A method for synchronization positioning and mapping based on millimeter wave communication multipath transmission signals

Abstract

The application belongs to the technical field of wireless positioning, and particularly relates to a synchronization positioning and mapping method based on millimeter wave communication multipath transmission signals. By establishing a multi-antenna virtual base station model, non-line-of-sight paths are equivalent to virtual line-of-sight paths transmitted by the multi-antenna virtual base station, the state of the virtual base station is described by position and azimuth angle, thereby simplifying the relationship between the geometric parameters of the non-line-of-sight transmission path and the state of the user, and the measurement information of the non-line-of-sight path transmission signals is more effectively utilized; secondly, the multi-antenna virtual base station is regarded as a map feature in a spatial radio frequency environment, a mobile user synchronization positioning and mapping algorithm based on millimeter wave multipath parameter measurement is proposed, the multipath parameter measurement set and the map feature set are modeled as a random finite set, and a probability hypothesis density filter is used to estimate the map, thereby significantly improving the synchronization positioning and mapping performance in the dense clutter and sensor missed detection scenarios.

Description

Technical Field

[0001] This invention belongs to the field of wireless positioning technology, specifically relating to a synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals. Background Technology

[0002] The integration of communication and sensing is one of the core visions of sixth-generation mobile communication (6G). In the future, base stations, terminals, and other devices in wireless networks will not only be able to communicate but will also possess environmental sensing capabilities similar to radar. By deeply integrating sensing and communication functions, a unified system architecture can be built, thereby improving spectrum and energy efficiency while reducing hardware costs. Millimeter-wave communication systems, with their massive MIMO antenna arrays and wide spectrum resources, provide the hardware foundation for achieving high-throughput data transmission and environmental sensing. Currently, research has utilized methods such as Bayesian filtering and deep learning to jointly estimate mobile user states and environmental maps. However, complex multipath propagation and noise interference in real-world environments can still lead to false alarms and missed detections, thus affecting the performance of related algorithms.

[0003] Electromagnetic waves encounter obstacles during propagation, resulting in reflection, scattering, and diffraction, thus causing multipath propagation of wireless signals. In traditional positioning systems, due to narrow signal bandwidth and limited multipath resolution, line-of-sight (LOS) and non-line-of-sight (NLOS) transmission signals often overlap, leading to decreased accuracy in LOS signal parameter estimation and limited positioning performance. In recent years, advancements in high-bandwidth millimeter-wave communication and massive MIMO (Multiple-Input Multiple-Output) beamforming technology have enabled high spatiotemporal resolution measurement methods, allowing for effective separation of different path components in mixed signals. This has made high-precision estimation of LOS and NLOS multipath parameters a reality. By effectively utilizing multipath information, not only can wireless positioning accuracy be significantly improved, but environmental characteristics can also be further perceived and reconstructed.

[0004] In multipath signal localization algorithms, the location of mobile users can be achieved by utilizing the constraint relationship between the geometric parameters of the LOS path and the positions of the transmitter and receiver. Simultaneously, by leveraging the constraint relationship between the geometric parameters of the NLOS path and the positions of the corresponding scatterers, the positions of the scatterers can be estimated, thus achieving the goal of Simultaneous Localization and Mapping (SLAM). The performance of traditional data association-based SLAM algorithms is highly dependent on the accuracy of data association and is sensitive to clutter generated by environmental interference and missed detections caused by sensor quality issues. Therefore, they perform poorly in complex environments with dense clutter and low detection probabilities. Random finite sets, as an emerging theoretical framework in the field of target tracking, can effectively address the data association difficulties faced by traditional target tracking methods in environments with dense clutter and missed detections, thereby improving the robustness and accuracy of the system in complex scenarios. Summary of the Invention

[0005] To overcome the problems in the prior art, this invention proposes a synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals.

[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0007] In a first aspect, the present invention provides a synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals, comprising the following steps:

[0008] A system model is established, which includes a millimeter-wave signal transmission model, a multi-antenna virtual base station model, and a recursive Bayesian filtering system model.

