A method and system for dynamic sensor scheduling for multi-region joint sensing

Abstract

This invention provides a sensor dynamic scheduling method and system for multi-region joint sensing, belonging to the field of vehicle-to-everything (V2X) communication scheduling technology. The method involves sorting the scheduling revenue values ​​of all idle regions in reverse order. If the number of idle regions is greater than a threshold M for the number of idle regions that can be scheduled each time, the M regions with the highest revenue are selected for scheduling; if the number of idle regions is less than or equal to M, all idle regions are scheduled. Feature extraction is performed on the sensors in the selected regions, and communication traffic and bandwidth allocation are performed on each sensor based on a confidence threshold binary search. Under the joint sensing framework of this invention, an information timeliness modeling mechanism is introduced. By evaluating the timeliness value of sensing data, the sensing timing and sensing communication traffic are dynamically adjusted to achieve refined management of communication resources. This effectively improves the overall performance of joint sensing in complex and dynamic communication environments, enhancing system robustness and sensing accuracy.

Description

Technical Field

[0001] This invention relates to the field of vehicle-to-everything (V2X) communication and scheduling technology, specifically to a sensor dynamic scheduling method and system for multi-regional joint sensing. Background Technology

[0002] With the development of autonomous driving, intelligent manufacturing, and smart cities, cooperative perception (CP) is gradually becoming a key means to improve the range and accuracy of environmental perception. In vehicle-to-everything (V2X) autonomous driving, different vehicles or roadside units can share perception information through vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), or road-to-infrastructure (I2I) communication, compensating for blind spots and limited perception distance in single-view perception, and further supporting collaborative decision-making and safe driving. In smart factories, various heterogeneous sensors (such as cameras, LiDAR, and ultrasonic sensors) collaborate to detect production line status, robotic arm movements, etc., to achieve refined monitoring and intelligent scheduling. In smart cities, roadside units (RSUs) distributed throughout the urban road network jointly perceive information such as traffic flow, pedestrian density, and environmental changes, providing support for environmental monitoring, traffic optimization, and emergency response.

[0003] However, in joint sensing scenarios, the information collected by different sensors contributes differently (i.e., has varying utility) to the accurate detection of relevant targets in the sensed area. Furthermore, due to the time-varying nature of the environment, multiple sensors sensing the same area exhibit differentiated time-varying utilities (i.e., timeliness) and spatial correlations. For example, at any given moment, some sensors may be able to observe the target object more clearly due to their advantageous position or angle, while other sensors may have poorer sensing performance due to obstruction or greater distance. Simultaneously, accurate and timely sensing is particularly important for certain critical objects (such as pedestrians crossing the road), and delayed or redundant information may degrade the overall sensing quality of the system. Moreover, when multiple sensors sense the same area, their sensed content has a certain degree of overlap; reasonable scheduling can avoid redundant transmission.

[0004] In practical applications, such as highway convoy collaboration, intelligent scheduling at urban intersections, multi-robot collaboration on intelligent manufacturing production lines, and joint monitoring of roadside units in smart cities, systems need to quickly and efficiently share key sensing information with limited communication resources, avoiding ineffective communication and ensuring the timeliness of sensing information. However, existing joint sensing technologies still have the following limitations: First, existing methods typically employ periodic broadcasting or simple triggering mechanisms for sensing scheduling, failing to make dynamic decisions based on the real-time utility of the information sensed by each sensor, leading to wasted communication resources or information lag. Second, existing solutions generally ignore the impact of changes in communication conditions on joint sensing performance, lacking methods for adaptively adjusting sensing data scheduling and transmission strategies under dynamic communication conditions. Third, there is a lack of a joint design covering the entire process of acquisition, scheduling, and fusion. Existing technologies often process multi-sensor data acquisition, information fusion, and sensing result broadcasting separately, lacking an integrated closed-loop process design from source acquisition, dynamic scheduling, fusion update, and broadcast publication.

