Intelligent Transportation Signal Edge Computing Resource Dynamic Allocation Method and Its System

Abstract

The present invention relates to the field of intelligent transportation technology, in particular to an intelligent transportation signal edge computing resource dynamic allocation method and system thereof. First, real-time and historical traffic data of multiple intersections are obtained to construct a traffic prediction model to predict future traffic flow changes; then, according to the prediction results, resource requirements are calculated to form an edge computing resource requirement matrix, and the resource allocation is dynamically adjusted to generate a resource slicing scheme; subsequently, the traffic flow gradient between intersections is analyzed to establish an edge node collaborative computing model to optimize the calculation task allocation and migration; finally, combined with the system operation and traffic control effects, a closed-loop feedback is formed to continuously optimize the resource allocation and collaborative strategy. Through the resource demand modeling driven by traffic flow prediction, the active pre-allocation of computing resources is realized, the resource allocation is changed from passive response to active pre-adaptation, and the system resource utilization rate is increased by 40% to 60%.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and particularly to a method and system for dynamically allocating edge computing resources for intelligent traffic signals, which are applied to the field of urban traffic management and control, and solve the problem of dynamic resource allocation of traffic signal control systems in an edge computing environment. Background Art

[0002] With the acceleration of urbanization, the urban traffic flow has been continuously increasing, and traditional fixed or simple adaptive traffic signal control systems have been difficult to meet the complex and changing traffic demands. In recent years, intelligent traffic signal control systems based on edge computing have received extensive attention. By deploying edge computing devices at traffic intersections, real-time processing of traffic data and intelligent control of traffic lights can be achieved, effectively alleviating traffic congestion problems.

[0003] However, existing edge computing traffic signal control systems generally have problems such as unreasonable resource allocation and untimely system response. Taking an edge computing traffic signal control system and control method disclosed in Chinese Patent CN109697866B as an example, this patent constructs a three-layer architecture including an information acquisition module, an information processing module, and an information optimization module, collects traffic data through radar, and uses the Q-learning algorithm for signal optimization. Although this system has improved the intelligent level of traffic signal control to a certain extent, there are still the following deficiencies in practical applications:

[0004] 1. Edge computing resources adopt a static allocation method and cannot flexibly adjust resource configuration according to the dynamic changes of traffic flow, resulting in insufficient computing resources during traffic peaks and resource waste during troughs;

[0005] 2. There is a lack of a direct cooperation mechanism between edge nodes, and all data needs to be relayed through the cloud center, increasing the system response delay and unable to achieve fast collaborative control of adjacent intersections;

[0006] 3. Only a fixed Q-learning algorithm is adopted, and the algorithm has insufficient adaptability and is difficult to cope with diverse traffic scenarios;

[0007] 4. The elastic recovery mechanism in case of system failures is not considered, reducing the system reliability;

[0008] 5. The task division between the edge and the cloud is fixed and cannot dynamically adjust the distribution of computing tasks according to network conditions and computing loads.

[0009] In fact, in a complex and changing urban traffic environment, traffic flow shows obvious spatio-temporal variation characteristics, and the computing demands at different intersections and different times are significantly different. Static resource allocation and fixed algorithm strategies are difficult to meet the requirements of intelligent transportation systems for real-time performance, reliability, and flexibility, greatly limiting the application effect of edge computing in the traffic field.

[0010] Therefore, it is urgent to develop an intelligent transportation signal edge computing resource allocation method and system that can dynamically allocate computing resources according to traffic flow prediction, achieve collaborative computing among edge nodes, and have the ability of dynamic task scheduling, so as to improve the intelligent level of traffic management and the operation efficiency of the system. Summary of the Invention

[0011] The object of the present invention is to provide an intelligent transportation signal edge computing resource dynamic allocation method and its system, aiming to overcome the above deficiencies existing in the prior art, and realize the dynamic allocation of edge computing resources, collaborative computing among edge nodes, and flexible scheduling of computing tasks based on traffic flow prediction, so as to improve the response speed, resource utilization rate and system reliability of the traffic signal control system.

[0012] The present invention proposes an intelligent transportation signal edge computing resource dynamic allocation method, including: obtaining real-time traffic state data and historical traffic data of multiple traffic intersections; predicting the traffic flow change trend in the future time window based on the real-time traffic state data and historical traffic data, and generating a traffic prediction model;

[0013] Calculating the resource demand mapping relationship according to the traffic prediction model, and determining the edge computing resource demand matrix;

[0014] Based on the edge computing resource demand matrix, dynamically adjusting the resource allocation strategy of the edge computing resource pool to generate a resource slicing scheme;

[0015] According to the resource slicing scheme, determining the traffic flow gradient relationship between intersections, and establishing an edge node collaborative computing model;

[0016] Based on the edge node collaborative computing model, dynamically determining the execution location and migration strategy of the computing task to achieve the optimal allocation of the computing task;

[0017] According to the system operation state and traffic control effect, forming a closed-loop feedback, and dynamically optimizing the resource allocation parameters and collaborative computing strategy.

[0018] Preferably, the obtaining of the real-time traffic state data and historical traffic data of multiple traffic intersections specifically includes:

[0019] Collecting real-time traffic state data through multi-modal sensors deployed at traffic intersections, where the multi-modal sensors include radar sensors, video sensors and acoustic sensors;

[0020] Preprocessing the data collected by the multi-modal sensors, including data cleaning, outlier detection and data alignment;

[0021] Extracting the historical traffic data of the traffic intersections from the traffic management database;

[0022] Fuse the preprocessed real-time traffic state data with the historical traffic data to generate a standardized traffic state vector.

[0023] Preferably, based on the real-time traffic state data and historical traffic data, predicting the traffic flow change trend in a future time window, and generating a traffic prediction model specifically includes:

[0024] Construct a multi-dimensional traffic state vector, including parameters such as vehicle speed, vehicle density, and queue length;

[0025] Introduce a time series feature extraction matrix, including time feature factors such as weekdays / weekends, weather conditions, and special events;

[0026] Use a deep learning model to analyze the temporal variation law of the traffic state vector and predict the traffic state in the next 5 to 30 minutes;

[0027] Calculate the traffic state change rate based on the prediction results to determine the future traffic flow trend.

[0028] Preferably, according to the traffic prediction model, calculating the resource demand mapping relationship and determining the edge computing resource demand matrix specifically includes:

[0029] Establish a mapping function between the traffic state and the computing resource demand, and convert the traffic state parameters into CPU, memory, network bandwidth, and storage resource demands;

[0030] Determine the resource demand elasticity coefficient according to the complexity of different traffic scenarios;

[0031] Calculate the resource demand change curve of each intersection in the future time window;

[0032] Generate a resource demand prediction matrix including the time dimension to describe the resource demand distribution of each intersection node in the future period.

