paper proposes a multipath estimation algorithm based on ?-level constraint differential evolution with the Minimum Error Entropy(MEE)criterion.The MEE criterion is adopted as the objective function.The statistical information of the estimated error and the prior information of the multipath parameters are used as constraints in the algorithm.The proposed algorithm can ensure that the estimation results have the least randomness while converging to the true value.The MEE-based ?RDE can be applied to both Gaussian and non-Gaussian noises.The research contents of this paper come from the National Natural Science Foundation of China(No.61603267,No.61503271)and the Natural Science Foundation of Shanxi Province(No.20140210022-7).The results of this studyprovide a new direction for multipath estimation,which is useful and instructive in terms of theory research and application.