Multi-layer cloud architecture intelligent traffic resource allocation method, storage medium and electronic equipment

Abstract

The invention provides a multi-layer cloud architecture intelligent traffic resource allocation method, a storage medium and an electronic device, the multi-layer cloud architecture comprises an edge cloud, a region cloud and a center cloud, and the method comprises the following steps: the edge cloud performs resource allocation on a received edge task by adopting a genetic algorithm; the regional cloud adopts a particle swarm optimization algorithm to carry out resource allocation on the received regional tasks; the central cloud performs resource allocation on the received central task by adopting a hybrid genetic algorithm and a simulated annealing algorithm; and the central cloud monitors the resource allocation results of the edge cloud and the regional cloud, and dynamically adjusts the resource allocation of the edge cloud and the regional cloud according to the monitoring results. According to the technical scheme, the resource utilization rate is effectively improved, the task completion time is shortened, and the real-time response capability and the overall performance of the intelligent traffic system are remarkably improved.

Description

Technical Field

[0001] The present application relates to the technical field of path planning during vehicle driving, and specifically to a multi-layer cloud architecture intelligent traffic resource allocation method, storage medium and electronic device. Background Art

[0002] Multi-layer cloud architecture (including edge cloud, regional cloud and central cloud) has gradually become the development trend of intelligent transportation systems. Edge cloud is deployed at the edge nodes of the transportation network, which can process real-time data nearby and significantly reduce transmission delay; regional cloud is located in a specific geographical area, responsible for managing and coordinating multiple edge cloud nodes to achieve regional data aggregation and analysis; central cloud is responsible for global data integration and long-term analysis. Although multi-layer cloud architecture has alleviated the bottleneck of a single central cloud to a certain extent, the existing methods still have the following shortcomings in resource allocation and task scheduling:

[0003] Unbalanced resource allocation: The heterogeneity and dynamic change characteristics of resources at each layer of the cloud make it difficult for existing resource allocation methods to achieve global optimization, often resulting in overload or idleness of some cloud resources.

[0004] Low task scheduling efficiency: Traditional task scheduling algorithms are difficult to balance the real-time requirements and task complexity at all levels, resulting in high task processing delays and affecting the overall system performance.

[0005] Lack of global coordination mechanism: Existing methods mostly use independent optimization and lack a collaborative optimization mechanism between levels. This makes it difficult to dynamically adapt to changes in real-time traffic conditions, reducing the system's responsiveness and reliability. Summary of the invention

[0006] The technical problem to be solved by this application is the deficiencies in resource allocation and task scheduling of the multi-layer cloud architecture that can only be applied in the transportation field in the prior art, and thus a multi-layer cloud architecture intelligent transportation resource allocation method, storage medium and electronic device are provided.

[0007] In a first aspect, the technical solution of the present application provides a multi-layer cloud architecture intelligent traffic resource allocation method, wherein the multi-layer cloud architecture includes an edge cloud, a regional cloud, and a central cloud, and the method includes:

[0008] The edge cloud uses a genetic algorithm to allocate resources for edge tasks it receives;

[0009] The regional cloud uses a particle swarm optimization algorithm to allocate resources for the regional tasks it receives;

[0010] The central cloud uses a hybrid genetic algorithm and a simulated annealing algorithm to allocate resources for the central tasks it receives;

[0011] The central cloud monitors the resource allocation results of the edge cloud and the regional cloud, and dynamically adjusts the resource allocation of the edge cloud and the regional cloud according to the monitoring results.

[0012] In some embodiments, the multi-layer cloud architecture intelligent transportation resource allocation method, the edge cloud uses a genetic algorithm to allocate resources for edge tasks received by it, including:

[0013] Randomly assigning the edge tasks to different edge clouds to obtain a set of initial resource allocation schemes;

[0014] Evaluate the fitness of each of the initial resource allocation schemes to obtain the execution time and resource utilization of each edge cloud executing edge tasks;

[0015] Select the initial resource allocation plan with the largest fitness evaluation result as the benchmark allocation plan;

[0016] After crossover, iteration, mutation and iteration processing are performed on the benchmark allocation scheme, several variant allocation schemes are obtained;

[0017] The fitness of each of the variant allocation schemes is evaluated, and the variant allocation scheme with the largest fitness evaluation result is used as the final edge cloud resource allocation scheme.

[0018] In some schemes, the multi-layer cloud architecture intelligent transportation resource allocation method is described, wherein the fitness of each of the initial resource allocation schemes is evaluated to obtain the execution time and resource utilization of each edge cloud executing edge tasks, and the following function is used as the fitness evaluation function of the edge cloud resource allocation scheme:

[0019] F1=ω 11 ·D 11 +ω 12 ·(1-U 11 );

[0020] Among them, ω 11 Indicates the weight value of task completion time, D 11 represents the edge task, ω 12 represents the edge cloud resource utilization weight value, U 11 Indicates the resource utilization of the edge cloud.