[0033] Preferably, based on the edge computing resource demand matrix, dynamically adjust the resource allocation strategy of the edge computing resource pool to generate a resource slicing scheme, specifically including:

[0034] Construct a virtualization layer of the edge computing resource pool to abstract physical resources into a dynamically allocable resource pool;

[0035] Design a resource pool description structure, including node identification, total resource amount, and available resource amount;

[0036] Define a resource allocation utility function to evaluate the efficiency of different resource allocation schemes;

[0037] Based on the principle of maximizing resource utility, calculate the optimal resource slicing strategy;

[0038] Generate a resource scheduling instruction set and perform dynamic slicing allocation of resources.

[0039] Preferably, according to the resource slicing scheme, determining the traffic flow gradient relationship between intersections and establishing an edge node collaborative computing model specifically includes:

[0040] Calculate the traffic flow change rate between adjacent intersections and generate a traffic flow gradient matrix;

[0041] Based on the traffic flow gradient matrix, identify intersection groups with high collaboration requirements;

[0042] Construct an edge node affinity matrix to quantify the collaboration efficiency between nodes;

[0043] Design a collaborative computing task description structure, including task priority, resource requirements, and deadline;

[0044] Establish a dynamic collaborative computing graph to determine the set of collaborative edge nodes and their communication relationships.

[0045] Preferably, based on the edge node collaborative computing model, dynamically determine the execution location and migration strategy of computing tasks to achieve optimal allocation of computing tasks, specifically including:

[0046] Define a task complexity evaluation vector to quantify the computational intensity, memory requirements, and timeliness of tasks;

[0047] Real-time monitor the load status of edge nodes and generate a load status matrix;

[0048] Construct a task migration decision model to determine the optimal execution location of tasks according to task complexity, node load, and collaboration relationship;

[0049] Establish a low-latency point-to-point communication channel to support data transmission during task migration;

[0050] Execute the task offloading instruction to dynamically allocate computing tasks to local execution, adjacent node execution, or cloud execution.

[0051] Preferably, according to the system operation status and traffic control effect, form a closed-loop feedback to dynamically optimize resource allocation parameters and collaborative computing strategies, specifically including:

[0052] Design a system performance index structure, including response time, resource utilization rate, energy consumption, and signal control efficiency;

[0053] Construct an objective function that balances the global and local to evaluate the overall system efficiency;

[0054] Introduce an adaptive weight vector to dynamically adjust the weights of each objective according to traffic conditions;

[0055] Optimize resource allocation parameters, collaboration scope, and task priorities based on the performance evaluation results;

[0056] Generate a new round of system tuning instructions to achieve a closed-loop feedback of prediction - allocation - collaboration - scheduling - evaluation - optimization.

[0057] Preferably, the method further includes:

[0058] Monitor the health status of edge nodes and construct a distributed health monitoring network;

[0059] Predict potential failure risks based on node performance anomaly metrics;

[0060] When a node failure or performance anomaly is detected, automatically trigger the function migration mechanism to migrate critical tasks to healthy nodes;

[0061] Implement an elastic degradation operation strategy according to the available resource status to ensure the priority execution of core functions;

[0062] Record the failure information and recovery process to form a system robustness knowledge base.

[0063] An intelligent transportation signal edge computing resource dynamic allocation system that executes the method includes:

[0064] A multi-modal data acquisition module for obtaining real-time traffic status data and historical traffic data of multiple traffic intersections;

[0065] A traffic flow prediction module for predicting the traffic flow change trend in a future time window based on the real-time traffic status data and historical traffic data, and generating a traffic prediction model;

[0066] A resource demand mapping module for calculating the resource demand mapping relationship according to the traffic prediction model and determining the edge computing resource demand matrix;

[0067] A resource dynamic allocation module for dynamically adjusting the resource allocation strategy of the edge computing resource pool based on the edge computing resource demand matrix and generating a resource slicing scheme;

[0068] An edge collaborative computing module for determining the traffic flow gradient relationship between intersections according to the resource slicing scheme and establishing an edge node collaborative computing model;

[0069] A task scheduling execution module for dynamically determining the execution location and migration strategy of computing tasks based on the edge node collaborative computing model to achieve optimal allocation of computing tasks;

[0070] A closed-loop feedback optimization module for forming a closed-loop feedback according to the system operation status and traffic control effect, and dynamically optimizing resource allocation parameters and collaborative computing strategies.

[0071] The beneficial effects of the present invention include:

[0072] 1. Through traffic flow prediction-driven resource demand modeling, proactive pre-allocation of computing resources is achieved, transforming resource allocation from passive response to proactive pre-adaptation, and the system resource utilization rate is increased by 40% - 60%;

[0073] 2. A multi-granularity resource pool dynamic slicing technology is constructed, enabling computing resources to flow flexibly according to traffic demands, significantly improving the resource allocation efficiency of the system;

[0074] 3. The edge collaborative computing mechanism based on traffic flow gradient realizes direct collaboration between adjacent intersections, and the system response time is reduced by 50% - 70%, especially in the case of edge node collaboration;

[0075] 4. The computing task dynamic offloading and migration decision engine enables the system to flexibly adjust the task execution location according to the load situation, improving the computing efficiency by 25% - 35%;

[0076] 5. The closed-loop feedback optimization system of edge-cloud collaboration realizes continuous optimization of system parameters, and the overall traffic efficiency is improved by 15% - 25%;

[0077] 6. A fault prediction and automatic recovery mechanism is introduced, and the system fault recovery time is shortened by more than 80%, greatly improving the system reliability. Description of the Drawings

[0078] Figure 1 is the overall architecture diagram of the intelligent traffic signal edge computing resource dynamic allocation system provided by the embodiment of the present invention;

[0079] Figure 2 is the flowchart of the intelligent traffic signal edge computing resource dynamic allocation method provided by the embodiment of the present invention;

[0080] Figure 3 is the structural schematic diagram of the traffic flow prediction module provided by the embodiment of the present invention;

[0081] Figure 4 is the working flowchart of the resource dynamic allocation module provided by the embodiment of the present invention;

[0082] Figure 5 is the performance comparison diagram between the embodiment of the present invention and the prior art. Detailed Embodiments

[0083] The following will describe in detail the specific embodiments of the present invention with reference to the drawings. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. For clarity, the detailed description of well-known technologies is omitted below.

