A method for tracking a moving target based on a smart reflective surface

Abstract

The application provides a moving target tracking method based on an intelligent reflecting surface, first, the motion state evolution linear model of the target is constructed by using the motion trajectory information of the moving target, and the motion state prediction can be obtained; second, the signal receiving model containing multiple sampling signals, time delay and Doppler frequency offset at the current moment is constructed based on the single antenna transmitting and receiving system model; third, the current target state information is estimated by using the extended Kalman filter or particle filter to process the nonlinear received signal based on the current received signal; finally, the next moment state information is predicted and the intelligent reflecting surface phase is regulated in advance according to the current state estimation of the extended Kalman filter / particle filter and the state evolution model, the next moment target state estimation accuracy is improved, and the moving target tracking is realized. The application can be used for intelligent networked vehicle communication and positioning tracking, and can also be used for intelligent factory motion operation robot communication and positioning tracking.

Description

A Moving Target Tracking Method Based on Intelligent Reflective Surface Technical Field

[0001] This invention relates to a motion target tracking method based on a smart reflective surface, integrating sensing and communication. Background Technology

[0002] In contemporary wireless communication sensing, traditional pilot-based target tracking methods require complex signal design and significant computational resources. Integrated sensing and communication design can achieve simultaneous sensing and communication, and improve the performance of both. Fan Liu and Jinhong Yuan proposed a radar-based integrated sensing and communication target tracking model. By predicting the target's motion state in two-dimensional coordinates, they achieve beamforming prediction at the radar end, and then use extended Kalman filtering for state estimation, improving the accuracy of positioning and tracking. However, radar-based tracking systems require the gain provided by massive MIMO antennas, resulting in high algorithm complexity and resource consumption. Smart reflectors, as a 6G enabling technology, are considered to greatly enhance communication and positioning. Therefore, a smart reflector-assisted integrated sensing and communication system expands the communication area and fully utilizes additional channel state information to further improve tracking performance.

[0003] The intelligent reflector-assisted sensing system mainly consists of two parts: one is the original sensing model composed of radar and base stations, primarily used for communication and sensing; the other is the deployed intelligent reflector, used to add additional communication links and virtual line-of-sight links, enhancing channel gain. Clearly, compared to other sensing models, this model fully utilizes the additional channel state information and adjustable phase characteristics provided by the intelligent reflector, increasing the signal-to-noise ratio and resolution of the received signal.

[0004] Currently, the main problems with intelligent reflector-assisted sensing and communication systems are: with the deployment of intelligent reflectors, complex channels emerge, such as the cascaded channels between the target, intelligent reflector, and base station. Although carefully designed estimation algorithms exist, the accuracy and complexity of channel estimation still increase significantly. Furthermore, this model still performs sensing at the radar end and estimation and tracking at the base station end, resulting in time delays and hindering the effective utilization of the phased array characteristics of the intelligent reflector. Thanks to the passive, low-power characteristics and phased array characteristics of intelligent reflectors, signal estimation and target tracking can be achieved at the intelligent reflector end. Summary of the Invention

[0005] This invention builds upon the existing intelligent reflector-assisted sensing and communication model by using an intelligent reflector instead of radar for perception and target tracking. It derives an extended three-dimensional state evolution model for moving targets and estimates the target's state information through extended Kalman filtering / particle filtering. Based on this, an adaptive phase control mechanism for the intelligent reflector is designed. This mechanism predicts the next moment's motion position information based on the current target motion state estimate and the state evolution model, and uses this prediction for intelligent reflector phase control. This matches the signal phase at the next moment to maximize the received signal strength, thereby improving the accuracy of the target state estimation. This model fully utilizes the phase control characteristics of the intelligent reflector, improving tracking accuracy and robustness with low power consumption and low complexity.

[0006] The technical solution is as follows:

[0007] A dynamic target tracking method based on an intelligent reflective surface, characterized in that the method includes:

[0008] Step 1: Constructing the moving target state evolution model and signal observation model, and estimating the target state using extended Kalman filtering / particle filtering:

[0009] First, by utilizing the trajectory information of the moving target itself, a linear model of the target's motion state evolution is constructed, which can be used to predict the motion state.

[0010] Secondly, based on the single-antenna transmit-receive system model, a signal reception model is constructed that includes multi-sampled signals, time delay, and Doppler frequency offset at the current moment;

[0011] Based on the received signal at the current moment, the nonlinear received signal is processed by extended Kalman filtering or particle filtering to estimate the target state information at the current moment.

[0012] Step 2: Adaptive intelligent reflector phase control strategy based on prediction information:

[0013] Finally, based on the current state estimation and state evolution model of extended Kalman filter / particle filter, the state information of the next moment is predicted and the phase of the intelligent reflector is adjusted in advance to improve the accuracy of target state estimation in the next moment and realize moving target tracking.

