CN112084026A - Low-energy-consumption edge computing resource deployment system and method based on particle swarm - Google Patents

Abstract

The present invention provides a low-energy consumption edge computing resource deployment system based on particle swarm, including a network resource energy consumption model storage module, a network resource deployment model storage module and a network resource deployment module, which are used to obtain computing nodes in the edge computing resources. The optimal number of and/or the optimal number of storage nodes. The present invention proposes a particle swarm-based low-energy edge computing resource deployment system by designing an SDN-based edge computing service model and an objective function to minimize the energy consumption of the entire network, which is based on the constraint relationship between various resources. In order to achieve the purpose of saving the energy consumption of the whole network, the system of the present invention reduces the number of storage nodes and computing nodes, and saves the energy consumption of the whole network.

Description

Low-energy edge computing resource deployment system and method based on particle swarm

technical field

The present invention relates to the technical field of communication resource allocation, in particular to a low energy consumption edge computing resource deployment system based on particle swarms; meanwhile, the present invention also relates to a low energy consumption edge computing resource deployment method based on particle swarms.

Background technique

In the 5G network environment, in order to reduce the delay of user services, edge computing technology has been successfully applied to solve this problem. With the growth of user services, the demand for edge computing resources is increasing. How to minimize the energy consumption of the entire network on the premise of meeting user needs has become an urgent problem to be solved. Existing research mainly adopts virtual machine migration, game theory, optimization theory and other methods to solve the problems of low business execution efficiency and energy consumption.

For example, in Sustainable Service Allocation Using a Metaheuristic Technique in a Fog Server for Industrial Applications, Mishra.S.K and others put forward a sustainable service solution for fog server nodes with the goal of improving the efficiency of user task execution, which reduces the energy consumption of fog nodes and users The execution time of the task. In the Hybrid method for minimizing service delay in edge cloud computing through VM migration and transmission power control, Rodrigues.T.G solves the problem that the server performance in edge computing cannot meet the needs of high response speed and low latency business in the manufacturing industry. Migration and network flow control technology, a business response model with high-speed transmission and fast processing is proposed, which better solves the problem of low performance of edge computing servers. In Delay-constrained hybrid computation offloading with cloud and fog computing, Meng X et al. took the user task delay requirement as the constraint, modeled the calculation amount and communication amount of the task in combination, and proposed a scheduling mechanism to minimize the user task delay to reduce the It reduces the computation and communication delay of user tasks. In the Decentralizedcomputation offloading game for mobile cloud computing, CHEN Xu. et al. proposed an optimization algorithm for offloading the computing requirements of user tasks based on game theory, taking the user task execution time limit as the constraint, which better solved the problem of low user task execution efficiency. In the Utility-aware Resource Allocation for Edge Computing, Xu.J et al. aimed at reducing resource consumption, and proposed a resource allocation mechanism for collaboration between cloud service providers and edge service providers. The total resource overhead is calculated, and the scalability and reliability of the allocation mechanism are verified. In A bandwidth allocation scheme to meet flow requirements in mobile edge computing, with the goal of maximizing the task transmission rate, Ito.Y et al. designed the information transmission bandwidth resource allocation mechanism in the mobile edge computing network, and realized the provision of corresponding services according to the specific needs of users. The bandwidth resource allocation scheme can meet the different needs of different users.

Through the analysis of existing research, it can be seen that many research results have been achieved in the field of resource allocation in the edge computing environment. However, the existing research mainly focuses on solving different problems such as delay, rate, and resource amount, ignoring the constraint relationship between various resources.

Therefore, under the premise of satisfying user needs, developing a low-energy edge computing method based on the constraint relationship between various resources to save the energy consumption of the entire network has become an urgent problem to be solved by those skilled in the art.

SUMMARY OF THE INVENTION

In view of this, the present invention aims to propose a low energy consumption edge computing resource deployment system based on particle swarm, so as to reduce the number of storage nodes and computing nodes, optimize the resource allocation scheme, and thus achieve the purpose of saving the energy consumption of the whole network.

In order to achieve the above object, the technical scheme of the present invention is achieved in this way:

A particle swarm-based low-energy edge computing resource deployment system for obtaining the optimal number of computing nodes and/or the optimal number of storage nodes in the edge computing resources; including

The network resource energy consumption model storage module is used to store the network resource energy consumption model including the total energy consumption calculation formula and the whole network energy consumption minimization objective function;

The network resource deployment model storage module is used to store the network resource deployment model constructed by the particle swarm optimization algorithm. In the particle swarm optimization algorithm, the deployment scheme of network resources is used as the particle position, and the optimization method of the network resource deployment scheme is used as the movement of particles. speed;

A network resource deployment module, configured to invoke the network resource energy consumption model and the network resource deployment model, take the network resource energy consumption model as a judgment condition, perform an iterative calculation on the network resource deployment model, and output the network resource deployment model solution, the network resource deployment solution includes the optimal number of storage nodes and/or the optimal number of computing nodes.

