Multi-station cooperative sensing method based on cellular-free sensing fusion architecture

Abstract

The invention relates to a multi-station cooperative sensing method based on a non-cellular sensing fusion architecture, and the method comprises the steps: S1, constructing a distributed non-cellular sensing fusion architecture integrating communication and sensing functions, which comprises a BBU and a head-end device; s2, constructing a weighted Kalman filtering fusion algorithm for fusing the trajectory data of the plurality of head-end devices to obtain target position estimation, and reducing uncertainty through a fused covariance matrix; s3, configuring communication parameters between the head end equipment and the BBU; s4, configuring sensing parameters of the BBU, and setting a trajectory processing algorithm and parameters; and S5, fusing the tracks of the head end devices by adopting a weighted Kalman filtering fusion algorithm according to the track data obtained from the different head end devices, optimizing target position estimation, and outputting a fusion track after dynamically adjusting a fusion weight and a covariance matrix. Compared with the prior art, the method has the advantages of improving the precision and stability of the system and the like.

Description

Technical Field

[0001] The present invention relates to a non-cellular collaborative sensing system, and in particular to a multi-station collaborative sensing method based on a non-cellular synaesthesia fusion architecture. Background Art

[0002] With the development of wireless communications and information technology, cellular-free networks and integrated interaware systems (ISACs) have become research hotspots for intelligent communication and perception systems. Traditional cellular networks rely on a single base station for data transmission and perception, making them susceptible to signal attenuation, bandwidth limitations, and coverage blind spots. In complex environments, insufficient base station coverage leads to signal loss and delay, making them unable to meet high real-time requirements. Cell-free networks, through the collaboration of multiple base stations for perception and data transmission, overcome the limitations of traditional cellular networks. They can provide better coverage over large areas, reduce signal blind spots, and enhance target detection accuracy and perception range. However, practical applications still face technical challenges such as data fusion, bandwidth, and latency. Fusion of multi-site perception data is difficult due to environmental differences and hardware heterogeneity. Efficiently integrating this data for accurate target positioning is crucial. Furthermore, efficient operation of the system under low-bandwidth and low-latency conditions remains a performance bottleneck. Furthermore, the system is prone to false alarms when there are no targets in the airspace, compromising system accuracy and reliability. Traditional perception systems are prone to generating false alarms when there are no targets in the airspace, resulting in unnecessary waste of computing resources and alarms. How to maintain high-precision detection and positioning in complex environments and avoid false alarms remains one of the core issues that multi-station collaborative perception systems must solve.

[0003] Therefore, current technical research urgently needs to develop a new multi-station collaborative perception method that can give full play to the collaborative advantages of each site, overcome the above problems while ensuring system stability, and improve the system's perception accuracy, data transmission efficiency, and real-time response capabilities. Summary of the Invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned existing technologies and provide a multi-station collaborative perception method based on a cellular-free interawareness fusion architecture to realize cellular-free multi-node joint target positioning and tracking, and utilize the method of inter-node perception information fusion to enhance the target position estimation capability, and to achieve stable network connection and data transmission between devices, ensuring that no false alarms are generated when there is no target in the airspace, effectively integrating heterogeneous perception data between different sites, and ensuring high-precision target positioning and tracking.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] According to one aspect of the present invention, a multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture is provided, the method comprising the following steps:

[0007] Step S1, building a distributed non-cellular synaesthesia fusion architecture integrating communication and perception functions, including a baseband processing unit (BBU) and a head-end device;

[0008] Step S2: construct a weighted Kalman filter fusion algorithm to fuse the trajectory data of multiple head-end devices to obtain target position estimation and reduce uncertainty through the fused covariance matrix;

[0009] Step S3: Configure the communication parameters between the headend device and the BBU to ensure normal communication between the two.

[0010] Step S4: configure the BBU's sensing parameters and set the trajectory processing algorithm and parameters;

[0011] In step S5, the trajectory data obtained from different head-end devices are fused using the weighted Kalman filter fusion algorithm constructed in step S2 to optimize the target position estimation, dynamically adjust the fusion weights and covariance matrix, and then output the fused trajectory.

[0012] As an optimal technical solution, the distributed cellular-free synaesthesia fusion architecture of step S1 also includes a user-centered processing unit UCDU and a central control unit, wherein the central control unit is connected to multiple UCDUs, one UCDU is connected to multiple BBUs, and one BBU is connected to multiple head-end devices.

[0013] As a preferred technical solution, the head-end device adopts an active antenna unit AAU, which is responsible for the transmission and reception of radio frequency signals. The BBU is responsible for baseband processing of communication and perception signals, receiving radio frequency signals from the AAU, and converting them into digital signals.

