High-precision multi-antenna detection and parameter extraction method for complex urban internet of vehicles channel

Abstract

The application claims a kind of complex urban vehicle networking channel high-precision multi-antenna detection and parameter extraction method.The method is based on the modular instrument bus architecture to build distributed detection hardware system, using external GPS signal to tame high stability rubidium atomic clock, realize the transmission end synchronization accuracy better than 10ns.Through the execution of whole link closed loop calibration containing measured template function and phase compensation vector, eliminate hardware non-ideal characteristics interference.Design satisfies the time division multiplexing control timing of channel coherence time physical constraint.In the parameter extraction stage, introduce pseudo-inverse adaptive initialization and space vector acceleration search mechanism to improve SAGE algorithm, and cooperate with nuclear power density adaptive clustering and multipath component distance tracking algorithm, realize the automatic continuous capture of the space-time evolution trajectory of multipath cluster under the belt vegetation shelter.Experiments show that the method time delay estimation error is better than 0.2ns, angle estimation error is better than 0.5 degrees, processing efficiency is improved by more than 30%.

Description

Technical Field

[0001] This invention relates to the field of vehicle-to-vehicle (V2V) wireless communication technology, and more specifically, to a high-precision multi-antenna detection and multipath parameter extraction method for complex urban V2V channels. Background Technology

[0002] As 5G evolves into 6G, autonomous driving, driverless vehicles, and intelligent transportation systems place stringent demands on the reliability and low latency of V2V communication. The 5.9GHz band, as a core dedicated band for intelligent transportation systems, requires crucial research into its channel characteristics in complex urban environments (such as areas with strip-like vegetation obstruction, high-density buildings, and high numbers of motor vehicles).

[0003] However, traditional channel probing methods have significant limitations when dealing with such environments:

[0004] Non-stationary characteristics are difficult to capture: the strip vegetation on both sides of the road causes the communication link to switch frequently between line-of-sight and partially obstructed states, and the multipath clusters exhibit a dynamic "birth and death" process, making it difficult for traditional systems to achieve high-precision spatiotemporal tracking.

[0005] Synchronization accuracy bottleneck: In high-mobility vehicle-to-vehicle scenarios, frequency deviation and time drift at the transmitting and receiving ends can severely disrupt the phase coherence between channel snapshots, leading to distortion of the extracted multipath parameters.

[0006] Non-ideal hardware characteristics interference: deviations in the bandpass characteristics of the RF front end, filter group delay ripple, and cable phase inconsistencies can introduce a large number of "pseudo-multipath" components.

[0007] The contradiction between computational efficiency and accuracy: Traditional multipath parameter extraction algorithms (such as the standard SAGE algorithm) rely on manual initialization when the number of multipaths is unknown, and convergence is slow and computational overhead is huge when processing massive snapshot data.

[0008] Therefore, there is an urgent need for a comprehensive solution that integrates high-synchronization hardware, full-link closed-loop calibration, high-dynamic timing design, and adaptive acceleration algorithms. Summary of the Invention

[0009] This invention aims to solve the problems of the prior art mentioned above. It proposes a high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) communication channels. The technical solution of this invention is as follows:

[0010] A high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) communication channels includes the following steps:

[0011] Step 1): Build and deploy a high-precision multi-antenna time-domain detection hardware system based on the NI PXIe platform, and establish a sub-sampling level transmit / receive synchronization mechanism;

[0012] Step 2): Construct and apply a full-link error closed-loop calibration module to perform hardware distortion compensation on the original observed signal through a measured data-driven approach;

[0013] Step 3): Construct an array measurement and control timing module for high-mobility urban scenarios, and perform multi-channel data acquisition that meets the physical constraints of channel coherence time;

[0014] Step 4): Construct an improved SAGE algorithm module based on adaptive initialization and space vector acceleration to recursively extract the physical parameters of multipath components;

[0015] Step 5): Construct a dynamic cluster evolution feature extraction unit, and use the kernel power density algorithm and multipath component distance algorithm to aggregate and continuously track the extracted multipath components, so as to realize the quantitative description of the non-stationary birth and death and power evolution law of channel multipath clusters.

[0016] Furthermore, in step 1), the high-precision multi-antenna time-domain detection hardware system adopts a distributed architecture, consisting of a transmitter (TX), a receiver (RX), and a synchronization support subsystem. At the transmitter, a vector signal transceiver generates a Zadoff-Chu constant envelope pseudo-random sequence with a center frequency of 5.9 GHz and an effective bandwidth of 40 MHz as the detection waveform. The specific synchronization process is as follows:

[0017] a. Establishment of frequency and time bases:

[0018] High-stability rubidium atomic clocks tamed by GPS 1PPS signals are configured at both the transmitting and receiving ends to provide an external reference frequency of 10MHz for the system. The relative time offset between the transmitting and receiving ends at the RF local oscillator level is locked within 10ns through a phase-locked loop to ensure the phase coherence of the channel snapshot during long-term measurement.

[0019] b. Subsampling-level synchronization maintenance:

[0020] By utilizing the high short-term stability of the rubidium atomic clock, the system enters hold mode in environments with weak GPS signal obstruction, enabling the system to have a nanosecond-level time delay resolution foundation, thereby supporting the subsequent sub-sampling level extraction of scattering paths of vegetation in complex urban areas.

[0021] The synchronization support subsystem is the "nerve center" of the high-precision detection system. This system uses independently configured GPS signals at both the transmitting and receiving ends to discipline the rubidium atomic clock in real time, combining the high short-term stability of the atomic clock with the long-term accuracy of the GPS signal. Through an internal phase-locked loop, the subsystem injects an externally provided 10MHz reference frequency into the local RF front-end, thus strictly locking the time offset between the transmitting and receiving ends within 10ns. This allows the system to maintain extremely high phase coherence even when the measurement vehicle is moving through complex urban areas and faces momentary external signal blockage, providing a precise time and frequency reference for capturing the generation and extinction of non-stationary multipath events and power evolution patterns.

[0022] Furthermore, the detection excitation sequence adopts the Zadoff-Chu sequence, with a sequence length of... The resampling rate is 80MHz; the element spacing of the 16-element uniform linear array is set to 25.4mm, which is half the wavelength corresponding to the 5.9GHz carrier frequency; the time synchronization accuracy of the transmitting and receiving ends is locked within 10ns, which means that the frequency drift of the local oscillator at the transmitting and receiving ends is eliminated by the GPS discipline mechanism, so that the delay estimation deviation between channel snapshots within a long measurement period is kept at the subsampling level.