Further, the total energy consumption calculation formula is the following formula

为存储节点能耗，

为动态计算能耗，

为静态计算能耗，

为内容传输能耗，

为计算传输能耗。In the formula, F is the total energy consumption of the node, ka ∈ KA is the content service request, _k ^{b ∈} _KB is the computing service request,

Energy consumption for storage nodes,

To dynamically calculate energy consumption,

Calculate energy consumption for static,

energy consumption for content transmission,

To calculate the transmission energy consumption.

Further, the objective function of minimizing the energy consumption of the whole network is the following formula:

为存储节点的最优数量，

为计算节点的最优数量，N为存储节点的最大数量，M为计算节点的最大数量。In the formula,

is the optimal number of storage nodes,

is the optimal number of computing nodes, N is the maximum number of storage nodes, and M is the maximum number of computing nodes.

Further, the iterative calculation includes the following steps:

Step 1, parameter initialization; the parameters of the initialization include the number of iterations MG, the particle swarm scale O, the initial position X _i of the randomly generated particles, and the initial velocity V _i of the randomly generated particles;

Step 2: Calculate the initial position; obtain the minimized energy consumption of the entire network according to the initialized parameters obtained in step 1 and the calculation formula of the total energy consumption of each node; obtain the deployment of network resources according to the minimized energy consumption of the entire network and the optimal initial position X _i of the particle, set the optimal initial position X _i as the global optimal initial position X _gb , and set the initial position X _i of each particle as the individual optimal initial position X _pb ;

Step 3: Update the particle velocity and particle position respectively; determine whether the initial position of each particle meets the constraints, if so, update the particle velocity and particle position respectively; if not, randomly generate new particle positions and particles speed;

Step 4: Update the new global optimal initial position and the individual optimal initial position respectively; when f(X _i )<f(X _pb ), set X _pb =X _i ; when f(X _pb )<f( X _gb ), set X _gb =X _pb ;

Step 5: Determine whether the end condition is reached; if yes, output the optimal X _i , if not, return to Step 3.

Further, the constraints are the maximum number of storage nodes and the maximum number of computing nodes.

At the same time, the present invention also proposes a particle swarm-based low-energy-consumption edge computing resource deployment method.

In order to achieve the above object, the technical scheme of the present invention is achieved in this way:

A particle swarm-based low-energy edge computing resource deployment method, obtaining the optimal number of computing nodes and/or the optimal number of storage nodes in the edge computing resource; comprising the following steps

a. Store the network resource energy consumption model including the calculation formula of total energy consumption and the objective function of minimizing energy consumption of the whole network;

b. Store the network resource deployment model constructed by the particle swarm optimization algorithm. In the particle swarm optimization algorithm, the deployment scheme of network resources is used as the particle position, and the optimization method of the network resource deployment scheme is used as the moving speed of the particles;

c. Invoke the network resource energy consumption model and the network resource deployment model, take the network resource energy consumption model as a judgment condition, perform an iterative calculation on the network resource deployment model, and output a network resource deployment scheme. The resource deployment scheme includes the optimal number of storage nodes and/or the optimal number of computing nodes.

Further, the total energy consumption calculation formula is the following formula

为存储节点能耗，

为动态计算能耗，

为静态计算能耗，

为内容传输能耗，

为计算传输能耗。In the formula, F is the total energy consumption of the node, ka ∈ KA is the content service request, _k ^{b ∈} _KB is the computing service request,

Energy consumption for storage nodes,

To dynamically calculate energy consumption,

Calculate energy consumption for static,

energy consumption for content transmission,

To calculate the transmission energy consumption.

Further, the objective function of minimizing the energy consumption of the whole network is the following formula:

为存储节点的最优数量，

为计算节点的最优数量，N为存储节点的最大数量，M为计算节点的最大数量。In the formula,

is the optimal number of storage nodes,

is the optimal number of computing nodes, N is the maximum number of storage nodes, and M is the maximum number of computing nodes.

