Navigation multipath signal estimation method based on particle filtering
By combining the extended Kalman filtering and genetic algorithm, the problems of particle degradation and diversity reduction in particle filtering are solved, high-precision estimation of navigation signals is achieved, and navigation positioning accuracy is improved.
Patent Information
- Application Number
- CN202411240821.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-05
- Publication Date
- 2024-12-31
AI Technical Summary
The existing navigation signal estimation method based on particle filtering has problems such as particle degradation and reduced particle diversity, resulting in slow convergence speed or inability to converge, affecting the accuracy of navigation signals.
Using a combination of extended Kalman filtering (EKF) and genetic algorithms, the diversity and adaptability of particle swarms are ensured by initializing particle parameters, updating particle weights, and using elite retention strategies and genetic algorithms to resample, cross-operation and mutation operations. Improve the accuracy of parameter estimation.
High-precision parameter estimation of direct signals and multipath signals is realized, reducing the impact of multipath signals on navigation signal performance and improving navigation positioning accuracy.
Smart Images
Figure CN119224796A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of satellite navigation and positioning, and in particular to a navigation multipath signal estimation method based on particle filtering. Background Art
[0002] The Global Navigation Satellite System (GNSS) plays a vital role in modern society. It helps users achieve accurate positioning, navigation and timing services by transmitting high-precision spatiotemporal information to the ground. It is widely used in many fields such as transportation, emergency rescue and military. With the continuous progress of society, users' demand for navigation and positioning is growing, and the importance of satellite navigation systems is becoming increasingly prominent.
[0003] However, satellite navigation systems face many challenges in practical applications, among which the impact of multipath signals on navigation signals is particularly significant. The multipath effect refers to the fact that during the propagation of navigation signals, the navigation signal is reflected or refracted by obstacles, causing the signal to reach the receiver along multiple paths, resulting in distortion of the navigation signal and carrier phase distortion, making it difficult for the receiver to accurately capture and track the signal, ultimately affecting the positioning accuracy. The main manifestations of multipath signals are changes in signal propagation delay and signal amplitude, which will change with the environment of the receiver and have uncertainty in time and position. With the continuous advancement of urbanization and the increase in the density of high-rise buildings, the impact of complex urban environments on navigation signals is becoming increasingly significant. Reflections from buildings, signal shielding, and the high density of objects exacerbate the multipath effect, significantly reducing the positioning accuracy of the navigation system. Dynamically changing urban environments, such as traffic flow and pedestrian movement, further exacerbate this impact.
[0004] There are three main methods for processing multipath signals: digital signal processing, data post-processing, and antenna-based processing. Digital signal processing technology is a hot topic in multipath signal processing technology. Digital signal processing technology uses the principle of parameter estimation to estimate multipath signals based on the relevant output values of the received signal, thereby obtaining direct signals. Currently, particle filtering is often used for signal estimation. Although particle filtering has strong flexibility and applicability, particle filtering has problems such as particle degradation and sample exhaustion, which may lead to slow convergence or failure to converge. Summary of the invention
[0005] The purpose of the present invention is to provide a navigation multipath signal estimation method based on particle filtering, aiming to solve the problems of particle degradation and reduced particle diversity in the existing signal estimation methods based on particle filtering, and to achieve high-precision parameter estimation of direct signals and multipath signals, thereby reducing the impact of multipath signals on navigation signal performance.
[0006] To achieve the above object, the present invention provides a navigation multipath signal estimation method based on particle filtering, comprising the following steps:
[0007] Step 1: Model and sample the navigation signal received by the receiver;
[0008] Step 2: Initialize the sampled signal and set parameters;
[0009] Step 3: Use EKF algorithm to update particle parameters;
[0010] Step 4: Use the probability density function of Gaussian distribution to obtain the weight of each particle;
[0011] Step 5: Use the elite retention strategy to retain high-weight particles;
[0012] Step 6: Use genetic algorithm to resample and select particles;
[0013] Step 7: Crossover operation generates a new particle swarm;
[0014] Step 8: Perform mutation operation, combine the number of iterations with the fitness of the particle, converge to obtain the optimal particle, and obtain the parameter estimation value.
[0015] Optionally, the navigation signal received in step 1 is composed of a direct signal and a multipath signal, and the sampled signal is expressed as follows:
[0016]
[0017] Where A0 represents the guidance signal amplitude, A i represents the multipath signal amplitude, c(t) is the pseudo code sequence, k is the kth sampling point, k i is the relative signal delay, T s is the sampling time of the signal, cos(ω0kT s ) is the carrier, M represents the number of multipath signals, n(kT s ) is random noise.
