A user association and resource allocation method and device for power multi-layer edge computing network

Abstract

The present invention discloses a method and device for user association and resource allocation for a multi-layer edge computing network for electric power, comprising the following steps: analyzing the communication performance of the multi-layer edge computing network; constructing four delays in edge computing tasks based on the analyzed communication performance, including local computing delay, upload delay of unloaded part of the data, processing delay of the edge server, and return delay of the processing result of the edge processor; constructing an optimization problem of minimizing the average delay of all power terminal task calculations based on the four delays in the edge computing task, solving the optimization problem, and obtaining the result of user association and resource allocation. To reasonably plan the connection relationship between the power terminal and the edge server, the allocation of computing power and spectrum, so as to achieve the minimum delay of edge computing.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication technology, and specifically relates to a user association and resource allocation method and device for a power multi-layer edge computing network. Background Art

[0002] With the rapid development of emerging technologies such as the Internet of Things (IoT), big data, and artificial intelligence, traditional computing network architectures are unable to cope with centralized processing of large amounts of power grid data, resulting in high network latency, bandwidth consumption, and other issues. Edge computing, as a distributed architecture, deploys computing resources at the edge of the network so that part of the data of the power terminal is calculated on the edge node to reduce the burden on the central server of the power grid, improve computing efficiency, and reduce network latency. The key to multi-layer edge computing in power is to offload tasks from the power terminal device layer and the edge layer. According to different scenario requirements, it flexibly chooses to perform computing tasks between different layers to achieve optimal resource allocation and maximize performance.

[0003] In the offloading allocation task of edge computing, many factors need to be considered. On the one hand, due to the uneven channel quality between users and base stations, the information exchange between power terminals and base stations will cause transmission delays. Different terminals need to select the most suitable base station to upload tasks to achieve the best overall performance; on the other hand, the performance of terminals and base stations is limited, and the terminal needs to have a task allocation relationship with the selected base station. Therefore, it is necessary to comprehensively consider the delays and performance limitations in each process while trying to improve performance and reduce overall delays as much as possible.

[0004] There are many strategies and algorithms to solve the problem of task offloading in edge computing networks. Starting from the communication delay and computing delay, the resources of power terminals and base stations are dispatched to reduce the delay or reduce energy consumption.

[0005] In order to solve the problem of idle resources due to the existence of idle remote edge servers in the edge computing network offloading environment, Reference 1 (Computational Task Offloading Algorithm in Multi-edge Node Scenario) proposes an improved shuffle frog offloading strategy, which considers minimizing the weighted sum while controlling the transmission energy consumption and the overall execution time of the model. Each update process is divided into decision vector update and offloading vector update. The decision vector updates the decision information through XOR and mutation operations, that is, offloading part of the tasks of the multi-task terminal to the edge server. The offloading vector is updated based on the adaptive step size of the old offloading vector to distribute the assigned tasks to different edge servers, and iterates continuously until convergence. Reference 2 (Task Offloading Strategy with Caching Mechanism in Mobile Edge Computing) considers the situation under the constraints of edge cache capacity, resource request capacity, etc., and optimizes the energy consumption cost of the device while ensuring the task execution delay. In this problem, the delay caused by communication is mainly composed of the upload and download rate of data, and the computing delay is mainly composed of the local computing delay and the overall computing delay of the edge. The goal is to minimize the energy consumption cost of the device while ensuring the terminal service quality.

[0006] Some existing solutions do not have a system model that takes into account factors such as base station computing power, energy, and spectrum allocation. When there are too many offloaded computing tasks, the edge server cannot handle a large number of tasks at the same time, resulting in an increase in the overall computing delay at the edge, thereby reducing the overall communication efficiency. At the same time, the edge server is power-limited. When the offloaded task is transmitted back, if the power is insufficient, the signal-to-noise ratio of the transmitted signal will be too low, causing signal distortion. The divisibility of power terminal tasks requires the terminal to consider the computing power, power and other conditions of the base station when selecting different edge servers to select the appropriate offloading strategy. Therefore, in a multi-layer edge computing network, the various limitations of terminals and edge servers make it a huge challenge to reasonably associate users and allocate resources. Summary of the invention

[0007] In view of the above, the purpose of the present invention is to provide a user association and resource allocation method and device for a power multi-layer edge computing network to rationally plan the connection relationship between power terminals and edge servers, as well as the allocation of computing power, power and spectrum, so as to achieve minimum latency for edge computing.