[0005] In summary, current joint sensing technologies primarily focus on the fusion of sensing information itself or the performance of sensing algorithms, without fully considering the impact of the timeliness of sensing data on the final joint sensing performance. Sensing data exhibits natural latency and staleness; delayed information is still broadcast and fused, potentially leading to decision-making errors. Currently, there is a lack of mechanisms for dynamically scheduling the timing of sensing information dissemination or acquisition. Existing systems largely ignore the impact of changing communication conditions on sensing system performance, assuming ideal or fixed communication environments. In practical applications, communication bandwidth, latency, packet loss rate, and channel occupancy rates constantly change. These factors directly affect the transmission quality and timeliness of sensing information. However, there is currently a lack of methods for adaptively adjusting sensing data scheduling and transmission strategies under dynamic communication conditions, and no systematic design exists for adjusting sensing task scheduling according to changes in communication conditions. Existing technologies often process multi-sensor data acquisition, information fusion, and sensing result broadcasting separately, lacking an integrated closed-loop process design from source acquisition, dynamic scheduling, fusion update, to broadcast release. Although current sensing systems integrate data from multiple sources, they lack system design for "which key data to prioritize when bandwidth is tight" and "when to collect and when to broadcast," resulting in low resource utilization efficiency and insufficient timeliness. Summary of the Invention

[0006] The purpose of this invention is to provide a sensor dynamic scheduling method and system for multi-region joint sensing, so as to solve at least one of the technical problems existing in the background art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides a sensor dynamic scheduling method for multi-region joint sensing, characterized in that it includes:

[0009] Based on the channel conditions, bandwidth, traffic constraints, and information age of the idle area under time slot k, the scheduling benefit value of all idle areas and their corresponding optimal area traffic are obtained; there are multiple sensors in the idle area.

[0010] Sort all idle area scheduling revenue values ​​in reverse order. If the number of idle areas is greater than the threshold M for the number of idle areas that can be scheduled each time, select the M areas with the highest revenue for scheduling; if the number of idle areas is less than or equal to M, schedule all idle areas.

[0011] Based on the scheduling benefit value and communication constraints of the scheduling area, the optimal communication volume for this scheduling is determined; feature extraction is performed on the selected area sensors, and communication volume and bandwidth allocation are performed on each sensor based on confidence threshold binary search.

[0012] As a further limitation of the first aspect of the present invention, each sensor in the selected area receives a scheduling start signaling and initiates scheduling for the area; each sensor extracts sensing features and generates a spatial confidence map; an initial value of the spatial confidence threshold is generated according to the communication volume constraint; each sensor generates a communication mask according to the threshold and calculates the communication volume value corresponding to the mask; it is determined whether the communication volume meets the communication volume constraint. If the total feedback communication volume exceeds the preset communication constraint, the confidence threshold is increased; if the communication volume is insufficient, the threshold is decreased.

[0013] As a further limitation of the first aspect of the present invention, the threshold is continuously adjusted by an iterative search algorithm until the difference between the sum of the communication quantities of each sensor and the communication constraint value is less than the tolerance error, thus obtaining the final threshold.

[0014] As a further limitation of the first aspect of the present invention, the communication volume required for each sensor's transmission sub-block is calculated, and bandwidth is allocated to each sensor based on the channel conditions and communication volume of each sensor to optimize the timeliness of the sensing task.

[0015] As a further limitation of the first aspect of the present invention, each sensor performs sensory feature extraction and generates a spatial confidence map, including: after receiving a scheduling signal, each sensor collects raw sensory data; the raw sensory data is processed by a feature extraction network to obtain feature representation; the feature representation is processed by a neural network to generate a task-oriented spatial confidence map.

[0016] As a further limitation of the first aspect of the present invention, each sensor generates a communication mask based on a threshold and calculates the corresponding communication value of the mask, including: generating a communication mask based on a spatial confidence threshold, setting the sub-block mask greater than or equal to the threshold to one, and setting the sub-block mask less than the threshold to zero. Each sensor calculates the communication value corresponding to the non-zero elements of the communication mask.

[0017] Secondly, the present invention provides a sensor dynamic scheduling system for multi-region joint sensing, comprising:

[0018] The calculation module is used to calculate the scheduling benefit value of all idle areas and the corresponding optimal area communication volume based on the channel conditions, bandwidth, communication volume constraints and information age of the idle areas under time slot k; there are multiple sensors in the idle areas;

[0019] The selected module is used to sort all idle area scheduling revenue values ​​in reverse order. If the number of idle areas is greater than the threshold M of the number of idle areas that can be scheduled each time, the M areas with the highest revenue are selected for scheduling; if the number of idle areas is less than or equal to M, all idle areas are scheduled.

[0020] The allocation module is used to solve for the optimal communication volume in the current scheduling based on the current scheduling benefit value, communication volume constraints, and the communication volume of the previous scheduling in the scheduling area; it performs feature extraction on the selected area sensors and performs communication volume allocation and bandwidth allocation for each sensor based on confidence threshold binary search.

