A method for task offloading and resource allocation in connected car scenarios

Abstract

The present invention provides a method for task offloading and resource allocation in a connected vehicle scenario, comprising: step S1, constructing a mobile edge computing system model for the Internet of Vehicles to receive semantic information tasks sent by each user vehicle; step S2, constructing a priority evaluation model to evaluate each semantic information task based on task urgency and task importance to obtain a task priority; step S3, processing each semantic information task to obtain a task processing result and transmitting it back to the corresponding user vehicle; step S4, constructing a comprehensive utility function as the fitness function of the model; step S5, introducing a particle swarm algorithm to define particles to characterize bandwidth and task allocation, and introducing an inertia factor and a learning factor to dynamically regulate the particle iterative update process to guide the particles to search for the optimal solution and obtain the best task offloading and resource allocation solution. The beneficial effect is that the present invention can achieve efficient optimization of task offloading and resource allocation, and improve the overall performance and reliability of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle networking communications, and in particular to a method for task offloading and resource allocation in a connected vehicle scenario. Background Art

[0002] With the rapid development of Internet of Vehicles (IoV) technology, vehicles have transformed from simple transportation tools into intelligent mobile terminals. IoV relies on vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications to complete perception, computing, and decision-making tasks. However, existing communication and computing resource allocation methods face challenges when dealing with computationally intensive and latency-sensitive tasks, mainly manifested in the following aspects:

[0003] 1. Limitations of traditional communication methods

[0004] Existing communication methods based on bit-level information transmission require high bandwidth for large data volumes and cannot effectively meet the time-sensitive and high-accuracy requirements of connected vehicles. Long-distance transmission to centralized cloud servers often reduces the quality of service (QoS) of the task due to large transmission delays.

[0005] 2. Deployment limitations of edge computing

[0006] Although roadside unit (RSU) edge computing servers have been applied to offloading tasks, edge server resources are limited and are prone to congestion under high task loads. Overloading of edge servers can lead to a decline in the overall performance of the system.

[0007] 3. Simplification of task scheduling priority

[0008] In existing research, task priority is usually determined only based on arrival time or maximum tolerable delay, ignoring the importance and urgency of the task. This may result in high-priority tasks not receiving a timely response, affecting user experience and the overall utility of the system. Summary of the Invention

[0009] The technical problem to be solved by the present invention is to achieve efficient optimization of task offloading and resource allocation, and improve the overall performance and reliability of the system. In order to overcome the defects of the above-mentioned existing technologies (or related technologies), the present invention provides a task offloading and resource allocation method in a connected car scenario.

[0010] The present invention provides a method for task offloading and resource allocation in a connected vehicle scenario, comprising:

[0011] Step S1: constructing a connected vehicle mobile edge computing system model in a connected vehicle scenario, wherein the connected vehicle mobile edge computing system model includes a roadside unit, multiple user vehicles, and multiple idle vehicles, and receiving semantic information tasks sent by each user vehicle through the roadside unit;

[0012] Step S2: constructing a priority evaluation model to define the task urgency and task importance of each of the semantic information tasks, and based on the task urgency and task importance, evaluating each of the semantic information tasks using the priority evaluation model to obtain a corresponding task priority;

[0013] Step S3, processing each of the semantic information tasks in order based on the task priority, dividing each of the semantic information tasks into multiple subtasks and assigning them to each of the idle vehicles and the roadside units for processing, obtaining corresponding task processing results, and transmitting them back to the user vehicle corresponding to the semantic information task;

[0014] Step S4, obtaining the delay data, semantic accuracy data, and social contribution rate during the processing of each of the semantic information tasks in step S3, constructing a comprehensive utility function based on the task priority, the delay data, the semantic accuracy data, and the social contribution rate of each of the semantic information tasks as the fitness function of the IoV mobile edge computing system model to quantify task offloading and resource allocation;

[0015] In step S5, a particle swarm algorithm is introduced to define particles to characterize bandwidth and task allocation, and an inertia factor and a learning factor are introduced to dynamically regulate the particle iterative update process to guide the particles to search for the optimal solution, obtain the best task offloading and resource allocation solution, and maximize the fitness function.