[0023] The 16-element uniform linear array has an element spacing of 25.4 mm, which is half the wavelength corresponding to the 5.9 GHz carrier frequency, in order to eliminate phase ambiguity in angle estimation.

[0024] Furthermore, step 2): constructing and applying a full-link error closed-loop calibration module to perform hardware distortion compensation on the original observed signal through a measured data-driven approach specifically includes:

[0025] a. Measured template function calibration unit:

[0026] The transmitter output port and receiver input port are directly connected back-to-back via an RF coaxial cable; the end-to-end integrated impulse response waveform, including transmit pulse shaping, RF front-end group delay ripple, power amplifier weak nonlinearity, and receiver filter impulse response, is obtained by actual measurement; the main lobe profile and side lobe details of the measured waveform are used to construct an actual template function to replace the traditional ideal mathematical Delta function as the benchmark for subsequent path stripping.

[0027] b. Antenna array spatial signature calibration unit:

[0028] The complex gain response and three-dimensional radiation pattern of a 16-element uniform linear array in the 5.9 GHz band were measured in a microwave anechoic chamber far-field environment. Using the first antenna channel as a reference, the phase deviation vectors of the remaining 15 channels relative to the reference channel were calculated. The original received signal is expressed by the following formula. Compensation was performed to obtain high-fidelity space observation signals. :

[0029] .

[0030] Furthermore, step 3): Constructing an array measurement and control timing module for high-mobility urban scenarios, and performing multi-channel data acquisition that satisfies the physical constraints of channel coherence time, specifically includes:

[0031] a. Quasi-static constraint calculation for scanning period:

[0032] Real-time vehicle speed sensing Calculate the maximum Doppler frequency shift And set the full array scan cycle. The following physical boundary constraints must be met:

[0033]

[0034] in, For the channel coherence time, by The time limit is kept within 1.09ms to ensure that all 16 receiving antennas complete polling within a single channel quasi-static window, thus guaranteeing the effectiveness of the spatial feature vector.

[0035] b. Time slot redundancy and multipath isolation design:

[0036] The scan period is divided into 17 equal-width slots, and a "beginning and end masking" strategy is adopted within each slot: a 12.8-inch band is set at the beginning of the slot. The silence protection period is designed to absorb nonlinear glitches during RF switch switching; a time width greater than the maximum expected multipath delay in the environment is set at the end of the time slot. An isolation zone is established to prevent residual multipath energy from leaking into adjacent channels.

[0037] Furthermore, in step 4), an improved SAGE algorithm module based on adaptive initialization and space vector acceleration is constructed. This module is the core means to achieve "super-resolution" detection. Specifically, it operates through the following technical logic:

[0038] Adaptive initialization unit: It performs least-squares deconvolution using the pseudo-inverse matrix of the probe sequence, automatically locks the initial correlation peak, determines the number of multipaths, and provides an initial time delay estimate without manual intervention.

[0039] Dynamic threshold determination unit: It adopts a dual threshold criterion based on background noise statistics and the strongest path power backoff (e.g., -30dB) to automatically filter out hardware artifacts and random noise, ensuring the physical authenticity of the extracted path.

[0040] Space Vector Acceleration Unit: In the maximization step (M-Step), the linear algebraic properties of the receiving array manifold are utilized to replace the traditional angle-by-angle nested loop search by performing an inner product operation between the spatial eigenvectors and the guiding matrix.

[0041] Recursive iterative processing: The "expectation step (E-Step)" and "maximization step (M-Step)" are alternately iterated. The signal is reconstructed and stripped based on the measured template function until the residual energy converges. Finally, a channel multipath parameter library with nanosecond-level accuracy and subdegree-level accuracy is output.

[0042] This module receives the high-fidelity signal processed in step 2) and extracts multipath parameters through deconvolution initial value locking and matrix operation acceleration mechanisms. The specific calculation steps are as follows:

[0043] a. Pseudo-inverse adaptive initialization unit:

[0044] Construct a cyclic matrix containing different time delay shifts using the transmitter reference ZC sequence. And calculate its pseudo-inverse matrix. The initial channel impulse response estimate is obtained quickly through least-squares deconvolution operation. As shown in the following formula:

[0045]

[0046] b. Dynamic dual-threshold judgment unit:

[0047] according to Statistical characteristics are used to set detection thresholds ,in, This is the noise floor threshold based on statistics of the no-signal area of ​​the channel. For a power threshold that backs down 30dB relative to the strongest path energy of the current snapshot, only when The peak value is greater than When the time is right, it is automatically determined to be a valid multipath and its initial number of paths is locked. Compared with the initial delay;

[0048] c. Space Vector Accelerated Search Unit:

[0049] In the maximization step (M-Step) of the SAGE algorithm, the linear algebraic properties of the array manifold are utilized to preconstruct horizontal angles covering the entire spatial domain. Guiding matrix The multipath arrival angle is extracted by replacing the traditional nested loop search of scalars per angle with a one-time high-dimensional complex matrix inner product operation as shown in the following formula. :

[0050]

[0051] In the formula, The spatial feature vector after time-domain projection. It is the conjugate transpose of the guiding matrix.

[0052] Furthermore, the full array scan cycle in step 3) The design logic is as follows: based on the maximum Doppler frequency shift caused by vehicle movement. Determine the channel coherence time ;Will The time limit is kept within 1.09ms, and the constraints are met. The protection interval within the time slot; It is set to be greater than 500ns to isolate multipath energy crosstalk between adjacent spatial channels.

[0053] Furthermore, in step 4), the pseudo-inverse adaptive initialization unit calculates the initial channel impulse response estimate using the following formula. : ;

[0054] in, A cyclic matrix consisting of probe sequences. Its pseudo-inverse matrix; the dynamic dual-threshold determination refers to taking the maximum value of the environmental noise floor statistics and the value that is 30dB back off from the strongest path power relative to the current snapshot as the determination threshold for effective multipath. ;

[0055] The space vector accelerated search refers to using the linear algebraic properties of array manifolds to replace the angle-wise scalar nested loop search with a one-time high-dimensional complex matrix inner product operation as shown in the following formula: ;in, For a pre-constructed guidance matrix covering the horizontal angles across the entire airspace, the multipath arrival angle is locked by finding the maximum index of the matrix inner product result.