[0018] Optionally, the execution process of step 2 is specifically as follows: first, a sampled local pseudocode signal is generated, and based on the local pseudocode signal, local delay signals with different delays are generated to form a local signal group; according to the characteristics of the multipath signal, four parameters to be estimated, namely, the direct signal amplitude, the multipath signal amplitude, the direct signal delay, and the multipath signal delay are defined; the number of particles is defined as Np, the number of iterations is defined as N times, the signal initial value, the state matrix, and the process noise matrix are defined, and the measurement noise variance is obtained according to the correlation results between the received signal and the local signal group.
[0019] Optionally, during the execution of step 3, the EKF algorithm and the latest observation information are used to obtain the sample mean and covariance, and the obtained parameters are used to update the particle parameters; the formula used by the EKF algorithm is as follows:
[0020]
[0021] P k|k-1 =FP - k|k T +Q
[0022]
[0023] P k|k =(IK k H k ) k|k-1
[0024] in, is the prior state transfer estimate, F is the state transfer matrix, w k-1 is the process noise, is the posterior state estimate, P k|k and P k|k-1 are the posterior and prior error covariances at the current moment, P - k|k is the posterior covariance matrix of the previous moment, Q is the process noise covariance matrix, z k represents the actual measurement value at time k, H represents the measurement matrix, v k is the measurement noise, K k is the Kalman gain, R k represents the measurement covariance matrix, and h(·) is the nonlinear measurement function.
[0025] Optionally, the probability density function expression in step 4 is
[0026]
[0027] Where n is the dimension of the state vector, μ is the measurement prediction value, R is the measurement noise covariance matrix, and x is the actual measurement value.
[0028] Optionally, the execution process of step 6 is specifically to resample the particles using a genetic algorithm, select the particles based on particle weights using a roulette wheel selection algorithm, select particles according to a ratio of particle weights, and select particles with high weights.
[0029] Optionally, the expression of the crossover operation in step 7 is as follows:
[0030]
[0031] where F max is the maximum mutation rate, F min is the minimum mutation rate, I is the current number of iterations, and M is the maximum number of iterations.
[0032] Optionally, the expression of the mutation operation in step 8 is as follows:
[0033]
[0034] Among them, G is the initial mutation rate, g i is the fitness of the current particle, and g is the average fitness.
[0035] The present invention provides a navigation multipath signal estimation method based on particle filtering, which combines the extended Kalman filter (EKF) and the genetic algorithm to jointly optimize the parameter estimation of the particle filter (PF). First, in the PF importance sampling stage, the EKF algorithm and the latest observation information are used to obtain the sample mean and covariance, which provides a reasonable recommended density for the particle filter and can recalculate the weight of each particle. Then, in the resampling stage of the particle filter, the improved genetic algorithm is used to copy and eliminate the particle set to avoid the problem of the particle filter falling into the local extreme value. The method proposed by the present invention can solve the particle degradation and particle diversity reduction problems existing in the particle filter, and realize the high-precision parameter estimation of the direct signal and the multipath signal, thereby reducing the influence of the multipath signal on the navigation signal performance. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0037] Figure 1 The present invention is a flowchart of the steps of a navigation multipath signal estimation method based on particle filtering.
[0038] Figure 2 It is a comparison diagram of multipath signal estimation results in the absence of noise according to a specific embodiment of the present invention.
[0039] Figure 3 It is a schematic diagram of the noise-free time mean square error result of a specific embodiment of the present invention.
[0040] Figure 4 It is a schematic diagram of multipath signal estimation results when SNR=-20dB in a specific embodiment of the present invention.
[0041] Figure 5 It is a schematic diagram of the mean square error result when SNR=-20dB in a specific embodiment of the present invention.
[0042] Figure 6It is a schematic diagram of particle weights of the PF algorithm after 500 iterations in a specific embodiment of the present invention.
[0043] Figure 7 It is a schematic diagram of particle weights of the algorithm of the present invention after 500 iterations in a specific embodiment of the present invention. DETAILED DESCRIPTION
[0044] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.
[0045] See also Figure 1 The present invention provides a navigation multipath signal estimation method based on particle filtering, comprising the following steps:
[0046] S1: Modeling and sampling of navigation signals received by the receiver;
[0047] S2: Initialization of the sampled signal and parameter setting;
[0048] S3: Update particle parameters using EKF algorithm;
[0049] S4: Use the probability density function of Gaussian distribution to obtain the weight of each particle;
[0050] S5: Using elite retention strategy to retain high-weight particles;
[0051] S6: Resample and select particles using genetic algorithm;
[0052] S7: Crossover operation generates a new particle swarm;
[0053] S8: Perform mutation operation, combine the number of iterations with the fitness of the particle, converge to obtain the optimal particle, and obtain the parameter estimation value.