[0084] Referring to Figure 1 , the intelligent transportation signal edge computing resource dynamic allocation system provided by the present invention includes: a multi-modal data acquisition module 11, a traffic flow prediction module 12, a resource demand mapping module 13, a resource dynamic allocation module 14, an edge collaborative computing module 15, a task scheduling and execution module 16, and a closed-loop feedback optimization module 17. These modules work together to form a complete intelligent transportation signal control system.

[0085] Referring to Figure 2 , the intelligent transportation signal edge computing resource dynamic allocation method provided by the present invention includes the following steps: obtaining real-time traffic state data and historical traffic data of multiple traffic intersections; predicting the traffic flow change trend in a future time window based on the real-time traffic state data and historical traffic data to generate a traffic prediction model; calculating a resource demand mapping relationship according to the traffic prediction model to determine an edge computing resource demand matrix; dynamically adjusting the resource allocation strategy of an edge computing resource pool based on the edge computing resource demand matrix to generate a resource slicing scheme; determining the traffic flow gradient relationship between intersections according to the resource slicing scheme to establish an edge node collaborative computing model; dynamically determining the execution location and migration strategy of a computing task based on the edge node collaborative computing model to achieve optimized allocation of the computing task; forming a closed-loop feedback based on the system operation state and traffic control effect to dynamically optimize resource allocation parameters and collaborative computing strategies.

[0086] Embodiment 1

[0087] First, the implementation method of obtaining real-time traffic state data and historical traffic data of multiple traffic intersections will be described in detail.

[0088] In a preferred embodiment of the present invention, the multi-modal data acquisition module 11 acquires real-time traffic state data through multi-modal sensors deployed at traffic intersections. The multi-modal sensors include radar sensors, video sensors, and acoustic sensors. Preferably, the radar sensor operates in the 24GHz frequency band, with a detection range of 150 meters and an accuracy of ±0.5 meters, and can accurately obtain vehicle position, speed, and quantity information; the video sensor uses a 4K resolution camera and supports edge image processing, and can identify vehicle models, license plates, and traffic events; the acoustic sensor is used to detect the sound characteristics of emergency vehicles to assist in judging special traffic conditions.

[0089] The data acquisition module 11 preprocesses the collected multi-modal data, including data cleaning, outlier detection, and data alignment. Specifically, in the data cleaning process, the moving window median filtering algorithm is used with a window size of 5 time units (usually 5 seconds), which can effectively filter sensor noise; for outlier detection, a statistics-based method is adopted, and when the data deviates from the mean by more than 3 standard deviations, it is identified as an outlier and replaced with the adjacent valid value; for data alignment, the timestamp matching method is used to ensure the temporal consistency of data from different sensors, and the alignment accuracy is better than 50 milliseconds.

[0090] Meanwhile, the system extracts the historical traffic data of the traffic intersection from the traffic management database, which usually includes traffic flow records in the past 6 months and is stored at a granularity of 15 minutes. The historical data is classified by weekdays / weekends / holidays and marked with influencing factors such as weather conditions and special events.

[0091] In addition, the data acquisition module 11 fuses the preprocessed real-time traffic state data with the historical traffic data to generate a standardized traffic state vector. In addition, the data acquisition module 11 fuses the preprocessed real-time traffic state data with the historical traffic data to generate a standardized traffic state vector. Preferably, the traffic state vector is defined as:

[0092] ,

[0093] where represents the average vehicle speed of the th lane (unit: km / h), represents the vehicle density of the th lane (unit: vehicles / 100m), represents the queue length of the th lane (unit: vehicles). In practical applications, a four-way intersection usually contains 12 lanes, so typically has a value of 12.

[0094] Embodiment 2

[0095] Next, the implementation method of predicting the traffic flow change trend in the future time window and generating a traffic prediction model based on the real-time traffic state data and the historical traffic data will be described in detail.

[0096] In an embodiment of the present invention, the traffic flow prediction module 12 constructs a multi-dimensional traffic state vector, which includes parameters such as vehicle speed, vehicle density, and queue length. At the same time, a time series feature extraction matrix F is introduced, which includes time feature factors such as weekdays / weekends, weather conditions, and special events:

[0097] ,

[0098] Among them, represents the date type (weekday = 1, weekend = 2, holiday = 3), represents the time period (morning rush hour = 1, flat peak = 2, evening rush hour = 3, night = 4), represents the weather condition (sunny = 1, cloudy = 2, rainy = 3, snowy to represents other time characteristic factors, such as special events, etc. In practical applications, The typical value of is 8, covering the main time characteristics affecting traffic flow.

[0099] Preferably, the traffic flow prediction module 12 uses a deep learning model to analyze the temporal variation law of the traffic state vector and predict the traffic state in the next 5 to 30 minutes. Specifically, a prediction model combining long short-term memory network (LSTM) with an attention mechanism is adopted, and its structure is as Figure 3 shown. This model includes an input layer, two LSTM hidden layers (each layer contains 128 neurons), an attention layer, and an output layer. The model input is a sequence of traffic state vectors in the past 60 minutes (at 5-minute intervals, a total of 12 time points), and the output is the predicted traffic state at the next 6 time points (5 minutes, 10 minutes, 15 minutes, 20 minutes, 25 minutes, and 30 minutes).

[0100] The calculation formula of the attention mechanism is:

[0101] ,

[0102] ,

[0103] ,

[0104] Among them, is the hidden state of the encoder, is the previous state of the decoder, , and are learnable parameters is the bias vector, is the attention weight, is the context vector. Based on the prediction results, the traffic flow prediction module 12 calculates the traffic state change rate and determines the future traffic flow trend. The traffic state change rate is defined as:

[0105] ,

[0106] Among them, represents the traffic state vector at time t, represents the predicted traffic state vector at time \(t + \Delta t\), is the prediction time interval.

[0107] In practical applications, the training of the prediction model uses a historical dataset. The first 90% of the data is used for training, and the last 10% is used for validation. Preferred training parameters include: batch size of 64, learning rate of 0.001, number of training epochs of 200, and early stopping strategy of stopping training if the validation loss does not decrease for 10 consecutive epochs. In this way, the model achieved performance metrics in actual tests with an average prediction error of less than 7% for 5-minute predictions, less than 12% for 15-minute predictions, and less than 18% for 30-minute predictions.

[0108] Embodiment III

[0109] The following details the implementation method of calculating the resource demand mapping relationship and determining the edge computing resource demand matrix according to the traffic prediction model.

[0110] In a preferred embodiment of the present invention, the resource demand mapping module 13 establishes a mapping function between traffic states and computing resource demands, converting traffic state parameters into CPU, memory, network bandwidth, and storage resource demands. Specifically, the resource demand mapping function \(R(ST,F)\) is defined as:

[0111] ,

[0112] where \(CPU\) represents the processor demand (unit: number of cores), \(MEM\) represents the memory demand (unit: GB), \(NET\) represents the network bandwidth demand (unit: Mbps), and \(STOR\) represents the storage demand (unit: GB).

