System energy-saving method and device in multi-user scenarios in mobile edge computing networks

Abstract

The present invention discloses a system energy-saving method and device in a multi-user scenario in a mobile edge computing network. According to the characteristics of user requests, the impact of the allocation of computing, network, and cache resources and the base station sleep status on the energy consumption of the base station edge system is comprehensively considered, and a model of user mobility, edge cache, user delay energy consumption, edge system energy consumption and user request quantity is established. Resource allocation and base station sleep strategy are carried out for the purpose of minimizing the energy consumption of the edge system, while meeting the user's delay requirements. A base station sleep and resource allocation decision algorithm based on the edge system energy consumption is proposed, thereby achieving the goal of optimal energy consumption of the base station edge system.

Description

Technical Field

[0001] The present invention relates to wireless communication technology, mobile communication, mobile edge computing and other related fields, and in particular to energy-saving methods and devices for edge systems in multi-user scenarios in mobile edge computing networks. Background Art

[0002] With the further development and commercialization of IoT (Internet of Things) and 5G technologies, various compute-intensive and latency-sensitive applications are gradually becoming part of daily life. This has led to an explosive growth in the amount of user data requested per unit time, placing tremendous pressure on traditional core networks. Mobile edge computing, with its proximity to user terminals, provides users with computing, network, and cache resources to meet their service and latency requirements, and has been widely researched.

[0003] Existing research has largely focused on how mobile edge computing can reduce latency and energy consumption for users, while ignoring the needs of service providers. While mobile edge computing reduces latency for end users, it also increases energy consumption for edge systems. The size of computing tasks offloaded to edge servers determines the computing energy consumed by edge servers. The rate at which these tasks are offloaded affects the transmission energy consumed by base stations. The size of the content read from edge caches by users is related to the cache energy consumed by edge servers. Furthermore, the distribution and sleep state of base stations influence user choices and latency. Therefore, different offloading strategies and base station sleep strategies are crucial for determining the energy consumption of edge systems. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a system energy-saving method in a multi-user scenario in a mobile edge computing network. For different types of user requests, the impact of computing, caching and allocation of network resources on the energy consumption of the edge system is comprehensively considered, and different base station sleep schemes are selected to achieve the purpose of system energy saving.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] The present invention proposes a system energy saving method in a multi-user scenario in a mobile edge computing network, comprising the following steps:

[0007] S1. Define user requests, edge system parameters, and computation offloading decision vectors in a multi-user scenario for base station mobile edge computing. The details are as follows:

[0008] S11. The user moves randomly in the cell covered by the macro base station and can only send one user request at a time. Define the user request (N is the total number of user requests), d iIndicates the size of the user's computing task, c i represents the computing resources required to process computing tasks, Indicates the network resources required by the user, Indicates the computational latency requirement, Indicates the network delay requirement;

[0009] S12, the macro base station is located in the center of the cell, and the rest of the small base stations are distributed throughout the cell. Each base station has a corresponding edge server, defining the edge base station system BS j ={w j ,f j ,B j ,CS j}, j∈{1,2,…,M} (M is the total number of base stations), w j Indicates the channel bandwidth that the base station can provide, f j Indicates the size of computing resources that the server can provide, B j Indicates the network bandwidth that the base station can provide; CS j Indicates the cache space size of the server;

[0010] S13. Users can choose local, edge, or cloud processing based on the characteristics of the computing task, and define two sets of decision variables for executing the computing task. Binary variables When it is 1, it means that the computing task generated by the user is processed in the local terminal, and the binary variable When it is 1, it means that the local computing tasks generated by the user are offloaded to the edge for processing. If it is 0, it means that the local computing tasks generated by the user are offloaded to the cloud for processing. The decision variable b i (1) When all 0s are present, the decision variable b i (2) Only then is the decision made whether it is 0 or 1.