Further, the iterative calculation includes the following steps:

c1, parameter initialization; the parameters of the initialization include the number of iterations MG, the particle swarm scale O, the initial position X _i of the randomly generated particles, and the initial velocity V _i of the randomly generated particles;

c2. Calculate the initial position; obtain the minimized energy consumption of the entire network according to the initialized parameters obtained in step c1 and the calculation formula of the total energy consumption of each node; obtain the deployment scheme of network resources according to the minimized energy consumption of the entire network and the optimal particle initial position X _i , set the optimal initial position Xi as the global optimal initial position X _{gb , and set the initial position X i of} each particle as the individual optimal initial position X _pb ;

c3. Update the particle velocity and particle position respectively; judge whether the initial position of each particle meets the constraints, if so, update the particle velocity and particle position respectively; if not, randomly generate new particle position and particle velocity ;

c4. Update the new global optimal initial position and the individual optimal initial position respectively; when f(X _i )<f(X _pb ), set X _pb =X _i ; when f(X _pb )<f(X _gb ), set X _gb =X _pb ;

c5. Determine whether the end condition is reached; if so, output the optimal X _i , if not, return to step c3.

Further, the constraints are the maximum number of storage nodes and the maximum number of computing nodes.

Compared with the prior art, the present invention has the following advantages:

1. The present invention designs a resource allocation scheme under the constraints of energy consumption and bandwidth with the joint goal of minimizing the energy consumption of the entire network and the occupancy of network bandwidth. First, by designing an edge computing service model based on SDN and the objective function of minimizing the energy consumption of the entire network, a low-energy edge computing resource deployment system based on particle swarm is proposed, which is based on the constraint relationship between various resources. Then, the purpose of saving the energy consumption of the whole network is achieved by optimizing the resource allocation scheme on the Analysis shows that the present invention has better application effect and performance, and better solves the problem of high energy consumption of network resources under different service request arrival rate environments.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can also be obtained from these drawings without any creative effort. In the attached image:

FIG. 1 is a schematic structural diagram of a particle swarm-based low-energy edge computing resource deployment system according to Embodiment 1 of the present invention;

2 is a schematic flowchart of a particle swarm-based low-energy-consumption edge computing resource deployment method according to Embodiment 2 of the present invention;

3 is a schematic diagram of comparing the number of storage nodes in the particle swarm-based low-energy-consumption edge computing method according to Embodiment 3 of the present invention;

FIG. 4 is a schematic diagram of comparing the number of storage nodes in the particle swarm-based low-energy-consumption edge computing method according to Embodiment 3 of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present invention.

In the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear" , "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", etc. Based on the orientation or positional relationship shown in the drawings, it is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be It is understood as a limitation of the present invention.

In the present invention, unless otherwise expressly specified and limited, the terms "installed", "connected", "connected", "fixed" and other terms should be understood in a broad sense, for example, it may be a fixed connection or a detachable connection , or integrally connected; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or the internal communication between the two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

Example 1

This embodiment relates to a particle swarm-based low-energy edge computing resource deployment system, which is used to obtain the optimal number of computing nodes and/or the optimal number of storage nodes in the edge computing resources; as shown in FIG. 1 , including

The network resource energy consumption model storage module is used to store the network resource energy consumption model including the calculation formula of total energy consumption and the objective function of minimizing the energy consumption of the whole network. At present, SDN-based edge computing technology has become the main technology and development trend. In the edge computing architecture of SDN, there are three kinds of devices: controller, repeater, and remote server. The controller adopts the active-standby redundancy mode to manage the resources of the whole network. Under the management of the controller, the transponder realizes the functions of calculation, storage and transmission of the whole network data. There are three types of forwarders: network nodes, storage nodes, and computing nodes. Among them, the storage node provides network transmission function and data storage function, and the computing node provides network transmission function and service calculation function.

The service usage process under the SDN-based edge computing architecture mainly includes three processes: the control process, the user's use of storage resources, and the user's use of computing resources.

(1) Control flow: The controller interacts with the repeater and the remote server to register and configure the resource status.

(2) User's process of using storage resources: the user initiates a content access request to the nearest storage node; the storage node checks whether there is any content resource required by the user; if not, requests the content resource from the remote server and stores the returned content resource ; The storage node returns the requested resource to the user.

(3) The user's use of computing resources process: the user submits a computing request to the nearest computing node, and the computing node calculates the user's computing task and returns the result to the user.

Through the analysis of the service usage process under the SDN-based edge computing architecture, it can be seen that in order to meet the needs of users, it is necessary to increase the number of storage nodes and computing nodes deployed in the network.

表示时间t内服务请求的到达数量，每个内容服务请求的存储内容大小使用表示。在计算服务请求方面，使用

表示时间t内服务请求的到达数量，每个服务请求的任务执行时长使用

表示，服务过程中需要的通信量使用

表示。Under the SDN-based edge computing architecture, service types include content service requests and computing service requests. Among them, the content service request is represented by ka ^{∈ KA, and the computing service request is represented by k b} _∈ ^KB _. In terms of content service requests, use

Indicates the arriving number of service requests within time t, and the storage content size of each content service request is expressed using the representation. In computing service requests, use

Indicates the number of service requests arriving in time t, and the task execution time of each service request is used

Indicates the traffic usage required during the service

express.