[0113] Preferably, the formula for calculating the CPU resource demand is:

[0114] ,

[0115] where is the CPU resource coefficient, with a typical value of 0.02 cores / (km / h·veh / 100m), is the adjustment factor based on time characteristics, with a value range of [0.8, 1.5]. For example, during the morning and evening rush hours takes the value of 1.5, indicating that more CPU resources are needed to handle complex traffic conditions. Similarly, the formula for calculating the memory resource demand is:

[0116] ,

[0117] where is the memory resource coefficient, with a typical value of 0.05 GB / veh, is an adjustment factor based on time characteristics, with a value range of [0.9, 1.3].

[0118] The formula for calculating the network bandwidth requirement is:

[0119] ,

[0120] where is the network bandwidth coefficient, with a typical value of 0.2 Mbps / (vehicle / 100 m), is an adjustment factor based on time characteristics, with a value range of [0.7, 1.4].

[0121] The formula for calculating the storage requirement is:

[0122] ,

[0123] where is the storage coefficient, with a typical value of 0.01 GB / (vehicle / 100 m·hour), is an adjustment factor based on time characteristics, with a value range of [0.9, 1.2], is the data retention duration (unit: hour), with a typical value of 24 hours.

[0124] According to the complexity of different traffic scenarios, the resource demand mapping module 13 determines the resource demand elasticity coefficient . Preferably, the elasticity coefficient is defined according to the traffic flow change rate as:

[0125] ,

[0126] where is the volatility sensitivity coefficient, with a typical value of 0.5, represents the relative change rate of the traffic state. When the traffic state changes violently, the value increases, and the system reserves more resources to cope with traffic fluctuations.

[0127] The resource demand mapping module 13 calculates the resource demand change curve of each intersection within the future time window. For the time point , the resource demand is calculated as:

[0128] ,

[0129] Finally, the resource demand mapping module 13 generates a resource demand prediction matrix including the time dimension , describing the resource demand distribution of each intersection node within the future time period:

[0130] ,

[0131] where Indicates the prediction time point, with typical values of 5 minutes, 10 minutes, 15 minutes, 20 minutes, 25 minutes, and 30 minutes, i.e., .

[0132] In practical applications, the system updates the resource demand prediction matrix every 3 minutes to ensure that the resource allocation strategy can respond promptly to changes in traffic conditions. When the predicted change in resource demand exceeds 20% of the currently allocated resources, the resource reallocation process is triggered.

[0133] Example 4

[0134] The following details the implementation method of dynamically adjusting the resource allocation strategy of the edge computing resource pool and generating a resource slicing scheme based on the edge computing resource demand matrix.

[0135] In an embodiment of the present invention, the resource dynamic allocation module 14 first constructs a virtualization layer for the edge computing resource pool, abstracting physical resources into a dynamically allocable resource pool. As Figure 4 shown, the virtualization layer of the resource pool abstracts the physical computing resources (CPU, memory, network, and storage) of each edge node to form a unified virtual resource pool, facilitating flexible allocation by the system.

[0136] Preferably, the resource dynamic allocation module 14 designs a resource pool description structure RP, including node identification, total resource quantity, and available resource quantity:

[0137] ,

[0138] Among them, NodeID is the unique identifier of the edge node, CPUTotal and CPUAvail respectively represent the total number of CPU cores and the available number of CPU cores, MEMTotal and MEMAvail respectively represent the total memory and the available memory (unit: GB), NETTotal and NETAvail respectively represent the total network bandwidth and the available network bandwidth (unit: Mbps), and STORTotal and STORAvail respectively represent the total storage space and the available storage space (unit: GB).

[0139] The resource dynamic allocation module 14 defines a resource allocation utility function U(RP, S, RM) to evaluate the efficiency of different resource allocation schemes:

[0140] ,

[0141] Among them, is the resource slicing strategy matrix, represents the resource utilization utility, represents the resource matching degree utility, represents the load balancing utility, , and are weight coefficients, and .

[0142] In practical applications, typical weight settings are . Preferably, the resource utilization utility The calculation formula is:

[0143] ,

[0144] where , , and respectively represent the CPU, memory, network, and storage resource amounts allocated by the slicing strategy .

[0145] The resource matching degree utility The calculation formula is:

[0146] ,

[0147] where represents the resource amount allocated by the slicing strategy, represents the predicted resource demand amount in the resource demand matrix. The load balancing utility The calculation formula is:

[0148] ,

[0149] where is the number of edge nodes, is the load rate of node (the ratio of allocated resources to total resources), is the average load rate of all nodes. Based on the principle of maximizing resource utility, the resource dynamic allocation module 14 solves the optimal resource slicing strategy through a genetic algorithm

[0150] ,

[0151] The key parameter settings of the genetic algorithm include: the population size is 100, the chromosome encoding uses real number encoding, the selection operation uses tournament selection, the crossover probability is 0.8, the mutation probability is 0.1, and the number of iterations is 100.

[0152] Finally, the resource dynamic allocation module 14 generates a resource scheduling instruction set to perform resource dynamic slicing allocation.

[0153] Preferably, the resource scheduling adopts a two-phase commit protocol to ensure the atomicity and consistency of resource allocation. In the first phase, the system sends resource reservation requests to all relevant nodes; in the second phase, after confirming that all nodes can meet the requests, a commit instruction is sent to officially execute the resource allocation.

[0154] In addition, to prevent the system from becoming unstable due to frequent resource adjustments, a resource allocation cooling period (typical value is 5 minutes) and a resource change threshold (re-allocation is triggered only when the resource demand change exceeds 20%) are set. In actual tests, the resource dynamic allocation mechanism increased the system resource utilization rate from 43% to 87%, significantly improving the system's resource utilization efficiency.

[0155] Embodiment Five

[0156] The following details the implementation method of determining the traffic flow gradient relationship between intersections and establishing an edge node collaborative computing model according to the resource slicing scheme.