[0021] Thirdly, the present invention provides a non-transitory computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the sensor dynamic scheduling method for multi-region joint sensing as described in the first aspect.

[0022] Fourthly, the present invention provides a computer device including a memory and a processor, wherein the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the sensor dynamic scheduling method for multi-region joint sensing as described in the first aspect.

[0023] Fifthly, the present invention provides an electronic device, comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the sensor dynamic scheduling method for multi-region joint sensing as described in the first aspect.

[0024] The beneficial effects of this invention are as follows: Under the joint sensing framework, an information timeliness modeling mechanism is introduced. By evaluating the timeliness value of sensing data, the timing of sensing and the amount of sensing communication are dynamically adjusted, thereby achieving refined management of communication resources. In complex and dynamic communication environments, it can effectively improve the overall performance of joint sensing and enhance the robustness and accuracy of the system.

[0025] The advantages of additional aspects of the invention will be set forth more clearly in the following description or will be learned by practice of the invention. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a schematic diagram of a multi-region joint sensing scenario as described in an embodiment of the present invention.

[0028] Figure 2 This is a flowchart of the sensor dynamic scheduling method for multi-region joint sensing as described in an embodiment of the present invention.

[0029] Figure 3 This is a flowchart illustrating the communication traffic allocation and bandwidth allocation process based on confidence threshold binary search as described in an embodiment of the present invention. Detailed Implementation

[0030] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0031] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0032] It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as here.

[0033] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, and / or groups thereof.

[0034] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0035] To facilitate understanding of the present invention, the present invention will be further explained and described below with reference to the accompanying drawings and specific embodiments. However, the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0036] Those skilled in the art should understand that the accompanying drawings are merely schematic diagrams of embodiments, and the components in the drawings are not necessarily essential for implementing the present invention.

[0037] This invention addresses multi-sensor joint sensing scenarios and proposes a time-sensitive communication scheduling method. Within the joint sensing framework, this method introduces an information timeliness modeling mechanism. By evaluating the timeliness value of sensed data, it dynamically adjusts sensing timing and communication volume, achieving refined management of communication resources. This method effectively improves the overall performance of joint sensing in complex and dynamic communication environments, enhancing system robustness and sensing accuracy.

[0038] Example 1

[0039] In this embodiment 1, a sensor dynamic scheduling system for multi-region joint sensing is first provided, including: a calculation module, used to calculate the scheduling benefit value of all idle regions and their corresponding optimal region communication volume based on the channel conditions, bandwidth, communication volume constraints, and information age of the idle regions under time slot k; there are multiple sensors in the idle regions. A selection module is used to sort all idle region scheduling benefit values ​​in reverse order. If the number of idle regions is greater than the threshold M of the number of idle regions that can be scheduled each time, the M regions with the highest benefit are selected for scheduling; if the number of idle regions is less than or equal to M, all idle regions are scheduled. An allocation module is used to solve for the optimal communication volume for the current scheduling based on the scheduling benefit value of the scheduled region, the communication volume constraints, and the communication volume of the previous scheduling; feature extraction is performed on the sensors in the selected regions, and communication volume and bandwidth allocation are performed on each sensor based on a confidence threshold binary search.

[0040] In this embodiment, the above-described system is used to implement a dynamic sensor scheduling method for multi-region joint sensing, including: calculating the scheduling benefit value of all idle regions and their corresponding optimal region communication volume based on the channel conditions, bandwidth, communication volume constraints, and information age of the idle regions under time slot k; there are multiple sensors in the idle regions; the scheduling benefit values ​​of all idle regions are sorted in reverse order; if the number of idle regions is greater than the threshold M of the number of idle regions that can be scheduled each time, the M regions with the largest benefits are selected for scheduling; if the number of idle regions is less than or equal to M, all idle regions are scheduled; the optimal communication volume for the current scheduling is calculated based on the scheduling benefit value of the scheduled region, the communication volume constraints, and the communication volume of the previous scheduling; feature extraction is performed on the selected region sensors, and communication volume and bandwidth allocation are performed on each sensor based on a confidence threshold binary search.