表示，计算方法如公式(1)。In the edge computing network environment, energy consumption includes storage energy consumption, computing energy consumption, and transmission energy consumption. Storage energy consumption is mainly consumed by storage nodes, using

Representation, the calculation method is as formula (1).

In the formula, p _ca is the energy consumption of the average storage consumption, and the unit is J/(bitgs).

和静态计算能耗

两种。动态计算能耗的计算方法如公式(2)。静态计算能耗的计算方法如公式(3)。Computing energy consumption is mainly consumed by computing nodes, including dynamic computing energy consumption

and static computing energy consumption

two kinds. The calculation method of dynamic calculation energy consumption is as formula (2). The calculation method of static calculation energy consumption is as formula (3).

为时间t内服务请求的到达数量。

为服务请求的任务执行时长。

为静态环境下VM的数量。p_static为执行计算任务的虚拟机副本在静态下的平均功耗，单位J/s。In the formula, p _active is the average power consumption of the virtual machine copy executing the computing task under the computing dynamics, the unit is J/(bitgs).

is the arrival number of service requests within time t.

The duration of the task execution time requested for the service.

is the number of VMs in a static environment. p _static is the average power consumption of the virtual machine copy performing the computing task under static state, in J/s.

使用公式(4)计算。计算传输能耗

使用公式(5)计算。Transmission energy consumption is mainly consumed by network nodes, including content transmission energy consumption and computing transmission energy consumption. Content transmission energy consumption

Calculated using formula (4). Calculate transmission energy consumption

Calculated using formula (5).

为存储服务请求的内容大小，p_link为链路的能耗参数，单位J/bit。p_node为网络节点的能耗参数，单位J/bit。

为内容服务请求到服务节点的平均跳数。

为计算服务过程中需要的通信量。

为计算服务请求到服务节点的平均跳数。In the formula,

is the content size of the storage service request, and p _link is the energy consumption parameter of the link, in J/bit. p _node is the energy consumption parameter of the network node, in J/bit.

Average number of hops to service nodes for content service requests.

To calculate the amount of traffic required in the service process.

Average number of hops to service nodes for computing service requests.

Based on the above analysis, the total energy consumption is defined as Equation (6).

Therefore, the present invention defines the objective function of minimizing energy consumption as formula (7).

为存储节点的最优数量，

为计算节点的最优数量，N为存储节点的最大数量，M为计算节点的最大数量。In the formula,

is the optimal number of storage nodes,

is the optimal number of computing nodes, N is the maximum number of storage nodes, and M is the maximum number of computing nodes.

The network resource energy consumption model includes the above-mentioned equations (6) and (7) and their operation and execution processes.

The network resource deployment model storage module is used to store the network resource deployment model constructed by the particle swarm optimization algorithm. In the particle swarm optimization algorithm, the deployment scheme of network resources is used as the particle position, and the optimization method of the network resource deployment scheme is used as the movement of particles. speed. The analysis of the intelligent search algorithm shows that the particle swarm optimization algorithm is a highly efficient global random search algorithm, which has been successfully applied to the solution of optimization problems. Preferably, the nodes in this embodiment include storage nodes and computing nodes, that is, network resources to be deployed include storage nodes and computing nodes. Therefore, in this embodiment, the particle swarm algorithm is used to derive the optimal number of optimal storage nodes and the optimal number of computing nodes.

The main idea of the particle swarm optimization algorithm is to describe the deployment scheme of network resources as the position of a particle; wherein, the deployment scheme of network resources is the number of storage nodes and the number of computing nodes. The particle then moves in these solution spaces in a direction determined by the particle's historical best position X _pb and the neighborhood's historical X _gb best position. Therefore, the key content includes four aspects: position, speed, position update, and speed update. A detailed description is given below.

其中，元素

表示第j个任务获取资源的网络资源的编号。(1) Location: The deployment scheme of network resources is used as the particle location. Assuming that the deployment scheme of the ith network resource includes n tasks to be completed, the particle position of the solution is expressed as

Among them, the element

Indicates the number of the network resource from which the jth task obtains the resource.

则该解的粒子移动速度表示为

其中，元素

取值为1和0。当

表示当前任务需要更新获取的网络资源的编号。当

表示当前任务不需要更新获取的网络资源的编号。(2) Speed: The optimization method of the deployment scheme of network resources is taken as the moving speed of the particles. Assume that the deployment scheme of the i-th network resource is

Then the particle moving speed of the solution is expressed as

Among them, the element

Takes values 1 and 0. when

Indicates the number of the network resource acquired by the current task that needs to be updated. when

Indicates the number of the acquired network resource that does not need to be updated for the current task.