[0021] In some schemes, the multi-layer cloud architecture intelligent transportation resource allocation method, the regional cloud uses a particle swarm optimization algorithm to allocate resources to the regional tasks it receives, including:

[0022] Divide the received regional tasks into a plurality of particle tasks, wherein in each particle task, some tasks are configured to the regional cloud, and the remaining tasks are allocated to the next level edge cloud of the regional cloud;

[0023] Obtain the fitness evaluation result of each particle task, and obtain the particle task with the lowest fitness evaluation result as the benchmark particle allocation plan;

[0024] The benchmark particle allocation scheme is updated and iterated according to the particle swarm optimization algorithm, and after the iteration is completed, the iteration result with the lowest fitness is selected as the final regional cloud resource allocation scheme.

[0025] In some schemes, the multi-layer cloud architecture intelligent transportation resource allocation method is described, wherein the fitness evaluation result of each particle task is obtained, and the particle task with the lowest fitness evaluation result is obtained as the benchmark particle allocation scheme, and the following function is used as the fitness evaluation function of the regional cloud resource allocation scheme:

[0026] F2=w 21 ·D 12 +w 22 ·(1-U 12 );

[0027] Among them, ω 21 Represents the weight value of the task delay time, D 12 represents the total delay of all regional tasks, ω 22 represents the total utilization weight value of the regional cloud and the corresponding edge cloud, U 12 Represents the total utilization of the regional cloud and the corresponding edge cloud.

[0028] In some schemes, the multi-layer cloud architecture intelligent transportation resource allocation method, the central cloud uses a hybrid genetic algorithm and a simulated annealing algorithm to allocate resources for the central tasks it receives, including:

[0029] A set of initial allocation schemes of central tasks are obtained by using genetic algorithm, in which different central tasks are allocated to the central cloud and the corresponding regional clouds;

[0030] The fitness of each initial allocation scheme is calculated by considering task delay time, resource utilization efficiency and load balancing conditions;

[0031] Select the initial allocation plan with the largest fitness result, perform degradation, mutation and iteration, and obtain the optimal initial allocation plan and the suboptimal allocation plan;

[0032] Performing a simulated annealing optimization algorithm on the optimal initial allocation scheme and the suboptimal allocation scheme to obtain an annealing optimized allocation scheme;

[0033] If the fitness difference between the annealing optimization allocation scheme and the optimal initial allocation scheme is less than a set threshold, the annealing optimization allocation scheme is used as the final central cloud resource allocation scheme.

[0034] In some schemes, the multi-layer cloud architecture intelligent transportation resource allocation method, the central cloud monitors the resource allocation results of the edge cloud and the regional cloud, and dynamically adjusts the resource allocation of the edge cloud and the regional cloud according to the monitoring results, including:

[0035] The central cloud obtains the traffic data collected in real time by the edge cloud and the regional cloud and the resource utilization of the edge cloud and the regional cloud;

[0036] evaluating whether the traffic condition is abnormal based on the traffic data;

[0037] Evaluate whether the resource utilization exceeds the set upper limit based on the resource utilization of the edge cloud and the regional cloud;

[0038] If the traffic condition is abnormal or the resource utilization rate exceeds the set upper limit, the resource allocation results of the edge cloud and the regional cloud will be readjusted.

[0039] In a second aspect, the technical solution of the present application provides a computer-readable storage medium on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the steps of the multi-layer cloud architecture intelligent transportation resource allocation method described in any technical solution of the first aspect are implemented.

[0040] In a third aspect, the technical solution of the present application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the multi-layer cloud architecture intelligent transportation resource allocation method described in any technical solution of the first aspect.

[0041] In a fourth aspect, the technical solution of the present application provides a computer device / equipment / system, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the multi-layer cloud architecture intelligent traffic resource allocation method described in any technical solution of the first aspect.

[0042] Compared with the prior art, the above technical solution provided by this application has the following technical effects:

[0043] The multi-layer cloud architecture intelligent transportation resource allocation method, storage medium and electronic device provided in this application adopt appropriate resource allocation algorithms in the edge cloud, regional cloud and central cloud respectively, and use the central cloud to achieve global collaborative optimization of resource allocation, thereby effectively improving resource utilization, reducing task completion time, and significantly improving the real-time response capability and overall performance of the intelligent transportation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of a multi-layer cloud architecture intelligent transportation resource allocation method according to an embodiment of the present application;

[0045] Figure 2 A block diagram of the scheduling principle of the multi-layer cloud architecture intelligent traffic resource allocation method according to an embodiment of the present application;

[0046] Figure 3 A schematic diagram of the hardware connection relationship of a computer device / equipment / system for executing a multi-layer cloud architecture intelligent traffic resource allocation method according to an embodiment of the present application. DETAILED DESCRIPTION

[0047] The specific implementation of the present application is further described below with reference to the accompanying drawings.

[0048] It is easy to understand that according to the technical solution of the present application, without changing the essential spirit of the present application, a variety of structural modes and implementation modes that can be replaced by those skilled in the art can be replaced with each other. Therefore, the following specific implementation modes and drawings are only exemplary descriptions of the technical solution of the present application, and should not be regarded as the entirety of the present application or as a limitation or restriction on the technical solution of the application.