[0008] To achieve the above-mentioned purpose of the invention, an embodiment provides a user association and resource allocation method for a power multi-layer edge computing network, which is applied to a multi-layer edge computing network composed of a power terminal, a small base station or a macro base station equipped with an edge server, and includes the following steps:

[0009] Analyze the communication performance of multi-layer edge computing networks;

[0010] Based on the analyzed communication performance, four delays in edge computing tasks are constructed, including local computing delay, upload delay of unloaded data, processing delay of edge server, and return delay of edge processor processing results;

[0011] Based on the four delays in edge computing tasks, an optimization problem is constructed to minimize the average delay of all power terminal task calculations;

[0012] Calculate reference quantities for selecting terminals and base stations based on channel quality and computing resources;

[0013] By optimizing user association, spectrum allocation, computing power allocation, and task offloading strategies, the average latency of all terminal task calculations can be minimized, the optimization problem can be solved, and the results of user association and resource allocation can be obtained.

[0014] Preferably, analyzing the communication performance of the multi-layer edge computing network includes:

[0015] Power terminal k selects small base station s to connect to, using variable a k,s To represent the connection status, where s=0 represents the connection status with the macro base station. Each terminal can only select one base station for connection, but one base station can connect with multiple terminals, so:

[0016]

[0017] Each small base station and macro base station are allocated orthogonal channels with bandwidth B. Within each small base station, the connected terminals reuse uplink channels and downlink channels. Define is the spectrum ratio of the uplink channel allocated by base station s to terminal k, then the uplink rate of terminal k It is expressed as:

[0018]

[0019] Among them, p k represents the transmission power of terminal k, h k,s represents the channel power gain between terminal k and base station s, N0 represents the power spectrum density of noise, and the sum of the spectrum allocated by base station s to all associated terminals cannot exceed the actual resource limit

[0020]

[0021] definition is the spectrum ratio of the downlink channel allocated to terminal k by base station s, and the downlink rate obtained by terminal k It is expressed as:

[0022]

[0023] Among them, p k,s is the power allocated by base station s to terminal k. Due to the limited power of base stations and the limited spectrum resources of downlink channels, the sum of the power of all base stations and the sum of the frequency resources of downlink channels cannot exceed the total energy P of the base stations in practice. s and total spectrum resources

[0024]

[0025] Preferably, four delays in edge computing tasks are constructed based on the analyzed communication performance, including:

[0026] Terminal k has d k bit computing task, which is calculated according to the ratio ρ k Split, 0≤ρ k ≤1, the amount of computing task data in the terminal is ρ k d k , the remaining data is processed by the edge server of the base station. The amount of data processed is determined by the task. It is assumed that the amount of data processed is proportional to the amount of data in the original computing task. k d k ;

[0027] Local computing latency Refers to the time spent on computing tasks stored locally at the power terminal:

[0028]

[0029] Among them, f k represents the computing resources of terminal k, Indicates the computing power required to process 1 bit of computing task;

[0030] Offload some data upload delays It refers to the time it takes for data to be transmitted from the terminal to the base station when the terminal uploads the data to the base station:

[0031]

[0032] Edge server processing delay It refers to the time required for the base station to process the task by relying on the computing resources allocated to the terminal after the offload task arrives at the base station:

[0033]

[0034] The latency of edge processor processing results being sent back It refers to the time it takes for the edge server to transmit the task result data back to the terminal after the calculation is completed:

[0035]

[0036] Among them, f k,s It represents the computing power allocated by base station s to terminal k.

[0037] Preferably, an optimization problem of minimizing the average delay of all power terminal task calculations is constructed based on the four delays in the edge computing task, including:

[0038]

[0039] Where K is the number of terminals.