[0016] Compared with the prior art, the task offloading and resource allocation method in the connected car scenario of the present invention has the following advantages:

[0017] In the present invention, step S1 is used to construct a mobile edge computing system model of the Internet of Vehicles and collect semantic information tasks, step S2 is used to construct a priority evaluation model and evaluate task priorities, step S3 is used to allocate and process semantic information tasks, step S4 is used to construct a comprehensive utility function, and step S5 is used to introduce a particle swarm algorithm, an inertia factor, and a learning factor, and search and solve the optimal task offloading and resource allocation scheme. The importance and urgency of the semantic information tasks are taken into consideration, so that resource allocation is tilted towards semantic information tasks with greater social contribution and importance rewards, effectively improving system efficiency and reliability. In addition, the present invention optimizes the inertia weight parameters and learning factors of the traditional particle swarm algorithm, so that the optimization capability is improved, and efficient optimization of task offloading and resource allocation is achieved.

[0018] In a possible implementation, in step S1, each of the user vehicles collects surrounding environment information, data volume, computing power, and maximum delay through an on-board device and generates the semantic information task after semantic encoding.

[0019] In a possible implementation, in step S2, the task priority is obtained by the following calculation formula:

[0020] ;

[0021] in,

[0022] Indicates the priority of the task;

[0023] Indicate the importance of the task;

[0024] A weight indicating the importance of the task;

[0025] Indicate the urgency of the task;

[0026] A weight representing the urgency of the task.

[0027] In a possible implementation, in step S3, the total number of the idle vehicles is M, the semantic information task is decoded and divided equally into M+1 subtasks, the M subtasks are randomly assigned to each idle vehicle for processing to obtain a corresponding first subresult, the remaining subtask is processed by the roadside unit to obtain a corresponding second subresult, and then the first subresults and the second subresults are merged to form the task processing result.

[0028] In a possible implementation, in step S4, the comprehensive utility function is constructed by the following expression:

[0029] ;

[0030] in,

[0031] represents the comprehensive utility function;

[0032] Indicates preset parameters;

[0033] Indicates the delay data;

[0034] Indicates the priority of the task;

[0035] representing said semantic accuracy data;

[0036] Indicates the social contribution rate.

[0037] In a possible implementation, step S5 includes:

[0038] Step S51, introduce the particle swarm algorithm, define particles to represent bandwidth and task allocation, set the number of particles to be L, particle number ∈{1,…,L}, the position of each particle is randomly generated in the initial state and represented as a matrix with N rows and (1+M+1) columns, and the idle vehicle assigned to the task is defined as a task vehicle, wherein the first column of the matrix represents the bandwidth allocated to all task vehicles, and the remaining M+1 columns represent the task allocation ratio of the M idle vehicles and the roadside units;

[0039] Step S52: In the particle iterative optimization process, the inertia factor and the learning factor are introduced to collaboratively control and update the particle velocity, wherein the learning factor includes a first learning factor for describing the particle's own cognitive experience and a second learning factor for describing the particle's social experience;

[0040] Step S53, during the particle iterative optimization process, obtaining and comparing the particle fitness values ​​in each iterative optimization process to dynamically adjust the first learning factor and the second learning factor;

[0041] Step S54: determine whether the particle iterative optimization process meets the preset cutoff condition:

[0042] If so, the optimal particle position is determined and the best task offloading and resource allocation scheme is determined;

[0043] If not, return to step S51.

[0044] In a possible implementation, in step S51, the first The particle Position:

[0045] ;

[0046] in,

[0047] Indicates the The particle location;

[0048] Indicates the The bandwidth allocated to each of the task vehicles during the iteration;

[0049] Indicates the The task allocation ratio of the M idle vehicles in the iteration;

[0050] Indicates the The task allocation ratio of the roadside unit at the iteration.

[0051] In a possible implementation, in step S52, the particle velocity is updated using the following calculation formula:

[0052] ;

[0053] in,

[0054] Indicates the The particle velocity at the iteration;

[0055] Indicates the The particle velocity at the iteration;

[0056] Indicates the The inertia factor at the iteration;

[0057] Indicates the The first learning factor at the iteration;

[0058] Represents a function that generates random numbers uniformly distributed between [0, 1];

[0059] Indicates the The current optimal solution of the particle at the iteration;

[0060] Indicates the The second learning factor at the iteration;

[0061] Indicates the The global optimal solution of particles at the iteration.