[0049] The directional terms such as up, down, left, right, front, back, front, back, top, bottom, etc. mentioned or may be mentioned in this specification are defined relative to the structures shown in the drawings. They are relative concepts and may change accordingly according to different positions and different usage conditions. Therefore, these or other directional terms should not be interpreted as restrictive terms.

[0050] This embodiment provides a multi-layer cloud architecture intelligent traffic resource allocation method, such as Figure 1 and Figure 2 As shown, the multi-layer cloud architecture includes an edge cloud, a regional cloud and a central cloud, and the method includes:

[0051] S100: The edge cloud uses a genetic algorithm to allocate resources for edge tasks it receives.

[0052] In the edge cloud, a genetic algorithm (GA) is used to allocate resources for low-latency and high-real-time tasks. The genetic algorithm optimizes the task allocation scheme of the edge cloud through selection, crossover and mutation operations to ensure fast response of tasks.

[0053] S200: The regional cloud uses a particle swarm optimization algorithm to allocate resources for the regional tasks it receives.

[0054] The regional cloud uses the Particle Swarm Optimization (PSO) algorithm to optimize the allocation of tasks with medium latency requirements. The PSO algorithm quickly finds the optimal utilization plan of regional resources through collaborative search by particle swarms, thereby improving the processing efficiency of the regional cloud.

[0055] S300: The central cloud uses a hybrid genetic algorithm and a simulated annealing algorithm to allocate resources for the central tasks it receives.

[0056] For high-latency and complex computing tasks, the central cloud adopts a hybrid genetic algorithm and simulated annealing algorithm (HybridGA-SA), combining the global search capability of GA with the local optimization capability of SA to ensure efficient scheduling of complex tasks and optimal utilization of resources.

[0057] S400: The central cloud monitors the resource allocation results of the edge cloud and the regional cloud, and dynamically adjusts the resource allocation of the edge cloud and the regional cloud according to the monitoring results.

[0058] The central cloud regularly monitors the resource utilization of each layer of cloud, identifies load imbalance through a global collaborative optimization mechanism, dynamically adjusts task allocation, achieves multi-layer cloud load balancing and dynamic redistribution of resources, and improves the overall stability and response speed of the system.

[0059] The multi-layer cloud architecture intelligent transportation resource allocation method provided by the above solution uses genetic algorithm, particle swarm optimization algorithm and hybrid genetic-simulated annealing algorithm to achieve more accurate and efficient resource allocation and task scheduling for task scheduling and resource allocation at different levels according to the needs and characteristics of different levels in the multi-layer cloud architecture. In addition, the above solution realizes the unified management and optimization of resources and tasks at each layer in the multi-layer cloud architecture through the global collaborative optimization mechanism, which significantly improves the overall performance and responsiveness of the system.

[0060] Furthermore, in some solutions, the edge cloud in step S100 uses a genetic algorithm to allocate resources for the edge tasks it receives, including:

[0061] S101: Randomly assign the edge tasks to different edge clouds to obtain a set of initial resource allocation schemes.

[0062] S102: Evaluate the fitness of each of the initial resource allocation schemes to obtain the execution time and resource utilization of each edge cloud executing edge tasks. Preferably, the following function is used as the fitness evaluation function of the edge cloud resource allocation scheme:

[0063] F1=ω 11 ·D 11 +ω 12 ·(1-U 11 );

[0064] Among them, ω 11 Indicates the weight value of task completion time, D 11 represents the edge task, ω 12represents the edge cloud resource utilization weight value, U 11 Indicates the resource utilization of the edge cloud.

[0065] S103: Selecting the initial resource allocation scheme with the largest fitness evaluation result as the benchmark allocation scheme.

[0066] S104: performing crossover, iteration, mutation and iteration processing on the reference allocation scheme to obtain a plurality of variant allocation schemes.

[0067] S105: Evaluate the fitness of each of the variant allocation schemes, and use the variant allocation scheme with the largest fitness evaluation result as the final edge cloud resource allocation scheme.

[0068] The resource allocation process of the above edge cloud is explained below through an example.

[0069] The following edge tasks and their requirements are set:

[0070] Task T11: Traffic monitoring (requirements: 2CPU, 1GB memory);

[0071] Task T12: Traffic light control (requirements: 1CPU, 0.5GB memory);

[0072] Task T13: Data analysis (requirements: 2CPU, 2GB memory);

[0073] Task T14: Accident detection (requirements: 1CPU, 1GB memory);

[0074] Task T15: Event prediction (requirements: 1CPU, 2GB memory).

[0075] Resource availability of edge cloud:

[0076] Edge Cloud 1 : 4CPU, 4GB memory;

[0077] Edge Cloud 2 : 3CPU, 2GB memory;

[0078] Edge Cloud 3 : 2CPU, 3GB memory;

[0079] The process of resource allocation using genetic algorithm:

[0080] Population initialization:

[0081] A set of initial resource allocation schemes (chromosomes) is randomly generated, that is, each chromosome represents an initial resource allocation scheme. For example, suppose there are 5 chromosomes:

[0082] Chromosome 1: T11→e1 ,T12→e 1 ,T13→e 2 ,T14→e 1 ,T15→e 3 ;

[0083] Chromosome 2: T11→e 2 ,T12→e 1 ,T13→e 1 ,T14→e 3 ,T15→e 1 ;

[0084] Chromosome 3: T11→e 1 ,T12→e 2 ,T13→e 2 ,T14→e 1 ,T15→e 3 ;

[0085] Chromosome 4: T11→e 3 ,T12→e 2 ,T13→e 1 ,T14→e 2 ,T15→e 1 ;

[0086] Chromosome 5: T11→e 1 ,T12→e 3 ,T13→e 1 ,T14→e 3 ,T15→e 2 .