[0014] Through the above process, the constructed three-dimensional motion state evolution model can accurately estimate the state transition, and the signal observation vector and extended Kalman filter or particle filter can achieve target state estimation. At the same time, based on the current state estimation and the state evolution model, the predicted state information can be obtained, and the phase of the intelligent reflective surface can be pre-adjusted to adapt to the signal sent by the target at the next moment, thereby improving the accuracy of target state estimation and realizing low-cost and low-complexity motion target perception and tracking.

[0015] This invention can be used for communication and location tracking of intelligent connected vehicles, and also for communication and location tracking of robots operating in intelligent factories. Attached Figure Description

[0016] Figure 1. Schematic diagram of the three-dimensional motion state evolution model of a moving target;

[0017] Figure 2 shows a simulation diagram of the distance approximation based on angle approximation and Taylor expansion;

[0018] Figure 3. Schematic diagram of high dynamic target state information tracking and estimation performance;

[0019] Figure 4. Mean square error of target state information under different numbers of intelligent reflective surface elements;

[0020] Figure 5. Mean square error of target state information under different receiver noise power. Detailed Implementation

[0021] This invention selects a smart reflector to replace radar in a sensor-integrated system for moving target tracking. Utilizing the passive reflection and large number of reflective elements of the smart reflector, the strength of the received signal can be improved with low power consumption. However, this increases the complexity of channel estimation, especially when facing moving targets, as time-varying channels lead to increased phase modulation complexity of the smart reflector. Therefore, this invention employs an adaptive phase modulation strategy for the smart reflector to achieve low-power, low-complexity moving target tracking.

[0022] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0023] Figure 1 shows a simplified three-dimensional coordinate motion evolution model. Based on the motion relationship shown in Figure 1, the motion target state evolution model is obtained.

[0024] Figure 2 shows the accuracy simulation results based on the simple angle approximation and the Taylor expansion approximation. The results show that the distance approximation based on the Taylor expansion is almost equal to the true value and has high accuracy.

[0025] Figure 3 shows the simulation results of single-shot moving target tracking, including azimuth, pitch, and distance. These three physical quantities determine the position of the object in space. The results show that the predicted intelligent reflector phase modulation strategy can achieve effective tracking, and the method based on extended Kalman filtering performs better.

[0026] Figure 4 illustrates the simulation results of tracking performance under different numbers of smart reflective surface elements. The results show that the prediction-based phase control strategy performs better under multiple elements, but worse under random elements. This is because multiple elements amplify the signal after random phase processing, resulting in a significant decrease in estimation accuracy.

[0027] Figure 5 depicts the simulation results of tracking performance under different receiver noise powers. The results show that the lower the noise power, the better the performance.

[0028] Example

[0029] Step 1. Constructing the moving target state evolution model and signal observation model, and estimating the target state using extended Kalman filtering / particle filtering.

[0030] First, constructing the state evolution model of the moving target.

[0031] Step (11), considering the scenario, as shown in Figure 1, obtain the original adjacent time-mapping distance position l and pitch angle φ. n azimuth φ n Geometric relationship between azimuth deviation Δφ and displacement Δd

[0032]

[0033] Step (12) obtains the distance through approximation and geometric mapping.

[0034] d n sinθ n =d n-1 sinθ n-1 +Δdsinφ n-1 .

[0035] Step (12.1) applies simple sinθ n =sinθ n-1 Approximately, the distance is obtained as Step (12.2) uses Taylor expansion based on the Pythagorean theorem to obtain the distance as...

[0036]

[0037]

[0038]

[0039] d n =d n-1 +Δdsinφ n-1 sinθ n-1 .

[0040] Step (12.3) involves a simulation experiment. The results are shown in Figure 2. The Taylor approximation represented by the blue line almost coincides with the true value, indicating that the Taylor expansion based on the Pythagorean theorem has a higher approximation accuracy.

[0041] Step (13): Obtain the azimuth angle φ based on the obtained Taylor distance approximation. n Evolutionary model

[0042]

[0043]

[0044]

[0045] Step (14): Obtain the pitch angle θ based on the obtained Taylor distance approximation. n Evolutionary model

[0046]

[0047]

[0048] Step (15) assumes the target maintains uniform motion, therefore the velocity remains constant, i.e., v n ≈v n-1 .

[0049] Step (16): Obtain the channel amplitude gain β based on the distance approximation. n Evolutionary model

[0050]

[0051]

[0052]

[0053]

[0054] in This represents the corrected gain at time n.