[0056] Furthermore, step 5) specifically includes:

[0057] a. Adaptive Clustering: A kernel power density algorithm based on the Laplace-Gaussian dual-domain kernel function is applied. This algorithm achieves automatic clustering of multipath components through the following technical steps:

[0058] Iterate through all extracted multipath components, applying amplitude exponential weighting centered on the time delay and angle of each component. Calculate the local power density of each point in the feature space using the following formula: This weighted quantization mechanism ensures that the multipath components with higher energy levels dominate the clustering decision.

[0059] The algorithm automatically identifies the number of physical scatterers in space by searching for the maximum local density points (i.e., "cluster centers"). Compared with the traditional K-means algorithm, this algorithm does not require a preset number of clusters and can adaptively handle non-stationary processes with drastic changes in the number of scatterers in complex urban environments. It can effectively aggregate fragmented multipath components into "scattering clusters" with clear physical meaning, thus providing accurate macroscopic statistical units for subsequent quantification of the smooth evolution of power caused by strip vegetation shading. Adaptive optimization: The algorithm automatically identifies the number of physical scatterers in space by searching for the maximum local density points (i.e., "cluster centers").

[0060] Step 5) also includes quantitative analysis of the tracking trajectory, specifically: based on the full life cycle trajectory of the multipath cluster obtained by tracking, linear regression fitting is performed on the trend of multipath cluster power evolution with snapshot in the logarithmic domain, and the slope parameter reflecting the change of cluster power fade-in and fade-out caused by the shading of the strip vegetation is extracted.

[0061] A high-precision multi-antenna detection system for complex urban vehicle-to-everything (V2X) channels implementing any of the methods described herein, comprising:

[0062] Transmitter module: Used to generate and radiate radio frequency detection signals carrying ZC sequences based on a high-stability rubidium atomic clock reference;

[0063] Receiver module: includes a 16-element uniform linear array, RF switches and acquisition board, used to perform TDM data acquisition;

[0064] Central processing module: stores computer programs for executing the end-to-end calibration, improved SAGE algorithm, kernel power density clustering, and multipath component distance tracking logic as described in claim 1.

[0065] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of any of the methods described.

[0066] The advantages and beneficial effects of this invention are as follows:

[0067] 1. High-precision time-frequency synchronization and super-resolution detection capabilities:

[0068] This invention constructs a high-precision multi-antenna time-domain detection system based on the NI PXIe platform. Employing a rubidium atomic clock (SRS FS725) in conjunction with real-time GPS signal training, it achieves a transceiver synchronization accuracy better than 10 ns, effectively eliminating phase drift in high-dynamic scenarios. By utilizing a 40MHz bandwidth Zadoff-Chu sequence to achieve nanosecond-level latency resolution, and combining it with a 16-element uniform linear array to achieve 5° spatial angular resolution, it can accurately capture weak multipath scattering components caused by strip vegetation in complex urban environments, providing a high-quality raw data foundation for refined vehicle-to-vehicle (V2V) modeling.

[0069] 2. Significant improvement in algorithm efficiency and extraction accuracy:

[0070] The improved SAGE algorithm proposed in this invention solves the problems of traditional algorithms' strong dependence on manual initial values ​​and susceptibility to local optima by using a pseudo-inverse adaptive initialization unit. Combined with a space vector accelerated search unit, angle optimization is transformed from scalar traversal to matrix inner product operations, improving the parameter extraction efficiency of a single channel snapshot by more than 30%. Furthermore, in conjunction with kernel power density (KPD) adaptive clustering and multipath component distance (MCD) tracking algorithms, it achieves automated and continuous capture of the spatiotemporal evolution trajectories of multipath clusters in more than 50,000 channel snapshots, avoiding the "fragmentation" phenomenon in the multipath parameter extraction process.

[0071] 3. Physical consistency of nonstationary evolution modeling:

[0072] The non-stationary cluster delay line (CDL) model constructed in this invention overcomes the limitation of traditional models (such as the 3GPP series) that simplify multipath cluster birth and death to "instantaneous jumps". By introducing a Markov evolution mechanism with dual time scales of birth and death time and sampling time, and combining it with a logarithmic domain linear smooth transition function fitted by actual measurements, a continuous fade-in / fade-out process of cluster power is realized. Verification results show that the simulated channel and the measured channel agree well on key statistical characteristics such as root mean square (RMS) delay spread and angle spread, and the accuracy is significantly better than the 3GPP standard CDL model, which can more realistically characterize the non-stationary channel flicker characteristics when vehicles pass through vegetated areas.

[0073] 4. Digital system-aided engineering design and optimization:

[0074] The independently developed channel analysis system for non-stationary cluster evolution in urban environments integrates data cleaning, parameter extraction, modeling and simulation, and visualization functions. Through a digital twin interactive interface, the system can intuitively reproduce the dynamic evolution process of multipath clusters under the obstruction of strip vegetation in complex urban areas. It quantifies the impact of vegetation scattering on path loss (adding an additional 15-25dB of loss), spatiotemporal dispersion, and spatial diversity gain, providing an intuitive and efficient scientific tool for algorithm development, array deployment, and link robustness assessment of vehicle-to-everything (V2X) multi-antenna systems.

[0075] The core innovation of this invention:

[0076] 1. Synergy between subsampling-level time-frequency synchronization and measured template calibration

[0077] Conventional techniques typically employ ideal mathematical functions (such as pulse signals) as the reconstruction benchmark for multipath stripping and rely on the device's internal crystal oscillator for synchronization. This invention achieves 10ns-level synchronization by disciplining a rubidium atomic clock in real time using GPS signals, providing an extremely precise physical prerequisite for phase coherence. Building upon this, it innovatively proposes "measured template calibration," which utilizes a direct cable connection to acquire waveforms containing actual hardware distortion. High-precision synchronization ensures the accuracy of the waveform, while the measured template ensures the cleanliness of the stripping. The synergy of these two techniques improves the thoroughness of the SAGE algorithm in the residual stripping stage by more than 15dB, enabling the extraction of weak multipath paths (such as vegetation scattering paths) masked by hardware sidelobes—something conventional ideal models cannot achieve.