[0087] The fitness of each of the initial resource allocation schemes is evaluated, that is, the fitness of each chromosome is evaluated. The function used for fitness evaluation should take into account the execution time of the task, resource utilization, etc. In this application, F1=ω is used. 11 ·D 11 +ω 12 ·(1-U 11 ) as the fitness evaluation function of the edge cloud resource allocation scheme. For example: Chromosome 1 will lead to D 11 =50ms, U 11 =0.8, the fitness is:

[0088] F1=1·50+2·(1-0.8)=50+0.4=50.4;

[0089] The tournament selection method is used to select chromosomes with high fitness as parents, that is, the initial resource allocation scheme with the largest fitness evaluation result is selected as the benchmark allocation scheme, for example, chromosome 1 and chromosome 3. After crossover, iteration, mutation and iteration processing of the benchmark allocation scheme, several variant allocation schemes are obtained, including:

[0090] Crossover: Generate new offspring chromosomes through crossover operation. For example, crossover chromosome 1 and chromosome 3 to generate a new chromosome: T11→e 1 ,T12→e 1 ,T13→e 2 ,T14→e 3 ,T15→e 3 .

[0091] Mutation: Randomly mutate the task allocation in some sub-chromosomes. For example, if the allocation of a task is changed, the corresponding sub-chromosome will mutate to: T11→e 2 ,T12→e 1 ,T13→e 1 ,T14→e 3 ,T15→e 3 ;

[0092] Iteration: Repeat the fitness evaluation, selection, crossover and mutation process until the maximum number of iterations is reached or the fitness no longer changes significantly (i.e., the fitness has reached the maximum). Once the maximum number of iterations is reached or the fitness is stable, the iteration stops. Select the chromosome with the highest fitness from the last generation as the final task allocation solution. Assume that the chromosome finally selected is: T11→e 1 ,T12→e 1 ,T13→e 2 ,T14→e 1 ,T15→e 3 .

[0093] Output resource usage: Update edge cloud resources based on the final task allocation plan:

[0094] Edge Cloud 1 : Assign tasks T11, T12, T14; use: 2CPU+1CPU+1CPU=4CPU; use memory: 1GB+0.5GB+1GB=2.5GB;

[0095] Edge Cloud 2 : Assign task T13; Use: 2CPU; Use memory: 2GB;

[0096] Edge Cloud 3 : Assign task T15; Use: 1CPU; Use memory: 2GB.

[0097] Finally, all tasks T11 to T15 are successfully assigned to edge cloud nodes, ensuring that resources are maximized while meeting the needs of the tasks.

[0098] Further preferably, in some schemes of the multi-layer cloud architecture intelligent transportation resource allocation method, the regional cloud in step S200 uses a particle swarm optimization algorithm to allocate resources to the regional tasks it receives, including:

[0099] S201: Divide the received regional task into a plurality of particle tasks, wherein in each particle task, a number of tasks are configured to the regional cloud, and the remaining tasks are allocated to the next level edge cloud of the regional cloud.

[0100] S202: Obtain the fitness evaluation result of each particle task, and obtain the particle task with the lowest fitness evaluation result as the benchmark particle allocation plan. The following function is used as the fitness evaluation function of the regional cloud resource allocation plan:

[0101] F2=w 21 ·D 12 +w 22 ·(1-U 12 );

[0102] Among them, ω 21 Represents the weight value of the task delay time, D 12 represents the total delay of all regional tasks, ω 22 represents the total utilization weight value of the regional cloud and the corresponding edge cloud, U 12 Represents the total utilization of the regional cloud and the corresponding edge cloud.

[0103] S203: updating and iterating the reference particle allocation scheme according to the particle swarm optimization algorithm, and selecting the iteration result with the lowest fitness as the final regional cloud resource allocation scheme after the iteration is completed.

[0104] The resource allocation process of the above regional cloud is explained below through an example.

[0105] Set the system environment of the regional cloud to:

[0106] Regional cloud: Available resources: 5CPU, 10GB memory; Edge cloud e 1 : Available resources: 3CPU, 4GB memory; edge cloud e 2 : Available resources: 4CPU, 6GB memory.

[0107] Assume that the regional tasks include three, and each task has specific resource requirements and latency requirements:

[0108] Task T21: Real-time video processing (requirements: 3CPU, 5GB memory, delay requirement: 20ms);

[0109] Task T22: Image recognition analysis (requirements: 2CPU, 2GB memory, latency requirement: 15ms);

[0110] Task T23: Data flow monitoring (requirements: 2CPU, 3GB memory, delay requirement: 25ms).