是指速度V_i+1中

的任务使用的资源，在位置X_i中需要进行调整。(3) Position update: position update is performed based on the position at the previous moment and the speed at the current moment, and the calculation method is as shown in formula (8). The formula (8) indicates that the position X _i+1 is the product operation of the position X _i at the previous moment and the velocity V _i+1 at the current moment. Among them, the product operation

refers to the velocity V _i+1 in

The resources used by the tasks of , need to be adjusted in location X _i .

是指对应元素在指定概率下的相加运算。X_pbΘX_i和X_gbΘX_i之间的运算符Θ是指减操作，用于比较两个粒子位置的异同。当两个粒子的相同位置上的元素相同时，减操作结果为1；当两个粒子的相同位置上的元素不同时，减操作结果为0。(4) Speed update: the speed is updated based on the position and speed at the last moment, and the calculation method is as formula (9). Equation (9) indicates that the velocity V _i+1 is a sum operation of three values of P ₁ V, P ₂ (X _pb ΘX _i ), and P ₃ (X _gb ΘX _i ). Among them, P ₁ , P ₂ , and P ₃ are three constants, representing the probability of whether the three values are updated, and P ₁ +P ₂ +P ₃ =1. and operation

Refers to the addition operation of corresponding elements under the specified probability. The operator Θ between X _pb ΘX _i and X _gb ΘX _i refers to the subtraction operation and is used to compare the similarities and differences of two particle positions. When the elements at the same position of the two particles are the same, the result of the subtraction operation is 1; when the elements at the same position of the two particles are different, the result of the subtraction operation is 0.

A network resource deployment module, configured to invoke the network resource energy consumption model and the network resource deployment model, take the network resource energy consumption model as a judgment condition, perform an iterative calculation on the network resource deployment model, and output the network resource deployment model solution, the network resource deployment solution includes the optimal number of storage nodes and/or the optimal number of computing nodes. The network resource deployment model includes the above-mentioned formulas (8) and (9) and their operation and execution processes. The iterative calculation includes the following five steps: initializing the parameters, calculating the initial position, updating the particle velocity and particle position, updating the global optimal initial position and individual optimal initial position, and judging whether the end condition is reached. specifically,

The input of the method is the number of tasks n, the maximum number of storage nodes N, the maximum number of computing nodes M, the number of iterations MG and the size of the particle swarm O, and the output is the optimal particle position X _i . Among them, the number of iterations MG and the size of particle swarm O are self-defined. The value of the iteration number MG is generally between 20 and 30. The value of particle swarm size N is generally between 5 and 10. Specific steps are as follows:

Step 1. Parameter initialization. The initialization parameters include the number of iterations MG, the particle swarm scale O, the initial position X _i of the randomly generated particles, and the initial velocity V _i of the randomly generated particles. Since the size of the particle swarm is O, in step 1, O "initial positions X _i of randomly generated particles" and O "initial speeds V _i of randomly generated particles" can be obtained.

Step 2: Calculate the initial position. According to the aforementioned "use the deployment scheme of network resources as particle positions", therefore, from O "randomly generated initial positions X _i of particles" in step 1, 0 "deployment schemes of network resources" can be obtained. In this embodiment, the network resource energy consumption model is called through the following steps: the network resource deployment module reads the network resource energy consumption model from the network resource energy consumption model storage module, and according to the network resource energy consumption model contains The instruction reads O "network resource deployment plans", and obtains O energy consumptions of the whole network according to the calculation strategy described by Equation (6). After obtaining O energy consumptions of the whole network by calling the network resource energy consumption model, the "network resource deployment plan" corresponding to the minimized minimum energy consumption of the whole network is set as the optimal initial particle position X _i . The optimal initial position X _i is set as the global optimal initial position X _gb , and the initial position X _i of each particle is set as the individual optimal initial position X _pb .

和计算节点的最优数量

即以能耗最小化的目标函数公式(6)和(7)作为判断条件。如符合约束条件，则使用公式(8)、公式(9)分别对粒子速度、粒子位置进行更新；如不符合约束条件，随机生成新的粒子位置和粒子速度。本实施例中，通过以下步骤调用所述网络资源部署模型：网络资源部署模块从网络资源部署模型存储模块读取所述网络资源部署模型，并根据所述网络资源能耗模型包含的指令读取O个“随机生成粒子的初始位置X_i”和O个“随机生成粒子的初始速度V_i”，根据式(8)和(9)描述的计算策略，分别对粒子速度、粒子位置进行更新。步骤三得到更新后的粒子位置，更新后的粒子位置对应新的网络资源部署方案。Step 3: Update the particle velocity and particle position respectively. First, determine whether the position of each particle meets the constraints, and perform corresponding operations. Preferably, the constraints are the optimal number of computing nodes and the optimal number of storage nodes. In this embodiment, taking the network resource energy consumption model as the judgment condition means: judging whether the initial position of each particle after the update conforms to the optimal number of storage nodes