[0078] 2. Dynamic timing design that satisfies coherent time constraints

[0079] Traditional Time Division Multiplexing (TDM) sampling often focuses only on the sampling rate, neglecting the mathematical constraints between vehicle speed and the consistency of spatial feature vectors. This invention strictly sets the full array scan period to be within 0.6 times the channel coherence time (1.09 ms). This design transforms "vehicle dynamics parameters" into "array measurement and control constraints." It ensures that the channel state is physically "quasi-static" during polling of the 16 antennas, thus guaranteeing the validity of the spatial phase data. This is the fundamental reason why angle parameters with 0.5-degree accuracy can be extracted subsequently.

[0080] 3. Adaptive Extraction Based on Pseudo-Inverse Initialization and Space Vector Acceleration

[0081] The traditional SAGE algorithm is widely recognized as "extremely time-consuming" and "reliant on human experience (preset number of paths)." This invention introduces pseudo-inverse matrix deconvolution for initial value locking and utilizes parallel matrix inner products to replace scalar search. Pseudo-inverse initialization automates the problems of "where to start calculating" and "how many paths to calculate," while vector acceleration addresses the problem of "how to calculate quickly." This combination makes it possible to process more than 50,000 massive snapshots, reducing the processing time from several weeks to several days, achieving a significant leap in both efficiency and accuracy.

[0082] 4. Clustering and Trajectory Quantization Tracking Based on Dual-Domain Kernel Functions

[0083] Existing channel models (such as the 3GPP standard) often treat multipath clusters as "instantaneously generated and instantaneously disappearing" pulses, and clustering typically employs the K-means algorithm with a preset number of clusters. This invention innovatively designs a Laplace-Gaussian dual-domain kernel function (matching different distribution characteristics of delay and angle) and utilizes logarithmic domain linear regression to quantize power evolution. This is the most profound characterization of the non-stationary characteristics of the channel in this invention. The KPD algorithm adaptively determines the number of clusters, MCD tracks the evolution trajectory, and finally, by fitting the slope, it finds the physical law of "smooth power fade-in and fade-out." This algorithm combination corrects the physical distortion of the international standard model in vegetation-occupied scenarios. Attached Figure Description

[0084] Figure 1 This invention provides the overall logical architecture and signal flow diagram of the detection system according to a preferred embodiment;

[0085] Figure 2 This is a physical composition diagram of the hardware detection platform described in this invention;

[0086] Figure 3 This is a schematic diagram of the connection of the measured template function calibration system described in this invention;

[0087] Figure 4 A waveform comparison graph showing the actual template function and the theoretical template function;

[0088] Figure 5 The antenna array calibration scene and the extracted 3D radiation pattern results;

[0089] Figure 6 A TDM (Transmission and Control) timing diagram designed for high-mobility urban scenarios;

[0090] Figure 7 This is a comparison chart of the computational efficiency of the improved SAGE algorithm of this invention and the traditional algorithm;

[0091] Figure 8 This is a schematic diagram illustrating the iterative convergence characteristics of the improved SAGE algorithm of this invention;

[0092] Figure 9 This is a verification diagram showing the measured results of the angular resolution of this detection system;

[0093] Figure 10 This is a comparison chart of the RMS delay spread of the model of this invention, the 3GPP standard model, and the measured channel.

[0094] Figure 11 This is a comparison chart of channel capacity in different urban scenarios obtained by evaluating the method of this invention. Detailed Implementation

[0095] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0096] The technical solution of the present invention to solve the above-mentioned technical problems is:

[0097] Figure 1 The diagram shows the overall logical architecture and signal flow of the detection system of this invention. It illustrates the distributed detection architecture based on the NI PXIe platform and details the complete radio frequency link from ZC sequence generation and up-conversion amplification to 16-element ULA reception and acquisition, demonstrating the hardware integration method of the system in the 5.9GHz band.

[0098] Figure 2 This is a physical diagram of the hardware detection platform described in this invention, showing the physical layout of the distributed detection system's chassis, rubidium atomic clock, and radio frequency link.

[0099] Figure 3 This is a schematic diagram of the connection of the measured template function calibration system described in this invention. The diagram shows the hardware connection method of the transmitter and receiver performing "back-to-back" closed-loop calibration through an RF coaxial cable, which is a prerequisite for obtaining a template function containing non-ideal characteristics (fingerprints) of the RF front end.

[0100] Figure 4 The graph shows a waveform comparison between the actual template function and the theoretical template function. The solid line in the graph represents the measured correlation peak waveform, which includes hardware distortion. This measured template replaced the ideal mathematical template, improving the energy stripping efficiency of the subsequent SAGE algorithm in the signal reconstruction stage by more than 15dB.

[0101] Figure 5 This image shows the antenna array calibration scenario and the extracted 3D radiation pattern. It fully demonstrates the phase offset calibration of a 16-element linear array in a microwave anechoic chamber environment. The calibration experiment process. By correcting the amplitude-phase inconsistency between channels, the accuracy of spatial angle estimation was ensured, and phase ambiguity caused by differences in cable length was eliminated.

[0102] Figure 6 This is a timing diagram for TDM (Time Management Device) telemetry and control designed for high-mobility urban scenarios. The diagram clearly defines the allocation logic of 17 equal-width time slots within a scan cycle. The diagram details the 12.8-second time slot front-end for each acquisition slot. The quiet period design is used to absorb transient glitches generated by RF switch switching, ensuring the purity and quasi-static characteristics of data during multi-channel acquisition.

[0103] Figure 7 This figure compares the computational efficiency of the improved SAGE algorithm of this invention with that of the traditional algorithm. It shows a quantitative comparison of the computational time consumption of the traditional SAGE algorithm and the improved algorithm based on space vector acceleration proposed in this invention when processing measured raw IQ data snapshots of different sizes (e.g., 200GB to 1000GB). The horizontal axis represents the size of the processed channel snapshots, and the vertical axis represents the total algorithm execution time (hours). The figure clearly shows that as the data volume increases, the improved algorithm, thanks to its matrix parallel operation advantage, exhibits a significantly lower slope in its computational time growth curve compared to the traditional algorithm, demonstrating an execution efficiency improvement of approximately 30% to 40% in the context of massive data.