[0040] Preferably, the process of solving the optimization problem includes:

[0041] Step 1: Initialize all variables in the optimization problem and randomly allocate all computing resources;

[0042] Step 2: Optimize the user association problem, keep other variables fixed, and optimize the connection variable a k,s , that is, selecting the base station with the lowest computation and backhaul delay cost after association offloading to achieve user association;

[0043] Step 3: Optimize the offloading decision problem, fix other variables, and optimize the computing task offloading ratio ρ k , the problem has a closed-form solution, the optimal ρ k The local computing delay of the terminal and the overall processing delay of the edge server are kept consistent, that is:

[0044]

[0045] Step 4: Optimize variables related to resource allocation separately p k,s 、f k,s , including: using the interior point method to find the optimal variables And variables p k,s ; For variable f k,s , there exists a closed-form solution, which makes the base station keep the edge offloading task calculation delay consistent for all users associated with it. For base station m, suppose there are l terminals associated with it, denoted as j1, j2, ..., j l , then the computing power allocation of base station m satisfies:

[0046]

[0047] Step 5, repeat steps 2 to 4 until the given cycle number I is reached.

[0048] Preferably, the method further proposes a new user association scheme, that is, constructing a low-complexity reference quantity for measuring the selection of power terminals and base stations based on channel quality and computing power resources, and the terminal selects the base station with the smallest reference quantity to achieve user association, including:

[0049]

[0050] Among them, M k,s represents the reference quantity selected by terminal k and base station s, SNR k,s Represents the signal-to-noise ratio between the terminal and the base station, which is used to measure the channel quality and satisfies B is the total bandwidth, p k is the terminal's transmit power, N0 is the power spectrum density of the noise, θ k is an empirical parameter.

[0051] Preferably, the process of using channel quality and computing resources to construct a reference quantity for measuring the selection of power terminals and base stations to perform user association includes:

[0052] The base station obtains channel information and its own remaining computing resources and sends them to the terminal;

[0053] The terminal uses empirical parameters θ and channel quality H k,s and the remaining computing resources of each base station f k Calculate the reference quantity M k,s ;

[0054] Terminal selection reference quantity M k,s The smallest base station to achieve user association.

[0055] To achieve the above-mentioned purpose of the invention, an embodiment also provides a user association and resource allocation device for a multi-layer edge computing network for electric power, including a memory and one or more processors, wherein the memory stores executable code, and the device is applied to a multi-layer edge computing network composed of electric power terminals, small base stations or macro base stations equipped with edge servers. When the one or more processors execute the executable code, they are used to implement the above-mentioned user association and resource allocation method for the multi-layer edge computing network for electric power.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] When optimizing the calculation of user association and resource allocation, we comprehensively consider multiple constraints such as user association, task offloading, power, computing power, and spectrum, and introduce a user-associated base station strategy that weights channel quality and computing power resources. Based on this, we rationally plan the connection relationship between power terminals and edge servers, as well as the allocation of computing power, power, and spectrum, to achieve the minimum latency of edge computing. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] 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 creative work.

[0059] Figure 1 It is a flow chart of a method for user association and resource allocation for a power multi-layer edge computing network provided by an embodiment;

[0060] Figure 2 is a task processing process diagram of an edge computing network provided by an embodiment;

[0061] Figure 3 This is a schematic diagram of edge computing in the power Internet of Things provided by an embodiment;

[0062] Figure 4 is a flow chart for solving the optimization problem provided by the embodiment;

[0063] Figure 5 It is another solution flow chart of the optimization problem provided by the embodiment. DETAILED DESCRIPTION

[0064] To make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation methods described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0065] The inventive concept of the present invention is: Edge computing is an architecture that moves computing and data storage from a central server to the edge of a network. Its emergence improves overall resource utilization efficiency and data processing speed. However, in the multi-layer edge computing network of the electric power Internet of Things, the computing power of the power terminal is limited, and the computing power and power of the edge server are limited. To this end, the present invention proposes a user association and resource allocation method and system for a multi-layer edge computing network for electric power to rationally plan the connection relationship between the terminal and the edge server, as well as the allocation of computing power, power and spectrum, so as to achieve the minimum delay of edge computing.