and the optimal number of compute nodes

That is, the objective function formulas (6) and (7) of energy consumption minimization are used as judgment conditions. If the constraints are met, the particle velocity and particle position are updated using formula (8) and formula (9) respectively; if the constraints are not met, new particle positions and particle speeds are randomly generated. In this embodiment, the network resource deployment model is invoked through the following steps: the network resource deployment module reads the network resource deployment model from the network resource deployment model storage module, and reads the network resource energy consumption model according to the instructions contained in the network resource energy consumption model. O "initial positions X _i of randomly generated particles" and O "initial velocities V _i of randomly generated particles", according to the calculation strategies described by equations (8) and (9), respectively update the particle velocity and particle position. In step 3, the updated particle positions are obtained, and the updated particle positions correspond to the new network resource deployment scheme.

Step 4: Update the new global optimal initial position and the individual optimal initial position respectively. When f(X _i )<f(X _pb ), set X _pb =X _i ; the network resource deployment module reads the network resource deployment model from the network resource deployment model storage module, and according to the network resource energy consumption model The included instruction reads "the new network resource deployment scheme corresponding to the updated particle position", and obtains the entire network energy consumption f(X _i ) corresponding to the new network resource deployment scheme according to the calculation strategy described in equation (6). Compare the whole network energy consumption f(X _i ) corresponding to the new network resource deployment scheme and the whole network energy consumption f(X _pb ) corresponding to the individual optimal initial position X _pb , if f(X _i )<f(X _pb ), Then, the updated particle position is set as the individual optimal initial position X _pb . Then, when f(X _pb )<f(X _gb ), set X _gb =X _pb , where f(X _pb ) is the network-wide energy consumption f(X _i corresponding to the new network resource deployment scheme in this step ), compare the energy consumption f(X _gb ) of the whole network corresponding to the optimal initial position X _gb of the whole network, if f(X _pb )<f(X _gb ), then set the updated particle position as the global optimum Initial position X _gb .

和计算节点的最优数量

若否，则返回步骤三，继续迭代直至输出最优的X_i，即可得到“网络资源的部署方案”，即存储节点的最优数量

和计算节点的最优数量

Step 5: Determine whether the global optimal initial position X _gb obtained in step 4 has reached the end condition. The end condition is the maximum number of iterations MG. If yes, then output the global optimal initial position X _gb obtained in step 4 as the optimal X _i , where the optimal X _i is the updated particle position obtained in step 3, and the updated particle position corresponds to the new one. Therefore, the optimal network resource deployment scheme, that is, the optimal number of storage nodes, can be output.

and the optimal number of compute nodes

If not, go back to step 3, continue to iterate until the optimal X _i is output, and then the "network resource deployment plan" can be obtained, that is, the optimal number of storage nodes

and the optimal number of compute nodes

Preferably, a network resource deployment and allocation module may also be included: after the network resource deployment and allocation module receives the network resource deployment scheme output by the network resource deployment module (that is, the optimal number of storage nodes and/or the optimal number of computing nodes), Performs the allocation of this network resource deployment.

Example 2

This embodiment relates to a particle swarm-based low-energy-consumption edge computing resource deployment method, as shown in FIG. 2 , including the following steps:

a. Store the network resource energy consumption model including the calculation formula of total energy consumption and the objective function of minimizing energy consumption of the whole network;

b. Store the network resource deployment model constructed by the particle swarm optimization algorithm. In the particle swarm optimization algorithm, the deployment scheme of network resources is used as the particle position, and the optimization method of the network resource deployment scheme is used as the moving speed of the particles;

c. Invoke the network resource energy consumption model and the network resource deployment model, take the network resource energy consumption model as a judgment condition, perform an iterative calculation on the network resource deployment model, and output a network resource deployment scheme. The resource deployment scheme includes the optimal number of storage nodes and/or the optimal number of computing nodes.

It should be noted that, the method in this embodiment can be implemented by using the corresponding modules, devices, etc. in the particle swarm-based low-energy edge computing resource deployment system in Embodiment 1, and those skilled in the art can refer to the technical solutions of the system The flow of steps for implementing the method, that is, the implementation in the system can be understood as a preferred example for implementing the method, which is not repeated here.