[0041] Joint sensing, as a multi-sensor, multi-view collaborative sensing method, allows different sensors or terminals to share raw or processed sensing information via wireless channels, compensating for the limitations of a single sensor or single terminal's field of view and improving the overall sensing performance of the system. For example, in vehicle-road cooperative sensing systems, vehicles and roadside units need to efficiently share sensing data to support accurate environmental perception and real-time decision-making. Due to the complex and time-varying sensing environment and the dynamic changes in communication conditions such as bandwidth, latency, and channel occupancy, the transmission and scheduling of sensing information face challenges. How to prioritize the transmission of critical sensing data under limited communication resources, ensuring the timeliness, accuracy, and effectiveness of information, has become a key issue in improving system response speed and security. Therefore, how to effectively schedule the transmission of sensed information by sensors to support real-time collaborative perception and decision-making, based on the importance of different sensing areas and the correlation of different sensor data under changing sensing environments and communication conditions, has become a core challenge in improving the efficiency of joint sensing systems.

[0042] In this embodiment, the above-described sensor dynamic scheduling method for multi-region joint sensing is illustrated in the following scenario diagram. Figure 1As shown, the system includes multiple sensing areas, multiple sensor devices, and a central server. Each area is monitored collaboratively by multiple sensors (approximately 2-4), including sensors such as LiDAR and cameras. The central server schedules sensors by time slot. At the beginning of each time slot, the central server selects which sensors in which areas to schedule based on the area status, the age of the area's information, channel conditions, and historical scheduling data, and allocates optimal communication volume and bandwidth to each sensor. The scheduling decision aims to optimize the long-term average information timeliness index for all areas. This information timeliness index is obtained by modeling the functional relationship between actual sensing performance (such as average detection accuracy) and information timeliness (such as information age), which can quantify the impact of the timeliness of sensing data on the performance of sensing tasks such as target detection. This index is used as the optimization target to improve the overall sensing quality and response capability of the system in multi-area sensing tasks.

[0043] Constrained by the central server's computing power, a maximum of M sensors from different regions are scheduled at a time. Sensors in the selected regions generate spatial confidence maps (representing the importance scores of different sub-blocks within the region), and the central server allocates transmission bandwidth to each sensor. Each sensor transmits its perceived features to the central server for feature fusion, used for tasks such as target detection, path planning, and environmental detection. The system continuously performs environmental perception, forming a closed-loop optimization.

[0044] like Figure 2 As shown in the figure, the specific process of the sensor dynamic scheduling method for multi-region joint sensing described in this embodiment is as follows:

[0045] 1. Multi-regional communication resource constraint design based on timeliness modeling

[0046] 1a. Construct an information timeliness model to quantify the changing patterns of perceived data value;

[0047] First, by modeling the functional relationship between the actual performance P of the sensed data (e.g., detection accuracy), information timeliness h(k), and the amount of sensed data b(k), P = f(h(k), b(k)), the impact of data timeliness (e.g., information age) on the sensing effect is characterized, serving as the basis for scheduling optimization. For example, the information timeliness h(k), the amount of sensed data b(k), and the actual performance P of the sensed data satisfy a hybrid decay model:

[0048] f(h(k),b(k))=-α·e -βh(k) +γb(k) -δ +∈,

[0049] Where α,β,γ,δ,∈ are non-negative system parameters, determined by the joint sensing scenario, and can be obtained through simulation fitting.

[0050] 1b. Design average communication resource constraints for each region, taking into account regional importance.

[0051] Based on the locational importance of each region (e.g., traffic density, target distribution frequency, whether it is a key monitoring area, etc.), a corresponding average traffic constraint is configured for each region. The configuration principles include:

[0052] Prioritize importance: Regions with higher importance have more critical sensing data and should receive a higher average communication resource ceiling;

[0053] Timeliness-driven: In areas with higher requirements for information timeliness, more communication resources are allocated to ensure the freshness and accuracy of the sensed data;

[0054] Optimal allocation under total resource constraints: Under the premise of limited total system bandwidth or communication resources, the average communication budget of each region is determined by optimization algorithms (such as static weighted allocation and dynamic resource scheduling);

[0055] Dynamic adjustability: Based on changes in traffic flow or external events, the weights and resource allocation strategies of each area can be periodically adjusted to adapt to environmental dynamics.

[0056] The average traffic constraint can be expressed as:

[0057]

[0058] Where w a B represents the importance weight of region a. total Represents the total communication resources of the system, satisfying:

[0059]

[0060] 1c. At the beginning of each time slot, execute the joint sensing scheduling strategy.

[0061] In each time slot, the scheduling system dynamically executes multi-region joint sensing scheduling tasks based on the set communication resource constraints, optimizing the overall system timeliness and task performance.

[0062] 2. The central server selects the scheduling region.

[0063] 2a. Determine if the area is idle: If the area is in either the sensor feature extraction or feature transmission stage, it indicates that the area is not idle and waits for the next time slot to make a judgment on it; if the area is idle, calculate its scheduling benefit.