[0104] Figure 8 This diagram illustrates the iterative convergence characteristics of the improved SAGE algorithm of this invention. The figure depicts the convergence trajectory of the multipath parameters (complex amplitude and angle of arrival) as the number of iterations changes. The horizontal axis represents the number of iterations, and the vertical axis represents the normalized values ​​of the parameter estimates. The curves in the figure show that, due to the introduction of the prediction unit based on pseudo-inverse adaptive initialization, the algorithm does not need to start from a random state for optimization. After 3 to 5 iterations, the physical parameters of each multipath can enter the stable convergence interval, proving that the algorithm has extremely high numerical stability while maintaining accuracy.

[0105] Figure 9 This figure verifies the measured results of the detection system's angular resolution. It shows the system's resolution performance using a channel simulator in a two-path scenario with a set spatial angular difference of 5°. The figure is presented in the form of an angular power spectrum (APS), with the horizontal axis representing the angle of arrival (degrees) and the vertical axis representing the normalized power (dB). The results show that at extremely close angular intervals, the angular spectrum extracted by this system clearly presents two independent, narrow, deep peaks, and the sidelobe level is effectively suppressed. This verifies, from the perspective of hardware and algorithm collaboration, that the system possesses sub-degree (better than 0.5°) angular resolution.

[0106] Figure 10This figure compares the RMS delay spread of the proposed model, the 3GPP standard model, and the measured channel. The graph uses a cumulative distribution function (CDF) to show the statistical matching degree of the three models on the root mean square delay spread (DS). The horizontal axis represents the RMS delay spread value (ns), and the vertical axis represents the cumulative probability. The measured curve in the figure almost coincides with the curve of the non-stationary model of this invention, while the curve of the 3GPP standard model shows a significant deviation due to the lack of consideration for the smooth energy transition caused by vegetation obstruction. This comparison intuitively demonstrates the superiority of the dual-timescale evolution mechanism proposed in this invention in characterizing the dispersion characteristics of complex urban channels.

[0107] Figure 11 This figure presents a comparison of channel capacity in different urban scenarios evaluated using the method of this invention. It shows the channel capacity distribution of a single-input multiple-output (SIMO) system calculated based on measured parameters under open line-of-sight (LOS) and vegetated occlusion (OLOS) scenarios. The horizontal axis represents channel capacity (bps / Hz), and the vertical axis represents cumulative distribution probability. The figure quantitatively compares the capacity performance under the two scenarios, revealing that although strip vegetation introduces path loss, the multipath diffuse scattering component it induces increases the spatial diversity gain of the channel. This results in a wider statistical dynamic range for the capacity distribution in the OLOS scenario, providing a decision-making basis for the deployment of vehicle-to-everything (V2X) systems in complex environments.

[0108] This invention provides a high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) channels. This method achieves high-precision characterization of complex urban multipath channels through the synergistic effects of hardware synchronization compensation, end-link closed-loop calibration, timing design that meets coherence time constraints, and an improved SAGE algorithm. The method comprises five core steps: hardware system construction and synchronization, end-link error calibration, high-mobility telemetry and control timing design, improved SAGE parameter extraction, and multipath clustering and tracking. The specific implementation details of each step are as follows:

[0109] Example 1: Construction of the detection hardware system and synchronous implementation of the subsampling level

[0110] This embodiment addresses high-vehicle-to-vehicle (V2V) scenarios in urban areas, establishing a distributed detection platform based on a modular instrument bus architecture. The specific implementation process is as follows:

[0111] 1. Hardware architecture setup

[0112] Reference Figure 1The detection system consists of a transmitter (TX) and a receiver (RX). The transmitter, mounted on the first test vehicle, is a core NI PXIe vector transceiver used to generate a 40MHz bandwidth Zadoff-Chu (ZC) constant envelope pseudo-random sequence with a center frequency set at 5.9GHz. The receiver, mounted on the second test vehicle, employs a 16-element uniform linear array (ULA) with an element spacing of... Set to 25.4mm (corresponding to half wavelength of 5.9GHz), and use a high-speed RF switch to perform time-division multiplexing (TDM) signal acquisition.

[0113] 2. Subsampling-level synchronization mechanism

[0114] Reference Figure 2 To eliminate phase deviation caused by local oscillator frequency drift at both ends of the receiver in non-stationary channel measurements, the system connects a high-stability rubidium atomic clock (SRS FS725) tamed by GPS 1PPS signals to both the TX and RX ends. Through a phase-locked loop, an external 10MHz reference clock is injected into the local oscillator to ensure that the relative time offset between the transmitter and receiver at the radio frequency level is always locked within 10ns. Even when the vehicle travels through an area obstructed by vegetation, causing a brief loss of GPS signal lock, the rubidium atomic clock still maintains its superior stability compared to other instruments. The short-term stability enters a maintenance mode, maintaining phase coherence between snapshots, providing a physical basis for subsequent nanosecond-level time delay resolution and sub-degree-level angle estimation.

[0115] Example 2: End-to-End Error Closed-Loop Calibration and Hardware Distortion Compensation

[0116] This embodiment eliminates the interference of non-ideal hardware characteristics on multipath parameter estimation through a measured data-driven approach. The specific process is as follows:

[0117] 1. Experimental template function acquisition

[0118] Reference Figure 3 Before performing field testing, the transmitter output port and receiver input port were connected back-to-back in a closed loop via an RF coaxial cable. The end-to-end integrated response, including transmit pulse shaping, RF front-end bandpass characteristic deviation, power amplifier nonlinearity, and receiver filter impulse response, was measured. (Refer to...) Figure 4 The measured correlation peak waveform (solid line) exhibits significant hardware distortion in the main lobe contour and side lobe details compared to the mathematically ideal template (dashed line). In this embodiment, the measured waveform is stored in the algorithm benchmark library as a benchmark template for subsequent SAGE algorithm signal reconstruction, thereby reducing the residual energy during path stripping by more than 15dB.

[0119] 2. Antenna Array Spatial Signature Calibration

[0120] Reference Figure 5 A 16-element uniform linear array was measured in a microwave anechoic chamber in the far-field environment. Using antenna channel 1 as a reference, the phase deviation vectors of the remaining channels relative to the reference channel were calculated. In actual measurements, the original received signal is analyzed using formula (1). Phase compensation was performed to obtain high-fidelity space observation signals. .

[0121] This step effectively eliminates phase inconsistencies caused by varying RF cable lengths or differences in devices between channels.