[0009] Virtual base stations are used as map features of the spatial radio frequency environment. The map feature set is modeled as a random finite set. Based on millimeter-wave multipath parameter measurements, Bayesian filtering is used to recursively estimate the joint posterior probability density function of mobile user state and map features. Particle filters are used to estimate the motion trajectory of mobile users. Based on the motion trajectory of mobile users, a probability hypothesis density (PHD) filter is used to estimate the environmental map features, resulting in mobile user trajectory estimates and environmental map estimates.

[0010] Furthermore, the millimeter-wave signal transmission model includes:

[0011] The millimeter-wave signal emitted by the base station is transmitted through different multipath channels and then reaches the mobile user receiver antenna array. The millimeter-wave multipath transmission signal is a superposition signal of line-of-sight path and first reflection non-line-of-sight path.

[0012] Furthermore, the multi-antenna virtual base station model includes:

[0013] Non-line-of-sight transmission signals are described as virtual line-of-sight transmission signals emitted by a virtual base station; the virtual base station is a mirror image of the physical base station with respect to the reflector surface, and the virtual base station is a multi-antenna system with the same antenna array as the physical base station; under the multi-antenna virtual base station model, the multipath parameters of non-line-of-sight are quantized and calculated as a function of the user's location and the virtual base station's state.

[0014] Furthermore, in the multi-antenna virtual base station model, position parameters and azimuth parameters are used together to describe the virtual base station state, and the azimuth of the virtual base station is regarded as the orientation angle of the virtual base station antenna array.

[0015] Furthermore, the system model of the recursive Bayesian filter includes: a state equation and a measurement equation;

[0016] The state equation is used to describe the relationship between the current user state and the historical user state.

[0017] The measurement equation is used to describe the relationship between millimeter-wave multipath parameters and mobile user status and virtual base station status.

[0018] Furthermore, the estimation of the motion trajectory of a mobile user using a particle filter includes:

[0019] Mobile user state particles are initialized by extracting them based on the prior probability of the mobile user's initial state. N Each particle yields a set of particles, all of which have equal weights. Each particle carries a map, and the feature set of all particles' maps is initialized to an empty set.

[0020] Based on the user's movement trajectory at the previous moment and the proposed distribution function at the current moment, the predicted distribution of the user trajectory is obtained through importance sampling. The proposed distribution function adopts the state transition model of mobile users.

[0021] Furthermore, the environmental map feature estimation based on the mobile user's motion trajectory using a probability hypothesis density filter includes:

[0022] The probability hypothesis density of the map is approximated using a Gaussian mixture model, resulting in... k -1 Moment Atlas;

[0023] A capacitive Kalman filter is used to generate the mean and covariance of the components of the nascent Gaussian mixture model, resulting in... k New map feature set at any time;

[0024] Will k -1 time map atlas and k By overlaying the feature sets of the newly generated map at each moment, we obtain k The probability hypothesis density of the map feature set at any given time;

[0025] Based on the non-line-of-sight measurement set, and using a Gaussian mixture model, the probability hypothesis density of the map carried by each particle is updated.

[0026] Furthermore, it also includes:

[0027] use k The measurement set at each moment, including the line-of-sight measurement set and the non-line-of-sight measurement set, is used to update the particle weights, thereby updating the mobile user state distribution.

[0028] Furthermore, the mobile user trajectory estimation and environmental map estimation include:

[0029] Based on the minimum mean square error criterion, the motion trajectory of mobile users is estimated by the particle weighted average method; at the same time, the radio frequency environment map is constructed by extracting the highest weight particles.

[0030] Compared with the prior art, the present invention has the following technical effects:

[0031] (1) This invention models the virtual base station as a multi-antenna system with the same antenna array configuration as the physical base station. The virtual base station antenna array can be understood as a mirror image of the physical base station antenna array with respect to the reflector. Furthermore, this invention uses position parameters and azimuth parameters to describe the virtual base station state. The azimuth angle of the virtual base station can be regarded as the orientation angle of the virtual base station antenna array. Using pose parameters to jointly model the virtual base station greatly simplifies the relationship between NLOS path geometry parameters, user state, and virtual base station state, and allows for more effective use of NLOS path measurement information.