[0014] As a preferred technical solution, step S2 specifically includes:

[0015] Assume the target status is:

[0016]

[0017] where x k ,y k represents the two-dimensional position of the target at time k, Indicates the speed of the target;

[0018] The state transition equation is:

[0019] x k =Fx k-1 +w k-1

[0020] Where F is the state transfer matrix, wk-1 is the process noise, assuming it is Gaussian noise with zero mean and covariance Q k , the observation equation is:

[0021] z k =Hx k +v k

[0022] where z k is the perception data, H is the observation matrix, v k is the observation noise, which is also assumed to be Gaussian noise with zero mean and covariance R k ;

[0023] Assume there are N sensing nodes, and the observation value of each node is Where i represents the i-th node. In the process of multi-node joint estimation, the observation values need to be weighted and fused according to the weight of each node. Assume that Z k Represents the set of observations of all nodes:

[0024]

[0025] The goal of weighted Kalman filtering is to use the observation data of all nodes to update the state estimate of the target and the covariance matrix P k .

[0026] As a preferred technical solution, the state estimation of the update target and the covariance matrix P k Specifically include:

[0027] First, calculate the observation vector The weighted average of

[0028]

[0029] where w i is the weighting coefficient corresponding to the i-th observation source, N is the number of base stations participating in the fusion, is the observation vector of the i-th base station at time k;

[0030] Then, calculate the weighted covariance matrix S k :

[0031]

[0032] in is the observation vector The weighted average of .

[0033] After obtaining the weighted average observation value and weighted covariance, the update steps specifically include:

[0034] 1) Calculate the gain K k

[0035]

[0036] 2) Update target state estimate:

[0037]

[0038] 3) Update the covariance matrix:

[0039] P k =(IK k H)P k-1

[0040] Where I represents the identity matrix.

[0041] As a preferred technical solution, the weighted Kalman filter fusion algorithm improves the accuracy of target positioning and tracking by weighting the observation data of multiple sensing nodes, and during the fusion process, takes into account the reliability and perception capabilities of different nodes and assigns different weights to each node.

[0042] As a preferred technical solution, step S3 specifically includes:

[0043] Step S301: After physically connecting the headend device to the BBU, check the indicator light on the headend device and use the Ping command to check the network connection between the BBU and the headend device.

[0044] Step S302, verify whether the IP address of the head-end device is correctly allocated;

[0045] Step S303, checking the power status of the head-end device;

[0046] Step S304: Check the R spectrum status, which is an important indicator for judging the signal quality between the head-end device and the BBU;

[0047] Step S305: Start the Jupyter service on the BBU to configure the parameters of the headend device;

[0048] Step S306: After the configuration is completed, the connection and communication status between the head-end device and the BBU is verified.

[0049] As a preferred technical solution, step S4 specifically includes:

[0050] Step S401, configuring physical layer sensing parameters through BBU configuration software;

[0051] Step S402, configuring the parameters of the trajectory processing algorithm through Matlab software;

[0052] Step S403: After the configuration is completed, enter the trajectory display interface to perform real-time display, stop display and playback display of the detected target.

[0053] As a preferred technical solution, step S5 specifically includes:

[0054] Use weighted Kalman filtering to fuse the trajectory data of multiple nodes. For each node i, its state estimation The process of updating k at each moment is as follows:

[0055]

[0056] in is the predicted state at time k, K i (k) is the Kalman gain, defined as:

[0057] K i (k) = P i (k - )H i (k) T (H i (k)P i (k - )H i (k) T +R i (k)) -1

[0058] Covariance matrix P i (k) The update rule is:

[0059] P i (k)=(IK i (k)H i (k))P i (k - )

[0060] Among them H i (k) represents the observation matrix of base station i at time k, P i (k - ) is the covariance matrix of the prior state estimation error of base station i at time k; the fused state estimation Expressed as:

[0061]

[0062] Weight w i The calculation formula is:

[0063]

[0064] in is the inverse matrix of the covariance matrix of the prior state estimation error of node i;

[0065] For each node i, the covariance matrix P i After fusion, the synthesized covariance matrix P fusion Calculated by weighted average method, the formula is as follows:

[0066]

[0067] Repeat the above steps until the target state estimation converges or the maximum number of iterations is reached.

[0068] As a preferred technical solution, in the actual system, the trajectory data of the node will change in real time, and the fusion process of the Kalman filter must also be executed in real time; whenever new data arrives, a weighted Kalman filter update will be executed.

[0069] Compared with the prior art, the present invention has the following advantages:

[0070] 1) The present invention uses a multi-station joint perception method to combine information from different stations. By sharing perception data and algorithm parameters, it can effectively reduce the blind spots of a single station and improve the detection accuracy and trajectory prediction capabilities of the target.