[0054] The following is a further explanation based on the specific implementation steps:
[0055] In step S1, the navigation signal received by the receiver is modeled. The signal received by the receiver consists of a direct signal and a multipath signal. The sampled signal is expressed as follows:
[0056]
[0057] Where A0 represents the guidance signal amplitude, A i represents the multipath signal amplitude, c(t) is the pseudo code sequence, k is the kth sampling point, k i is the relative signal delay, Ts is the sampling time of the signal, cos(ω0kT s ) is the carrier, M represents the number of multipath signals, n(kT s ) is random noise.
[0058] Step S2 is the initialization design process. First, a sampled local pseudocode signal is generated. Based on the local pseudocode signal, local delay signals with different delays are generated to form a local signal group. According to the characteristics of the multipath signal, four parameters to be estimated, namely, the direct signal amplitude, the multipath signal amplitude, the direct signal delay, and the multipath signal delay, are defined. The number of particles is defined as Np, the number of iterations is defined as N, the signal initial value, the state matrix, and the process noise matrix are defined, and the measurement noise variance is obtained according to the correlation results between the received signal and the local signal group.
[0059] In step S3, the sample mean and covariance are obtained using the EKF algorithm and the latest observation information, and the particle parameters are updated using the obtained parameters; the formula used by the EKF algorithm is as follows:
[0060]
[0061] P k|k-1 =FP - k|k T +Q
[0062]
[0063] P k|k =(IK k H k ) k|k-1 .
[0064] Step S4, using the probability density function of Gaussian distribution to obtain the weight of each particle.
[0065]
[0066] Where n is the dimension of the state vector, μ is the measurement prediction value, R is the measurement noise covariance matrix, and x is the actual measurement value. The smaller the difference between the measurement and the prediction, the greater the weight of the particle will be.
[0067] Step S5: Since the genetic algorithm may lose particles with high fitness during the evolution process, the elite retention strategy is used to directly copy the particle individuals with larger weights to the next generation without crossover and mutation operations, which can ensure that the advantages of high-weight particles are retained and improve the overall performance of the algorithm.
[0068] In step S6, the particles are resampled using a genetic algorithm, and the particles are selected based on particle weights using a roulette wheel selection algorithm. Particles are selected according to the proportion of particle weights, and particles with higher weights have a greater probability of being selected.
[0069] In the execution process of step S7, a crossover operation is used to randomly select a bunch of particles for exchange to generate a new particle group, so as to combine the characteristics of different particles to find a better solution. The crossover probability can be gradually reduced as the iteration proceeds to increase the stability of the population.
[0070]
[0071] where F max is the maximum mutation rate, F min is the minimum mutation rate, I is the current number of iterations, and M is the maximum number of iterations.
[0072] In step S8, a mutation operation is performed to introduce particle diversity, prevent the particle swarm from falling into a local optimal solution, and better explore the space. The number of iterations is combined with the fitness of the particles to flexibly adjust the mutation rate.
[0073]
[0074] Among them, G is the initial mutation rate, g i is the fitness of the current particle, and g is the average fitness. In terms of fitness, particles with low fitness are allowed to use high mutation rates, and particles with high fitness are allowed to use low mutation rates, thereby enhancing the exploratory nature of particles. In terms of the number of iterations, the mutation rate is gradually reduced to facilitate early exploration and later convergence. Finally, the optimal particle is obtained and the normalized weight of the particle is calculated to obtain the parameter estimation value.
[0075] Furthermore, in order to verify the effectiveness of the method proposed in the present invention, the present invention proposes a specific embodiment, which can be found in detail in Figures 2 to 6 , the estimation results of the present invention are compared with those of EKF and PF. Assume that there is a multipath signal, the state transfer matrix is a fourth-order unit matrix, and the actual direct signal amplitude, multipath signal amplitude, direct signal delay, and multipath signal delay are 1, 0.7, 0, and 0.3 respectively. The number of particles is 100 and the number of iterations is 500.
[0076] Figure 2 and Figure 3 This is a comparison chart of three algorithms when there is no noise. In the estimation of the four parameters of direct signal amplitude, multipath signal amplitude, direct signal delay, and multipath signal delay, the estimation accuracy of the method of the present invention is higher than that of the EKF and PF algorithms. The direct signal can be obtained by subtracting the estimated multipath signal from the received signal, and then the subsequent receiver capture, tracking, demodulation and other operations can be performed.