[0032] (2) This invention models the millimeter-wave multipath parameter measurement and environmental map feature (virtual base station) set as a random finite set, and uses Bayesian filtering to recursively estimate the joint posterior probability density function of the mobile user state and map features. In order to reduce computational complexity and ensure real-time performance requirements, this invention uses Rao-Blackwellized (RB) particle filters to achieve simultaneous localization and environmental mapping. That is, the state of the mobile user is first estimated using particle filters to obtain the mobile user's trajectory, and then the environmental map features are estimated using PHD filters based on the mobile user's trajectory. Finally, each particle is fused to obtain the mobile user trajectory estimate and the environmental map estimate. Attached Figure Description

[0033] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a schematic diagram of the multi-antenna virtual base station and multipath parameters of the present invention;

[0035] Figure 2 This is a flowchart of the RBPHD-SLAM algorithm of the present invention;

[0036] Figure 3 For vehicle movement trajectories and environmental maps;

[0037] Figure 4 For the simultaneous existence of LOS and NLOS paths, the root mean square error plot of vehicle state estimation is shown.

[0038] Figure 5 The root mean square error map of vehicle state estimation is shown when the LOS path is blocked and only the NLOS path exists.

[0039] Figure 6 The average GOSPA error map of the virtual anchor point map when both LOS and NLOS paths exist simultaneously;

[0040] Figure 7 The average GOSPA error map of the virtual anchor point map is shown when the LOS path is occluded and only the NLOS path exists. Detailed Implementation

[0041] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects of the technical solutions proposed according to the present invention are described in detail below with reference to the accompanying drawings and preferred embodiments. Specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form. Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0042] The purpose of this invention is to provide a synchronous positioning and mapping method based on non-line-of-sight (NLOS) signals in millimeter-wave communication. To fully exploit the location information in NLOS path transmission signals, these signals are described as virtual LOS path transmission signals emitted by a virtual base station. By establishing a virtual base station model, the constraints between multipath parameters TOA (Time of Arrival), AOD (Angle of Departure), and AOA (Angle of Arrival) and the virtual base station state are simplified. Furthermore, this invention considers the virtual base station (VA) as the source of the NLOS path transmission signal. Although the multipath parameters change during user movement, the signal source remains constant. Based on this, the virtual base station can be used as a map feature of the spatial radio frequency environment and recorded in the map, thereby forming a radio frequency environment map based on geometric features. Mobile users, using the multipath parameter measurement results of millimeter-wave signals transmitted by the base station, estimate their own state while constructing a radio frequency environment map characterized by the virtual base station.

[0043] In this embodiment, refer to Figures 1-7 This paper presents a synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals, including the following steps:

[0044] Step 100: Establish a system model, which includes a millimeter-wave signal transmission model, a multi-antenna virtual base station model, and a recursive Bayesian filtering system model;

[0045] Step 200: Using the virtual base station as a map feature of the spatial radio frequency environment, the map feature set is modeled as a random finite set. Based on millimeter-wave multipath parameter measurement, Bayesian filtering is used to recursively estimate the joint posterior probability density function of the mobile user state and map features. Particle filter is used to estimate the motion trajectory of the mobile user. Based on the motion trajectory of the mobile user, PHD filter is used to estimate the environmental map features, thus obtaining the mobile user trajectory estimate and the environmental map estimate.

[0046] The following is a detailed explanation of each of the above steps:

[0047] Step 100: Establish a system model, which includes a millimeter-wave signal transmission model, a multi-antenna virtual base station model, and a recursive Bayesian filtering system model.

[0048] This invention considers the application scenario of millimeter-wave MIMO mobile communication systems, where both base stations and mobile users are equipped with antenna arrays.

[0049] Step 110: Establish a millimeter-wave signal transmission model. The millimeter-wave transmission signal is a superposition of the LOS path transmission signal and the first-reflection NLOS path transmission signal.