[0071] 2) This invention utilizes an efficient connection and configuration method between the BBU and the headend device, simplifying the hardware deployment and maintenance costs of the system. By optimizing the connection and configuration process between devices, it reduces the complex hardware and software resource investment.

[0072] 3) The multi-station joint perception method provided by the present invention takes redundancy and fault tolerance mechanisms into consideration in system design; even if some sites fail, the other sites can still ensure the stable operation of the system, thereby improving the system's reliability and anti-interference ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 Schematic diagram of the cellular-free synaesthesia fusion architecture of the present invention;

[0074] Figure 2 A schematic diagram of setting physical layer sensing algorithm parameters in the BBU configuration software of the present invention;

[0075] Figure 3 This is a schematic diagram of the placement and connection status of the head-end device of the present invention;

[0076] Figure 4 Schematic diagram of the R spectrum test results of AAU of the present invention;

[0077] Figure 5Schematic diagram of the bandwidth test results of the AAU of the present invention;

[0078] Figure 6 A schematic diagram of the multi-node perception trajectory transmission results of the present invention;

[0079] Figure 7 This is a schematic diagram of the interface display result when there is no target in the airspace of the present invention;

[0080] Figure 8 Schematic diagram of the multi-node trajectory fusion result when there is a target in the airspace of the present invention;

[0081] Figure 9 It is a specific flow chart of the method of the present invention. DETAILED DESCRIPTION

[0082] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0083] This invention proposes a multi-station collaborative perception method based on a cellular-free interawareness fusion architecture, aiming to address the limitations of existing cellular communication systems in complex environments. Through the collaboration of multiple independent sites, the system can cover a wider area, improving target detection accuracy and perception range. The invention focuses on addressing the fusion of multi-site perception data, using an efficient algorithm to integrate data from different sites, thereby improving the accuracy and stability of the system. Furthermore, the system optimizes system parameters and avoids false alarms. This method is suitable for use in fields such as drones and intelligent transportation, and has broad application prospects.

[0084] like Figure 9 As shown, the present invention specifically includes the following steps:

[0085] Step 1: Build a cellular-free synaesthesia fusion system framework

[0086] The system comprises headend equipment, including the baseband unit (BBU), spatial extension units (SDUs), user (UE)-centric processing units (UCDUs), and a central control unit (CCU). The headend equipment utilizes active antenna units (AAUs), which transmit and receive RF signals and typically integrate an antenna array and RF processing unit. The BBU primarily handles baseband processing for communication and sensing signals. It receives RF signals from the AAUs and converts them into digital signals. The BBU also processes sensing signals, such as target detection and positioning. Each UCDU connects directly to multiple SDUs or via a switch, and each centralized control unit connects directly to multiple UCDUs or via a switch. Data flows between the BBU and the UE-centric processing unit utilize the enhanced Common Radio Interface (eCPRI). When base stations utilize the eCPRI fronthaul protocol, two methods are commonly used for synchronization. One method involves the BBU extracting clock and time from GPS, using this as the master clock / time to provide synchronization for the AAUs. The other method utilizes 1588 and SyncE switches to distribute the time / clock provided by a clock server to all BBUs and AAUs in the system, synchronizing the clocks and times of all units in the system. The main functions of the centralized control unit include determining the data flow correspondence between the BBU and the UE-centered processing unit, such as Figure 1 To support perception, perception signals must be embedded. The frame structure design embeds perception signals into frequency-domain signals and leverages the AAU's transceiver functionality designed in Option 7-2 to ensure transparent transmission of perception signals, supporting perception without interfering with the AAU's communication tasks.

[0087] Step 2: Construct a weighted Kalman filter fusion algorithm

[0088] The Weighted Kalman Filter (WKF) fusion algorithm aims to improve the accuracy of target positioning and tracking, especially in scenarios where multiple sensing nodes are jointly sensing. By weighting the sensing information of different nodes, the data of each node is given appropriate weight during the fusion process to ensure high-precision target estimation. Assume that the state of the target is:

[0089]

[0090] where x k ,y k represents the two-dimensional position of the target at time k, Represents the speed of the target. The state transition equation is:

[0091] x k =Fx k-1 +w k-1

[0092] Where F is the state transfer matrix, w k-1 is zero-mean Gaussian noise with covariance Q k The observation equation is:

[0093] z k =Hx k +v k

[0094] where z k is the perception data (i.e., observation value), H is the observation matrix, v k is the observation noise, which is also assumed to be Gaussian noise with zero mean and covariance R k Assume there are N sensing nodes, and the observation value of each node is Where i represents the i-th node. In the process of multi-node joint estimation, it is necessary to perform weighted fusion on the observations according to the weight of each node. Assume that Z k Represents the set of observations of all nodes:

[0095]

[0096] The goal of weighted Kalman filtering is to use the observation data of all nodes to update the state estimate of the target and the covariance matrix P k First, calculate the weighted mean of the weighted observations

[0097]

[0098] Then, calculate the weighted covariance matrix S k

[0099]

[0100] After obtaining the weighted average observation value and weighted covariance, the update steps are similar to those of conventional Kalman filtering. First, calculate the gain K k

[0101]

[0102] Then, update the target state estimate:

[0103]

[0104] Finally, update the covariance matrix:

[0105] P k =(IK k H)P k-1

[0106] The weighted Kalman filter improves target positioning and tracking accuracy by weighting the observation data from multiple sensing nodes. During the fusion process, each node is assigned a different weight based on its reliability and sensing capabilities. This approach effectively improves system robustness and accuracy in scenarios involving multi-node joint sensing.

[0107] The specific steps of the algorithm are as follows:

[0108] Input: Set the initial state estimate And the initial error covariance matrix P0. For each sensor node i, initialize the node observation noise variance Calculate weights:

[0109] Output: estimated values of target state variables such as position information and velocity and the state error covariance matrix P k

[0110] 1) Predict the current state estimation based on the system dynamic equation Calculate the predicted state error covariance matrix

[0111] 2) Get the observation of each sensor node i Calculate weighted average observation Calculate the weighted covariance matrix

[0112] 3) Calculate Kalman gain Update target state estimate Update the error covariance matrix

[0113] 4) Repeat steps 1-3) until the target state estimation converges or the maximum number of iterations is reached.

[0114] Step 3: Device connection, configuration and headend parameter initialization

[0115] 3.1 Device Connection

[0116] When connecting the headend and BBU, first connect the H3C device. To ensure proper data transmission and control information exchange, the H3C device should be plugged into the fourth port from the left using a data cable. This port should be capable of transmitting both data and control messages. Next, use optical fiber to connect the headend and BBU system. Fiber optic cables should be plugged into the first and second fiber optic ports of the device. Fiber optic cables are automatically assigned corresponding IP addresses depending on the port.

[0117] 3.2 Check the connection between the headend and BBU

[0118] After completing the physical connection, check the indicator lights on the headend device. Under normal circumstances, the green light on the left side of the headend device should flash, and the green light in the middle should be on, indicating that the device is connected properly. In addition, even if the indicator lights are normal, it is still possible that the ping function cannot be connected. To do this, the user needs to execute the following ping command in the command line to check the network connection between the BBU and the headend:

[0119] Command: ping <headend IP>

[0120] If the ping operation fails, it indicates that there is a problem with the connection between the devices. In this case, the user should perform a command line check to troubleshoot the network connection problem.

[0121] 3.3IP address query and verification

[0122] Since the headend device automatically assigns a new IP address each time it is powered on, and the last digit of the IP address assigned each time is automatically incremented, to verify whether the headend IP address is correctly assigned, you can execute the following command on the BBU device:

[0123] Command: cat / var / lib / dhcp / dhcpd.leases|greplease

[0124] This command will list all assigned IP addresses. Users can confirm the IP address of the head-end device based on the returned results. After querying the IP address of the head-end device, users should execute the Ping command to confirm whether the connection between the BBU and the head-end device is normal:

[0125] Command: ping <headend IP>

[0126] If the ping command returns a normal response, the connection between the headend and the BBU device has been successfully established. If the ping command fails, further troubleshooting is required based on the device connection status and configuration.

[0127] 3.4 Fiber Port and IP Address Allocation Rules

[0128] The IP address assignment for the headend device is closely related to the port into which the fiber is plugged. The specific assignment rules are as follows: When the fiber is plugged into the first port, the last digit of the assigned IP address is 0. When the fiber is plugged into the second port, the last digit of the assigned IP address is 1, and so on. For example, the IP address 10.3.1.xx represents the transmitting headend, and 10.3.0.xx represents the receiving headend. Each time the headend device is powered on or off, the BBU system reassigns an IP address to the headend device, and the last digit increments. Therefore, when configuring the IP address of the board instantiated by Jupyter, users should ensure that the IP address is adjusted according to the latest IP address assigned by the BBU to avoid configuration errors caused by IP address changes.

[0129] 3.5 Power Status Check

[0130] After connecting the devices, check the power status. Use the power status indicator to confirm that the operating frequency band is within the specified range. For example, if the power is within the 100 MHz band (-2 to -1), the system is operating normally. If the power is outside the specified range, it may be due to a configuration error or hardware failure. Follow the troubleshooting guide to ensure the power status is restored to normal.