[0062] In a possible implementation, in step S53, the first learning factor is adjusted using the following calculation formula:

[0063] ;

[0064] in,

[0065] Indicates the The first learning factor at the iteration;

[0066] Indicates the The first learning factor at the iteration;

[0067] Indicates the preset adjustable coefficient;

[0068] Indicates the The fitness of the particle's current optimal solution at the iteration;

[0069] Indicates the The fitness of the particle at the iteration.

[0070] In a possible implementation, in step S53, the second learning factor is adjusted using the following calculation formula:

[0071] ;

[0072] in,

[0073] Indicates the The second learning factor at the iteration;

[0074] Indicates the The second learning factor at the iteration;

[0075] Indicates the preset adjustable coefficient;

[0076] Indicates the The fitness of the particle's current optimal solution at the iteration;

[0077] Indicates the The fitness of the particle at the iteration. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 is a flow chart of the steps of the present invention;

[0079] Figure 2 This is a structural diagram of the vehicle network mobile edge computing system model of the present invention;

[0080] Figure 3 Schematic diagram comparing the utility values ​​of priority models at different system bandwidths of the present invention;

[0081] Figure 4 Schematic diagram comparing the utility values ​​of various algorithms for different numbers of idle vehicles of the present invention. DETAILED DESCRIPTION

[0082] First, those skilled in the art should understand that these embodiments are merely for explaining the technical principles of the embodiments of the present invention and are not intended to limit the scope of protection of the embodiments of the present invention. Those skilled in the art may make adjustments as needed to adapt to specific application scenarios.

[0083] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0084] See also Figure 1 , an embodiment of the present invention discloses a method for task offloading and resource allocation in a connected vehicle scenario, comprising:

[0085] Step S1, construct a vehicle network mobile edge computing system model including a single roadside unit (RSU), multiple user vehicles (UVs) and idle vehicles (IVs), such as Figure 2 As shown in the figure, the user vehicle collects surrounding environment information through the on-board equipment, generates a semantic information task (SIT) containing task attributes (such as surrounding environment information, data volume, computing amount, maximum delay, etc.) after semantic encoding, and offloads the semantic information task to the roadside unit through the V2I link;

[0086] Step S2: Design a priority assessment model that comprehensively considers the weights of task importance and urgency. By defining task urgency (related to delay constraints and computational effort) and task importance (related to vehicle and task type), determine task priority, provide a basis for subsequent resource allocation, and ensure that critical tasks are handled first. The calculation formula for task priority is: ,in, 、 Represents the importance and urgency of the task, 、 Representing their respective weights, the roadside unit determines the task priority of the semantic information task according to the priority evaluation model and processes them in sequence;

[0087] Step S3: When processing a semantic information task, the roadside unit first broadcasts the request information, and the idle vehicle then feeds back its own vehicle information to the roadside unit. Vehicle information for Indicates that Indicates the distance between the vehicle and the roadside unit, The roadside unit can provide CPU computing power to offload computations for vehicles. After decoding the semantic information task, it divides the semantic information task into M+1 subtasks according to the optimization algorithm. It distributes M subtasks to idle vehicles via the V2I link, and processes the remaining subtask itself. After completing the subtask, the idle vehicle sends the processing results back to the roadside unit, which then merges and sends them back to the user vehicle.

[0088] Step S4: To achieve precise optimization of task offloading and resource allocation in the Internet of Vehicles, a comprehensive utility function is constructed. This function comprehensively considers core elements such as the relative delay of the task, the accuracy of semantic understanding, the task importance reward value, and the social contribution rate. The relative delay reflects the timeliness of task processing, covering the delay of the vehicle unloading the task to the roadside unit, completing the subtask, and queuing at the roadside unit. The semantic understanding accuracy is determined based on the vehicle's collection of environmental information, encoding compression, and transmission processing flow, combined with the experimental relationship model between encoding parameters and accuracy, to ensure accurate communication and understanding of task semantics and improve processing quality. The task importance reward function sets a constant weight according to the importance of the task, highlighting the value of important tasks and ensuring their priority processing. The social contribution rate is based on task priority grouping, and the weighted sum of the number of tasks completed by different groups within the maximum allowable delay is counted to measure the contribution of semantic information tasks to society and guide resources to high-contribution tasks. This utility function provides a quantitative basis for optimizing task offloading and resource allocation, driving system performance improvement.