[0050] Considering a linearly time-varying multipath transmission channel, the millimeter-wave signal emitted by the base station reaches the receiver antenna array of the mobile user after transmission through the multipath channel, which includes the LOS path and the NLOS path. The short wavelength characteristic means that the propagation of millimeter waves mainly relies on the line-of-sight path and a few low-order reflection paths. Without loss of generality, it is assumed that the millimeter-wave multipath transmission signal is a superposition signal of the LOS path and the first-reflection NLOS path.

[0051] Specifically, the mobile user receiver antenna array... k Received signal at each moment It can be represented as:

[0052] ;

[0053] In the above formula, l =0 indicates the LOS path, others... Indicates the NLOS path; For path complex gain; Indicates the receiving antenna array in On the guide vector, Indicates AOA; Indicates the transmitting antenna array in On the guide vector, Indicates AOD; Indicates that the transmitter will send a signal. A translation was performed on the time axis, the amount of which is... , Indicates TOA; This represents additive white Gaussian noise.

[0054] Step 120: Establish a multi-antenna virtual base station model, and describe the NLOS path signal as a virtual LOS path signal emitted by the virtual base station.

[0055] The virtual base station is a mirror image of the physical base station with respect to the reflective surface. The virtual base station is a multi-antenna system with the same antenna array as the physical base station. The antenna array of the virtual base station can be understood as a mirror image of the antenna array of the physical base station with respect to the reflective surface.

[0056] The virtual base station state m is represented by two parameters: position and azimuth. , The horizontal and vertical coordinates of the virtual base station in two-dimensional space. Indicates the azimuth angle of the virtual base station antenna array, such as Figure 1 As shown.

[0057] In the multi-antenna virtual base station model, the multipath parameters of the NLOS path can be calculated by quantizing the user state and the virtual base station state as a function, specifically expressed as a vector containing three components:

[0058] ;

[0059] In the above formula, Indicates the transmission distance of the NLOS path; Indicates the AOD of the NLOS path; The AOA represents the NLOS path; This represents the pose vector of a mobile user. , The horizontal and vertical coordinates of a mobile user in a two-dimensional space. Indicates the azimuth angle of the mobile user; Indicates transpose; The equation for measuring the NLOS diameter; This represents the clock difference between the mobile user and the base station.

[0060] Step 130: Establish a system model for recursive Bayesian filtering, and estimate the mobile user state and the environment map with virtual base stations as feature points within the theoretical framework of recursive Bayesian filtering.

[0061] The system model of the recursive Bayesian filter includes a state equation and a measurement equation. The state equation describes the relationship between the current user state and the historical user state, and the measurement equation describes the relationship between the millimeter-wave multipath parameters and the mobile user state and the virtual base station state.

[0062] State equations describe the relationship between the current user state and historical user states. Given a mobile user... k The state at time -1 can be predicted based on a specified state transition model. k The state at time t, the prediction process can be represented by the transition probabilities as follows:

[0063] ;

[0064] Specifically, definition k Constantly moving the user's state ,in express k The user's pose vector is constantly shifted. , express k The user's position coordinates on the two-dimensional plane are constantly being moved. express k Constantly move the user's azimuth angle; express k Constant user line speed; express k The user's angular velocity is constantly moving. express kGiven the clock difference between the mobile user and the base station at any given time, the state transition model for the mobile user is as follows:

[0065] ;

[0066] in, This is the state transition function based on the mobile user motion model. The process noise is modeled as zero-mean Gaussian noise with covariance Q.

[0067] The measurement equations focus on the measurement process of millimeter-wave multipath parameters. Mobile users use channel estimation algorithms to measure the TOA, AOD, and AOA parameters of millimeter-wave multipath transmission signals. Due to environmental obstructions, limited communication distance, and noise, the multipath parameter measurement set contains a large amount of clutter and missed detections. This invention models millimeter-wave multipath parameter measurements as a Random Finite Set (RFS). Compared to random vectors, the elements of an RFS have no inherent order, and its cardinality is a random variable. Therefore, an RFS can naturally describe a multipath parameter measurement set where the eigenvalues ​​and number of parameters are unknown.

