Resource allocation method for vehicle-mounted edge computing unloading based on particle swarm optimization algorithm

Abstract

The invention relates to the technical field of vehicle-mounted edge computing unloading, in particular to a resource allocation method for vehicle-mounted edge computing unloading based on a particle swarm optimization algorithm, and the method comprises the steps: constructing a V2I computing task unloading model of multiple vehicles; the resource utilization rate of the server is used as an optimization target, and a Lagrangian multiplier method and a particle swarm optimization algorithm are used for resource allocation; and verification is carried out in the environment of urban macro cellular bidirectional straight streets. The method comprises the following steps: firstly, constructing a V2I calculation task unloading model, and establishing a task model, a communication model and a calculation model; after the model is determined, the optimization objective of constructing the calculation task is to maximize the resource utilization rate of an edge server and minimize the total cost of time and energy consumption; and in a resource allocation process, realizing allocation of communication resources by using a Lagrangian multiplier method, and realizing allocation of computing resources by using a particle swarm optimization algorithm. Simulation results show that the resource utilization rate of the edge server can be effectively improved to 98% or above.

Description

Technical Field

[0001] The present invention relates to the field of vehicle-mounted edge computing offloading technology, and in particular to a resource allocation method for vehicle-mounted edge computing offloading based on a particle swarm optimization algorithm. Background Art

[0002] The Internet of Vehicles (IoV) and intelligent transportation systems are important development directions of my country's strategy to build a strong transportation nation. IoV uses wireless communication technology to connect cars, pedestrians and road facilities, and realizes data intercommunication and sharing between people, cars, roads, networks and clouds, and ultimately realizes functions such as intelligent transportation, intelligent services and intelligent driving. Therefore, with the rapid advancement of IoV, intelligent connected vehicles (ICV) have ushered in rapid development. ICV is an organic combination of the Internet of Vehicles and smart cars. It is a new generation of cars equipped with advanced on-board sensors, controllers, actuators and other devices, and integrates modern communication and network technologies.

[0003] In real-world scenarios, the diverse new in-vehicle applications and large-scale data access of ICVs will generate a large number of computationally intensive tasks, which significantly increases the demand for computing resources. However, the computing, storage, and communication capabilities in a single vehicle onboard unit (OBU) are limited. Although vehicular cloud computing (VCC) has abundant computing and storage resources, it provides a solution to this challenge. However, the centralized cloud server is far away from the vehicle, which slows down the transmission rate, increases the time and energy costs, and ultimately leads to a decline in the service quality of the task.

[0004] In order to make up for the shortcomings of VCC, Vehicular Edge Computing (VEC) introduces the concept of mobile edge computing into the vehicle network. VEC refers to the deployment of cellular base stations and edge servers near the vehicle, pushing computing tasks from the cloud to the edge of the network, making data processing closer to the vehicle. This greatly shortens the link transmission between the vehicle and the server, while reducing system costs. However, as the main processing unit of edge computing, the computing resources of the edge server are very scarce. Moreover, with the increase in vehicle users, the limited computing resources of the VEC server also affect the delay and energy consumption of task execution. Summary of the invention

[0005] The purpose of the present invention is to provide a resource allocation method for vehicle-mounted edge computing offloading based on a particle swarm optimization algorithm, aiming to solve the resource allocation situation of edge servers in existing vehicle networks, improve their resource utilization and reduce computing task time and total energy consumption cost.

[0006] To achieve the above object, the present invention provides a resource allocation method for vehicle-mounted edge computing offloading based on a particle swarm optimization algorithm, comprising the following steps:

[0007] Step 1: Build a multi-vehicle V2I computing task offloading model;

[0008] Step 2: Use the Lagrange multiplier method and particle swarm optimization algorithm to allocate resources and obtain a resource allocation solution for computing task offloading;

[0009] Step 3: In the environment of a two-way straight street in an urban macrocell, the time energy consumption cost and resource utilization results of the computing data are verified based on the resource allocation scheme of the computing task offloading.

[0010] Optionally, the multi-vehicle V2I computing task offloading model includes a task model, a communication model and a computing model;

[0011] In the task model, the vehicle offloads the computing task to the VEC server deployed near the Uma base station. The VEC server returns the calculation result of the offloaded task to the corresponding vehicle after computing and processing.