[0064] 2b. Calculate the scheduling benefits for all idle regions: Based on the channel conditions of any idle region a under time slot k, such as signal-to-noise ratio (SNR) a The bandwidth of this region is W a The information about the age h in this area a(k). Define the information age h of this region. a (k) is h a (k)=kk m , where k m This is the last time this area was scheduled.

[0065] The optimal communication volume for this scheduling in all idle areas is obtained. Based on the characteristics of the joint sensing task and theoretical analysis, the optimal communication volume for region a is calculated as follows:

[0066]

[0067] Where F(h,b) is defined as the integral of the performance function with respect to the information age, that is:

[0068]

[0069] According to Shannon's formula, the channel transmission rate in this time slot is r. a (k)=W a ·log2(1+SNR a If the regional communication volume is b, and the feature extraction and feature transmission delay is μ(b), then the delay required for this scheduling task is d. a (k) is

[0070]

[0071] 2c. Selecting the scheduling region: Due to the limited computing power of the central server, a maximum of M regions can be scheduled at a time. Calculate the scheduling benefit for each region:

[0072] Assuming the region 'a' is scheduled, based on the characteristics of the joint sensing task and theoretical analysis, the scheduling benefit function is defined as follows:

[0073]

[0074] Among them, the last item This is a virtual queue term based on Lyapunov optimization theory, configured to ensure that the long-term average communication volume does not exceed a constraint value. Define V as an adjustable parameter, B a (k) is the virtual traffic queue for region a, which is updated at the end of each time slot according to the following rules:

[0075]

[0076] Sort all idle area scheduling revenue values ​​in reverse order. If the number of idle areas is greater than M, select the revenue value (i.e., U in step 2b). a (h aThe M regions with the largest (k) values ​​are scheduled; if the number of free regions is less than or equal to M, then all free regions are scheduled.

[0077] 2d. Optimal traffic allocation within the scheduling region:

[0078] Based on the scheduling revenue value of this scheduling region, communication constraints, and the communication volume of the previous scheduling, solve for the optimal communication volume for this scheduling. (See step 2b for the solution method).

[0079] 3. The central server allocates communication traffic and bandwidth to sensors in the selected area. The central server interacts with the sensors within the selected area via signaling, the sensors extract features, and the central server allocates communication traffic and bandwidth to each sensor based on a confidence threshold binary search. Figure 3 As shown.

[0080] 3a. The central server sends a scheduling start signal to each sensor in the selected area to initiate scheduling for that area.

[0081] 3b. Each sensor extracts sensory features and generates a spatial confidence map, including:

[0082] Upon receiving scheduling signals from the central server, each sensor immediately acquires raw sensing data. This raw sensing data is then processed by a feature extraction network to obtain feature representations. For example, 3D point cloud images acquired by LiDAR are processed by a PointPillars network, and raw images acquired by cameras are processed by a Fast R-CNN network. These feature representations are then used by a neural network to generate a task-oriented spatial confidence map. Taking object detection as an example, the detection confidence map generated by the detector head decoder represents the spatial confidence map, generating a confidence score for each sub-block within the region. Sub-blocks containing the target object are more critical than background sub-blocks, and therefore receive higher confidence scores.

[0083] 3c. Central server estimates confidence map threshold: The central server generates an initial value for the spatial confidence threshold based on constraints such as communication volume and sends it to each sensor.

[0084] 3d. Each sensor generates a communication mask based on a threshold and calculates the corresponding communication value for the mask:

[0085] Communication mask generation: Based on the spatial confidence threshold, each sensor sets the sub-block mask of the confidence map with a confidence level greater than or equal to the threshold to one, and sets the sub-block mask of the confidence level less than the threshold to zero.

[0086] Traffic calculation: Count the number N elements with a value of 1 in the mask. valid Multiplying this by the storage size 's' of each element (e.g., 4 bytes for floating-point numbers) gives the total communication volume of the feature map:

[0087] b = N valid ·s,

[0088] Where s represents the communication overhead (in bytes or bits) of a single valid feature element.

[0089] Each sensor sends its own communication value to the central server.

[0090] 3e. The central server determines whether the communication volume meets the communication volume constraint: if the total feedback communication volume exceeds the preset communication constraint, the confidence threshold is increased; if the communication volume is insufficient, the threshold is decreased. The threshold is continuously adjusted through an iterative search algorithm (such as binary search) until the difference between the sum of the communication volumes of each sensor and the communication constraint value is less than the tolerance error, thus obtaining the final threshold.