[0157] In the preferred embodiment of the present invention, the edge collaborative computing module 15 first calculates the traffic flow change rate between adjacent intersections to generate a traffic flow gradient matrix G:

[0158] ,

[0159] Among them, represents the traffic flow gradient between the th intersection and its adjacent intersection, and the calculation formula is

[0160] ,

[0161] Among them, represents the traffic flow flowing out of intersection (unit: vehicles / hour), represents the traffic flow flowing into intersection (unit: vehicles / hour), represents the distance between adjacent intersections (unit: km). Based on the traffic flow gradient matrix , the edge collaborative computing module 15 identifies groups of intersections with high collaboration requirements. Preferably, when the absolute value of the traffic flow gradient between two adjacent intersections exceeds the threshold (typical value is 100 vehicles / hour·km), it is determined that these two intersections have collaboration requirements. The system constructs a collaboration requirement graph , where the vertex set represents intersections, and the edge set represents pairs of intersections with collaboration requirements. The edge collaborative computing module 15 constructs an edge node affinity matrix to quantify the collaboration efficiency between nodes:

[0162] ,

[0163] Among them, represents the traffic correlation degree, represents the physical distance affinity, represents the resource complementarity degree, 、 and are weight coefficients, and . In practical applications, typical weight settings are .

[0164] Traffic correlation degree The calculation formula is:

[0165] ,

[0166] Among them, is the physical distance between node and node , and are the minimum and maximum cooperation distance thresholds respectively, and the typical values are and 2km respectively.

[0167] Resource complementarity degree The calculation formula is:

[0168] ,

[0169] Among them, and represent the available resources and the total resource amount of node respectively.

[0170] The edge collaborative computing module 15 designs a collaborative computing task description structure

[0171] ,

[0172] Among them, TaskID is the unique task identifier, Priority is the task priority (1 - 10, 10 is the highest), ResourceReq is the resource requirement description, Deadline is the task deadline, and Dependencies is the task dependency list.

[0173] Preferably, the collaborative computing tasks are divided into the following types:

[0174] Traffic status sharing task: Priority 8, low resource requirements, strict deadline (<100ms);

[0175] Signal coordination calculation task: Priority 9, medium resource requirements, medium deadline (<500ms);

[0176] Traffic prediction task: Priority 7, high resource requirements, loose deadline (<2s);

[0177] Resource coordination task: Priority 6, low resource requirements, medium deadline (<1s);

[0178] Finally, the edge collaborative computing module 15 establishes a dynamic collaborative computing graph to determine the set of collaborative edge nodes and their communication relationships. The collaborative computing graph is a weighted directed graph, where nodes represent edge computing nodes, edges represent collaborative relationships, and edge weights represent collaborative strengths (from the affinity matrix A). The system establishes a point-to-point (P2P) low-latency communication channel based on the collaborative computing graph to support direct data exchange between edge nodes.

[0179] In actual implementation, the P2P communication channel is implemented based on the UDP protocol, and the forward error correction (FEC) technology is used to improve reliability. The communication delay is controlled within 10 milliseconds. To optimize communication efficiency, the system uses incremental update and data compression technologies to reduce the amount of transmitted data. The typical bandwidth requirement is 2 - 5 Mbps for each P2P link.

[0180] The collaborative scope in the system is dynamically adjusted according to traffic conditions. In the congested state, the collaborative scope expands and can cover up to 10 intersections within 2 kilometers; in the unobstructed state, the collaborative scope shrinks and usually only includes 2 - 3 adjacent intersections. This dynamic collaboration mechanism enables the system to reduce unnecessary communication overhead while maintaining control effects.

[0181] Embodiment Six

[0182] The following details the implementation method of dynamically determining the execution location and migration strategy of computing tasks based on the edge node collaborative computing model to achieve the optimal allocation of computing tasks.

[0183] In the preferred embodiment of the present invention, the task scheduling execution module 16 first defines a task complexity evaluation vector C to quantify the computational intensity, memory requirements, and timeliness of tasks:

[0184] ,

[0185] Among them, represents the computational complexity (level 1 - 10, with 10 being the highest), represents the memory requirement (unit: MB), represents the network transmission volume (unit: KB), represents the timeliness requirement (unit: ms).

[0186] Preferably, the system predefines complexity templates for different types of tasks, such as:

[0187] Traffic data processing task: ;

[0188] Signal optimization calculation task: ;

[0189] Traffic prediction task: ;

[0190] Resource scheduling task: ;

[0191] The task scheduling execution module 16 monitors the load status of edge nodes in real time and generates a load status matrix . The load status of each node includes the following metrics:

[0192] ,

[0193] Among them, represents the CPU utilization rate (0% - 100%), represents the memory utilization rate (0% - 100%), represents the network bandwidth utilization rate (0% - 100%), represents the task queue length. The system updates the load status matrix once per second to ensure that task scheduling decisions are based on the latest load conditions.

[0194] The task scheduling execution module 16 constructs a task migration decision model , and determines the optimal execution location of the task according to task complexity, node load, and collaboration relationship:

[0195] ,

[0196] Among them, 0 represents local execution, 1 represents offloading to a neighboring node, and 2 represents offloading to the cloud. The decision-making process considers various factors such as task characteristics, resource availability, node affinity, and network conditions.

[0197] Preferably, the task migration decision adopts a hierarchical strategy. First, it determines whether the task is suitable for local execution. When the load of the local node is lower than the threshold (the typical value is 70%) and the task complexity is not high , local execution is selected; otherwise, it enters the next layer of decision-making.

[0198] In the second layer of decision-making, the system evaluates the feasibility of offloading the task to a neighboring node. Calculate the offloading utility function :

[0199] ,

[0200] Among them, is the local node, is the target node, , and are weight coefficients. When the maximum offloading utility exceeds the threshold (typical value is 0.6), select to offload the task to the corresponding neighboring node; otherwise, enter the third-layer decision-making.

[0201] In the third-layer decision-making, the system evaluates the benefits of offloading the task to the cloud. When the task calculation complexity is high or both the local and neighboring nodes are highly loaded (the load rates both exceed 80%), select to offload the task to the cloud for execution.

[0202] The task scheduling and execution module 16 establishes a low-latency peer-to-peer communication channel to support data transmission during the task migration process. The communication channel is implemented based on the QUIC protocol, featuring low latency and high reliability, and is suitable for the edge computing environment. To improve the transmission efficiency, the system compresses and differentially transmits the migrated data to reduce network overhead.

[0203] Finally, the task scheduling and execution module 16 executes the task offloading instruction to dynamically allocate the computing task to the optimal execution location.

[0204] In practical applications, the task dynamic scheduling mechanism significantly improves the computing efficiency of the system. The test results show that compared with the fixed task allocation, the dynamic scheduling reduces the average response time of the system by 68% and improves the computing resource utilization rate by 52%, especially more obvious during the traffic peak period.

[0205] Embodiment VII

[0206] The following details the implementation method of forming a closed-loop feedback and dynamically optimizing the resource allocation parameters and collaborative computing strategy according to the system operation status and traffic control effect.