[0068] Specifically, the present invention will k The set of millimeter-wave multipath parameter measurements at time t is represented as a random finite set. ,in, , These correspond to the measurement sets of LOS diameters and NLOS diameters, respectively.

[0069] Assuming mobile users can identify the LOS path, that is:

[0070] ;

[0071] ;

[0072] In the above formula, Indicates the measured value; The equation representing the measurement of the LOS diameter; Indicates the measurement noise of the LOS path; Represents the pose vector of the base station; express k The user's pose vector is constantly moved.

[0073] Assume that there exists at least one NLOS path measurement, i.e. ,in, It could originate from NLOS path measurements or clutter. Consider this first. Derived from map features ,Right now Indicates virtual base station The corresponding NLOS path observations, , The equation representing the measurement of the NLOS diameter, This represents the measurement noise of the NLOS path. Regarding clutter, this invention models it as having an intensity of... The Poisson point process. Furthermore, not all virtual base stations can generate measurements; sensors may miss detections. Therefore, a detection probability is introduced. To indicate when a mobile user is in a state At that time, virtual base station It can generate the probability of measurement.

[0074] Step 200: Use the virtual base station as a map feature of the spatial radio frequency environment (virtual base station), model the map feature set as a random finite set, and recursively estimate the joint posterior probability density function of the mobile user state and map features using Bayesian filtering based on millimeter-wave multipath parameter measurement; use a particle filter to estimate the motion trajectory of the mobile user, and use a PHD filter to estimate the environmental map features based on the motion trajectory of the mobile user, thus obtaining the mobile user trajectory estimate and the environmental map estimate.

[0075] During a mobile user's movement, the spatial radio frequency environment map is unknown, meaning the pose and number of virtual base stations are also unknown. This invention models the map feature (virtual base station) set as a random finite set, and... k The set of virtual base stations observed at any given time is represented as ,in, express k The first time observed l Each map feature. It should be noted that the description of map features is based on a multi-antenna virtual base station model.

[0076] This invention uses Bayesian filtering to recursively estimate the joint posterior probability density function of mobile user pose and map features. ,in, This represents the initial state of a mobile user. Mobile User State It is a random vector, a map feature set. It is a random finite set, which makes the joint posterior probability density function The calculation is very difficult.

[0077] To reduce computational complexity, this invention designs an RBPHD-SLAM algorithm based on a multi-antenna virtual base station model. The RBPHD-SLAM algorithm uses an RB particle filter to combine the joint posterior of the user's motion trajectory and the map. The process is decomposed into the product of the user's trajectory posterior and the map posterior conditioned on the trajectory. First, a particle filter is used to estimate the moving user's trajectory. Then, based on the trajectory, a PHD filter is used to estimate the environmental map feature set. The PHD filter models the map feature set as a Poisson random finite set, whose probability density can be described by its intensity function (also known as the PHD function). The PHD filter recursively estimates the PHD function of the map feature set. This significantly reduces the algorithm complexity. The RBPHD-SLAM algorithm flowchart is as follows: Figure 2 .

[0078] The specific implementation of the millimeter-wave communication RBPHD-SLAM algorithm based on a multi-antenna virtual base station model is as follows:

[0079] Step 210: Initialize the mobile user state particles.

[0080] At the initial moment, based on the prior probability of the mobile user's initial state. Extract N Each particle carries a map feature set that is initialized to an empty set, i.e., the map feature set... All particles have equal weights, with a value of 1 / N , thus obtaining the particle set , This indicates the initial state of the mobile user.

[0081] Step 220: Predict the movement trajectory of mobile users.

[0082] Based on the user's movement trajectory at the previous moment and the system's proposal distribution function at the current time The predicted distribution of user trajectories is obtained through importance sampling. ,in, Indicates from time 1 to k The movement trajectory of a mobile user in real time; Indicates time 1 to k- The movement trajectory of a mobile user at time 1; Indicates from time 1 to k Multipath measurement set at time -1; Indicates from time 1 to k Multipath measurement set at time; Indicates the first i The particles from time 1 to k The movement trajectory of a mobile user at time 1; Indicates the first i Particles k The user's state is constantly being moved.