[0111] Particle swarm optimization process:

[0112] Initialize the particle swarm: that is, different regional task allocation schemes. Each particle represents a possible regional task allocation scheme, for example:

[0113] Particle 1: T21→regional cloud, T22→e 1 ,T23→e 2 ;

[0114] Particle 2: T21→e 1 ,T22→Regional Cloud,T23→e 2 ;

[0115] Particle 3: T21→e 2 ,T22→e 1 ,T23→Regional cloud;

[0116] Particle 4: T21→regional cloud, T22→regional cloud, T23→e 1 ;

[0117] Particle 5: T21→e 2 ,T22→Regional Cloud,T23→e 1 ;

[0118] The fitness of each of the initial resource allocation schemes is evaluated to obtain the execution time and resource utilization of each edge cloud executing edge tasks: the fitness of each particle is calculated, considering the delay and resource utilization, with F1 = ω 11 ·D 11 +ω 12 ·(1-U 11 ) is used as the fitness evaluation function of the edge cloud resource allocation scheme. The calculation example is as follows:

[0119] Resource allocation of particle 1: T21 in the regional cloud: 3 CPUs, 5 GB memory; T22 in e 1 : Using 2CPU, 2GB memory; T23 in e 2 : Use 2CPU, 3GB memory.

[0120] Get regional cloud utilization:

[0121]

[0122] Get Edge Cloud 1 Utilization:

[0123]

[0124] Edge Cloud 2 Utilization:

[0125]

[0126]

[0127] Get total delay: 20ms+15ms+25ms=60ms;

[0128] The final fitness result is:

[0129] F2(particle 1)=1·60+2·(1-1.091)=60-0.182≈59.82.

[0130] Update and iteration: Use the PSO algorithm to update the speed and position of the particles. Each particle adjusts its direction to explore a more optimized task allocation. The PSO algorithm can be implemented using existing algorithms.

[0131] Check and correct: Check the task allocation results obtained after each round of iteration as candidate solutions to ensure that resource usage does not exceed the limit, and select the optimal solution as the current task allocation result according to the fitness value.

[0132] Achieve optimal decision: After multiple iterations, select the task allocation result with the lowest fitness as the final task allocation solution. For example, suppose particle 3 becomes the optimal solution after multiple rounds of updates.

[0133] The allocation of particle 3 is:

[0134] Mission T21 in edge cloud 2 ,Conditions met: sufficient resources (using 3CPU, 5GB),meet this edge cloud configuration;

[0135] Task T2 in edge cloud 1 , meet the conditions: sufficient resources (using 2CPU, 2GB);

[0136] Task T3 is in the regional cloud and meets the following conditions: sufficient resources (2 CPUs, 3 GB).

[0137] After PSO optimization iterations, a solution suitable for task resource allocation was finally found by reasonably calculating the trade-off between resource load and delay, which not only improved resource utilization but also maintained task execution efficiency.

[0138] Further preferably, in the multi-layer cloud architecture intelligent transportation resource allocation method, in step S300, the central cloud uses a hybrid genetic algorithm and a simulated annealing algorithm to allocate resources for the central tasks it receives, including:

[0139] S301: A set of initial allocation schemes for central tasks is obtained by using a genetic algorithm, in which different central tasks are allocated to a central cloud and corresponding regional clouds.

[0140] S302: Calculate the fitness of each initial allocation scheme by considering task delay time, resource utilization efficiency and load balancing conditions.

[0141] S303: Select the initial allocation plan with the largest fitness result, perform degradation, mutation and iteration, and obtain the optimal initial allocation plan and the suboptimal allocation plan.

[0142] S304: Performing a simulated annealing optimization algorithm on the optimal initial allocation scheme and the suboptimal allocation scheme to obtain an annealing optimized allocation scheme.

[0143] S305: If the fitness difference between the annealing optimization allocation scheme and the optimal initial allocation scheme is less than a set threshold, the annealing optimization allocation scheme is used as the final central cloud resource allocation scheme. When the fitness difference is less than the set threshold, it means that the fitness change is very small and it can be considered that there is basically no change.

[0144] The following example illustrates the resource allocation process of the central cloud. In the following example, the system environment is set as follows:

[0145] Central cloud: mainly handles complex computing tasks; available resources: 16CPU, 32GB memory.

[0146] Regional cloud R1: handles minor complex tasks or tasks that are transferred due to insufficient central resources; available resources: 10 CPU, 20 GB memory.

[0147] Regional cloud R2: also supports tasks that are insufficient for processing in the central cloud; available resources: 8 CPUs, 16 GB memory.

[0148] Setting the central task includes four complex tasks that need to be scheduled:

[0149] Task T31: Traffic flow prediction simulation (8CPU, 10GB memory, high latency tolerance).

[0150] Task T32: Domain Big Data Analysis (6CPU, 12GB memory, medium latency acceptance).

[0151] Task T33: Roadside perception data quality analysis (5CPU, 8GB memory, low latency).

[0152] Task T34: Automotive UBI evaluation (5CPU, 6GB memory, medium latency).