[0091] The central server sends the final threshold to each sensor.

[0092] 3f. Allocate transmission bandwidth to each sensor to optimize the timeliness of sensing tasks.

[0093] The required communication volume of each sensor's transmission sub-block is calculated. Based on the channel conditions for communication between each sensor and the central server and the communication volume of each sensor, bandwidth is allocated to each sensor to optimize the timeliness of the sensing task.

[0094] Each sensor's allocated bandwidth is sent to its corresponding sensor.

[0095] 4. Select a region and perform joint sensing tasks: Sensors within the selected region transmit features, which are then fused at the central server to complete the joint sensing task. Figure 3 As shown.

[0096] 4a. Each sensor transmits features based on its allocated bandwidth: Each sensor sends masked sensing features to the central server via a wireless network (such as the two generations of cellular vehicle-to-everything (V2X) standards, NR-V2X, etc.).

[0097] 4b. The central server fuses the sensor data from various sensors:

[0098] Based on the mid-term fusion model, fusion methods that can be adopted include: feature concatenation (concatenating multi-source features along the channel or spatial dimensions and then integrating them through convolution); attention-based fusion (assigning weights to features from different sources to guide the model to focus on more reliable inputs); and feature alignment (aligning features in a bird's-eye view or global coordinate system before fusion). Taking the commonly used feature concatenation fusion method as an example, let the feature maps from two sources be... and Where C1 and C2 are the number of channels in the feature map, and H and W are the spatial dimensions (height and width) of the feature map, feature concatenation stitches together multi-source features along the channel dimension or spatial dimension, resulting in the concatenated feature map F. cat It can be represented as:

[0099]

[0100] Then fuse them through convolutional layers:

[0101] F fused =σ(W conv *F cat +b),

[0102] Among them, W conv σ is the kernel weight, b is the convolution bias term, and σ(·) is the activation function.

[0103] 4c. Complete the joint perception task: The central server uses the fused features for subsequent tasks such as target detection, path planning, and environmental monitoring.

[0104] Example 2

[0105] In this second embodiment, a smart transportation scenario is used as an example to illustrate the sensor dynamic scheduling method for multi-area joint perception proposed in this embodiment: In a certain city system, multiple sensors (such as lidar) are deployed, distributed across 15 perception areas. The central server has a time-slot scheduling mechanism, performing scheduling once every 0.1 seconds. In a certain time slot, the central server, based on the current state of area a (such as the signal-to-noise ratio SNR of the transmission channel), performs scheduling. a ) and information age h a (k), calculate the regional scheduling revenue U a (h(k)) (Calculation method is shown in step 2c). Select the top 3 free regions in terms of revenue (e.g., region 2, region 7, region 11). The server allocates the optimal communication budget to the sensors in these regions. (Calculation method see step 2b) and bandwidth (bandwidth allocation method see step 3f). After each sensor extracts features, a spatial confidence map is generated (confidence map generation method see step 3b). A communication mask is generated based on the confidence threshold issued by the central server (communication mask generation method see step 3d), and only the features corresponding to high-confidence sub-blocks are transmitted. The threshold is dynamically adjusted through binary search to ensure that the total communication volume does not exceed the set budget. The server fuses all received multi-source features (feature fusion method see step 4b), performs target detection tasks, and uses the detection results for traffic light control and hazard warning, etc.

[0106] Example 3

[0107] This embodiment 3 provides a non-transitory computer-readable storage medium for storing computer instructions. When these computer instructions are executed by a processor, they implement the sensor dynamic scheduling method for multi-region joint sensing as described above. The method includes:

[0108] Based on the channel conditions, bandwidth, traffic constraints, and information age of the idle area under time slot k, the scheduling benefit value of all idle areas and their corresponding optimal area traffic are obtained; there are multiple sensors in the idle area.

[0109] Sort all idle area scheduling revenue values ​​in reverse order. If the number of idle areas is greater than the threshold M for the number of idle areas that can be scheduled each time, select the M areas with the highest revenue for scheduling; if the number of idle areas is less than or equal to M, schedule all idle areas.

[0110] Based on the scheduling benefit value of the current scheduling area, the communication volume constraints, and the communication volume of the previous scheduling, the optimal communication volume for the current scheduling is determined; feature extraction is performed on the sensors in the selected area, and communication volume allocation and bandwidth allocation are performed on each sensor based on a confidence threshold binary search.