[0012] In the communication model, the initial position and terminal position of the vehicle are both within the coverage of the same Uma base station, and all vehicles will not cause interference during task offloading;

[0013] The computing model consists of a local computing model and an edge computing model.

[0014] Optionally, during the execution of step 2, the Lagrange multiplier method is used to implement the allocation of communication resources, and the particle swarm optimization algorithm is used to implement the allocation of computing resources.

[0015] Optionally, resource utilization in the edge computing model is defined as:

[0016]

[0017] The allocation optimization problem in step 2 is modeled as P1:

[0018] P1:

[0019] stC1:a n ∈{0,1},n=1,2,……,N

[0020]

[0021] C4: 0 ≤ P n ≤P max

[0022]

[0023] Where A={a1,a2,…,a n ,…,a N} is the offloading decision set of all tasks; is the edge computing resource allocation set; P = {p1, p2, ..., p n ,…,p N} is the power resource allocation set; constraint C1 represents the unloading decision of vehicle n; constraint C2 indicates that the total execution delay cannot exceed the maximum tolerable delay T n,max ; Constraint C3 indicates the limitation of local CPU computing power; Constraint C4 indicates that the transmission power cannot exceed the maximum transmission power p max ; Constraint C5 indicates the range of edge computing resource allocation; Constraint C6 indicates that the total computing resource requirement of the offload task cannot exceed the computing resources provided by the edge server.

[0024] Optionally, the Lagrangian function constructed in step 2 is:

[0025]

[0026] Where λ = {λ n ≥0}, μ={μ n ≥0} is the Lagrange multiplier corresponding to the uplink transmission time constraint C2; ν={ν n ≥0} is the Lagrange multiplier corresponding to the maximum computing resource allocation constraint C6 of the VEC server.

[0027] Optionally, the resource allocation strategy process using the particle swarm optimization algorithm in step 2 includes the following steps:

[0028] Step 2.1: Initialization, determine the number of particles N, the number of iterations maxInter_PSO, the transmission power P, etc., and initially set the random position and speed of the particles;

[0029] Step 2.2: Calculate the fitness value of each particle according to the fitness function;

[0030] Step 2.3: Update the velocity and position of the particles, generate new particles, and perform boundary value processing on the new particles;

[0031] Step 2.4: Introduce the mutation operator to reinitialize the position of the particle;

[0032] Step 2.5: For each particle, compare its fitness value with the fitness value of the best position p_best it has experienced. If it is better, take it as the current individual optimal position;

[0033] Step 2.6: For each particle, compare its fitness value with the fitness value of the best position g_best experienced globally. If it is better, take it as the current global optimal position;

[0034] Step 2.7: The maximum number of iterations is used as the end condition for judgment. If it is not reached, return to step 2.2;

[0035] Step 2.8: Output the global optimal resource allocation plan.

[0036] Optionally, the result obtained by the particle swarm optimization algorithm adopted in the resource allocation scheme for offloading the computing tasks is a local optimal solution, and the computing result still needs to be verified multiple times.

[0037] The present invention provides a resource allocation method for vehicle-mounted edge computing offloading based on a particle swarm optimization algorithm, by constructing a V2I computing task offloading model for multiple vehicles; taking server resource utilization as the optimization target, using the Lagrange multiplier method and the particle swarm optimization algorithm for resource allocation; and verifying the time energy consumption cost and resource utilization results of the computing data in the environment of a two-way straight street of a macro cell in the city. Specifically, a V2I computing task offloading model is first constructed for the vehicle, and a task model, a communication model and a computing model are established based on the model; after the model is established, the optimization target of the computing task is constructed as follows: maximizing the resource utilization of the edge server and minimizing the total cost of time energy consumption; and in the resource allocation process, the Lagrange multiplier method is used to realize the allocation of communication resources, and the particle swarm optimization algorithm can be used to realize the allocation of computing resources. The simulation results show that the present invention realizes the reasonable allocation of computing resources and can effectively increase the resource utilization of the edge server to more than 98%. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 It is a step flow chart of a method for automobile edge computing offloading and resource allocation based on a particle swarm optimization algorithm of the present invention.