[0153] The hybrid optimization algorithm process includes:

[0154] Initialization and encoding: The initial state is performed using a genetic algorithm;

[0155] Chromosome encoding: Each chromosome represents a task allocation scheme (a combination of the central cloud and the regional cloud). For example, a chromosome is represented as a four-tuple (C, C, R1, R2), indicating that tasks T31 and T32 are allocated to the central cloud, T3 to the regional cloud R1, and T4 to the regional cloud R2.

[0156] Population initialization: Randomly generate multiple feasible chromosomes based on resource conditions.

[0157] Fitness calculation: Considering task delay, resource utilization efficiency and load balancing, the following function is used to determine the fitness result:

[0158] Fitness(i)=w 31 (1-Delay Factor(i))+w 32 Resource Utilization Factor(i);

[0159] Among them, Delay Factor is used to consider whether the task is satisfied within the acceptable delay range;

[0160] Resource Utilization Factor is used to measure resource utilization (considering the ratio of allocation and actual use).

[0161] w 31 and w 32 It is the corresponding weight value, which is set according to the demand.

[0162] Genetic algorithm execution, including:

[0163] Selection: Use tournament selection method to select the best;

[0164] Crossover: Perform a crossover operation (single-point or multi-point crossover) on the selected chromosomes to produce new offspring.

[0165] Mutation: Change certain genes with a small probability (change in task allocation) to introduce diversity.

[0166] Evolutionary updating includes: after several generations (generally set to dozens of generations), the global optimal solution is continuously approached and the optimal solution at the current stage is retained; multiple suboptimal solutions are determined for further refinement and optimization by the simulated annealing process.

[0167] Simulated annealing optimization, including:

[0168] Initial solution selection: Select the best chromosome in the genetic algorithm stage as the initial solution;

[0169] Temperature setting and annealing strategy: Set the initial temperature and use the annealing strategy to gradually reduce the temperature. The exponential temperature decay method is usually used.

[0170] Neighborhood search: Randomly sample surrounding allocations as new solutions, such as slightly adjusting the allocation location of a task.

[0171] Probability acceptance function: According to Determine whether to accept a new task allocation plan. Even if the new plan has slightly worse performance, it is still accepted with a certain probability to avoid local optimality.

[0172] Termination condition: If the minimum temperature is reached or the fitness does not increase significantly, the process will terminate.

[0173] Overall resource allocation result: Through the above process, the optimal resource allocation for the collaboration between the central cloud and the regional cloud is finally obtained as follows:

[0174] Tasks T31 and T32 are executed in the central cloud: the central cloud can bear high loads and handle complex tasks. Task T33 is transferred to regional cloud R1: due to latency and resource requirements, it is more efficient to process in the regional cloud. Task T34 is transferred to regional cloud R2: according to the load balancing needs, the pressure on the central cloud can be reasonably reduced.

[0175] The above hybrid optimization combining genetic algorithm and simulated annealing provides global optimal scheduling management for complex tasks, taking into account both efficient computing and smooth resource coordination, and solving the challenge of insufficient central cloud resources.

[0176] The above examples prove that this solution can take into account both task completion time and resource utilization. The comprehensive optimization objective function adopted at each level further ensures the balanced optimization of the system in many aspects.

[0177] Further preferably, in the multi-layer cloud architecture intelligent transportation resource allocation method, in step S400, the central cloud monitors the resource allocation results of the edge cloud and the regional cloud, and dynamically adjusts the resource allocation of the edge cloud and the regional cloud according to the monitoring results, including:

[0178] S401: The central cloud obtains the traffic data collected in real time by the edge cloud and the regional cloud and the resource utilization of the edge cloud and the regional cloud.

[0179] S402: Evaluate whether the traffic condition is abnormal based on the traffic data.

[0180] S403: Evaluate whether the resource utilization rate exceeds a set upper limit based on the resource utilization rates of the edge cloud and the regional cloud.

[0181] S404: If the traffic condition is abnormal or the resource utilization rate exceeds the set upper limit, the resource allocation results of the edge cloud and the regional cloud are readjusted. The abnormal traffic condition may be congestion, and the set upper limit may be confirmed based on empirical values, such as 98%, 99%, etc.

[0182] The following example illustrates the dynamic adjustment process of the central cloud. i Processing real-time traffic monitoring and signal control, regional cloud k Optimize traffic analysis and task allocation with medium latency, while the central cloud is responsible for task scheduling with high latency and complex calculations. These three levels of cloud computing structure need to be effectively globally coordinated and optimized to adapt to changes in real-time traffic conditions.

[0183] The specific steps of dynamic adjustment include:

[0184] Real-time data collection and communication: The central cloud regularly collects real-time data from various regional clouds and edge clouds, including: 1 ,e 2 ,e 3 :Real-time traffic flow data, signal light status, accident detection information. Regional Cloud k : Traffic analysis results obtained from edge clouds, task allocation and resource utilization of each edge cloud.

[0185] Situation analysis, including:

[0186] The central cloud analyzes the collected data and evaluates the current traffic conditions, for example, whether the traffic flow is abnormal (for example, a sudden congestion in a certain area). Analyze whether the resource utilization of the edge cloud is too high, resulting in delayed task response. At this stage, data analysis and machine learning techniques can be used to predict future traffic flow and identify potential congestion points.