[0122] Example 3: Measurement and Control Timing Design and Data Acquisition in High-Mobility Scenarios

[0123] To address the severe Doppler frequency shift caused by vehicles traveling at speeds exceeding 40 km / h in complex urban environments, this embodiment designs a measurement and control scheme that satisfies the quasi-static physical constraints of the channel:

[0124] 1. Calculation of physical constraints for scanning period

[0125] Reference Figure 6 The system calculates the channel coherence time in real time. At typical vehicle speeds Below, maximum Doppler frequency shift ,according to The calculated coherence time is approximately 1.9 ms. To ensure that the data acquired from all 16 antenna channels of the entire array are within the same quasi-static window, this embodiment limits the full array scan period. Strictly controlled within 1.09ms, meeting the requirements. The physical constraints ensure the phase coherence of the spatial eigenvectors.

[0126] 2. Time slot redundancy and isolation design

[0127] like Figure 6 As shown at the bottom, a complete scan cycle is divided into 17 slots. In this embodiment, a 12.8-inch sensor is set at the front end of each acquisition slot. The silent protection period is used to absorb transient glitches generated by RF switch switching; the time width at the end of the time slot is set to be greater than the maximum expected multipath delay in the environment. The isolation zone. Over 50,000 valid channel snapshots were collected in actual measurements, accumulating high-quality raw IQ data for non-stationary channel modeling.

[0128] Example 4: High-precision parameter extraction using an improved SAGE algorithm

[0129] This embodiment utilizes an adaptive initialization and space vector accelerated search mechanism to achieve efficient extraction of multipath physical parameters:

[0130] 1. Pseudo-inverse adaptive initialization

[0131] Constructing a cyclic matrix using a reference probe sequence The initial channel impulse response estimate is calculated using formula (3). .

[0132] in, Its pseudo-inverse matrix is ​​used. Through deconvolution operations, the system can automatically locate the initial correlation peak without manually pre-setting the number of multipaths. This is combined with dynamic dual-threshold judgment—taking the maximum value of the environmental noise floor statistics and the value that falls back 30dB from the strongest path power in the current snapshot as the threshold. Only when the peak value is greater than It is then determined to be an effective multipath.

[0133] 2. Space Vector Accelerated Search

[0134] In the maximization step (M-Step), the linear algebraic properties of the array manifold are utilized to pre-construct horizontal angles covering the entire spatial domain. Guiding matrix By replacing the traditional angle-by-angle scalar nested loop search with a one-time high-dimensional complex matrix inner product operation of formula (4), the multipath arrival angle is extracted. .

[0135] In the formula, This represents the spatial feature vector after temporal projection. This method improves the computational efficiency of massive snapshots by more than 30%.

[0136] Example 5: Multipath Clustering, Tracking, and Non-stationary Evolution Feature Extraction

[0137] This embodiment details how to convert the extracted discrete multipath parameters into channel model parameters with physical meaning:

[0138] 1. Implementation of KPD adaptive clustering:

[0139] Due to the highly uneven distribution of scatterers in complex urban areas, this invention employs the KPD algorithm to aggregate the tens of thousands of multipath components extracted in step 4. The algorithm utilizes the Laplace kernel function to handle dispersion in the angular dimension and the Gaussian kernel function to handle dispersion in the time delay dimension, calculating the power density at each multipath point using the following formula:

[0140]

[0141] This method can automatically identify scattering clusters with clear physical meaning, effectively avoiding the clustering bias caused by the inaccurate preset number of clusters in the traditional K-means algorithm.

[0142] 2. MCD-based continuous tracking of cluster trajectories:

[0143] To capture dynamic changes during the movement of drones / vehicles, this invention utilizes the MCD index to quantify the displacement and power changes of clusters between adjacent snapshots. By tracking 50,000 consecutive snapshots, the system can record the entire lifecycle trajectory of each cluster from birth to death.

[0144] 3. Extraction of linear fade-in / fade-out features under vegetation occlusion:

[0145] This is the core of the invention in characterizing non-stationary characteristics. In a typical scenario of strip vegetation shading, this embodiment extracts the slope (dB / ms) of the power fade-in / fade-out change by performing linear regression fitting on the tracking trajectory. The parameters generated in this step directly support the construction of the non-stationary cluster delay line channel model, improving the consistency between the simulation model and the measured values ​​in the dynamic evolution of delay spread and angle spread to over 95%.

[0146] Example 6: Performance Verification and Comparison of the Technical Solution of the Invention

[0147] To further verify the superiority of the detection method and parameter extraction algorithm described in this invention, this embodiment provides a performance comparison analysis based on measured data and a channel simulator:

[0148] 1. Verification of algorithm computational efficiency:

[0149] Reference Figure 7 The improved SAGE algorithm proposed in this invention, when processing the same scale of measured data (e.g., 1000 channel snapshots), shows a computational time comparison between the traditional SAGE algorithm and the proposed algorithm. The improved algorithm, due to the introduction of a space vector acceleration mechanism, achieves approximately a 35% improvement in computational efficiency. (Refer to...) Figure 8 The algorithm converges in about 5 iterations, and its log-likelihood function has a stable growth curve, proving that the adaptive initialization mechanism can effectively prevent the algorithm from getting trapped in local optima.

[0150] 2. Verification of high-precision resolution of the detection system:

[0151] Reference Figure 9 In this embodiment, a channel simulator was used to set up multipath components with a 5° angle difference for super-resolution performance testing. Experimental results show that this detection system can clearly distinguish arrival paths with extremely close physical locations in space. Statistical calculations show that the root mean square error (RMSE) of angle estimation in the 5.9GHz band is better than 0.5°, and the delay estimation error is better than 0.2ns. This proves that the end-to-end closed-loop calibration module described in step 2 effectively eliminates phase ambiguity introduced by cables and the RF front end.

[0152] 3. Comparison of consistency between non-stationary models and measured statistical characteristics:

[0153] Reference Figure 10 This embodiment compares the non-stationary CDL model constructed in this invention, the 3GPP standard CDL model, and the measured cumulative distribution function (CDF) of the channel on the root mean square (RMS) delay spread. The comparison results show that:

[0154] a. The 3GPP standard model, due to its quasi-static assumption, cannot characterize the smooth energy evolution caused by vegetation shading, and its CDF curve has a large systematic deviation from the measured value.