[0066] In the embodiment, the power multi-layer edge computing network is composed of power terminals, small base stations or macro base stations equipped with edge servers. Edge offloading in the multi-layer edge computing network is a process in which the terminal sends part of the task to the base station for processing according to its own computing power, and the base station returns the processed data to the terminal. Figure 2 As shown, it mainly includes the following four processes:

[0067] (1) The terminal selects a base station, which can be a small base station or a macro base station. The terminal calculates the proportion of local computing tasks and base station computing tasks based on its own computing resources and the computing resources, power, and spectrum resource allocation of the base station.

[0068] (2) The terminal calculates the locally retained computing tasks, generates local computing delay as a latency measurement indicator, and sends the remaining tasks to the base station. These data will pass through the uplink channel between the terminal and the base station, generating the upload delay of the offloaded tasks;

[0069] (3) The base station allocates its own computing power to calculate the computing tasks sent by different terminals based on its own computing resources and the relationship with the terminals, resulting in edge offloading task computing delay;

[0070] (4) The base station returns the processed computing task to the terminal. The processed computing task passes through the downlink channel between the base station and the terminal, generating a backhaul delay. The delay generated by offloading the computing task to the edge end is composed of the upload delay of the offload task, the edge offload task computing delay, and the backhaul delay.

[0071] This is different from traditional user association, which depends on channel quality. However, when it comes to edge computing tasks, the remaining computing power or the demand for computing power also needs to be considered in order to reduce the overall latency.

[0072] The user association and resource allocation method for the power multi-layer edge computing network provided in the embodiment can be used for Figure 3 The edge computing scenario in the power Internet of Things shown in the figure, in which multiple small base stations equipped with edge servers and macro base stations equipped with edge servers form a multi-layer heterogeneous edge computing network, and K power communication terminals (referred to as power terminals) are randomly distributed in the network. Each terminal has computing tasks to process, such as video rendering, compression, etc. Assume that these computing tasks are divisible, that is, the tasks can be split and calculated. Considering the limited computing power of the terminal, if all computing tasks are performed locally, it will bring large computing overhead and delay. Therefore, the terminal can offload part of the computing tasks to the edge server associated with it. It is assumed that there are S small base stations and 1 macro base station in the network. There is an overlapping coverage area between the small base stations. The terminal can choose to connect to a small base station or a macro base station within the overlapping range. There are multiple spectrum reuse methods for the communication between the base station and the terminal to reuse the uplink and downlink channels. The spectrum allocated by the base station and the terminal for communication affects the time of uploading and downloading computing tasks. The higher the allocated spectrum, the higher the communication rate.

[0073] In the above edge computing scenario, Figure 1As shown, the user association and resource allocation method for the power multi-layer edge computing network provided by the embodiment includes the following steps:

[0074] S1, analyzes the communication performance of the power multi-layer edge computing network.

[0075] In the embodiment, when performing communication performance analysis, the power terminal k selects the small base station s for connection, and uses the variable a k,s To represent the connection status, where s=0 represents the connection status with the macro base station. Each terminal can only select one base station for connection, but one base station can connect with multiple terminals, so:

[0076]

[0077] Each small base station and the macro base station are allocated orthogonal channels with bandwidth B. Therefore, the interference between small base stations and the interference between small base stations and macro base stations are not considered. Within the range of each small base station, the connected terminals reuse the uplink channel and the downlink channel, and define is the spectrum ratio of the uplink channel allocated by base station s to terminal k, then the uplink rate of terminal k It is expressed as:

[0078]

[0079] Among them, p k represents the transmission power of terminal k, h k,s represents the channel power gain between terminal k and base station s, N0 represents the power spectrum density of noise, and the sum of the spectrum allocated by base station s to all associated terminals cannot exceed the actual resource limit

[0080]

[0081] definition is the spectrum ratio of the downlink channel allocated to terminal k by base station s, and the downlink rate obtained by terminal k It is expressed as:

[0082]

[0083] Among them, p k,s is the power allocated by base station s to terminal k. Due to the limited power of base stations and the limited spectrum resources of downlink channels, the sum of the power of all base stations and the sum of the frequency resources of downlink channels cannot exceed the total energy P of the base stations in practice. s and total spectrum resources

[0084]

[0085] S2, based on the analyzed communication performance, constructs four delays in edge computing tasks, including local computing delay, upload delay of offloaded part of the data, processing delay of the edge server, and return delay of the processing result of the edge processor.