[0055] Secondly, the construction of the received signal model

[0056] Step (17), in constructing the signal reception model, mainly includes the received signal strength r n Delay τ n Doppler frequency offset μ n sampling signal

[0057] Step (17.1), the original received signal r n (t)

[0058]

[0059] Where κ is the smart reflector array factor gain, ψ is the phase shift from the target to the smart reflector, and p n It is the transmission power, H RC H represents the channel between the smart reflector and the receiving antenna. MR Ω represents the channel between the target and the smart reflector. n The phase shift of the smart reflector is represented by s. n Indicates the transmitted signal, n t This represents additive white Gaussian noise.

[0060] Step (17.2) involves applying a matched filter to the original signal r. n (t) is processed

[0061]

[0062]

[0063] Where β MR，n α represents the channel amplitude gain between the target and the intelligent reflector at time n. MR，n This represents the channel phase vector between the target and the intelligent reflector at time n.

[0064] Step (17.3), calculate the time delay τ n

[0065]

[0066] Where n τ This represents time delay noise.

[0067] Step (17.4) Calculate the Doppler frequency offset μ n

[0068]

[0069] Where n f f represents Doppler frequency offset noise. c Indicates the carrier frequency.

[0070] Next, extended Kalman filtering is used for state estimation.

[0071] Step (18): Set the parameter vector and observation vector in the extended Kalman filter process.

[0072] x = [φ, θ, d, v, β] T

[0073] y = [r1, ..., r n ,τ,μ] r

[0074] Step (19) simplifies the state evolution model and the received signal model.

[0075]

[0076] Step (20): Calculate the Jacobian matrix for the state evolution function and the receiver model function.

[0077] Step (20.1) Calculate the Jacobian matrix of the state evolution function.

[0078]

[0079] Step (20.2) Calculate the Jacobian matrix of the received signal function.

[0080]

[0081] Where r = r1, ..., r n .

[0082] Step (21) Define the motion state vector noise V S and signal observation vector noise V M

[0083]

[0084]

[0085] in This represents the variance of the corresponding azimuth, elevation, range, speed, and channel gain.

[0086] This represents the variance of the received signal, time delay, and Doppler band.

[0087] Step (22) involves using extended Kalman filtering, combined with the motion state estimation from the previous moment. Current received signal vector y n To achieve the current target state information The estimation. The steps of the extended Kalman filter are shown in the following equation.

[0088]

[0089]

[0090]

[0091]

[0092]

[0093] P n =(IK n H n )P n|n-1

[0094] The method of step (23) is parallel to that of step (22), that is, the state estimate can be obtained by step (22) or step (23).

[0095] In addition, step (23) can also be performed by combining particle filtering with the motion state estimate from the previous moment. Current received signal vector y n To achieve the current target state information The estimation. The particle filter process is shown in the following equation.

[0096]

[0097]

[0098]

[0099]

[0100] Where E(x) n ) represents finding x n The mathematical expectation, N p Indicates the number of particles. Represents particles The weight, Represents particles The normalized weights are p(·) and q(·), which represent the conditional probability and the proposal distribution, respectively.

[0101] Step 2. Adaptive intelligent reflector phase control strategy based on prediction information

[0102] Finally, intelligent reflector phase modulation and received signal construction based on target predicted state information.

[0103] Step (2.1) divides the phase of the smart reflector into two parts and designs them separately to reduce the complexity of phase control of the smart reflector.

[0104]

[0105] Step (2.2) involves designing the first half Ω of the phase of the smart reflector. RC，n， This is used to compensate for the fixed phase deviation between the smart reflector and the receiving antenna, making the channel between the smart reflector and the receiving antenna a fixed channel.

[0106]

[0107] Step (2.3) Design the latter half of the phase of the smart reflector. Based on the state estimate and state evolution model of the previous time step, the predicted state information for the next time step is obtained and used to set the phase of the adaptive smart reflector.

[0108]

[0109] in q i Represents the polar coordinates of the i-th element.

[0110] In step (2.4), a random phase modulation strategy is set as a control, thus obtaining a smart reflective surface phase modulation strategy based on randomness and adaptation.

[0111]

[0112] Among them It follows a uniform distribution in [0, 2π).

Claims

1. A dynamic target tracking method based on an intelligent reflective surface, characterized in that, The method includes: Step 1, constructing a moving target state evolution model and a signal observation model, and estimating the target state using extended Kalman filtering / particle filtering: First, using the moving target's own trajectory information, a moving target state evolution model is constructed to obtain motion state prediction; second, based on a single-antenna transmit-receive system model, a signal receiving model is constructed that includes multi-sampled signals, time delay, and Doppler frequency offset at the current moment; third, based on the received signal at the current moment, extended Kalman filtering or particle filtering is used to process the nonlinear received signal to estimate the target state information at the current moment; Step 2, an adaptive intelligent reflector phase control strategy based on prediction information: According to the current state estimation and state evolution model of extended Kalman filtering / particle filtering, the state information at the next moment is predicted, and the phase of the intelligent reflector is controlled in advance to improve the accuracy of target state estimation at the next moment and realize moving target tracking.