[0131] 3.6R spectrum status check

[0132] The R spectrum status is a key indicator of signal quality between the headend and BBU. Normally, the R spectrum should display a single correlation peak with high overlap. A clear peak with good overlap indicates normal device status. Multiple correlation peaks in the R spectrum may indicate poor signal quality or a configuration issue. Users should follow the solution to identify and remedy the cause of the abnormal R spectrum.

[0133] 3.7 Jump to the headend command check

[0134] In some cases, users may need to jump to the headend command line for more detailed diagnosis and inspection. By directly entering the headend command line, users can execute more system diagnostic commands to ensure that the connection and data transmission between devices are in normal working condition.

[0135] 3.8 Connecting the BBU

[0136] First, open the Mobaxterm remote terminal tool and connect to the BBU. In Mobaxterm, click the Session button in the upper left corner and enter the BBU's IP address in the pop-up window to connect to the BBU. After logging in, the system prompts you to enter the BBU password. Enter the correct password to successfully log in to the BBU.

[0137] 3.9 Start Jupyter Server

[0138] Start the Jupyter service: To configure the headend, you need to start the Jupyter service. Enter the following command in the command line interface of the BBU device to start the Jupyter service:

[0139] Command: sudo systemctl start jupyter.service

[0140] After starting the Jupyter service, you can access the BBU's web interface through a browser to configure headend information. Before executing the above command, ensure that the Jupyter service is installed on the BBU. If Jupyter is not installed, the system displays an error message. You need to install the Jupyter service. For installation steps, see the installation instructions.

[0141] 3.10 Configuring Headend Parameters

[0142] Access the BBU device's web interface. Open a browser and enter the BBU device's IP address in the browser address bar in the following format:

[0143] http: / / 192.168.30.59:8888 /

[0144] The user will enter the Jupyter web interface to configure headend parameters. After entering the BBU device's Jupyter web interface, the system will prompt you to enter a password. After entering the correct password, the user will enter the Jupyter configuration interface and prepare to configure the headend.

[0145] 3.11 Upload and configure files

[0146] After powering on the BBU for the first time, you need to upload the configuration file cap_poc0.ipynb. This file contains the initial parameters and settings required for headend configuration. After uploading the file, you can proceed with headend configuration. After the upload is successful, open the file cap_poc0.ipynb, which contains all the functions and parameter settings required to configure the headend. Upon opening the file, you will see the code blocks for the configuration steps.

[0147] 3.12 Follow the configuration steps step by step

[0148] The first step in the configuration process is to import the required functions. You need to execute the first code block in the file to import the function library and prepare for subsequent operations. You can execute the code by selecting the code block and pressing Ctrl+Enter. Next, you need to instantiate the board. This requires entering the IP address of the transceiver headend. The IP address of the transceiver headend can be queried via the command line. After entering the correct IP address, execute the corresponding code block to instantiate the board.

[0149] The third step in the configuration file configures the POC 1.5 headend case. Users need to adjust the appropriate parameter settings according to their needs and execute the configuration code blocks. The configuration process requires executing each code block in the file in sequence. Users can execute each code block step by step by selecting it and pressing Ctrl+Enter. After each step is executed, the system automatically performs the corresponding configuration.

[0150] 3.13 Complete the configuration

[0151] After completing the above steps, the headend device configuration is complete. Based on the configuration results, users can verify the connection and communication status between the headend and BBU to ensure that the device has been set up as expected.

[0152] Step 4: Configure BBU physical layer perception parameters and trajectory processing parameters

[0153] 4.1 Configuring physical layer perception parameters

[0154] First, the user needs to open the BBU configuration software sensing_ctrl to configure the parameters of the physical layer sensing algorithm. This software will be used to set the various configurations of the sensing algorithm to ensure that the BBU can correctly perform the physical layer sensing task. The user sets the parameters of the physical layer sensing algorithm in the BBU configuration software according to actual needs. Figure 2 The settings in are the same.

[0155] During configuration, ensure the following parameters are set correctly: Configure the IP address of the server (BBU) according to the actual network environment. Configure the IP address of the host computer (PC) according to the actual network environment. Configure the beam case index to 70.

[0156] After setting all parameters, click the Generate Parameters button to generate a new configuration file. Then, click the Restart button to import the configuration into the BBU. The BBU will then load the latest physical layer sensing algorithm configuration and apply it to actual operations. After the configuration is complete, click the Download and Display Current Configuration button to view the BBU's current algorithm parameter configuration. This allows you to verify that the configuration has been imported correctly and has taken effect.

[0157] 4.2 Configuration Notes

[0158] In this step, only the physical layer sensing parameters for a single sector (cell 0) are configured. The configuration of other sectors will be further adjusted based on actual needs. Please ensure that the IP addresses of the configured BBU and PC are adjusted according to actual conditions to ensure a stable and reliable network connection.