[0086] In the embodiment, the power terminal k has d k bit computing task, which is calculated according to the ratio ρ k Split, 0≤ρ k ≤1, the amount of computing task data in the terminal is ρ k d k , the remaining data is processed by the edge server of the base station. The amount of data processed is determined by the task. It is assumed that the amount of data processed is proportional to the amount of data in the original computing task. k d k Therefore, the latency of edge computing is divided into four parts: the first is the local computing latency, the second is the upload latency of the unloaded data, the third is the processing latency of the edge server, and the fourth is the return latency of the edge server processing results. They can be expressed as:

[0087] Local computing latency Refers to the time spent on terminal computing for locally retained computing tasks:

[0088]

[0089] Among them, f k represents the computing resources of terminal k, It indicates the computing power required to process 1 bit of computing task, which can be measured by the number of CPU cycles;

[0090] Offload some data upload delays It refers to the time it takes for data to be transmitted from the terminal to the base station when the terminal uploads the data to the base station:

[0091]

[0092] Edge server processing delay It refers to the time required for the base station to process the task by relying on the computing resources allocated to the terminal after the offload task arrives at the base station:

[0093]

[0094] The latency of edge processor processing results being sent back It refers to the time it takes for the edge server to transmit the task result data back to the terminal after the calculation is completed:

[0095]

[0096] Among them, fk,s Represents the computing resources allocated by base station s to terminal k.

[0097] Therefore, the time delay T of the power terminal k calculation task is k It can be expressed as the higher value between the time consumed by local terminal computing and the total time consumed by edge computing upload, calculation, and return, that is:

[0098]

[0099] S3, based on the four delays in edge computing tasks, constructs an optimization problem to minimize the average delay of all power terminal task calculations.

[0100] In the embodiment, in order to reduce the computing task latency of all power terminals, it is necessary to minimize the average latency of all terminal task computing by optimizing user association, spectrum allocation, computing power allocation and task offloading strategy. The terminal needs to weigh the channel and computing power resources, select the appropriate base station for connection, and select the optimal task offloading strategy. The optimization problem is as follows:

[0101]

[0102] S4, by optimizing user association, spectrum allocation, computing power allocation and task offloading strategy, minimizes the average delay of all terminal task calculations, solves the optimization problem, and obtains the results of user association and resource allocation.

[0103] In the embodiment, the optimization problem is a mixed integer nonlinear optimization problem, which is a non-convex and NP-hard problem. Figure 4 As shown in Figure 2, the process of solving the optimization problem includes:

[0104] Step 1: Initialize all variables in the optimization problem and randomly allocate all computing resources.

[0105] Step 2: Optimize the user association problem, keep other variables fixed, and optimize the connection variable a k,s , its essence is a selection problem, that is, selecting the base station with the lowest computation and backhaul delay cost after association offloading.

[0106] In the embodiment, based on the overall optimization process, a better user-associated base station selection scheme is proposed in the edge computing scenario in the power Internet of Things. Unlike the traditional user association that looks at the channel quality, the invention also considers the impact of the remaining computing power or computing power demand on the base station selection, thereby reducing the overall delay. In the user association sub-problem, the base station is selected to obtain the minimum transmission delay, that is:

[0107]

[0108] Step 3: Optimize the offloading decision problem, fix other variables, and optimize the computing task offloading ratio ρ k , the unloading decision problem has a closed-form solution, the optimal ρ k The local computing delay of the power terminal and the overall processing delay of the edge server are kept consistent, that is:

[0109]

[0110] Step 4: Optimize variables related to resource allocation separately p k,s 、f k,s , including: using the interior point method to find the optimal variables p k,s ; For variable f k,s , there exists a closed-form solution, which makes the base station keep the edge offloading task calculation delay consistent for all users associated with it. For base station m, suppose there are l terminals associated with it, denoted as j1, j2, ..., j l , then the computing power allocation of base station m satisfies:

[0111]

[0112] Step 5, repeat steps 2 to 4 until the given cycle number I is reached.