[0187] Task reallocation strategy development, including:

[0188] Based on the analysis results, the central cloud decides whether to adjust the task allocation strategy. For example, if it is found that the edge cloud 1 The resource utilization rate of e 2 and e 3 The utilization rate is low, so you can consider 1 Transfer some tasks. If the area cloud r k The traffic forecast shows that a major road is about to be congested. The central cloud needs to give priority to the traffic flow forecast tasks related to this, and may need to reallocate tasks within the region.

[0189] Global load balancing and resource reallocation, including:

[0190] Edge cloud adjustment: For example, 1 Task T processed 1 Partially transferred to edge cloud 2 This adjustment may need to be achieved by splitting and rearranging tasks, for example: 1 Part of the calculation is transferred to e 2 , to reduce e 1 of load.

[0191] Regional Cloud Adjustment: Regional Cloud k The task scheduling strategy needs to be adjusted accordingly to ensure that the overall system still maintains low latency. For example, the coordination strategy of traffic lights needs to be adjusted to better cope with the upcoming traffic flow, rather than just reacting based on static data.

[0192] In specific implementation, the following multi-objective optimization model can be used to perform the above dynamic adjustment. The multi-objective optimization model takes resource utilization U and task completion time D as the two main optimization objectives, and its objective function is designed as follows:

[0193] Minimize Z = α*D + β*(1-U);

[0194] in:

[0195] D: Task completion time, the goal is to minimize it.

[0196] 1-U: reflects the underutilization of resources, and the goal is to minimize underutilized resources.

[0197] α and β: weight coefficients, which adjust the importance of different optimization objectives according to actual needs.

[0198] Minimize D: Ensure that tasks can be completed in the shortest time possible and improve the system's responsiveness.

[0199] Minimize (1-U): that is, maximize U to ensure that resources are fully utilized and avoid resource waste.

[0200] Through this design, the model can shorten the task completion time while maximizing the efficiency of resource utilization, achieving a balance between the two.

[0201] The periodic communication and collaborative optimization between the central cloud, regional cloud and edge cloud form a closed loop. The high efficiency of information flow ensures the rapid response of decision-making. By continuously collecting new data feedback and evaluation, iteratively improving task allocation and resource utilization strategies, we ensure that the system continues to operate efficiently.

[0202] The above algorithm framework of this application has the ability of dynamic monitoring and adjustment, can respond to traffic conditions and resource changes in real time, improves the robustness and flexibility of the system, provides significant technical improvements in intelligent traffic resource allocation and task scheduling under a multi-layer cloud architecture, solves the shortcomings of existing technologies, and has high innovation and practicality.

[0203] An embodiment of the present application also provides a computer-readable storage medium having a computer program / instruction stored thereon, characterized in that when the computer program / instruction is executed by a processor, the steps of the multi-layer cloud architecture intelligent transportation resource allocation method described in any one of the above method embodiments are implemented.

[0204] An embodiment of the present application also provides a computer program product, including a computer program / instruction, characterized in that when the computer program / instruction is executed by a processor, the steps of the multi-layer cloud architecture intelligent transportation resource allocation method described in any one of the above method embodiments are implemented.

[0205] A computer device / equipment / system, such as Figure 3As shown, the electronic device includes at least one processor 31 and at least one memory 32, at least one of the memories 32 stores program information, and at least one of the processors 31 reads the program information and executes the multi-layer cloud architecture intelligent traffic resource allocation method described in any of the above method embodiments. The device may also include: an input device 33 and an output device 34. The processor 31, the memory 32, the input device 33 and the output device 34 can be connected in communication. The memory 32, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The processor 31 executes various functional applications and data processing by running the non-volatile software programs, instructions and modules stored in the memory 32, that is, the multi-layer cloud architecture intelligent traffic resource allocation method provided by any of the above schemes is implemented. The memory 32 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required by at least one function; the data storage area may store data created according to the use of the multi-layer cloud architecture intelligent traffic resource allocation method, etc. In addition, the memory 32 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 32 may optionally include a memory remotely arranged relative to the processor 31, and these remote memories may be connected to a device for executing the multi-layer cloud architecture intelligent traffic resource allocation method through a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The input device 33 may receive input user clicks, and generate signal inputs related to user settings and function controls of the multi-layer cloud architecture intelligent traffic resource allocation method. The output device 34 may include a display device such as a display screen. When the one or more modules are stored in the memory 32, when the one or more processors 31 are run, the multi-layer cloud architecture intelligent traffic resource allocation method in any of the above-mentioned method embodiments is executed.

[0206] As needed, the above technical solutions can be combined to achieve the best technical effect.

[0207] The above are only the principles and preferred embodiments of the present application. It should be noted that, for ordinary technicians in this field, on the basis of the principles of the present application, several other modifications can be made, which should also be considered as the protection scope of the present application.

Claims

1. A multi-layer cloud architecture intelligent traffic resource allocation method, characterized in that: The multi-layer cloud architecture includes an edge cloud, a regional cloud and a central cloud, and the method includes: The edge cloud uses a genetic algorithm to allocate resources for edge tasks it receives; The regional cloud uses a particle swarm optimization algorithm to allocate resources for the regional tasks it receives; The central cloud uses a hybrid genetic algorithm and a simulated annealing algorithm to allocate resources for the central tasks it receives; The central cloud monitors the resource allocation results of the edge cloud and the regional cloud, and dynamically adjusts the resource allocation of the edge cloud and the regional cloud according to the monitoring results.