[0159] 4.3 Open Matlab and load the trajectory processing program

[0160] First, the user needs to open the Matlab software, enter the Matlab command line interface, and run the following program files:

[0161] AsmoteSensingPanel.P

[0162] This file contains the relevant functions of the trajectory processing algorithm. After the user runs this file, he will enter the configuration interface for further settings.

[0163] 4.4 Configuring trajectory processing algorithm parameters

[0164] After the program runs, the user will enter the Configurations interface of Matlab. This interface is used to configure the relevant parameters of the trajectory processing algorithm. In the trajectory configuration interface, select the Load Config button to load the previously debugged configuration file. By loading the existing configuration, the user can ensure the consistency and accuracy of the trajectory processing algorithm parameters. The user needs to enter the SaveRVA folder, which contains the previously tested and saved point cloud data. In this folder, select the required file (for example, "Purple Mountain-Chaotian (calibrated)") and open the file. After opening the data file, the system will prompt whether to replay the trajectory. At this time, since only the parameters are configured, the user selects No, indicating that the trajectory will not be replayed.

[0165] 4.5 Obtain and verify the configuration

[0166] After the data is imported, the user needs to click the getConfig button to obtain the configuration file for the current case. This operation will display the configuration details, and the user can confirm whether the configuration is correct. The user needs to verify whether the following configuration items are correct: ensure that the IP address of the PC is configured correctly; confirm whether the relevant configuration parameters of the trajectory processing algorithm are correct. The engineering parameter configuration includes the mobile position of the device and the adjusted parameters. The user needs to re-measure the relevant engineering parameters, such as using the RTK handheld radio to measure the longitude and latitude, and using the attitude meter to measure the head pointing angle. In addition, parameters such as FOV (field of view), cluster configuration, and configuration ID also need to be adjusted according to the actual scenario use case.

[0167] 4.6 Displaying the Detected Target Trajectory

[0168] After configuration is complete, the user can enter the trajectory display interface. Click the Run button to begin displaying the target's motion trajectory. The system will now display the target's real-time motion. After clicking Run, the user will be prompted to save the trajectory data. If the user selects Yes, the system will save the point cloud data to the SaveRVA folder. If the user does not want to save the data, select No to simply observe the perception performance.

[0169] 4.7 Stop displaying tracks

[0170] When the user needs to stop the trajectory display, he can enter the Matlab command line window and press Ctrl+C to terminate the current trajectory display.

[0171] 4.8 Playback Track

[0172] To replay a previously recorded trajectory, click the Load button and select the data file in the SaveRVA folder (for example, "Purple Mountain - Chaotian (calibrated)"). After loading the file, the user can select the data to replay. Once the data is loaded, click the Replay button to replay the trajectory. The system will replay the target's motion trajectory based on the saved point cloud data, allowing the user to observe and analyze the trajectory data.

[0173] Step 5: Deploy the Kalman filter fusion algorithm

[0174] In multi-node joint target positioning and tracking, each node provides an independent trajectory. To improve the accuracy and precision of target position estimation, the weighted Kalman filter (WKF) algorithm can be used to fuse the trajectory data of different nodes. The specific process is as follows:

[0175] First, you need to build an MQTT Broker. Each node (sensor device) will send its own trajectory data as a message to the MQTT Broker. This process can be implemented through the MQTT client library, which then receives the MQTT message and fuses it. Subscribers (for example, a fusion server or a master node) are required to receive the trajectory data published by different nodes. Subscribers subscribe to messages on a specific topic through the MQTT client. The schematic diagram of MQTT sending and receiving is as follows: Figure 6 shown.

[0176] After receiving the trajectory data from different nodes, the subscriber can pass these data into the Kalman filter fusion algorithm for processing. The weighted Kalman filter (WKF) is used to fuse the trajectory data of multiple nodes. For each node i, its state estimation The process of updating k at each moment is as follows:

[0177]

[0178] in is the predicted state at time k, K i (k) is the Kalman gain, defined as:

[0179] K i (k) = P i (k - )H i (k) T (H i (k)P i (k - )H i (k) T +R i (k)) -1

[0180] Covariance matrix P i (k) The update rule is:

[0181] P i (k)=(IK i (k)H i (k))P i (k - )

[0182] When the trajectory data of multiple nodes need to be fused, it is assumed that the Kalman filter state estimation of each node are independent, and fusion can be performed based on these estimates. The weighted Kalman filter (WKF) method uses the estimated value of each node and its covariance to weightedly fuse their results. The core idea of the weighted Kalman filter is to calculate the weight through the covariance matrix of the node. The smaller the covariance of the node, the greater its weight, and vice versa. The fused state estimate can be expressed as:

[0183]

[0184] Weight w i The calculation formula is:

[0185]

[0186] For each node i, the covariance matrix P i After fusion, the synthesized covariance matrix P fusion It can be calculated by weighted average method, the formula is as follows:

[0187]

[0188] Repeat the above steps until the target state estimation converges or the maximum number of iterations is reached.