[0207] In the preferred embodiment of the present invention, the closed-loop feedback optimization module 17 designs the system performance index structure P, including response time, resource utilization rate, energy consumption, and signal control efficiency:

[0208] ,

[0209] Among them, ResponseTime represents the system response time (unit: ms), and ResourceUtil represents the resource utilization rate , PowerConsumption represents power consumption (unit: W), and SignalEfficiency represents signal control efficiency, which is evaluated by the traffic flow improvement rate. The closed-loop feedback optimization module 17 constructs an objective function that balances the global and local , and evaluates the overall system performance:

[0210] ,

[0211] Among them, represents the global objective function, represents the local objective function, and are weight coefficients, and .

[0212] The global objective function is defined as:

[0213] ,

[0214] Among them, is the maximum acceptable response time (typical value is 1000ms), is the optimal resource utilization rate (typical value is 85%), is the maximum power consumption benchmark value, and are weight coefficients, and . The local objective function is defined as:

[0215] ,

[0216] Among them, LocalResponseTime represents the local response time performance, LocalResourceBalance represents the local resource balance, LocalTaskCompletion represents the local task completion rate, , and are weight coefficients, and .

[0217] The closed-loop feedback optimization module 17 introduces an adaptive weight vector , and dynamically adjusts the weights of each objective according to the traffic conditions. Preferably, during the traffic peak period, the system increases the weights of the global objective and signal efficiency; during the traffic trough period, the system increases the weights of power consumption and local response time. The weight adjustment follows the following rules:

[0218] When the traffic flow exceeds 80% of the capacity, set (global priority); when the traffic flow is between 50% and 80% of the capacity, set (Balancing the global and local); when the traffic flow is below 50% of the capacity, set (Local priority);

[0219] Based on the performance evaluation results, the closed-loop feedback optimization module 17 optimizes the resource allocation parameters, cooperation scope, and task priorities. The optimization process uses the Bayesian optimization algorithm to balance exploration and exploitation and quickly converge to the optimal parameter settings.

[0220] The key steps of the Bayesian optimization algorithm include: constructing a Gaussian process model of the relationship between system performance and parameters; selecting the next set of parameters for evaluation based on an acquisition function (such as Expected Improvement); updating the model and iterating until convergence or reaching the maximum number of iterations;

[0221] The optimized parameters include:

[0222] Resource allocation parameters: resource demand elasticity coefficient , weight of the resource allocation utility function 、 、 ; Cooperative computing parameters: cooperation threshold , weight of the affinity matrix 、 、 ; Task scheduling parameters: local execution threshold , offloading utility threshold ; Finally, the closed-loop feedback optimization module 17 generates a new round of system tuning instructions to achieve a closed-loop feedback of prediction - allocation - cooperation - scheduling - evaluation - optimization. The tuning instructions are sent to each functional module through a message queue to update the system parameters. The system performs global optimization every 30 minutes and retains multiple sets of parameter configurations to handle different traffic scenarios.

[0223] In actual tests, the closed-loop feedback optimization mechanism significantly improves the overall performance of the system. As the running time increases, the system's adaptability to traffic conditions continuously enhances. After a 3-month test period, compared with the initial configuration, the system response time is reduced by 24%, the resource utilization rate is increased by 18%, the energy consumption is reduced by 15%, and the signal control efficiency is increased by 22%.

[0224] Embodiment VIII

[0225] The implementation method of the system fault detection and recovery mechanism is described in detail below.

[0226] In one embodiment of the present invention, the system monitors the health status of edge nodes and constructs a distributed health monitoring network. Each edge node 21, 22,..., 2n periodically sends a heartbeat signal (typical frequency is once every 5 seconds), which includes basic health metrics: CPU utilization, memory usage, storage space, network connection status, and temperature. At the same time, a mutual monitoring relationship is formed among the nodes, and each node is monitored by at least two other nodes to ensure the reliability of fault detection.

[0227] The system predicts potential fault risks based on node performance anomaly metrics. The anomaly detection uses a method combining statistics and machine learning, mainly considering the following metrics: CPU utilization continuously exceeds 95% for more than 2 minutes; memory usage suddenly increases by more than 30%; the heartbeat signal delay exceeds 3 times the expected value; the storage I / O error rate is higher than the threshold (typical value is 0.1%); the system temperature exceeds the warning value (typical value is 75°C);

[0228] When any metric is abnormal, the system calculates the fault risk score R:

[0229] ,

[0230] where, is the number of metrics, is the current metric value, is the normal threshold, is the critical threshold, is the metric weight. When the risk score exceeds the warning threshold (typical value is 0.7), the system starts preventive maintenance; when it exceeds the emergency threshold (typical value is 0.9), an emergency migration process is triggered.

[0231] When a node failure or performance anomaly is detected, the system automatically triggers a function migration mechanism to migrate critical tasks to healthy nodes. The migration process is based on task priorities. First, critical tasks such as traffic signal control are migrated, and then data processing and prediction tasks. The migration uses a state save and restore mechanism to ensure task continuity.

[0232] Preferably, the function migration process includes the following steps: identifying the set of critical tasks to be migrated; selecting the best target node according to the current collaborative computing graph; creating a task state snapshot, including the running environment, data context, and execution status; transmitting the state snapshot to the target node through a high-priority channel; reconstructing the task running environment and resuming execution at the target node;

[0233] In actual implementation, the entire migration process is controlled to be completed within 3 seconds to ensure that the traffic control function is not interrupted. The system maintains a task dependency graph to ensure that related tasks are migrated in the correct order and avoid function loss.

[0234] According to the available resources, the system implements an elastic degradation operation strategy to ensure the priority execution of core functions. The degradation strategies include: reducing the priority of non-critical tasks or suspending their execution; reducing the data sampling rate and processing accuracy; narrowing the scope of collaborative control and focusing on local optimization; adopting alternative algorithms with lower computational complexity;

[0235] During the degraded operation, the system still guarantees the basic traffic signal control function and maintains traffic order, but may reduce the control accuracy and optimization effect. When the system resources are restored, it automatically returns to the normal operation state.

[0236] The system records the fault information and the recovery process to form a system robustness knowledge base. Each fault event record includes: fault type, detection method, impact scope, recovery strategy, recovery time, and effect evaluation. The knowledge base is used to continuously improve the fault detection and recovery strategies and enhance the overall robustness of the system.

[0237] In actual tests, by simulating different types of node faults (such as hardware faults, network interruptions, resource exhaustion, etc.), the system demonstrated good elastic recovery capabilities. The average fault detection time was 1.2 seconds, the function migration completion time was 2.8 seconds, and the overall service interruption time was controlled within 4 seconds, meeting the reliability requirements of the traffic control system.

[0238] Embodiment Nine

[0239] The following details the implementation methods of the hardware and software architectures of the intelligent traffic signal edge computing resource dynamic allocation system.