2. The dynamic target tracking method based on an intelligent reflective surface as described in claim 1, characterized in that, Step 1, the construction of the moving target state evolution model, includes: Step 11, obtaining the original adjacent time-time mapped distance and position. azimuth Azimuth deviation and motion displacement geometric relationship Step 12, obtain the distance through approximation and geometric mapping. Step 12.1 Simple Application Approximately, the distance is obtained as Step 12.2 Based on the Pythagorean theorem, the Taylor expansion yields the distance as... Step 12.3 Conduct simulation experiments; Step 13, obtain the azimuth angle based on the obtained Taylor distance approximation. Evolutionary model Step 14: Obtain the pitch angle based on the obtained Taylor distance approximation. Evolutionary model Step 15, assuming the target maintains uniform motion, therefore the velocity remains constant, that is... Step 16: Obtain the channel amplitude gain based on the distance approximation. Evolutionary model in express The time-based correction gain.

3. The dynamic target tracking method based on an intelligent reflective surface as described in claim 2, characterized in that, Step 1, the construction of the received signal model, includes: Step 17, which mainly includes the received signal strength. Delay Doppler frequency deviation The sampled signal; Step 17.1, the original received signal in, It is the gain factor of the intelligent reflective surface array. It is the phase shift from the target to the smart reflector. It's the transmission power. This represents the channel between the smart reflector and the receiving antenna. This represents the channel between the target and the smart reflector. This indicates the phase shift of the smart reflective surface. Indicates the signal being sent. This represents additive white Gaussian noise; step 17.2, the original signal is processed through a matched filter. Process in express The channel amplitude gain between the target at any given time and the intelligent reflector. express The channel phase vector between the target and the intelligent reflector at any given time; Step 17.3, calculate the time delay. in Indicates time delay noise; Step 17.4, calculate Doppler frequency offset. in This indicates Doppler frequency deviation noise. Indicates the carrier frequency.

4. The dynamic target tracking method based on an intelligent reflective surface as described in claim 3, characterized in that, Step 1, extending Kalman filtering for state estimation, includes: Step 18, setting the parameter vector and observation vector in the extended Kalman filtering process. Step 19: Simplify the state evolution model and the received signal model. in and The noise represents the state evolution model and the received signal model, and the variance is equal to that in step 21. and Step 20: Calculate the Jacobian matrix for the state evolution function and the receiver model function; Step 20.1: Calculate the Jacobian matrix of the state evolution function. Step 20.2, calculate the Jacobian matrix of the received signal function. in Step 21: Define the motion state vector noise. and signal observation vector noise in This represents the variance of the corresponding azimuth, elevation, range, velocity, and channel gain. This represents the variance of the received signal, time delay, and Doppler frequency offset; Step 22, through extended Kalman filtering, combined with the motion state estimate from the previous moment... Current received signal vector To achieve the current target state information The estimation; the steps of the extended Kalman filter are shown in the following equation. Step 23 is parallel to step 22; that is, the state estimate is obtained using either step 22 or step 23. In step 23, particle filtering is used, combined with the motion state estimate from the previous moment. Current received signal vector To achieve the current target state information The estimation; the particle filtering process is shown in the following equation: in Expressing the request The mathematical expectation, Indicates the number of particles. Represents particles The weight, Represents particles Normalized weights, Let represent the conditional probability and the proposal distribution, respectively.

5. The dynamic target tracking method based on an intelligent reflective surface as described in claim 4, characterized in that, Step 2, intelligent reflector phase control based on target predicted state information, includes: Step 2.1, dividing the phase of the intelligent reflector into two parts and designing them separately to reduce the complexity of intelligent reflector phase control. Step 2.2, Design the first half of the phase of the intelligent reflective surface This is used to compensate for the fixed phase deviation between the smart reflector and the receiving antenna, making the channel between the smart reflector and the receiving antenna a fixed channel. Step 2.3, Design the latter half of the phase of the intelligent reflective surface Based on the state estimate and state evolution model of the previous moment, the predicted state information for the next moment is obtained and used to set the phase of the adaptive smart reflector. in ， Indicates the first Step 2.4 involves determining the polar coordinates of each element and setting a random phase modulation strategy as a control, thus obtaining a smart reflector phase modulation strategy based on randomness and adaptation. Among them , obey The uniform distribution.

Images