[0189] In a real system, the trajectory data of the node will change in real time, so the fusion process of the Kalman filter also needs to be executed in real time. Whenever new data arrives, a weighted Kalman filter update is performed:

[0190] 1. Get the current state estimate and covariance matrix for each node.

[0191] 2. Calculate weights based on the covariance of the nodes.

[0192] 3. Use the weights to perform weighted summation on the estimates of each node to obtain the fused state estimate.

[0193] 4. Update the fused covariance matrix.

[0194] In order to verify the performance advantages of the method of the present invention, an example process of the present invention is given below.

[0195] (1) Experimental setup and parameters, as shown in Table 1

[0196] Table 1

[0197]

[0198] The experimental setup is shown in Table 1. This experiment was conducted on the rooftop of A5 in Wireless Valley. This location was chosen to ensure that the experiment has a wide coverage range, reduce external interference, and ensure the accuracy of the measurement. After the experimental site layout is completed, ensure that the equipment and measurement tools are ready. The head-end device is installed at 90 degrees vertically upward to ensure effective target detection and data collection. Install the head-end device at the specified location and fix it at a predetermined angle (90 degrees vertically upward). Ensure that the various parameters of the head-end device have been set through the previous configuration steps. Select the RTK350 drone as the mobile target in the experiment. The drone needs to be preset according to the flight path required by the experiment and ensure that it has high-precision positioning capabilities. All necessary hardware and software configurations should be completed in advance.

[0199] Turn on the BBU and enable the perception algorithm using the configuration software. Ensure proper connection between the BBU and the headend. Use Matlab to configure trajectory processing parameters and load the debugged configuration file to ensure correct data acquisition and processing. Configure the trajectory in Matlab and load the point cloud data saved from the previous test. During configuration, select an appropriate file (such as "Purple Mountain - Chaotian (calibrated)") and configure trajectory playback to ensure correct display of trajectory data. After configuration, start the perception system and use Jupyter to monitor and analyze data in real time to ensure smooth target detection and trajectory playback. After starting the experiment, begin target data acquisition. The RTK350 drone will begin flying according to the preset flight path, while the headend will continuously collect wireless signals and target data according to the configured parameters. During data acquisition, monitor the system's operation using the Matlab command line to ensure real-time feedback of trajectory data and display of the target's motion trajectory. When the experimental objectives are completed, enter the Matlab command line window and press Ctrl+C to terminate data acquisition and stop trajectory display. Save the data file to the SaveRVA folder as needed.

[0200] (2) Equipment debugging and experimental process of multi-station collaborative perception system

[0201] In this experiment, two pairs of headend devices were selected for debugging and testing. Figure 3 First, the BBU and AAU successfully completed a ping test, ensuring a stable network connection between the devices. The test results confirmed no packet loss or excessive latency in data transmission between the BBU and AAU, demonstrating normal system operation. Subsequently, the two pairs of AAUs were tested and successfully acquired the target signal. The R spectrum exhibited a single correlation peak, indicating that the target signal was correctly identified and the system was operating stably. Figure 4 and Figure 5 The R spectra and bandwidths of two pairs of AAUs are shown, verifying the accurate identification and processing of the signals.

[0202] During BBU debugging, after configuration and startup, if Figure 7 As shown in the figure, when there is no target in the airspace, the BBU can correctly display the trajectory interface without false alarms, which verifies the normal operation of the system when there is no target in the airspace. Figure 8 As shown in the figure, the BBU successfully completed the multi-node trajectory fusion display task, capable of collecting and processing data at altitudes exceeding 400 meters. The display effect met expectations, and the system operated stably without any anomalies. This demonstrates that the BBU can stably detect and track targets and effectively avoid false alarms.

[0203] (3) The above description is only a specific implementation example of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0204] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0205] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be determined by the appended claims.

Claims

1. A multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture, characterized in that: The method comprises the following steps: Step S1, building a distributed non-cellular synaesthesia fusion architecture that integrates communication and perception functions, including a baseband processing unit (BBU) and a head-end device; Step S2: construct a weighted Kalman filter fusion algorithm to fuse the trajectory data of multiple head-end devices to obtain target position estimation and reduce uncertainty through the fused covariance matrix; Step S3: Configure the communication parameters between the headend device and the BBU to ensure normal communication between the two. Step S4: configure the BBU's sensing parameters and set the trajectory processing algorithm and parameters; In step S5, the trajectory data obtained from different head-end devices are fused using the weighted Kalman filter fusion algorithm constructed in step S2 to optimize the target position estimation, dynamically adjust the fusion weights and covariance matrix, and then output the fused trajectory.