[0083] In this implementation, the state transition model for mobile users is selected as the proposed distribution function:

[0084] ;

[0085] In the above formula, Indicates the first i The particles from time 1 to k The movement trajectory of a mobile user at time 1; Indicates from time 1 to k Multipath measurement set at time; Indicates the first i Particles k Constantly moving the user's state, Indicates the first i Particles k Move the user's state at time -1; Represents the state transition probability of a mobile user, describing the user's state transition from... k 1 hour has arrived k The pattern of state changes over time.

[0086] At this point, the predicted particle weights are:

[0087] ;

[0088] In the above formula, Indicates the first i Individual particles k Time based k Prediction weights for information at time 1; Indicates the first i Individual particles k The posterior weights at time 1.

[0089] Step 230: Map PHD prediction.

[0090] k PHD function for predicting map atlases at any time ,Depend on k -1 time-attribute PHD function and k PHD function for instantaneous new map feature set This is formed by superposition, denoted as Equation 1:

[0091] ;

[0092] To further reduce the complexity of the SALM algorithm, this invention uses a Gaussian Mixture (GM) model to approximate the map PHD, that is, to represent the distribution of map features by a weighted sum of multiple Gaussian components. The first... k -1 moment i The PHD function of the posterior map feature set carried by each particle Let it be denoted as Equation 2:

[0093] ;

[0094] in, express k The number of GM components in the Gaussian mixture model at time 1; Indicates the first j The weights of the GM components of a Gaussian mixture model; They represent the first j The mean and covariance of the GM components of a Gaussian mixture model; Let P represent the Gaussian probability density function with mean m and covariance matrix P.

[0095] To effectively generate new map features, an adaptive birth strategy is adopted. The main idea is to generate a new map feature for each measurement and determine the mean and variance of the new GM components based on the measurement values. The new map feature set is represented by the PHD function. Its formula is denoted as Formula 3:

[0096] ;

[0097] in, express k The number of GM components corresponding to the newly generated map features at any given time; This represents the weight of the GM component corresponding to the new map feature, and is usually set to a small number; Indicates observation The mean of the corresponding new map features; The covariance matrix represents the features of the newly generated map. Since the measurement equation is nonlinear, this invention uses a cubic Kalman filter (CKF) to generate the mean and covariance of the newly generated GM components.

[0098] Substituting equations two and three into equation one, we get k PhD in predicting map feature sets at specific times:

[0099] ;

[0100] In the above formula, expressk PhD in real-time map prediction; , express k Predict the number of GM components in the map PHD at any time. express k- Predict the number of GM components in the map PHD at time 1. express k The number of GM components corresponding to the newly generated map features at any given time; Indicates the first j The weights of each GM component.

[0101] Step 240: Map PHD Update.

[0102] Get NLOS measurement set at time Subsequently, based on the Gaussian mixture model (GM), the map PHD for each particle is updated as follows:

[0103] ;

[0104] in, Indicates the first i Individual particles k The probability hypothesis density (PHD) of the map feature set updated at each time step; Indicates the first i Individual particles k- The probability hypothesis density (PHD) of the map feature set updated at time 1; express k Time NLOS Measurement Set The cardinality; express The l One element; Indicates that it is located at Map features at the location in the user's state Time-based observations The likelihood function is usually modeled as a Gaussian distribution; Indicates the measured value The clutter intensity.

[0105] The first term in the above formula corresponds to the mobile user's state. The first term corresponds to the update when no observations are generated, while the second term corresponds to the update when observations are present. In the latter case, the observations may originate from clutter, which has been taken into account in the denominator. In practical applications, the parameters of the PHD for updating the map when observations are present can be estimated using any standard Gaussian filtering technique. This invention employs a Gaussian approximation method based on a first-order Taylor series, i.e., using capacitive Kalman filtering to complete the PHD update process.