[0111] Example 4

[0112] This embodiment 4 provides a computer device, including a memory and a processor, wherein the processor and the memory communicate with each other, and the memory stores program instructions that can be executed by the processor. The processor calls the program instructions to execute the sensor dynamic scheduling method for multi-region joint sensing as described above, the method including:

[0113] Based on the channel conditions, bandwidth, traffic constraints, and information age of the idle area under time slot k, the scheduling benefit value of all idle areas and their corresponding optimal area traffic are obtained; there are multiple sensors in the idle area.

[0114] Sort all idle area scheduling revenue values ​​in reverse order. If the number of idle areas is greater than the threshold M for the number of idle areas that can be scheduled each time, select the M areas with the highest revenue for scheduling; if the number of idle areas is less than or equal to M, schedule all idle areas.

[0115] Based on the scheduling benefit value of the current scheduling area, the communication volume constraints, and the communication volume of the previous scheduling, the optimal communication volume for the current scheduling is determined; feature extraction is performed on the sensors in the selected area, and communication volume allocation and bandwidth allocation are performed on each sensor based on a confidence threshold binary search.

[0116] Example 5

[0117] This embodiment 5 provides an electronic device, including: a processor, a memory, and a computer program; wherein, the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the sensor dynamic scheduling method for multi-region joint sensing as described above. The method includes:

[0118] Based on the channel conditions, bandwidth, traffic constraints, and information age of the idle area under time slot k, the scheduling benefit value of all idle areas and their corresponding optimal area traffic are obtained; there are multiple sensors in the idle area.

[0119] Sort all idle area scheduling revenue values ​​in reverse order. If the number of idle areas is greater than the threshold M for the number of idle areas that can be scheduled each time, select the M areas with the highest revenue for scheduling; if the number of idle areas is less than or equal to M, schedule all idle areas.

[0120] Based on the scheduling benefit value of the current scheduling area, the communication volume constraints, and the communication volume of the previous scheduling, the optimal communication volume for the current scheduling is determined; feature extraction is performed on the sensors in the selected area, and communication volume allocation and bandwidth allocation are performed on each sensor based on a confidence threshold binary search.

[0121] In summary, the sensor dynamic scheduling method and system for multi-region joint perception described in this invention significantly improves the system's response speed and decision-making accuracy by optimizing the multi-sensor joint perception scheduling strategy, achieving efficient data transmission and timeliness under limited communication resources. Through periodic broadcasting of cooperation requests, intelligent selection of cooperating devices, and dynamic allocation of communication traffic based on channel conditions, unnecessary data redundancy and wasted computing resources are avoided, improving the overall system's communication and computing efficiency. Furthermore, this technical solution, through precise coordination of feature extraction, fusion, and target detection of sensor data, enables multi-device collaborative perception to respond more quickly and accurately to dynamic environmental changes, greatly enhancing the system's perception capabilities and reliability in complex scenarios. This solution not only has significant advantages in transportation fields such as autonomous driving but also provides strong technical support for perception and decision-making in other high-demand scenarios such as smart cities, low-altitude economy, and smart factories, possessing broad application prospects and promotional value.

[0122] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0123] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0124] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0125] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment, whereby a series of operational steps are performed to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0126] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that, based on the technical solutions disclosed in the present invention, various modifications or variations that can be made by those skilled in the art without creative effort should be included within the scope of protection of the present invention.

Claims

1. A dynamic sensor scheduling method for multi-region joint sensing, characterized in that, include: According to time slot Considering the channel conditions, bandwidth, traffic constraints, and information age of the idle areas, the scheduling benefit value of all idle areas and their corresponding optimal area traffic are obtained. There are multiple sensors within the idle areas. The optimal area traffic is calculated as follows: ； Where b represents the regional communication volume, Indicates the lower limit of regional communication volume. Indicates the upper limit of regional communication volume; Indicates the age of the region; Indicates the time delay required for this scheduling task; This is the integral of the performance function with respect to the information age; The scheduling benefit value is: ； in, This is a virtual queue term based on Lyapunov optimization theory, configured to ensure that the long-term average communication volume does not exceed a constraint value. ; It is an adjustable parameter. For the region The virtual queue of communication traffic; f represents the impact of information age on perception, information timeliness. Perceived data volume Actual performance of perceived data Satisfies the mixed attenuation model: ； in These are non-negative system parameters; Sort all idle area scheduling revenue values ​​in reverse order. If the number of idle areas is greater than the threshold M for the number of idle areas that can be scheduled each time, select the M areas with the highest revenue for scheduling; if the number of idle areas is less than or equal to M, schedule all idle areas. Based on the scheduling benefit value of the current scheduling area, the communication volume constraints, and the communication volume of the previous scheduling, the optimal communication volume for the current scheduling is determined; feature extraction is performed on the sensors in the selected area, and communication volume allocation and bandwidth allocation are performed on each sensor based on a confidence threshold binary search.