2. The multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 1, characterized in that: The distributed cellular-free synaesthesia fusion architecture of step S1 further includes a user-centered processing unit UCDU and a central control unit. The central control unit is connected to multiple UCDUs, one UCDU is connected to multiple BBUs, and one BBU is connected to multiple head-end devices.

3. A multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 1 or 2, characterized in that: The head-end device uses an active antenna unit AAU, which is responsible for transmitting and receiving radio frequency signals. The BBU is responsible for baseband processing of communication and perception signals, receiving radio frequency signals from the AAU, and converting them into digital signals.

4. The multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 1, characterized in that: The step S2 specifically includes: Assume the target status is: where x k ,y k represents the two-dimensional position of the target at time k, Indicates the speed of the target; The state transfer equation is: x k =Fx k-1 +w k-1 Where F is the state transfer matrix, w k-1 is the process noise, assuming it is Gaussian noise with zero mean and covariance Q k , the observation equation is: z k =Hx k +v k where z k is the perception data, H is the observation matrix, w k is the observation noise, which is also assumed to be Gaussian noise with zero mean and covariance R k ; Assume there are N sensing nodes, and the observation value of each node is Where i represents the i-th node. In the process of multi-node joint estimation, the observation values need to be weighted and fused according to the weight of each node. Assume that Z k Represents the set of observations of all nodes: The goal of weighted Kalman filtering is to use the observation data of all nodes to update the state estimate of the target and the covariance matrix P k .

5. The multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 4, characterized in that: The updated target state estimate and the covariance matrix P k Specifically include: First, calculate the weighted mean of the weighted observations where w i is the weighting coefficient corresponding to the i-th observation source, N is the number of base stations participating in the fusion, is the observation vector of the i-th base station at time k; Then, calculate the weighted covariance matrix S k in It is an observation The weighted average of After obtaining the weighted average observation value and weighted covariance, the update steps specifically include: 1) Calculate the gain K k 2) Update target state estimate: 3) Update the covariance matrix: P k =(I-K k H)P k-1 Where I represents the identity matrix.

6. A multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 1 or 4, characterized in that: The weighted Kalman filter fusion algorithm improves the accuracy of target positioning and tracking by weighting the observation data of multiple sensing nodes, and during the fusion process, it gives each node a different weight considering the reliability and sensing ability of different nodes.

7. The multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 1, characterized in that: The step S3 specifically includes: Step S301: After physically connecting the headend device to the BBU, check the indicator light on the headend device and use the Ping command to check the network connection between the BBU and the headend device. Step S302, verify whether the IP address of the head-end device is correctly allocated; Step S303, checking the power status of the head-end device; Step S304: Check the R spectrum status, which is an important indicator for judging the signal quality between the head-end device and the BBU. Step S305: Start the Jupyter service on the BBU to configure the parameters of the headend device; Step S306: After the configuration is completed, the connection and communication status between the head-end device and the BBU is verified.

8. The multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 1, characterized in that: The step S4 specifically includes: Step S401, configuring physical layer sensing parameters through BBU configuration software; Step S402, configuring the parameters of the trajectory processing algorithm through Matlab software; Step S403: After the configuration is completed, enter the trajectory display interface to perform real-time display, stop display and playback display of the detected target.

9. The multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 5, characterized in that: The step S5 specifically includes: Use weighted Kalman filtering to fuse the trajectory data of multiple nodes. For each node i, its state estimation The process of updating k at each moment is as follows: in is the predicted state at time k, K i (k) is the Kalman gain, defined as: K i (k)=P i (k - )H i (k) T (H i (k)P i (k - )H i (k) T +R i (k)) -1 Covariance matrix P i (k) The update rule is: P i (k)=(I-K i (k)H i (k))P i (k - ) Among them H i (k) represents the observation matrix of base station i at time k, P i (k - ) is the covariance matrix of the prior state estimation error of base station i at time k; Fusion state estimation Expressed as: Weight w i The calculation formula is: in is the inverse matrix of the covariance matrix of the prior state estimation error of node i; For each node i, the covariance matrix P i After fusion, the synthesized covariance matrix P fusion Calculated by weighted average method, the formula is as follows: Repeat the above steps until the target state estimation converges or the maximum number of iterations is reached.

10. A multi-station collaborative sensing method based on a cellular-free synaesthesia fusion architecture according to claim 1 or 9, characterized in that: In actual systems, the trajectory data of nodes will change in real time, and the fusion process of the Kalman filter must also be executed in real time; whenever new data arrives, a weighted Kalman filter update will be performed.

Images