[0011] S2. Analyze the impact of user mobility on computing task offloading and edge system resources. The details are as follows:

[0012] S21. The user's movement changes the distance between them and the base station, thereby affecting the rate at which local computing tasks are uploaded to the edge system. The number of base stations the user is connected to also changes as they move, thereby increasing or decreasing the user's options for offloading computing tasks.

[0013] S22. After the user leaves the coverage area of ​​the base station, he will continue to occupy the resources of the base station and the edge server, thereby increasing the overall pressure of the edge system. Therefore, it is necessary to determine whether to stop providing edge resources to the user based on the location relationship between the user and the base station.

[0014] S3. Design an edge caching strategy based on the distribution of base stations, the size of the edge cache space, and the characteristics of user request content in the local database. The specific description is as follows:

[0015] S31. Macro base stations are located in the center of a cell and cover the entire cell, while small base stations are distributed throughout the cell. Each base station has a corresponding edge server, and the cache space in the small base station server is the same and smaller than that of the macro base station. Based on the latency differences of user requests, they are divided into video, data, and emerging service categories.

[0016] S32. Mobile edge computing is introduced because traditional networks cannot meet users' low-latency requirements. Therefore, it is necessary to consider user latency while reducing edge system energy consumption. In order to increase the number of users who choose to obtain requested content, it is necessary to increase the proportion of users who obtain content at the edge.

[0017] S33. When all base stations are in working state, the edge system can provide users with all files in the local database; when some base stations are dormant, the cached content in their corresponding edge servers will not be accessible to users; different user request file contents occupy different cache spaces and have different popularity. In order to avoid the edge servers being filled with popular video files that occupy a large cache space, multiple edge cache spaces are divided according to user request types to ensure that each edge server has the same number of various file contents; the dormancy of the base station will turn off the cache function of the corresponding server, so the popular content will be cached in the macro base station server that will not sleep, and the remaining content will be cached evenly in the small base station servers; the popularity of the content files requested by mobile users follows the Zipf distribution, and its expression is: Among them, p i represents the popularity of each requested content, K represents the total number of cacheable contents, and ξ is the Zipf parameter, which is generally between [0.32, 0.85].

[0018] S4. Combine user latency, base station energy consumption, and edge server energy consumption models to perform base station dormancy and resource allocation to achieve energy saving. The details are as follows:

[0019] S41. When users execute computing tasks, computing delays are generated. Local computing tasks can be executed locally, at the edge, or in the cloud. However, tasks generated after the user request reaches the server can only be executed at the edge or in the cloud. The latency of executing computing tasks locally is: The user delay caused by the edge server performing computing tasks is: The user delay caused by the cloud server executing the computing task is: They represent the computing resources provided by mobile terminals, edge servers, and cloud servers to users, respectively. i Indicates the computing resources required to process computing tasks, d i Indicates the size of the computing task generated by the user, r i With R i They represent the rates of uploading local computing tasks to the MEC server and cloud server, respectively, Represents the channel bandwidth provided by the base station and the cloud layer, p i 、h i , σ 2 denote the terminal transmit power, channel gain, and noise power respectively. Combined with the computation offloading decision variables, the user computation delay under different circumstances is defined as:

[0020] S42. When a user requests server content or uploads data, the user will get the uplink and downlink network rates provided by the base station. When the resources provided by the base station are less than the user's expected value, the user will experience a jerky experience. Therefore, the user's expected network bandwidth is defined as The actual network bandwidth provided by the base station or cloud layer to the user is The network delay is defined as:

[0021] S43. In order to express the user's acceptance of the computing delay and network delay generated during the resource allocation process, define the user delay tolerance Among them, α and β are non-negative parameters less than 1, which are used to indicate the preference of different user requests for computing resources or network resources;

[0022] S44. In the process of allocating bandwidth resources to users, the base station will generate corresponding energy consumption, which mainly includes the energy consumption of transmission computing tasks and the energy consumption of the base station's basic equipment, namely p0 is the transmission power of each resource block, p1 is the basic equipment power of the base station, η is the energy efficiency coefficient of the power amplifier, n i is the number of resource blocks consumed by the user, and T is the time size; w0 and B0 represent the channel bandwidth and network transmission rate occupied by a resource block respectively;