[0040] Figure 2 It is a schematic diagram of the structure of the multi-vehicle V2I computing task offloading model of the present invention.

[0041] Figure 3 It is a schematic diagram of the workflow of the particle swarm algorithm of the present invention to realize the computing resource allocation stage.

[0042] Figure 4 It is a schematic diagram of the effect of the change in the number of tasks on resource utilization in the experimental phase of a specific embodiment of the present invention.

[0043] Figure 5 It is a schematic diagram of the effect of the change in the number of tasks on the total system cost during the experimental phase of a specific embodiment of the present invention.

[0044] Figure 6 It is a schematic diagram of the effect of the number of iterations on resource utilization in the experimental phase of a specific embodiment of the present invention.

[0045] Figure 7 It is a schematic diagram of the effect of the number of iterations on the total system cost in the experimental phase of a specific embodiment of the present invention.

[0046] Figure 8 It is a schematic diagram of the effect of the task volume on resource utilization in the experimental phase of a specific embodiment of the present invention.

[0047] Fig. 9 It is a schematic diagram of the impact of the task volume on the total system cost in the experimental stage of a specific embodiment of the present invention.

[0048] Fig.10 It is a schematic diagram of the influence of the delay weight coefficient on the resource utilization rate in the experimental stage of a specific embodiment of the present invention.

[0049] Fig.11 It is a schematic diagram of the influence of the delay weight coefficient on the total system cost in the experimental stage of a specific embodiment of the present invention. DETAILED DESCRIPTION

[0050] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0051] See also Figure 1 The present invention provides a resource allocation method for vehicle-mounted edge computing offloading based on a particle swarm optimization algorithm, comprising the following steps:

[0052] S1: Construct a multi-vehicle V2I computing task offloading model;

[0053] S2: Use Lagrange multiplier method and particle swarm optimization algorithm to allocate resources and obtain resource allocation scheme for computing task offloading;

[0054] S3: In the environment of a two-way straight street of an urban macro cell, the time energy consumption cost and resource utilization results of the computing data are verified respectively based on the resource allocation scheme of the computing task offloading.

[0055] The following is a further explanation based on the specific implementation steps:

[0056] In step S1, the V2I computing task offloading model of multiple vehicles is as follows: Figure 2 As shown in the figure, under this model, the task model, communication model and calculation model are constructed respectively. The detailed process is as follows:

[0057] (1) Task Model

[0058] Assume that the model consists of a VEC server and N vehicles. The vehicle offloads the computing task to the VEC server deployed near the Uma base station (BS). The VEC server returns the calculation result of the offloaded task to the corresponding vehicle after computing. Among them, the VEC server has limited resources, and the total computing capacity is defined as F VEC A VEC server can provide services to multiple vehicles, and the computing resources allocated to each vehicle are also limited, defined as in, represents the computing power allocated to the nth vehicle by the VEC server. The total number of vehicles in the network is N, and the vehicle set is defined as V = {V1, V2, …, V n ,…,V N}. Assume that each vehicle has a computationally intensive and delay-sensitive task R n ={L n ,C n ,T n,max}. Among them, L n Represents task R n The input data size, C n Indicates completion of task R n The total number of CPU cycles required, T n,max Represents task R n The maximum time constraint.

[0059] Since the VEC server provides services for multiple vehicles, the computing tasks can be offloaded to the edge or local vehicle devices for execution. Therefore, a binary offloading scheme is considered, that is, the task cannot be split and can only be processed as a whole. The binary offloading decision variable of vehicle n is defined as a n ∈{0,1}. when a n =0, indicating that the vehicle V n Choose to perform local calculation; a n =1, indicating that the vehicle V nChoose to offload its computing tasks to the VEC server for calculation. Let A = {a n |n=1,2,…,n,…,N} represents the decision configuration of all vehicle computing tasks.