[0089] In step S5, the improved particle swarm optimization algorithm (ePSOO-AIU) is used to solve the task offloading and resource allocation problems. The particle representation bandwidth and task allocation are defined. The system utility function (comprehensive multi-factor) is used as the fitness function. The inertia factor and learning factor are dynamically adjusted during iteration to guide the particles to search for the optimal solution, obtain the best task offloading and resource allocation plan, and achieve optimal system performance.

[0090] Step S5 can be specifically divided into the following processing steps:

[0091] Step S51: The optimization goal of the system model of the present invention is to find the optimal bandwidth allocation and task allocation ratio so as to maximize the utility function. Let the number of particles be L and the particle number be ∈{1,…,L}, in the initial state, the position of each particle is randomly generated according to the constraints C1~C5, which is represented as a matrix with N rows and (1+M+1) columns, where the first column represents the bandwidth allocated to all task vehicles, and the next M+1 columns represent the task allocation ratio of M idle vehicles and a single roadside unit;

[0092] Step S52: using the comprehensive utility function as the fitness function of the algorithm;

[0093] Step S53, in the algorithm iterative optimization, the particle velocity is updated by the inertia factor and the first learning factor , the second learning factor Collaborative regulation, which is based on the concept of swarm intelligence optimization and simulates the foraging or evolutionary behavior of biological groups in nature;

[0094] Particle speed at Iterations, according to the formula:

[0095] renew;

[0096] in, It is a function that generates random numbers uniformly distributed between [0, 1]. Similar to the random exploration factor in biological behavior, it introduces uncertainty into the particle search path, avoids falling into a fixed pattern, and enhances the breadth of global exploration. is the current optimal solution of the particle; It is the global optimal solution, the crystallization of collective wisdom, and represents the overall optimal direction;

[0097] In order to ensure that the algorithm has global convergence capability in the early stage and local convergence capability in the later stage, the inertia factor The value of will gradually decrease during the algorithm iteration process. The specific update formula is:

[0098] ;

[0099] in, and are the current number of iterations and the maximum number of iterations respectively, and are the maximum and minimum values ​​of the inertia factor, respectively. This update formula ensures that in the early stage of the algorithm, the inertia factor The larger the value of , the greater the momentum of the particle when the position is updated, so that a wide range of searches can be carried out in the solution space, reducing the risk of falling into the local optimal solution. As the number of iterations increases, the inertia factor will gradually decrease, approaching , limiting the jumping range of particles, thus performing a refined search when approaching the optimal solution and improving the local convergence ability;

[0100] First learning factor and the second learning factor Describe the degree of influence of the particle's own cognitive experience and social experience on the particle's speed. During the iteration process, the particle's fitness value can be dynamically adjusted by comparing each iteration. and The specific update formula is:

[0101] ;

[0102] in 、 and Indicates in The fitness of the particle at the time of the update, the fitness of the particle's current optimal solution, and the fitness of the global optimal solution, and is an adjustable coefficient. According to the above formula, the self-cognition factor According to the particle's current solution Its individual optimal solution The gap and the current number of iterations Adjustment is made when the fitness of the particle Far below its own optimal solution hour, will be larger, so that particles can rely more on their own experience to conduct extensive exploration, which helps to stay away from the current local area and increase the search space. Increase, the denominator will gradually decrease The increase in the particle size can avoid the oscillatory behavior caused by over-reliance on self-cognition, thereby smoothly converging to the optimal region. The above formula is obtained by the fitness of the current solution of the particle. Fitness with the global optimal solution Dynamic adjustment of the gap , and introduced Item to control its changing speed, when the fitness of the particle and the global optimal solution When the gap is large, the social experience factor Increase, push the particles to rely more on global experience to move closer to the global optimal solution, promote the global search ability, and when the particles are close to the global optimal solution, Items will be reduced The change of makes the particles gradually slow down the search speed when approaching the optimal solution;

[0103] Step S54: The iteration continues until the preset cutoff condition is met, such as the maximum number of iterations is reached or the fitness value changes very little for multiple consecutive times, that is, the optimal particle position is found and the optimal solution for task offloading and resource allocation is determined; otherwise, the iterative optimization is continued to continuously improve the solution performance. Example 1

[0104] The system simulation is based on a system consisting of a single roadside unit, multiple user vehicles and multiple idle vehicles. Figure 2 The main simulation parameters and settings of the IoV system are shown in Table 1 below:

[0105] Table 1 Simulation parameters

[0106] ;

[0107] Unless explicitly stated, the default values ​​of the parameters are used. To comprehensively evaluate the effectiveness of the proposed AIU priority model, the following commonly used priority models are considered in combination with the proposed offloading strategy for performance comparison. These models do not consider the importance and urgency of the tasks;

[0108] Solution 1: ePSOO-ADC: RSU prioritizes tasks with smaller delay constraints based on the delay constraints of each task. Solution 2: ePSOO-ACS: RSU prioritizes tasks with fewer computing resources based on the computing resources required by the tasks. Solution 3: ePSOO-ATO: RSU prioritizes tasks that arrive first based on their arrival time.

[0109] In addition, to further evaluate the effectiveness and applicability of the proposed ePSOO-AIU algorithm, the following common offloading algorithms in the literature are considered for performance comparison: Algorithm 1: offloading strategy GAO based on genetic algorithm; Algorithm 2: offloading strategy ACO based on ant colony algorithm;

[0110] See also Figure 3 and Figure 4 , Figure 3 Schematic diagram comparing the utility values ​​of priority models at different system bandwidths of the present invention; Figure 4 This is a schematic diagram comparing the utility values ​​of various algorithms for different numbers of idle vehicles in the present invention. Figure 3 The system utility values ​​of each priority model at different uplink total bandwidths are shown. It can be seen that the utility values ​​of all priorities increase with the increase of total bandwidth and eventually stabilize. This is because increasing the uplink total bandwidth can improve the transmission rate, which is conducive to reducing task offloading time. When bandwidth resources are very abundant, this benefit will gradually decrease and the utility value will stabilize. In addition, the system utility of the offloading strategy based on the priority model of the present invention is always higher than that based on the other three priority models. This is because the model of the present invention takes into account the importance and urgency of the task, so that resource allocation is tilted towards tasks with greater social contribution and importance rewards, effectively improving system performance. Figure 4 The table shows a comparison of the system utility values ​​of the algorithm of the present invention and two other common algorithms at different numbers of idle vehicles. It can be seen that the utility values ​​of the algorithm of the present invention are significantly higher than those of the GA and ACO algorithms. The algorithm of the present invention optimizes the inertia weight parameters and learning factors of the traditional particle swarm algorithm, thereby improving the optimization ability. The genetic algorithm GA can obtain a better solution by performing operations such as selection, crossover and mutation on chromosomes, which is second only to the algorithm of the present invention. The ant colony algorithm guides the generation of solutions in a probabilistic manner by simulating the accumulation and volatilization of pheromones during the foraging process of ants. However, due to the slow update of pheromones and the easy falling into local optimality, its performance is poor, and the system performance is significantly lower than that of the GA algorithm and the algorithm of the present invention. In summary, the algorithm of the present invention has stronger adaptability to the vehicle network unloading scenario, showing superior system performance.

[0111] In the description of the present invention, the reference terms "one embodiment", "some embodiments", "in the present embodiment", "specific examples", or "some examples" mean that the specific features, mechanisms, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, mechanisms, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples without contradiction.

[0112] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for task offloading and resource allocation in a connected car scenario, characterized in that: The following steps are involved: Step S1: constructing a connected vehicle mobile edge computing system model in a connected vehicle scenario, wherein the connected vehicle mobile edge computing system model includes a roadside unit, multiple user vehicles, and multiple idle vehicles, and receiving semantic information tasks sent by each user vehicle through the roadside unit; Step S2: constructing a priority evaluation model to define the task urgency and task importance of each of the semantic information tasks, and based on the task urgency and task importance, evaluating each of the semantic information tasks using the priority evaluation model to obtain a corresponding task priority; Step S3, processing each of the semantic information tasks in order based on the task priority, dividing each of the semantic information tasks into multiple subtasks and assigning them to each of the idle vehicles and the roadside units for processing, obtaining corresponding task processing results, and transmitting them back to the user vehicle corresponding to the semantic information task; Step S4, obtaining the delay data, semantic accuracy data, and social contribution rate during the processing of each of the semantic information tasks in step S3, constructing a comprehensive utility function based on the task priority, the delay data, the semantic accuracy data, and the social contribution rate of each of the semantic information tasks as the fitness function of the IoV mobile edge computing system model to quantify task offloading and resource allocation; In step S5, a particle swarm algorithm is introduced to define particles to characterize bandwidth and task allocation, and an inertia factor and a learning factor are introduced to dynamically regulate the particle iterative update process to guide the particles to search for the optimal solution, obtain the best task offloading and resource allocation solution, and maximize the fitness function.