[0240] The system of the present invention includes multiple distributed edge nodes 31, 32,..., 3n and a cloud center node 40, forming a multi-level computing architecture. The edge nodes are deployed at traffic intersections and are directly connected to traffic signal lights and multi-modal sensors; the cloud center node provides global resource scheduling and optimization functions.

[0241] In terms of hardware, each edge node includes the following components: The computing unit uses a multi-core processor with the ARMCortex-A76 architecture, with a main frequency of 2.2GHz, supporting dynamic frequency adjustment; the storage system: configures a hierarchical storage architecture, including a cache (4GB) and a main storage (16GB); the network interface: integrates a 5G / WiFi6 dual-mode communication module, supporting low-latency P2P communication; the power management: adopts an intelligent power management system, supporting load-based dynamic power consumption control; the sensor interface: provides a standard interface to connect radars, cameras, and acoustic sensors; the cloud center node uses a high-performance server cluster, configured with IntelXeon processors, 128GB of memory, and 10Gbps network interfaces, providing powerful computing capabilities to support global optimization tasks.

[0242] In terms of software architecture, the system adopts a microservices architecture, mainly including the following layers: Operating System Layer: Based on a customized real-time operating system, supporting resource isolation and dynamic allocation; Middleware Layer: Includes a resource virtualization manager, a distributed computing framework, and a message queue system; Application Layer: Implements core functions such as traffic prediction, resource scheduling, and collaborative computing; The system uses container technology to achieve task encapsulation and migration. Each functional module is encapsulated as an independent container, facilitating flexible deployment and migration. The container image size is controlled within 100MB to ensure fast migration.

[0243] In terms of communication, the system constructs a three-layer communication network: Edge Internal Communication: Based on shared memory and local message queues, with a latency < 1ms; Inter-Edge Node Communication: P2P direct communication based on the QUIC protocol, with a latency < 10ms; Edge-Cloud Communication: Based on a secure VPN tunnel, supporting dynamic bandwidth allocation;

[0244] The system adopts a unified data format and interface standard to ensure seamless integration between modules. It uses the JSON format for data exchange and RESTAPI and gRPC to implement service calls.

[0245] In terms of security, the system implements multi-level security protection: Identity Authentication: A two-way authentication mechanism based on PKI; Communication Encryption: Adopts the TLS1.3 protocol to ensure secure data transmission; Access Control: Fine-grained role-based access control; Intrusion Detection: Real-time monitoring of network traffic to detect abnormal behaviors;

[0246] In terms of data management, the system adopts a hierarchical storage strategy: Real-time Data: Stored in an in-memory database, supporting millisecond-level access; Recent Data: Stored in a local SSD, retained for 24 hours; Historical Data: Compressed and uploaded to cloud storage for long-term analysis

[0247] In actual deployment, edge nodes are installed in traffic signal control cabinets or roadside boxes, and are connected to traffic signal controllers and multi-modal sensors through standard interfaces. The typical deployment density is 10 - 15 nodes per square kilometer, covering major road intersections.

[0248] Embodiment Ten

[0249] The following details the application effects and performance advantages of the present invention in actual traffic scenarios.

[0250] The intelligent transportation signal edge computing resource dynamic allocation system of the present invention has been tested and verified in typical traffic sections of multiple cities. Taking a central area of a certain city as an example, the test area includes 25 traffic intersections, with an average daily traffic volume of about 120,000 vehicle trips. The test lasts for 3 months, covering different traffic scenarios on weekdays, weekends, and holidays.

[0251] In terms of resource utilization efficiency, as shown in part a of the appendix Figure 5 The computing resource utilization rate of the system of the present invention reaches 87%, which is 102% higher than the static resource allocation method in Comparative Document 1 (CN109697866B). Through traffic prediction-driven dynamic resource allocation, the system can sense changes in traffic demand in advance and actively adjust resource allocation, significantly improving resource utilization efficiency.

[0252] In terms of system response time, as shown in part b of the appendix Figure 5 The average response time of the system of the present invention is 43 ms, which is 77% lower than that of Comparative Document 1 (CN109697866B). Especially during traffic peak periods, the improvement in response time is more obvious, ensuring that the system can respond to changes in traffic conditions in real time and adjust signal timing strategies in a timely manner.

[0253] In terms of edge collaboration efficiency, as shown in part c of the appendix Figure 5 The edge node collaboration delay of the system of the present invention is only 28 ms, which is 92% lower than that of Comparative Document 1 (CN109697866B). Based on the traffic gradient-based edge collaborative computing mechanism, adjacent intersections can directly exchange information and perform collaborative control, significantly improving the efficiency of multi-intersection coordinated control.

[0254] In terms of task scheduling flexibility, as shown in part d of the appendix Figure 5 The task scheduling flexibility score of the system of the present invention reaches 92 points (out of 100), which is 100% higher than that of Comparative Document 1 (CN109697866B). Through the computing task dynamic offloading and migration mechanism, the system can flexibly adjust the task execution location according to the load situation and make full use of distributed computing resources.

[0255] In terms of energy efficiency, as shown in part e of the appendix Figure 5 The energy efficiency of the present invention reaches 1.35 MIPS / W, which is 221% higher than that of Comparative Document 1 (CN109697866B). Through intelligent resource management and dynamic power consumption control, the system can significantly reduce energy consumption while ensuring performance.

[0256] In terms of traffic passing efficiency, as shown in part f of the appendix Figure 5 The system of the present invention reduces the average passing time of vehicles in the test area by 38%, which is 111% higher than that of Comparative Document 1 (CN109697866B). Through the optimized signal timing strategy and multi-intersection collaborative control, the system significantly improves traffic flow efficiency and reduces congestion.

[0257] In terms of system reliability, during the 3-month test period, the system of the present invention successfully handled 37 edge node abnormal events. The average detection time was 1.2 seconds, the function migration completion time was 2.8 seconds, and the service interruption time was controlled within 4 seconds, meeting the high reliability requirements of the traffic control system.

[0258] In terms of system scalability, the system of the present invention supports the rapid access of new nodes. The access time is shortened from the traditional 4 hours / node to 3 minutes / node, improving the flexible expansion ability of the system. Through standardized interfaces and an automated configuration synchronization mechanism, the system can be conveniently expanded to a larger-scale traffic network.

[0259] In summary, the intelligent transportation signal edge computing resource dynamic allocation method and system of the present invention significantly improve the performance and efficiency of the traffic signal control system through innovative technologies such as traffic prediction-driven resource allocation, multi-granularity resource slicing, edge collaborative computing, task dynamic scheduling, and closed-loop feedback optimization, providing strong support for intelligent transportation management.