2. The sensor dynamic scheduling method for multi-region joint sensing according to claim 1, characterized in that, Each sensor within the selected area receives the scheduling start signal and initiates scheduling for that area; each sensor extracts sensing features and generates a spatial confidence map; an initial value for the spatial confidence threshold is generated based on the communication volume constraints; each sensor generates a communication mask based on the threshold and calculates the corresponding communication volume value; it is determined whether the communication volume meets the communication volume constraints. If the total feedback communication volume exceeds the preset communication constraints, the confidence threshold is increased; if the communication volume is insufficient, the threshold is decreased.

3. The sensor dynamic scheduling method for multi-region joint sensing according to claim 2, characterized in that, The threshold is continuously adjusted through an iterative search algorithm until the difference between the sum of the communication volume of each sensor and the communication constraint value is less than the tolerance error, thus obtaining the final threshold.

4. The sensor dynamic scheduling method for multi-region joint sensing according to claim 2, characterized in that, The communication volume required for each sensor's transmission sub-block is calculated. Based on the channel conditions and communication volume of each sensor, bandwidth is allocated to each sensor to optimize the timeliness of the sensing task.

5. The sensor dynamic scheduling method for multi-region joint sensing according to claim 1, characterized in that, Each sensor extracts sensing features and generates a spatial confidence map, including: after receiving the scheduling signal, each sensor collects raw sensing data; the raw sensing data is processed by a feature extraction network to obtain feature representations; the feature representations are processed by a neural network to generate a task-oriented spatial confidence map.

6. The sensor dynamic scheduling method for multi-region joint sensing according to claim 1, characterized in that, Each sensor generates a communication mask based on a threshold and calculates the corresponding communication value for the mask, including: generating a communication mask based on a spatial confidence threshold, setting the sub-block mask with values ​​greater than or equal to the threshold to one, and setting the sub-block mask with values ​​less than the threshold to zero; and calculating the communication value corresponding to the non-zero elements of the communication mask.

7. A sensor dynamic scheduling system for multi-region joint sensing, characterized in that, include: The calculation module is used to calculate based on time slots. Considering the channel conditions, bandwidth, traffic constraints, and information age of the idle areas, the scheduling benefit value of all idle areas and their corresponding optimal area traffic are obtained. There are multiple sensors within the idle areas. The optimal area traffic is calculated as follows: ； Where b represents the regional communication volume, Indicates the lower limit of regional communication volume. Indicates the upper limit of regional communication volume; Indicates the age of the region; Indicates the time delay required for this scheduling task; This is the integral of the performance function with respect to the information age; The scheduling benefit value is: ； in, This is a virtual queue term based on Lyapunov optimization theory, configured to ensure that the long-term average communication volume does not exceed a constraint value. ; It is an adjustable parameter. For the region The virtual queue of communication traffic; f represents the impact of information age on perception, information timeliness. Perceived data volume Actual performance of perceived data Satisfies the mixed attenuation model: ； in These are non-negative system parameters; The selected module is used to sort all idle area scheduling revenue values ​​in reverse order. If the number of idle areas is greater than the threshold M of the number of idle areas that can be scheduled each time, the M areas with the highest revenue are selected for scheduling; if the number of idle areas is less than or equal to M, all idle areas are scheduled. The allocation module is used to solve for the optimal communication volume in the current scheduling based on the current scheduling benefit value, communication volume constraints, and the communication volume of the previous scheduling in the scheduling area; it performs feature extraction on the selected area sensors and performs communication volume allocation and bandwidth allocation for each sensor based on confidence threshold binary search.

8. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium is used to store computer instructions, which, when executed by a processor, implement the sensor dynamic scheduling method for multi-region joint sensing as described in any one of claims 1-6.

9. A computer device, characterized in that, The system includes a memory and a processor, which communicate with each other. The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the sensor dynamic scheduling method for multi-region joint sensing as described in any one of claims 1-6.

10. An electronic device, characterized in that, include: The device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the sensor dynamic scheduling method for multi-region joint sensing as described in any one of claims 1-6.

Images