[0113] In order to reduce the complexity of the algorithm, the example proposes a user association strategy based on channel quality and computing power based on the above steps. The specific method is as follows:

[0114] In the actual base station selection process, a reference quantity M based on channel quality and computing power can be constructed to measure the selection of power terminal k and base station s:

[0115]

[0116] Among them, M k,s represents the reference quantity selected by terminal k and base station s, SNR k,s represents the signal-to-noise ratio between terminal k and base station s, which is used to measure channel quality and satisfies θ k is an empirical parameter, through Calculate, ξ k Indicates the data compression ratio, represents the computing resources required by the base station to process the 1-bit task of terminal k, H k,s Indicates channel information.

[0117] Specifically, the terminal selects the base station with the smallest reference amount to achieve user association according to the channel information to different base stations and the remaining computing power of the corresponding base stations, including:

[0118] The base station obtains channel information and its own remaining computing resources and sends them to the terminal;

[0119] The terminal uses empirical parameters θ and channel quality H k,s and the remaining computing resources of each base station f k Calculate the reference quantity M k,s ;

[0120] Terminal selection reference quantity M k,s The smallest base station to achieve user association.

[0121] The solution process of the optimization problem based on the low-complexity user association strategy is as follows: Figure 5 shown.

[0122] Based on the same inventive concept, an embodiment further provides a user association and resource allocation device for a multi-layer edge computing network for electric power, including a memory and one or more processors, wherein an executable code is stored in the memory, and when one or more processors execute the executable code, the user association and resource allocation method for the multi-layer edge computing network for electric power is implemented, specifically including the following steps:

[0123] S1, analyzes the communication performance of the power multi-layer edge computing network;

[0124] S2, based on the analyzed communication performance, constructs four delays in edge computing tasks, including local computing delay, upload delay of offloaded data, processing delay of edge server, and return delay of edge processor processing results;

[0125] S3, based on the four delays in edge computing tasks, constructs an optimization problem to minimize the average delay of all power terminal task calculations;

[0126] S4, by optimizing user association, spectrum allocation, computing power allocation and task offloading strategy, minimizes the average delay of all terminal task calculations, solves the optimization problem, and obtains the results of user association and resource allocation.