Example 3

This embodiment relates to the performance verification of the particle swarm-based low-energy edge computing resource deployment system of Embodiment 1, and specifically uses US64 [12] to simulate the network topology environment (for the simulated network topology environment, please refer to the literature Choi N, Guan K, Kilper DC, et al. In-network caching effect on optimal energy consumption in content-centric networking [C]//2012 IEEE international conference on communications (ICC). IEEE, 2012: 2889-2894). Among them, US64 is a typical topology of a commercial network, which can effectively describe the characteristics of network nodes in a commercial network. In terms of performance parameters of network nodes, set the power consumption of static computing resources to 50W, set the energy consumption of storage nodes to 3* ^10-7 J/bit, and set the energy consumption of link resources to 1 ^{* 10-7} J/bit, set the energy consumption of network nodes to 2*10 ^-7 J/bit. In the number of service requests, set the number of service requests arriving per second from 1 to 25. The maximum number of storage nodes is set to 80, and the maximum number of compute nodes is set to 20.

和计算节点

的部署数量方面的优劣，将本发明系统RDAoPS与传统算法RDAoRN(Resource Deployment Algorithm based onRequest Number)进行了比较。其中，算法RDAoRN是根据网络拓扑的特点和服务请求的数量进行存储节点和计算节点的部署。本实施例在不同的服务请求到达率环境下，验证了两种算法的存储节点和计算节点数量。比较结果如图3和图4所示。In order to verify that the system of the present invention solves the storage node

and compute nodes

Compared with the traditional algorithm RDAoRN (Resource Deployment Algorithm based on Request Number), the system RDAoPS of the present invention has the advantages and disadvantages in terms of the number of deployments. Among them, the algorithm RDAoRN deploys storage nodes and computing nodes according to the characteristics of the network topology and the number of service requests. This embodiment verifies the number of storage nodes and computing nodes of the two algorithms under different service request arrival rate environments. The comparison results are shown in Figures 3 and 4.

In Figure 3, the X-axis represents the service request arrival rate increased from 5 to 30 per second, and the Y-axis represents the number of storage nodes. As can be seen from the figure, as the arrival rate of service requests increases, the number of storage nodes under both algorithms increases. This is because the service demand increases, and more storage nodes need to be deployed to meet the service demand. It can be seen from the comparison of the results of the two algorithms that when the service request arrival rate of the system of the present invention reaches 20 per second, the number of deployed storage nodes tends to be stable. This shows that the location of the storage node deployed by the present invention is relatively reasonable.

In Figure 4, the X-axis also represents the service request arrival rate, and the Y-axis represents the number of computing nodes. As can be seen from the figure, as the arrival rate of service requests increases, the number of computing nodes under both algorithms increases. This is because the service demand increases, and more computing nodes need to be deployed to meet the service demand. It can be seen from the comparison of the results of the two algorithms that when the service request arrival rate of the system of the present invention reaches 25 per second, the number of deployed computing nodes tends to be stable. This shows that the positions of the computing nodes deployed by the present invention are relatively reasonable.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the scope of the present invention. within the scope of protection.

Claims (10)

1. A low-energy-consumption edge computing resource deployment system based on particle swarm is characterized in that: the method comprises the steps of obtaining the optimal number of computing nodes and/or the optimal number of storage nodes in the edge computing resource; comprises that

The network resource energy consumption model storage module is used for storing a network resource energy consumption model comprising a total energy consumption calculation formula and a whole network energy consumption minimization objective function;

the system comprises a network resource deployment model storage module, a particle swarm optimization algorithm and a particle deployment model generation module, wherein the network resource deployment model storage module is used for storing a network resource deployment model constructed by the particle swarm optimization algorithm, and the particle swarm optimization algorithm takes a deployment scheme of network resources as the positions of particles and takes an optimization mode of the network resource deployment scheme as the moving speed of the particles;

and the network resource deployment module is used for calling the network resource energy consumption model and the network resource deployment model, carrying out iterative computation on the network resource deployment model by taking the network resource energy consumption model as a judgment condition, and outputting a network resource deployment scheme, wherein the network resource deployment scheme comprises the optimal number of the storage nodes and/or the optimal number of the computing nodes.

2. The particle swarm-based low energy edge computing resource deployment system of claim 1, wherein: the total energy consumption calculation formula is as follows

Wherein F is total node energy consumption, k^a∈K_AServing the request for content, k^b∈K_BIn order to compute the request for a service,

in order to store the energy consumption of the nodes,

in order to dynamically calculate the energy consumption,

in order to calculate the energy consumption statically,

the energy consumption for the transmission of the content is,

to calculate the transmission energy consumption.

3. The particle swarm-based low energy edge computing resource deployment system of claim 1, wherein: the energy consumption minimization objective function of the whole network is as follows

In the formula (I), the compound is shown in the specification,

for the optimal number of storage nodes to be used,

for the optimal number of compute nodes, N is the maximum number of storage nodes, and M is the maximum number of compute nodes.