[0155] b. The model of this invention uses the linear fade-in / fade-out slope extracted in step 5 to perform time-varying weight quantization of cluster power. The generated simulation data shows good agreement with the measured curves in both LOS and OLOS scenarios. Calculations show that its statistical characteristic matching degree is more than 15% higher than that of the 3GPP standard model.

[0156] 4. Value of Vehicle-to-Everything (V2X) Channel Capacity Assessment and Engineering Guidance:

[0157] Reference Figure 11 This embodiment demonstrates the channel capacity CDF curve calculated using parameters extracted using the method of this invention. The results quantitatively reveal that in complex urban environments, although linear vegetation obstruction (OLOS) introduces approximately 15-25 dB of path loss, the abundant lateral diffuse scattering components generated by the vegetation increase the channel's equivalent degrees of freedom (EDOF), making the spatial diversity gain of the OLOS scenario superior to that of the open line-of-sight (LOS) scenario under the same signal-to-noise ratio conditions. This finding provides a direct physical layer basis for the anti-fading design and array deployment optimization of multi-antenna systems in vehicle-to-everything (V2X) networks, demonstrating the practical value of this invention in assisting the construction of intelligent transportation systems.

[0158] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions.

[0159] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0160] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0161] The above embodiments should be understood as illustrative only and not as limiting the scope of protection of the present invention. After reading the description of the present invention, those skilled in the art can make various alterations or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) communication channels, characterized in that, Includes the following steps: Step 1): Build and deploy a high-precision multi-antenna time-domain detection hardware system based on the NI PXIe platform, and establish a sub-sampling level transmit / receive synchronization mechanism; Step 2): Construct and apply a full-link error closed-loop calibration module to perform hardware distortion compensation on the original observed signal through a measured data-driven approach; Step 3): Construct an array measurement and control timing module for high-mobility urban scenarios, and perform multi-channel data acquisition that meets the physical constraints of channel coherence time; Step 4): Construct an improved SAGE algorithm module based on adaptive initialization and space vector acceleration to recursively extract the physical parameters of multipath components; Step 5): Construct a dynamic cluster evolution feature extraction unit, and use the kernel power density algorithm and multipath component distance algorithm to aggregate and continuously track the extracted multipath components, so as to realize the quantitative description of the non-stationary birth and death and power evolution law of channel multipath clusters.

2. The high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) channels according to claim 1, characterized in that, In step 1), the high-precision multi-antenna time-domain detection hardware system adopts a distributed architecture, consisting of a transmitter (TX), a receiver (RX), and a synchronization support subsystem. At the transmitter, a vector signal transceiver generates a Zadoff-Chu constant envelope pseudo-random sequence with a center frequency of 5.9 GHz and an effective bandwidth of 40 MHz as the detection waveform. The specific synchronization process is as follows: a. Establishment of frequency and time bases: High-stability rubidium atomic clocks tamed by GPS 1PPS signals are configured at both the transmitting and receiving ends to provide an external reference frequency of 10MHz for the system. The relative time offset between the transmitting and receiving ends at the RF local oscillator level is locked within 10ns through a phase-locked loop to ensure the phase coherence of the channel snapshot during long-term measurement. b. Subsampling-level synchronization maintenance: By utilizing the high short-term stability of rubidium atomic clocks, the system enters hold mode in environments with weak GPS signal obstruction, enabling the system to have a nanosecond-level time delay resolution foundation, thereby supporting the subsequent sub-sampling level extraction of scattering paths from vegetation in complex urban areas. The synchronization support subsystem is the "nerve center" of the high-precision detection system. This system disciplines the rubidium atomic clock in real time by using independently configured GPS signals at both the transmitting and receiving ends, combining the high short-term stability of the atomic clock with the long-term accuracy of the GPS signal. Through the internal phase-locked loop, the subsystem injects the externally provided 10MHz reference frequency into the local radio frequency front end, thereby strictly locking the time offset between the transmitting and receiving ends within 10ns. This allows the system to maintain extremely high phase coherence even when the measurement vehicle is moving in complex urban areas and faces instantaneous external signal blockage, providing a precise time and frequency reference for capturing the non-stationary multipath generation and extinction and power evolution laws.

3. The high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) channels according to claim 2, characterized in that, The detection excitation sequence adopts the Zadoff-Chu sequence, with a sequence length of [missing information]. The resampling rate is 80MHz; the element spacing of the 16-element uniform linear array is set to 25.4mm, which is half the wavelength corresponding to the 5.9GHz carrier frequency; the time synchronization accuracy of the transmitting and receiving ends is locked within 10ns, which means that the frequency drift of the local oscillator at the transmitting and receiving ends is eliminated by the GPS discipline mechanism, so that the delay estimation deviation between channel snapshots within a long measurement period is kept at the subsampling level. The 16-element uniform linear array has an element spacing of 25.4 mm, which is half the wavelength corresponding to the 5.9 GHz carrier frequency, in order to eliminate phase ambiguity in angle estimation.

4. The high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) channels according to claim 1, characterized in that, Step 2): Constructing and applying a full-link error closed-loop calibration module to perform hardware distortion compensation on the original observed signal using a measured data-driven approach specifically includes: a. Measured template function calibration unit: The transmitter output port and receiver input port are directly connected back-to-back via an RF coaxial cable; the end-to-end integrated impulse response waveform, including transmit pulse shaping, RF front-end group delay ripple, power amplifier weak nonlinearity, and receiver filter impulse response, is obtained by actual measurement; the main lobe profile and side lobe details of the measured waveform are used to construct an actual template function to replace the traditional ideal mathematical Delta function as the benchmark for subsequent path stripping. b. Antenna array spatial signature calibration unit: The complex gain response and three-dimensional radiation pattern of a 16-element uniform linear array in the 5.9 GHz band were measured in a microwave anechoic chamber far-field environment. Using the first antenna channel as a reference, the phase deviation vectors of the remaining 15 channels relative to the reference channel were calculated. The original received signal is expressed by the following formula. Compensation was performed to obtain high-fidelity space observation signals. : 。 5. The high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) communication channels according to claim 1, characterized in that, Step 3): Constructing an array measurement and control timing module for high-mobility urban scenarios, performing multi-channel data acquisition that satisfies the physical constraints of channel coherence time, specifically including: a. Quasi-static constraint calculation for scanning period: Real-time vehicle speed sensing Calculate the maximum Doppler frequency shift And set the full array scan cycle. The following physical boundary constraints must be met: in, For the channel coherence time, by The time limit is kept within 1.09ms to ensure that all 16 receiving antennas complete polling within a single channel quasi-static window, thus guaranteeing the effectiveness of the spatial feature vector. b. Time slot redundancy and multipath isolation design: The scan period is divided into 17 equal-width slots, and a "beginning and end masking" strategy is adopted within each slot: a 12.8-inch band is set at the beginning of the slot. The silence protection period is designed to absorb nonlinear glitches during RF switch switching; a time width greater than the maximum expected multipath delay in the environment is set at the end of the time slot. An isolation zone is established to prevent residual multipath energy from leaking into adjacent channels.