2. The multi-layer cloud architecture intelligent traffic resource allocation method according to claim 1, characterized in that: The edge cloud uses a genetic algorithm to allocate resources for edge tasks received by it, including: Randomly assigning the edge tasks to different edge clouds to obtain a set of initial resource allocation schemes; Evaluate the fitness of each of the initial resource allocation schemes to obtain the execution time and resource utilization of each edge cloud executing edge tasks; Select the initial resource allocation plan with the largest fitness evaluation result as the benchmark allocation plan; After crossover, iteration, mutation and iteration processing are performed on the benchmark allocation scheme, several variant allocation schemes are obtained; The fitness of each of the variant allocation schemes is evaluated, and the variant allocation scheme with the largest fitness evaluation result is used as the final edge cloud resource allocation scheme.

3. The multi-layer cloud architecture intelligent traffic resource allocation method according to claim 2, characterized in that: The fitness of each of the initial resource allocation schemes is evaluated to obtain the execution time and resource utilization of each edge cloud executing the edge task, and the following function is used as the fitness evaluation function of the edge cloud resource allocation scheme: F1=ω 11 ·D 11 +ω 12 ·(1-U 11 ); Among them, ω 11 represents the weight value of task completion time, D 11 represents the edge task, ω 12 represents the edge cloud resource utilization weight value, U 11 Indicates the resource utilization of the edge cloud.

4. The multi-layer cloud architecture intelligent traffic resource allocation method according to claim 1, characterized in that: The regional cloud uses a particle swarm optimization algorithm to allocate resources for the regional tasks it receives, including: Divide the received regional tasks into a plurality of particle tasks, wherein in each particle task, some tasks are configured to the regional cloud, and the remaining tasks are allocated to the next-level edge cloud of the regional cloud; Obtain the fitness evaluation result of each particle task, and obtain the particle task with the lowest fitness evaluation result as the benchmark particle allocation plan; The benchmark particle allocation scheme is updated and iterated according to the particle swarm optimization algorithm, and after the iteration is completed, the iteration result with the lowest fitness is selected as the final regional cloud resource allocation scheme.

5. The multi-layer cloud architecture intelligent traffic resource allocation method according to claim 4, characterized in that: The fitness evaluation result of each particle task is obtained, and the particle task with the lowest fitness evaluation result is obtained as the benchmark particle allocation scheme, and the following function is used as the fitness evaluation function of the regional cloud resource allocation scheme: F2=w 21 ·D 12 +w 22 ·(1-U 12 ); Among them, ω 21 Represents the weight value of the task delay time, D 12 represents the total delay of all regional tasks, ω 22 represents the total utilization weight value of the regional cloud and the corresponding edge cloud, U 12 Represents the total utilization of the regional cloud and the corresponding edge cloud.

6. The multi-layer cloud architecture intelligent traffic resource allocation method according to claim 1, characterized in that: The central cloud uses a hybrid genetic algorithm and a simulated annealing algorithm to allocate resources for the central tasks it receives, including: A set of initial allocation schemes of central tasks are obtained by using genetic algorithm, in which different central tasks are allocated to the central cloud and the corresponding regional clouds; The fitness of each initial allocation scheme is calculated by considering task delay time, resource utilization efficiency and load balancing conditions; Select the initial allocation plan with the largest fitness result, perform degradation, mutation and iteration, and obtain the optimal initial allocation plan and the suboptimal allocation plan; Performing a simulated annealing optimization algorithm on the optimal initial allocation scheme and the suboptimal allocation scheme to obtain an annealing optimized allocation scheme; If the fitness difference between the annealing optimization allocation scheme and the optimal initial allocation scheme is less than a set threshold, the annealing optimization allocation scheme is used as the final central cloud resource allocation scheme.

7. The multi-layer cloud architecture intelligent traffic resource allocation method according to any one of claims 1 to 6, characterized in that: The central cloud monitors the resource allocation results of the edge cloud and the regional cloud, and dynamically adjusts the resource allocation of the edge cloud and the regional cloud according to the monitoring results, including: The central cloud obtains the traffic data collected in real time by the edge cloud and the regional cloud and the resource utilization of the edge cloud and the regional cloud; evaluating whether the traffic condition is abnormal based on the traffic data; Evaluate whether the resource utilization exceeds the set upper limit based on the resource utilization of the edge cloud and the regional cloud; If the traffic condition is abnormal or the resource utilization rate exceeds the set upper limit, the resource allocation results of the edge cloud and the regional cloud will be readjusted.

8. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the steps of the multi-layer cloud architecture intelligent transportation resource allocation method described in any one of claims 1-7 are implemented.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the steps of the multi-layer cloud architecture intelligent transportation resource allocation method described in any one of claims 1-7 are implemented.

10. A computer device / equipment / system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the multi-layer cloud architecture intelligent traffic resource allocation method according to any one of claims 1 to 7.

Images