[0060] (2) Communication model

[0061] When vehicle data is offloaded to the VEC server, delay and energy consumption will be generated in the uplink. In this model, it is assumed that the initial position and terminal position of the vehicle are within the coverage of the same BS. In addition, each vehicle uses an orthogonal uplink transmission channel to each other, and all vehicles will not cause interference during the task offloading process. Therefore, the uplink data transmission rate r of the process of vehicle n offloading the task to the VEC server is n Expressed as:

[0062]

[0063] Where B represents the uplink channel bandwidth from the vehicle to the VEC server, N0 represents the Gaussian white noise power from the vehicle to the VEC server, and P n represents the uplink transmission power from the vehicle to the VEC server, g n =(d 3D ) -δ represents the channel gain between the vehicle terminal and the BS, and δ = 4 represents the path loss constant. 3D represents the 3D distance between the vehicle and the target BS, which is defined as:

[0064]

[0065] Among them, h BS Indicates the height of BS; h UT Indicates the deployment height of the intelligent connected vehicle antenna; d 2D Represents the plane distance from the vehicle to the BS.

[0066] (3) Computational model

[0067] The computing model consists of a local computing model and an edge computing model.

[0068] 3-1) Local computing model

[0069] When vehicle n chooses to perform computing tasks locally, the CPU, as an important component of the on-board equipment, is responsible for computing and processing. Its computing power is determined by the CPU cycle frequency. (cycles / s). Therefore, task R n Local computing time for:

[0070]

[0071] Energy consumption is another key indicator for measuring edge computing efficiency. CPU power consumption is a superlinear function of computing frequency and can be expressed as Where k = 10 -11 is a constant related to the hardware architecture. Therefore, task R n Local computing energy consumption for:

[0072]

[0073] According to formulas (3) and (4), the total cost of local calculation can be expressed as:

[0074]

[0075] Among them, 0<ω1<1 means that task R n The weight coefficient of the time required to complete the calculation, 0<ω2<1 means that the task R n The weight coefficient of the energy consumption required for calculation satisfies ω1+ω2=1. The value of the weight is dynamically adjusted according to the needs of the vehicle application. For example, for applications with high latency requirements, a larger ω1 will be set to reduce latency.

[0076] 3-2) Edge computing model

[0077] When vehicle n chooses to calculate task R n When offloading to the VEC server for execution, the offloading process is divided into three stages: sending, computing, and receiving. The sending stage refers to the vehicle n offloading the computing task to the BS through the V2I method, and its transmission time is and energy consumption It is expressed as:

[0078]

[0079] The calculation phase refers to the use of the VEC server connected to the BS to calculate the task R n Calculate the delay time It can be expressed as:

[0080]

[0081] in, Indicates that the VEC server is task R n The allocated computing resources meet

[0082] The receiving phase refers to when the VEC server completes the calculation processing, the vehicle is responsible for receiving the calculation results sent back by it. The amount of data received by the vehicle from the VEC server is much smaller than the original amount of data, and the two have a large difference in order of magnitude. Therefore, the downlink transmission speed is fast, the occupied bandwidth is small, and the required delay is short. Therefore, the present invention ignores the time and energy overhead of the receiving phase when establishing the calculation model.

[0083] Finally, Task R n The total latency, total energy consumption, and total cost of edge computing can be expressed as:

[0084]

[0085] In step S2, the server resource utilization is taken as the optimization target, and the Lagrange multiplier method and particle swarm optimization algorithm are used for resource allocation.

[0086] Specifically, the computing resources of edge servers are expensive and limited, so their resources should be fully utilized. Resource utilization is defined as:

[0087]

[0088] The optimization problem is modeled as P1:

[0089]

[0090] stC1:a n ∈{0,1},n=1,2,……,N

[0091]

[0092] C4: 0 ≤ P n ≤P max

[0093]

[0094] Where A={a1,a2,…,a n ,…,a N} is the offloading decision set of all tasks; is the edge computing resource allocation set; P = {p1, p2, ..., p n ,…,p N} is the power resource allocation set. Constraint C1 represents the unloading decision of vehicle n; constraint C2 indicates that the total execution delay cannot exceed the maximum tolerable delay T n,max ; Constraint C3 indicates the limitation of local CPU computing power; Constraint C4 indicates that the transmission power cannot exceed the maximum transmission power p max; Constraint C5 indicates the range of edge computing resource allocation; Constraint C6 indicates that the total computing resource requirement of the offload task cannot exceed the computing resources provided by the edge server.