[0106] Step 250: Particle weight update.

[0107] use k Multipath measurement set of time Including LOS measurement and NLOS measurement The particle weights are updated, thereby updating the mobile user state distribution. Analysis shows that the... i The updated weights of each particle It is given by the following formula:

[0108] ;

[0109] In the above formula, Indicates the first i In the state of individual particles k Conditional probability density of apparent LOS measurement at time t; Indicates that in the given first i Particles k The mobile user's movement trajectory at time 1 and time 1 to k When the measurement set is at time 1, k Conditional probability density of non-line-of-sight NLOS measurements at time; Indicates the first i Individual particles k Time based k Prediction weights for information at time 1.

[0110] in, It is the contribution of LOS measurement. The likelihood function for this set is derived from NLOS measurements, and there are various strategies available for calculating it. In this embodiment, a Poisson-Dobernal hybrid update is used for probabilistic inference.

[0111] ;

[0112] When the number of effective particles is too low, the system will perform particle resampling.

[0113] Step 260: Mobile user movement trajectory and map estimation.

[0114] This invention estimates the movement trajectory of mobile users based on the minimum mean square error criterion and the particle weighted average method; at the same time, it constructs a radio frequency environment map by extracting the highest weighted particles.

[0115] like Figure 3As shown, consider a two-dimensional vehicular network scenario, which includes a ring road, a millimeter-wave base station (BS), and four virtual base stations (VAs), whose pose vectors are [0 m; 0 m; 0 m]. 0 ], [200 m; 0 m; 180 0 ],[0m; 200 m; 0 0 ], [-200 m; 0 m; 180 0 ],[0 m; -200 m; 0 0 A moving vehicle is traveling on a road. The vehicle's state transition function is:

[0116] ;

[0117] In the above formula, express k Vehicle status at time -1 express k The vehicle's azimuth at time -1; express k The linear velocity of the vehicle at time -1; express k The vehicle's angular velocity at time -1; The measurement time interval. Detection probability of the NLOS path. The value is set to 0.9, the clutter number follows a Poisson distribution, and the mean is set to [value missing]. .

[0118] Figure 4 and Figure 5 This paper demonstrates how the root mean square error (RMSE) of the proposed method for estimating vehicle position, clock deviation, and azimuth angle changes over time in the following two scenarios: Figure 4 Both the LOS and NLOS paths exist simultaneously. Figure 5 In the case of LOS path obstruction, only NLOS path exists. It can be seen that even with LOS obstruction, the proposed method achieves a vehicle position estimation error of less than 1m, a clock difference estimation error of approximately 0.3m, and an azimuth angle estimation error of less than 0.30. With LOS present, the vehicle positioning error decreases to below 0.4m, and the estimation errors of clock difference and heading angle also decrease, but the performance improvement is relatively limited.

[0119] Figure 6 and Figure 7 This demonstrates the average Generalized Optimal Sub-Pattern Assignment (GOSPA) error of the virtual base station map, where... Figure 6This corresponds to the situation where both the LOS path and the NLOS path exist simultaneously; while Figure 7 This corresponds to the case where the LOS path is occluded, and only the NLOS path exists. It can be seen that as the vehicle moves through the environment, the average GOSPA error gradually decreases, and from the third time step onwards, the vehicle can consistently observe all virtual base stations. Compared to the vehicle state estimation performance, the improvement in environment map estimation performance brought by the existence of the LOS path is limited, because the LOS path and the pose of the virtual base stations are not directly correlated.