2. The task offloading and resource allocation method according to claim 1, characterized in that: In step S1, each of the user vehicles collects surrounding environment information, data volume, computing power, and maximum delay through on-board equipment and generates the semantic information task after semantic coding.

3. The task offloading and resource allocation method according to claim 1, characterized in that: In step S2, the task priority is obtained by the following calculation formula: ； in, Indicates the priority of the task; Indicate the importance of the task; A weight indicating the importance of the task; Indicate the urgency of the task; A weight representing the urgency of the task.

4. The task offloading and resource allocation method according to claim 1, characterized in that: In step S3, the total number of the idle vehicles is M. After decoding, the semantic information task is divided into M+1 subtasks. The M subtasks are randomly assigned to each idle vehicle for processing to obtain a corresponding first subresult. The remaining subtask is processed by the roadside unit to obtain a corresponding second subresult. Subsequently, the first subresults and the second subresults are merged to form the task processing result.

5. The task offloading and resource allocation method according to claim 1, characterized in that: In step S4, the comprehensive utility function is constructed by the following expression: ； in, represents the comprehensive utility function; Indicates preset parameters; Indicates the delay data; Indicates the priority of the task; representing said semantic accuracy data; Indicates the social contribution rate.

6. The task offloading and resource allocation method according to claim 1, characterized in that: The step S5 comprises: Step S51, introduce the particle swarm algorithm, define particles to represent bandwidth and task allocation, set the number of particles to be L, particle number ∈{1,…,L}, the position of each particle is randomly generated in the initial state and represented as a matrix with N rows and (1+M+1) columns, and the idle vehicle assigned to the task is defined as a task vehicle, wherein the first column of the matrix represents the bandwidth allocated to all task vehicles, and the remaining M+1 columns represent the task allocation ratio of the M idle vehicles and the roadside units; Step S52: In the particle iterative optimization process, the inertia factor and the learning factor are introduced to collaboratively control and update the particle velocity, wherein the learning factor includes a first learning factor for describing the particle's own cognitive experience and a second learning factor for describing the particle's social experience; Step S53, during the particle iterative optimization process, obtaining and comparing the particle fitness values ​​in each iterative optimization process to dynamically adjust the first learning factor and the second learning factor; Step S54: determine whether the particle iterative optimization process meets the preset cutoff condition: If so, the optimal particle position is determined and the best task offloading and resource allocation scheme is determined; If not, return to step S51.

7. The task offloading and resource allocation method according to claim 6, characterized in that: In step S51, the first The particle Position: ； in, Indicates the The particle location; Indicates the The bandwidth allocated to each of the task vehicles during the iteration; Indicates the The task allocation ratio of the M idle vehicles in the iteration; Indicates the The task allocation ratio of the roadside unit at the iteration.

8. The task offloading and resource allocation method according to claim 6, characterized in that: In step S52, the particle velocity is updated using the following calculation formula: ； in, Indicates the The particle velocity at the iteration; Indicates the The particle velocity at the iteration; Indicates the The inertia factor at the iteration; Indicates the The first learning factor at the iteration; Represents a function that generates random numbers uniformly distributed between [0, 1]; Indicates the The current optimal solution of the particle at the iteration; Indicates the The second learning factor at the iteration; Indicates the The global optimal solution of particles at the iteration.

9. The task offloading and resource allocation method according to claim 6, characterized in that: In step S53, the first learning factor is adjusted using the following calculation formula: ； in, Indicates the The first learning factor at the iteration; Indicates the The first learning factor at the iteration; Indicates the preset adjustable coefficient; Indicates the The fitness of the particle's current optimal solution at the iteration; Indicates the The fitness of the particle at the iteration.

10. The task offloading and resource allocation method according to claim 6, characterized in that: In step S53, the second learning factor is adjusted using the following calculation formula: ； in, Indicates the The second learning factor at the iteration; Indicates the The second learning factor at the iteration; Indicates the preset adjustable coefficient; Indicates the The fitness of the particle's current optimal solution at the iteration; Indicates the The fitness of the particle at the iteration.

Images