[0095] In addition, in the resource allocation process, there is the allocation of communication resources and computing resources. First, the Lagrange multiplier method is used to achieve the deterministic allocation of communication resources. Secondly, the particle swarm optimization algorithm is used to achieve the optimal allocation of computing resources. The detailed process is as follows:

[0096] First, when allocating communication and computing resources in edge offloading, let a n =1, then the optimization problem P1 can be simplified as:

[0097]

[0098]

[0099] (1) Computing edge communication resource allocation

[0100] The communication resource allocation problem is to allocate the transmission power in edge computing offloading. According to the constraint C2 in the optimization problem P1.2, we know When the constraint condition takes the equal sign, we can get:

[0101]

[0102] Let t n for The upper bound of but Therefore, the optimization problem P3 can be rewritten as:

[0103]

[0104] Taking the first-order derivative and the second-order derivative of the variable t in the objective function of P3.1, we get:

[0105]

[0106] Depend on It can be seen that the objective function of formula (16) is a concave function. Therefore, the present invention adopts the Lagrange multiplier method to solve the optimal solution of power allocation. First, the constructed Lagrange function is:

[0107]

[0108] Where λ = {λ n ≥0}, μ={μ n ≥0} is the Lagrange multiplier corresponding to the uplink transmission time constraint C2; ν={ν n≥0} is the Lagrange multiplier corresponding to the maximum computing resource allocation constraint C6 of the VEC server.

[0109] The dual problem can be expressed as:

[0110]

[0111] Based on the hierarchical optimization decomposition method, the dual problem of formula (20) is divided into two layers. The inner layer is the minimization of the transmission time; the outer layer is the maximization problem of the Lagrange multiplier.

[0112] In addition, for the Lagrangian solution algorithm, it is proved that the local optimal solution is basically consistent with the global optimal solution, and the subgradient algorithm is an effective method for solving the Lagrangian problem. Therefore, the present invention uses the subgradient algorithm to iteratively update the Lagrangian multiplier to solve the optimal solution of the optimization variable. The iterative formula is:

[0113]

[0114] μ n (t+1)=[μ n (t)+β(t)(T n，max -t n )] + (twenty two)

[0115]

[0116] Among them, [z] + =max{z,0}, t is the iteration index, β(t) represents the iteration positive step length.

[0117] Therefore, according to the above conditions, the optimal solution can be obtained for:

[0118]

[0119] (2) Computing edge computing resource allocation

[0120] When the transmission power is known, the computing resource allocation problem can be further optimized on the constrained optimization problem P3.1 as follows:

[0121]

[0122] The allocation problem of edge computing resources belongs to a multi-objective optimization problem. The present invention uses the particle swarm optimization (PSO) algorithm to obtain the computing resource allocation scheme when each vehicle decides to perform edge computing.

[0123] PSO is an iterative swarm intelligence algorithm that randomly searches for the global optimal solution through cooperation and competition among groups. Applying it to edge computing resource allocation can obtain a better amount of computing resources for each vehicle. In the PSO algorithm, the amount of resources allocated to each vehicle is called a particle, and the current position of each particle is x = {x1, x2, ..., x i ,…,x n}, the velocity is v={v1,v2,…,v i ,…,v n}. Let p_best represent the optimal position of the particle itself, and x_best represent the best position searched by the entire population. Then in the tth iteration, the speed and position update formula of particle i is expressed as:

[0124]

[0125] Where c1 is the individual learning factor and c2 is the group learning factor, which control the moving direction of the particle. r1 and r2 are random numbers in the interval [0,1]. w is the inertia weight, which indicates the influence of the previous velocity of the particle on the current velocity. It is of great significance in the global and local search capabilities of the equilibrium algorithm. The present invention adopts the inertia weight definition formula proposed by Shi and Eberhart:

[0126]

[0127] w max and w min represents the maximum and minimum values ​​of the inertia weight, t is the current number of iterations, and T is the maximum number of iterations of the population. By dynamically adjusting the inertia weight w, the global search capability can be effectively improved, so that the convergence of particles is also enhanced. In addition, the present invention adopts a floating-point encoding method and uses the objective function of the problem to be solved as a fitness function to evaluate the quality of each solution. Therefore, for the unloading decision X, its fitness function is f(X) = Z(X). Among them, the flow chart of the computing resource allocation strategy based on the PSO algorithm is as follows Figure 3 The specific steps are as follows:

[0128] Step 1: Initialization. Determine the number of particles N, the number of iterations maxInter_PSO, the transmission power P, etc., and initially set the random position and speed of the particles;

[0129] Step 2: Calculate the fitness value of each particle according to the fitness function (objective function);

[0130] Step 3: Update the speed and position of the particles, generate new particles, and perform boundary value processing on the new particles;

[0131] Step 4: Introduce the mutation operator to reinitialize the position of the particle;

[0132] Step 5: For each particle, compare its fitness value with the fitness value of the best position p_best it has experienced. If it is better, take it as the current individual optimal position;

[0133] Step 6: For each particle, compare its fitness value with the fitness value of the best position g_best experienced globally. If it is better, take it as the current global optimal position;

[0134] Step 7: Use the maximum number of iterations as the end condition for judgment. If it is not reached, return to the second step;

[0135] Step 8: Output the globally optimal computing resource allocation plan.

[0136] Step S3: in an environment of a two-way straight street in an urban macro cell, the time energy consumption cost and resource utilization results of the computing data are verified based on the resource allocation scheme for offloading the computing task.

[0137] Specifically, based on Matlab software, simulation experiments are carried out under different parameter conditions such as the number of tasks, the number of PSO algorithm iterations, the amount of task data, and the weight coefficients of delay and energy consumption to verify the performance of the proposed computation offloading and resource allocation scheme. The parameter settings are shown in Table 1.

[0138] Table 1 Parameter settings

[0139]

[0140]

[0141] For further information, see Figures 4 to 11 , the present invention is verified by simulation results:

[0142] (1) Figure 4 and Figure 5 The effects of the number of tasks on resource utilization and total system cost are shown. When the total computing power of the VEC server is 30 GHz, the number of tasks set in the experiment is 4, 6, 8, 10, 12, 14 and 16 respectively. From the simulation results, the resource utilization of the VEC server always tends to be stable and can reach more than 99%. Moreover, compared with the Local-VEC and LSGA algorithms, the total system cost of the algorithm proposed in the present invention is the smallest, and it grows steadily with the increase of the number of tasks.

[0143] (2) Figure 6 and Figure 7The influence of the number of iterations in the PSO algorithm on resource utilization and total system cost is shown respectively. The number of iterations in the PSO algorithm is 60, 80, 100, 120, 140 and 160 (times), the total number of tasks N = 10, the weight coefficient of time and energy consumption ω1 = ω2 = 0.5, L n =4. Among them, since the task in the Local-VEC algorithm is to perform the calculation task only on the local vehicle, and the PSO algorithm does not exist in the LSGA algorithm, the number of iterations of the PSO algorithm has no effect on their experimental results. However, it can be observed that when the number of iterations is small, the algorithm proposed in the present invention is still stable in terms of resource utilization and total system cost.

[0144] (3) Figure 8 and Fig. 9 The influence of the workload of vehicle tasks on resource utilization and total system cost is shown respectively. The workload of the task is taken between [3,6], the total number of tasks N = 10, and the weight coefficient of time and energy consumption ω1 = ω2 = 0.5. Figure 8 It can be seen that as the workload of tasks increases, resource utilization tends to be stable. Fig. 9 It can be seen that the total system cost of the algorithm proposed in the present invention is the smallest, and there is an obvious upward trend when the task volume is greater than 5.

[0145] (4) Fig.10 and Fig.11 The effects of different delay weight coefficients on resource utilization and total system cost are shown respectively. Take the number of tasks N = 10, the number of iterations in the PSO algorithm T = 200, Ln = 4, and the delay weight coefficient between [0.1, 1]. From the simulation results, it can be seen that with the increase of the delay weight coefficient, the algorithm proposed in the present invention makes the resource utilization of the VEC server always tend to be stable, and the total system cost is also significantly better than other algorithms.

[0146] The experimental results fully prove that the particle swarm optimization algorithm proposed in the present invention can not only make full use of the resources of the edge server, but also effectively reduce the total cost of time and energy consumption.