[0077] Figure 4 and Figure 5 This is a comparison chart of the three algorithms when the SNR is -20dB. In the estimation of the four parameters of direct signal amplitude, multipath signal amplitude, direct signal delay, and multipath signal delay, the estimation accuracy of the method of the present invention is still good.
[0078] Figure 6 The particle weight changes of the PF algorithm after 500 iterations are given. The weight distribution of the particles in the figure is relatively sparse, with fewer particles with large weights and more particles with low weights. The particle degradation problem may be more serious. Figure 7 This is the estimation method proposed by the present invention. As can be seen from the figure, the weight distribution of particles is relatively uniform, which ensures the diversity of particles, reduces the problem of particle degradation, and improves the robustness and accuracy of estimation.
[0079] The present invention selects a reasonable suggested density for particle filtering by utilizing EKF and adopts a genetic algorithm to increase particle diversity.
[0080] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.
Claims
1. A navigation multipath signal estimation method based on particle filtering, characterized in that: The following steps are involved: Step 1: Model and sample the navigation signal received by the receiver; Step 2: Initialize the sampled signal and set parameters; Step 3: Use EKF algorithm to update particle parameters; Step 4: Use the probability density function of Gaussian distribution to obtain the weight of each particle; Step 5: Use the elite retention strategy to retain high-weight particles; Step 6: Use genetic algorithm to resample and select particles; Step 7: Crossover operation generates a new particle swarm; Step 8: Perform mutation operation, combine the number of iterations with the fitness of the particle, converge to obtain the optimal particle, and obtain the parameter estimation value.
2. The particle filtering-based navigation multipath signal estimation method according to claim 1, wherein: The navigation signal received in step 1 consists of a direct signal and a multipath signal. The sampled signal is expressed as follows: Where A0 represents the guidance signal amplitude, A i represents the multipath signal amplitude, c(t) is the pseudo code sequence, k is the kth sampling point, k i is the relative signal delay, T s is the sampling time of the signal, cos(ω0kT s ) is the carrier, M represents the number of multipath signals, n(kT s ) is random noise.
3. The particle filtering-based navigation multipath signal estimation method according to claim 2, wherein: The execution process of step 2 is specifically as follows: first, a sampled local pseudocode signal is generated; based on the local pseudocode signal, local delay signals with different delays are generated to form a local signal group; according to the characteristics of the multipath signal, four parameters to be estimated, namely, the direct signal amplitude, the multipath signal amplitude, the direct signal delay, and the multipath signal delay, are defined; the number of particles is defined as Np, the number of iterations is defined as N, the signal initial value, the state matrix, and the process noise matrix are defined; and the measurement noise variance is obtained according to the correlation results between the received signal and the local signal group.
4. The particle filtering-based navigation multipath signal estimation method according to claim 3, characterized in that: During the execution of step 3, the EKF algorithm and the latest observation information are used to obtain the sample mean and covariance, and the obtained parameters are used to update the particle parameters; the formula used by the EKF algorithm is as follows: P k|k-1 =FP - k|k T +Q P k|k =(I-K k H k )P k|k-1 。 in, is the prior state transfer estimate, F is the state transfer matrix, w k-1 is the process noise, is the posterior state estimate, P k|k and P k|k-1 are the posterior and prior error covariances at the current moment, P - k|k is the posterior covariance matrix of the previous moment, Q is the process noise covariance matrix, z k represents the actual measurement value at time k, H represents the measurement matrix, v k is the measurement noise, K k is the Kalman gain, R k represents the measurement covariance matrix, and h(·) is the nonlinear measurement function.
5. The particle filtering-based navigation multipath signal estimation method according to claim 4, characterized in that: The probability density function expression in step 4 is Where n is the dimension of the state vector, μ is the measurement prediction value, R is the measurement noise covariance matrix, and x is the actual measurement value.
6. The particle filtering-based navigation multipath signal estimation method according to claim 5, characterized in that: The execution process of step 6 is specifically to resample the particles using the genetic algorithm, select the particles based on the particle weights using the roulette wheel selection algorithm, select the particles according to the proportion of the particle weights, and select the particles with high weights.
7. The particle filtering-based navigation multipath signal estimation method according to claim 6, characterized in that: The expression of the crossover operation in step 7 is as follows: where F max is the maximum mutation rate, F min is the minimum mutation rate, I is the current number of iterations, and M is the maximum number of iterations.
8. The particle filtering-based navigation multipath signal estimation method according to claim 7, characterized in that: The expression of the mutation operation in step 8 is as follows: Among them, G is the initial mutation rate, g i is the fitness of the current particle, and g is the average fitness.