[0120] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals, characterized in that, Includes the following steps: A system model is established, which includes a millimeter-wave signal transmission model, a multi-antenna virtual base station model, and a recursive Bayesian filtering system model. The millimeter-wave signal transmission model includes: the millimeter-wave signal emitted by the base station is transmitted through different multipath channels and then reaches the mobile user receiver antenna array. The millimeter-wave communication multipath transmission signal is a superposition signal of line-of-sight transmission signal and a first-reflection non-line-of-sight transmission signal. The virtual base station is used as a map feature of the spatial radio frequency environment. The map feature set is modeled as a random finite set. Based on the millimeter-wave multipath parameter measurement, the joint posterior probability density function of the mobile user state and map features is recursively estimated using Bayesian filtering. The motion trajectory of the mobile user is estimated using a particle filter. Based on the motion trajectory of the mobile user, the environmental map is estimated using a probability hypothesis density filter. The mobile user trajectory estimate and the environmental map estimate are obtained. The multi-antenna virtual base station model includes: describing non-line-of-sight paths as virtual line-of-sight paths emitted by the virtual base station; the virtual base station is a mirror image of the physical base station with respect to the reflector surface, and the virtual base station is a multi-antenna system with the same antenna array as the physical base station; under the multi-antenna virtual base station model, the multipath parameters of the non-line-of-sight paths are quantized and calculated through a function of user state and virtual base station state; In the multi-antenna virtual base station model, position parameters and azimuth parameters are used together to describe the virtual base station state, and the azimuth angle of the virtual base station is regarded as the orientation angle of the virtual base station antenna array. The system model of the recursive Bayesian filter includes a state equation and a measurement equation; the measurement equation is used to describe the relationship between millimeter-wave multipath parameters and the mobile user state and the virtual base station state. The virtual base station state m is represented by two parameters: position and azimuth. , The horizontal and vertical coordinates of the virtual base station in two-dimensional space. This represents the azimuth angle of the virtual base station antenna array. In the multi-antenna virtual base station model, the multipath parameters of the NLOS path are calculated through a function quantization of the user state and the virtual base station state, specifically expressed as a vector containing three components: ； In the above formula, Indicates the transmission distance of the NLOS path; Indicates the AOD of the NLOS path; The AOA represents the NLOS path; This represents the pose vector of a mobile user. , The horizontal and vertical coordinates of a mobile user in a two-dimensional space. Indicates the azimuth angle of the mobile user; Indicates transpose; The equation for measuring the NLOS diameter; This represents the clock difference between the mobile user and the base station.

2. The synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals according to claim 1, characterized in that, In the system model of the recursive Bayesian filter, the state equation is used to describe the relationship between the current user state and the historical user state.

3. The synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals according to claim 1, characterized in that, The method of estimating the motion trajectory of a mobile user using a particle filter includes: The mobile user state particles are initialized by extracting N particles based on the prior probability of the mobile user's initial state to obtain a particle set, and all particles have equal weights; each particle carries a map, and the map feature set of all particles is initialized to an empty set. Based on the user's movement trajectory at the previous moment and the proposed distribution function at the current moment, the predicted distribution of the user trajectory is obtained through importance sampling. The proposed distribution function adopts the state transition model of mobile users.

4. The synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals according to claim 1, characterized in that, The environmental map estimation based on the mobile user's motion trajectory using a probability hypothesis density filter includes: The probability hypothesis density of the map is approximated using a Gaussian mixture model, resulting in... k -1 Moment Atlas; A capacitive Kalman filter is used to generate the mean and covariance of the components of the nascent Gaussian mixture model, resulting in... k New map feature set at any time; Will k -1 time point map feature set and k By overlaying the feature sets of the newly generated map at each moment, we obtain k The probability hypothesis density of the map feature set at any given time; Based on the non-line-of-sight measurement set, and using a Gaussian mixture model, the probability hypothesis density of the map carried by each particle is updated.

5. The synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals according to claim 3, characterized in that, Also includes: use k The measurement set at each moment, including the line-of-sight measurement set and the non-line-of-sight measurement set, is used to update the particle weights, thereby updating the mobile user state distribution.

6. The synchronous positioning and mapping method based on millimeter-wave communication multipath transmission signals according to claim 1, characterized in that, The mobile user trajectory estimation and environmental map estimation include: Based on the minimum mean square error criterion, the motion trajectory of mobile users is estimated by the particle weighted average method; at the same time, the radio frequency environment map is constructed by extracting the highest weight particles.

Images