4. The particle swarm-based low energy edge computing resource deployment system of any of claims 1-3, wherein: the iterative computation comprises the following steps:

step one, initializing parameters; the initialized parameters comprise iteration number MG, particle swarm size O and initial position X of randomly generated particles_iInitial velocity V of randomly generated particles_i；

Step two, calculating an initial position; obtaining the minimum total network energy consumption according to the initialized parameters obtained in the step one and the total energy consumption calculation formula of each node; obtaining a deployment scheme of network resources and an optimal particle initial position X according to the minimized whole network energy consumption_iSetting the optimal initial position X_iSet as the global optimum initial position X_gbThe initial position X of each particle_iSet as the individual optimum initial position X_pb；

Step three, respectively updating the particle speed and the particle position; judging whether the initial position of each particle meets the constraint condition, if so, respectively updating the particle speed and the particle position; if not, randomly generating a new particle position and a new particle speed;

respectively updating the new global optimal initial position and the individual optimal initial position; when f (X)_i)＜f(X_pb) When, set up X_pb＝X_i(ii) a When f (X)_pb)＜f(X_gb) When, set up X_gb＝X_pb；

Step five, judging whether an ending condition is reached; if yes, outputting the optimal X_iAnd if not, returning to the step three.

5. The particle swarm-based low energy edge computing resource deployment system of claim 4, wherein: the constraint conditions are the maximum number of storage nodes and the maximum number of calculation nodes.

6. A low-energy-consumption edge computing resource deployment method based on particle swarm is characterized in that: obtaining the optimal number of computing nodes and/or the optimal number of storage nodes in the edge computing resources; comprises the following steps

a. Storing a network resource energy consumption model comprising a total energy consumption calculation formula and a whole network energy consumption minimization objective function;

b. storing a network resource deployment model constructed by a particle swarm optimization algorithm, wherein the deployment scheme of network resources is used as the positions of particles in the particle swarm optimization algorithm, and the optimization mode of the network resource deployment scheme is used as the moving speed of the particles;

c. and calling the network resource energy consumption model and the network resource deployment model, carrying out iterative computation on the network resource deployment model by taking the network resource energy consumption model as a judgment condition, and outputting a network resource deployment scheme, wherein the network resource deployment scheme comprises the optimal number of storage nodes and/or the optimal number of computing nodes.

7. The particle swarm-based low-energy-consumption edge computing resource deployment method of claim 6, wherein: the total energy consumption calculation formula is as follows

Wherein F is total node energy consumption, k^a∈K_AServing the request for content, k^b∈K_BIn order to compute the request for a service,

in order to store the energy consumption of the nodes,

in order to dynamically calculate the energy consumption,

in order to calculate the energy consumption statically,

the energy consumption for the transmission of the content is,

to calculate the transmission energy consumption.

8. The particle swarm-based low-energy-consumption edge computing resource deployment method of claim 6, wherein: the energy consumption minimization objective function of the whole network is as follows

In the formula (I), the compound is shown in the specification,

for the optimal number of storage nodes to be used,

for the optimal number of compute nodes, N is the maximum number of storage nodes, and M is the maximum number of compute nodes.

9. The particle population-based low energy edge computing resource deployment method of any one of claims 6-8, wherein: the iterative computation comprises the following steps:

c1, initializing parameters; the initialized parameters comprise iteration number MG, particle swarm size O and initial position X of randomly generated particles_iInitial velocity V of randomly generated particles_i；

c2, calculating an initial position; obtaining the minimum total network energy consumption according to the initialized parameters obtained in the step c1 and the total energy consumption calculation formula of each node; obtaining a deployment scheme of network resources and an optimal particle initial position X according to the minimized whole network energy consumption_iSetting the optimal initial position X_iSet as the global optimum initial position X_gbThe initial position X of each particle_iSet as the individual optimum initial position X_pb；

c3, respectively updating the particle speed and the particle position; judging whether the initial position of each particle meets the constraint condition, if so, respectively updating the particle speed and the particle position; if not, randomly generating a new particle position and a new particle speed;

c4, updating the new global optimal initial position and the individual optimal initial position respectively; when f (X)_i)＜f(X_pb) When, set up X_pb＝X_i(ii) a When f (X)_pb)＜f(X_gb) When, set up X_gb＝X_pb；

c5, judging whether the ending condition is reached; if yes, outputting the optimal X_iIf not, return to step c 3.

10. The particle swarm-based low-energy-consumption edge computing resource deployment method of claim 9, wherein: the constraint conditions are the maximum number of storage nodes and the maximum number of calculation nodes.

Images