6. The high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) communication channels according to claim 1, characterized in that, Step 4): Constructing an improved SAGE algorithm module based on adaptive initialization and space vector acceleration. This module is the core means to achieve "super-resolution" detection; it operates through the following technical logic: Adaptive initialization unit: It performs least squares deconvolution using the pseudo-inverse matrix of the probe sequence, automatically locks the initial correlation peak, determines the number of multipaths, and provides an initial time delay estimate without manual intervention; Dynamic threshold determination unit: It adopts a dual threshold criterion based on noise floor statistics and strongest path power backoff to automatically filter out hardware artifacts and random noise, ensuring the physical authenticity of the extracted path; Space Vector Acceleration Unit: In the maximization step (M-Step), the linear algebraic properties of the receiving array manifold are utilized to replace the traditional angle-by-angle nested loop search by the inner product operation of the spatial eigenvector and the guiding matrix; Recursive iterative processing: The "Expectation Step (E-Step)" and "Maximization Step (M-Step)" are alternately iterated. The signal is reconstructed and stripped based on the measured template function until the residual energy converges. Finally, a channel multipath parameter library with nanosecond-level and subdegree-level accuracy is output. This module receives the high-fidelity signal processed in step 2) and extracts multipath parameters through deconvolution initial value locking and matrix operation acceleration mechanisms. The specific calculation steps are as follows: a. Pseudo-inverse adaptive initialization unit: Construct a cyclic matrix containing different time delay shifts using the transmitter reference ZC sequence. And calculate its pseudo-inverse matrix. The initial channel impulse response estimate is obtained quickly through least-squares deconvolution operation. As shown in the following formula: b. Dynamic dual-threshold judgment unit: according to Statistical characteristics are used to set detection thresholds ,in, This is the noise floor threshold based on statistics of the no-signal area of ​​the channel. For a power threshold that backs down 30dB relative to the strongest path energy of the current snapshot, only when The peak value is greater than When the time is right, it is automatically determined to be a valid multipath and its initial number of paths is locked. Compared with the initial delay; c. Space Vector Accelerated Search Unit: In the maximization step (M-Step) of the SAGE algorithm, the linear algebraic properties of the array manifold are utilized to preconstruct horizontal angles covering the entire spatial domain. Guiding matrix The multipath arrival angle is extracted by replacing the traditional nested loop search of scalars per angle with a one-time high-dimensional complex matrix inner product operation as shown in the following formula. : In the formula, The spatial feature vector after time-domain projection. It is the conjugate transpose of the guiding matrix.

7. The high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) channels according to claim 4, characterized in that, Step 3) Full array scan cycle The design logic is as follows: based on the maximum Doppler frequency shift caused by vehicle movement. Determine the channel coherence time ;Will The time limit is kept within 1.09ms, and the constraints are met. The protection interval within the time slot; It is set to be greater than 500ns to isolate multipath energy crosstalk between adjacent spatial channels.

8. The high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) communication channels according to claim 1, characterized in that, In step 4), the pseudo-inverse adaptive initialization unit calculates the initial channel impulse response estimate using the following formula. : ; in, A cyclic matrix consisting of probe sequences. Its pseudo-inverse matrix; the dynamic dual-threshold determination refers to taking the maximum value of the environmental noise floor statistics and the value that is 30dB back off from the strongest path power relative to the current snapshot as the determination threshold for effective multipath. ; The space vector accelerated search refers to using the linear algebraic properties of array manifolds to replace the angle-wise scalar nested loop search with a one-time high-dimensional complex matrix inner product operation as shown in the following formula: ;in, For a pre-constructed guidance matrix covering the horizontal angles across the entire airspace, the multipath arrival angle is locked by finding the maximum index of the matrix inner product result.

9. A high-precision multi-antenna detection and parameter extraction method for complex urban vehicle-to-everything (V2X) communication channels according to claim 1, characterized in that, Step 5) specifically includes: a. Adaptive Clustering: A kernel power density algorithm based on the Laplace-Gaussian dual-domain kernel function is applied. This algorithm achieves automatic clustering of multipath components through the following technical steps: Iterate through all extracted multipath components, and apply amplitude exponential weighting centered on the time delay and angle of each component; calculate the local power density of each point in the feature space using the following formula: This weighted quantization mechanism ensures that the multipath components with higher energy levels dominate the clustering decision. The algorithm automatically identifies the number of physical scatterers in space by searching for the maximum local density point (i.e., the "cluster center"); b. Cluster trajectory tracking: The multipath component distance index is applied to quantify the geometric similarity of component clusters between adjacent channel snapshots, so as to realize the continuous tracking of physical clusters on the time axis; Step 5) also includes quantitative analysis of the tracking trajectory, specifically: based on the full life cycle trajectory of the multipath cluster obtained by tracking, linear regression fitting is performed on the trend of multipath cluster power evolution with snapshot in the logarithmic domain, and the slope parameter reflecting the change of cluster power fade-in and fade-out caused by the shading of the strip vegetation is extracted.

10. A high-precision multi-antenna detection system for complex urban vehicle-to-everything (V2X) channels implementing the method of any one of claims 1 to 9, characterized in that, include: Transmitter module: Used to generate and radiate radio frequency detection signals carrying ZC sequences based on a high-stability rubidium atomic clock reference; Receiver module: includes a 16-element uniform linear array, RF switches and acquisition board, used to perform TDM data acquisition; Central processing module: stores computer programs for executing the end-to-end calibration, improved SAGE algorithm, kernel power density clustering, and multipath component distance tracking logic as described in claim 1.

Images