[0147] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A resource allocation method for vehicle-mounted edge computing offloading based on particle swarm optimization algorithm, characterized in that: The following steps are involved: Step 1: Build a multi-vehicle V2I computing task offloading model; Step 2: Use the Lagrange multiplier method and particle swarm optimization algorithm to allocate resources and obtain a resource allocation solution for computing task offloading; Step 3: In the environment of a two-way straight street in an urban macrocell, the time energy consumption cost and resource utilization results of the computing data are verified based on the resource allocation scheme of the computing task offloading.

2. The resource allocation method for vehicle-mounted edge computing offloading based on particle swarm optimization algorithm as claimed in claim 1, characterized in that: The V2I computing task offloading model for multiple vehicles includes task model, communication model and computing model; In the task model, the vehicle offloads the computing task to the VEC server deployed near the Uma base station. The VEC server returns the calculation result of the offloaded task to the corresponding vehicle after computing and processing. In the communication model, the initial position and terminal position of the vehicle are both within the coverage of the same Uma base station, and all vehicles will not cause interference during task offloading; The computing model consists of a local computing model and an edge computing model.

3. The resource allocation method for vehicle-mounted edge computing offloading based on particle swarm optimization algorithm as claimed in claim 2, characterized in that: During the execution of step 2, the Lagrange multiplier method is used to implement the allocation of communication resources, and the particle swarm optimization algorithm is used to implement the allocation of computing resources.

4. The resource allocation method for vehicle-mounted edge computing offloading based on particle swarm optimization algorithm as claimed in claim 3 is characterized in that: Resource utilization in the edge computing model is defined as: The allocation optimization problem in step 2 is modeled as P1: P1： s.t.C1:a n ∈{0,1},n=1,2,……,N C2： C3: C4: 0≤P n ≤P max C5： C6： Where A={a1,a2,…,a n ,…,a N } is the offloading decision set of all tasks; is the edge computing resource allocation set; P = {p1, p2, ..., p n ,…,p N } is the power resource allocation set; constraint C1 represents the unloading decision of vehicle n; constraint C2 indicates that the total execution delay cannot exceed the maximum tolerable delay T n,max ; Constraint C3 indicates the limitation of local CPU computing power; Constraint C4 indicates that the transmission power cannot exceed the maximum transmission power p max ; Constraint C5 indicates the range of edge computing resource allocation; Constraint C6 indicates that the total computing resource requirement of the offload task cannot exceed the computing resources provided by the edge server.

5. The resource allocation method for vehicle-mounted edge computing offloading based on particle swarm optimization algorithm as claimed in claim 4, characterized in that: The Lagrangian function constructed in step 2 is: Where λ = {λ n ≥0}, μ={μ n ≥0} is the Lagrange multiplier corresponding to the uplink transmission time constraint C2; ν={ν n ≥0} is the Lagrange multiplier corresponding to the maximum computing resource allocation constraint C6 of the VEC server.

6. The resource allocation method for vehicle-mounted edge computing offloading based on particle swarm optimization algorithm as claimed in claim 5, characterized in that: The resource allocation strategy process using the particle swarm optimization algorithm in step 2 includes the following steps: Step 2.1: Initialization, determine the number of particles N, the number of iterations maxInter_PSO, the transmission power P, etc., and initially set the random position and speed of the particles; Step 2.2: Calculate the fitness value of each particle according to the fitness function; Step 2.3: Update the velocity and position of the particles, generate new particles, and perform boundary value processing on the new particles; Step 2.4: Introduce the mutation operator to reinitialize the position of the particle; Step 2.5: For each particle, compare its fitness value with the fitness value of the best position p_best it has experienced. If it is better, take it as the current individual optimal position; Step 2.6: For each particle, compare its fitness value with the fitness value of the best position g_best experienced globally. If it is better, take it as the current global optimal position; Step 2.7: The maximum number of iterations is used as the end condition for judgment. If it is not reached, return to step 2.2; Step 2.8: Output the global optimal resource allocation plan.

7. The resource allocation method for vehicle-mounted edge computing offloading based on particle swarm optimization algorithm as claimed in claim 6, characterized in that: The result obtained by the particle swarm optimization algorithm used in the resource allocation scheme for computing task offloading is a local optimal solution, and the calculation result still needs to be verified multiple times.

Images