[0127] The user association and resource allocation device for the electric power multi-layer edge computing network provided in the embodiment, at the hardware level, includes not only a processor and a memory, but also hardware required for other services such as an internal bus, a network interface, and a memory. The memory is a non-volatile memory, and the processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to implement the user association and resource allocation method for the electric power multi-layer edge computing network described in S1-S4 above. Of course, in addition to software implementation methods, the present invention does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0128] The specific implementation methods described above provide a detailed description of the technical solutions and beneficial effects of the present invention. It should be understood that the above is only the most preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, supplements and equivalent substitutions made within the scope of the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A user association and resource allocation method for a multi-layer edge computing network for electric power, characterized in that: The method is applied to a power multi-layer edge computing network composed of a power terminal and a small base station or a macro base station equipped with an edge server, and comprises the following steps: The communication performance of the multi-layer edge computing network of the power industry is analyzed, including: the power terminal k selects the base station s to connect to, and uses the variable a k,s To indicate the connection status, each terminal can only select one base station to connect, but one base station can connect with multiple terminals, then: Each small base station and macro base station are allocated orthogonal channels with bandwidth B. Within each small base station, the connected power terminals reuse uplink channels and downlink channels. Define is the spectrum ratio of the uplink channel allocated by base station s to terminal k, then the uplink rate of terminal k It is expressed as: Among them, p k represents the transmission power of terminal k, h k,s represents the channel power gain between terminal k and base station s, N0 represents the power spectrum density of noise, and the sum of the spectrum allocated by base station s to all associated terminals cannot exceed the actual resource limit definition is the spectrum ratio of the downlink channel allocated to the power terminal k by the base station s, and the downlink rate obtained by the terminal k It is expressed as: Among them, p k,s is the power allocated to the power terminal k by the base station s. Due to the limited power of the base station and the limited spectrum resources of the downlink channel, the sum of the power of all base stations and the sum of the frequency resources of the downlink channel cannot exceed the total energy P of the base station in practice. s and total spectrum resources Based on the analyzed communication performance, four delays in edge computing tasks are constructed, including local computing delay, upload delay of unloaded data, processing delay of edge server, and return delay of edge server processing results, including: power terminal k has d k bits of computing tasks, which are calculated according to the ratio ρ k Split, 0≤ρ k ≤1, the amount of computing task data in the terminal is ρ k d k , the remaining data is processed by the edge server of the base station. The amount of data processed is determined by the task. It is assumed that the amount of data processed is proportional to the amount of data in the original computing task. k d k ; Local computing latency Refers to the time spent on terminal computing for locally retained computing tasks: Among them, f k represents the remaining computing resources of terminal k, Indicates the computing power required to process 1 bit of computing task; Offload some data upload delays It refers to the time it takes for data to be transmitted from the terminal to the base station when the terminal uploads the data to the base station: Edge server processing delay It refers to the time required for the base station to process the task by relying on the computing resources allocated to the terminal after the offload task arrives at the base station: The latency of the edge server’s processing results being sent back It refers to the time it takes for the edge server to transmit the task result data back to the terminal after the calculation is completed: Among them, f k,s represents the computing power allocated to terminal k by base station s; Based on the four delays in edge computing tasks, the optimization problem of minimizing the average delay of all power terminal task calculations is constructed, which is expressed as: Where K is the number of terminals; By optimizing user association, spectrum allocation, computing power allocation, and task offloading strategies, the average latency of all terminal task calculations is minimized, the optimization problem is solved, and the results of user association and resource allocation are obtained, including: Step 1: Initialize all variables in the optimization problem and randomly allocate all computing resources; Step 2: Optimize the user association problem, keep other variables fixed, and optimize the connection variable a k,s , that is, selecting the base station with the lowest computation and backhaul delay cost after association offloading to achieve user association; Step 3: Optimize the offloading decision problem, fix other variables, and optimize the computing task offloading ratio ρ k , the problem has a closed-form solution, the optimal ρ k The local computing delay of the power terminal and the overall processing delay of the edge server are kept consistent, that is: Step 4: Optimize variables related to resource allocation separately p k,s 、f k,s , including: using the interior point method to find the optimal variables p k,s ; For variable f k,s , there exists a closed-form solution, which makes the base station keep the edge offloading task calculation delay consistent for all users associated with it. For base station m, suppose there are l terminals associated with it, denoted as j1, j2, ..., j l , then the computing power allocation of base station m satisfies: Step 5, repeat steps 2 to 4 until the given cycle number I is reached.

2. The user association and resource allocation method for a power multi-layer edge computing network according to claim 1 is characterized in that: The method also proposes a new user association scheme, that is, a low-complexity reference quantity for measuring the selection of power terminals and base stations is constructed based on channel quality and computing power resources, and the terminal selects the base station with the smallest reference quantity to achieve user association, including: Among them, M k,s represents the reference quantity selected by terminal k and base station s, SNR k,s Represents the signal-to-noise ratio between the terminal and the base station, which is used to measure the channel quality and satisfies θ k is an empirical parameter, H k,s represents the channel information, f s Indicates the remaining computing resources allocated by the base station.

3. The user association and resource allocation method for a power multi-layer edge computing network according to claim 2 is characterized in that: The process of using channel quality and computing resources to build a reference quantity for measuring power terminals and base station selection for user association includes: The base station obtains channel information and its own remaining computing resources and sends them to the terminal; The terminal is based on the empirical parameter θ k , channel information H k,s and the remaining computing resources allocated to each base station f s Calculate the reference quantity M k,s ; Terminal selection reference quantity M k,s The smallest base station to achieve user association.

4. A user association and resource allocation device for a power multi-layer edge computing network, comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: The device is applied to a multi-layer edge computing network composed of power terminals, small base stations or macro base stations equipped with edge servers. When the one or more processors execute the executable code, it is used to implement the user association and resource allocation method for the power multi-layer edge computing network described in any one of claims 1-3.

Images