[0260] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent transportation signal edge computing resource dynamic allocation method, characterized in that Including: Obtain real-time traffic status data and historical traffic data of multiple traffic intersections; Based on the real-time traffic status data and historical traffic data, predict the traffic flow change trend in the future time window and generate a traffic prediction model; According to the traffic prediction model, calculate the resource demand mapping relationship and determine the edge computing resource demand matrix; Based on the edge computing resource demand matrix, dynamically adjust the resource allocation strategy of the edge computing resource pool and generate a resource slicing scheme; According to the resource slicing scheme, determine the traffic flow gradient relationship between intersections and establish an edge node collaborative computing model; Based on the edge node collaborative computing model, dynamically determine the execution location and migration strategy of computing tasks to achieve optimal allocation of computing tasks; According to the system operation status and traffic control effect, form a closed-loop feedback and dynamically optimize the resource allocation parameters and collaborative computing strategy; According to the resource slicing scheme, determine the traffic flow gradient relationship between intersections and establish an edge node collaborative computing model specifically including: Calculate the traffic flow change rate between adjacent intersections and generate a traffic flow gradient matrix; Based on the traffic flow gradient matrix, identify intersection groups with high collaboration requirements; Construct an edge node affinity matrix to quantify the collaborative efficiency between nodes; Design a collaborative computing task description structure including task priority, resource requirements, and deadline; Establish a dynamic collaborative computing graph to determine the set of collaborative edge nodes and their communication relationships.

2. The method according to claim 1, wherein The obtaining of real-time traffic status data and historical traffic data of multiple traffic intersections specifically includes: Collect real-time traffic status data through multi-modal sensors deployed at traffic intersections, and the multi-modal sensors include radar sensors, video sensors, and acoustic sensors; Preprocess the data collected by the multi-modal sensors, including data cleaning, outlier detection, and data alignment; Extract the historical traffic data of the traffic intersections from the traffic management database; Fuse the preprocessed real-time traffic status data with the historical traffic data to generate a standardized traffic status vector.

3. The method according to claim 1, characterized in that Based on the real-time traffic status data and historical traffic data, predicting the traffic flow change trend in the future time window and generating a traffic prediction model specifically includes: Construct a multi-dimensional traffic status vector including vehicle speed, vehicle density, and queue length parameters; Introduce a time series feature extraction matrix including weekday / weekend, weather condition, and special event time feature factors; Use a deep learning model to analyze the temporal change law of the traffic status vector and predict the traffic status in the next 5 to 30 minutes; Calculate the traffic status change rate based on the prediction result and determine the future traffic flow trend.

4. The method according to claim 1, wherein According to the traffic prediction model, calculating the resource demand mapping relationship and determining the edge computing resource demand matrix specifically includes: Establish a mapping function between traffic status and computing resource demand, and convert traffic status parameters into CPU, memory, network bandwidth, and storage resource requirements; Determine the resource demand elasticity coefficient according to the complexity of different traffic scenarios; Calculate the resource demand change curve of each intersection in the future time window; Generate a resource demand prediction matrix including the time dimension to describe the resource demand distribution of each intersection node in the future time period.

5. The method according to claim 1, wherein Based on the edge computing resource demand matrix, dynamically adjust the resource allocation strategy of the edge computing resource pool. The specific steps for generating a resource slicing scheme include: Construct a virtualization layer for the edge computing resource pool to abstract physical resources into a dynamically allocable resource pool; Design a resource pool description structure that includes node identifiers, total resource amounts, and available resource amounts; Define a resource allocation utility function to evaluate the efficiency of different resource allocation schemes; Based on the principle of maximizing resource utility, calculate the optimal resource slicing strategy; Generate a resource scheduling instruction set to perform dynamic resource slicing allocation.

6. The method according to claim 1, characterized in that, Based on the edge node collaborative computing model, dynamically determine the execution location and migration strategy of computing tasks to achieve optimal allocation of computing tasks. The specific steps include: Define a task complexity evaluation vector to quantify the computational intensity, memory requirements, and timeliness of tasks; Real-time monitor the load status of edge nodes to generate a load status matrix; Construct a task migration decision model to determine the optimal execution location of tasks according to task complexity, node load, and collaborative relationships; Establish a low-latency peer-to-peer communication channel to support data transmission during task migration; Execute task offloading instructions to dynamically allocate computing tasks to local execution, neighboring node execution, or cloud execution.

7. The method according to claim 1, characterized in that According to the system operation status and traffic control effect, form a closed-loop feedback to dynamically optimize resource allocation parameters and collaborative computing strategies. The specific steps include: Design a system performance index structure that includes response time, resource utilization rate, energy consumption, and signal control efficiency; Construct an objective function that balances the global and local to evaluate the overall system performance; Introduce an adaptive weight vector to dynamically adjust the weights of each objective according to traffic conditions; Based on the performance evaluation results, optimize resource allocation parameters, collaborative scope, and task priorities; Generate a new round of system tuning instructions to achieve a closed-loop feedback of prediction - allocation - collaboration - scheduling - evaluation - optimization.

8. The method according to claim 1, wherein The method further includes: Monitor the health status of edge nodes and construct a distributed health monitoring network; Based on node performance anomaly indicators, predict potential failure risks; When a node failure or performance anomaly is detected, automatically trigger a function migration mechanism to migrate critical tasks to healthy nodes; Implement an elastic degradation operation strategy according to the available resource status to ensure that core functions are executed first; Record failure information and recovery processes to form a system robustness knowledge base.

9. An intelligent transportation signal edge computing resource dynamic allocation system for implementing the method according to any one of claims 1-8, characterized in that It includes: A multi-modal data acquisition module for obtaining real-time traffic status data and historical traffic data of multiple traffic intersections; A traffic flow prediction module for predicting the traffic flow change trend in the future time window based on the real-time traffic status data and historical traffic data to generate a traffic prediction model; A resource demand mapping module for calculating resource demand mapping relationships according to the traffic prediction model to determine the edge computing resource demand matrix; A resource dynamic allocation module for dynamically adjusting the resource allocation strategy of the edge computing resource pool based on the edge computing resource demand matrix to generate a resource slicing scheme; An edge collaborative computing module, which is used to determine the traffic flow gradient relationship between intersections according to the resource slicing scheme and establish an edge node collaborative computing model; A task scheduling and execution module, which is used to dynamically determine the execution location and migration strategy of computing tasks based on the edge node collaborative computing model to achieve optimal allocation of computing tasks; A closed-loop feedback optimization module, which is used to form a closed-loop feedback according to the system operation state and traffic control effect and dynamically optimize the resource allocation parameters and collaborative computing strategy.

Images