[0023] S45. When executing the user's computing task, the edge server will generate energy consumption for task computing and some basic computing, namely When the edge server caches user requested content or the user reads cached content, it will generate corresponding cache energy consumption, that is, p2 is the basic computing power of the edge server, p m is the maximum computing power of the edge server, f iThe size of computing resources obtained by the user, f m is the maximum computing resource that can be allocated to the server, p3 is the power generated by the cached content, and e c Energy consumption per bit of content read by the user, d i e The size of the content read from the edge server for the user;

[0024] S46, according to the number of base stations and solve different base station sleep schemes respectively, under the determined base station sleep condition, establish an optimization problem with the goal of minimizing the energy consumption of the edge system, that is, min(E 1 +E 2 +E 3 ) and solves it through an improved intelligent algorithm; each base station sleep situation corresponds to a resource allocation scheme with the lowest energy consumption of the edge system, and each resource allocation scheme corresponds to different user delay tolerance. Ultimately, the system energy consumption and the lowest base station sleep and resource allocation scheme that meet the user delay tolerance are selected.

[0025] Furthermore, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the specific steps of the aforementioned method when the computer program is executed by a processor.

[0026] At the same time, the present invention also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the specific steps of the method of the present invention when executing the computer program.

[0027] The present invention adopts the above technical solution, which has the following technical effects compared with the prior art:

[0028] The present invention comprehensively considers the impact of the allocation of computing, network, and cache resources and the base station sleep status on the energy consumption of the base station edge system based on the characteristics of user requests, and establishes models of user mobility, edge cache, user delay energy consumption, edge system energy consumption, and the number of user requests. It allocates resources and implements base station sleep strategies for the purpose of minimizing the energy consumption of the edge system, improving the access rate of the edge server, reducing the delay of user requests, and achieving the optimal energy consumption of the base station edge system. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a working scenario diagram of the overlapping macro base station and small base station provided by the present invention.

[0030] Figure 2 It is a flowchart of base station sleep and resource allocation provided by the present invention. DETAILED DESCRIPTION

[0031] The following describes in detail the specific implementation of the edge system energy-saving method of the present invention, with reference to the accompanying drawings. Obviously, the described embodiments are only a portion of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments derived by persons of ordinary skill in the art without inventive effort are within the scope of protection of the present invention.

[0032] First, from an analytical perspective, the main issues that need to be considered when implementing the present invention are given:

[0033] Step 1: Design an appropriate edge caching strategy based on the distribution of base station edge servers and the characteristics of user requests. Cache different content for each edge server based on the popularity of the user requested content and the type and number of files in the local database. This strategy mainly includes:

[0034] (a) Edge Cache Analysis: When all base stations are in operation, their corresponding edge servers can cache all user-requested files and related content. However, when some base stations are dormant, the cached content of the edge servers corresponding to these dormant base stations cannot be accessed by users.

[0035] (b) User request analysis: Considering the differences in parameters in different user requests, user requests are divided into video, data, and emerging business categories;

[0036] (c) Caching Strategy Design: In a multi-user scenario using mobile edge computing (MEC), the macro base station serves as the center of the cell and provides full cell coverage. Small base stations are scattered throughout the cell but have smaller coverage areas. Therefore, to increase the user service ratio at the edge, popular content is cached in the macro base station MEC server, while the remaining content is evenly distributed among the small base station MEC servers. The base station edge servers within the cell share the content in their cache, and the popularity of the content in the cache follows a Zipf distribution.

[0037] Step 2: Analyze the impact of user mobility, user latency requirements, and the number of user requests on edge system energy consumption. Combined with the edge system energy consumption model, the energy consumption minimization problem in a given base station sleep scenario is transformed into a mixed integer nonlinear programming problem. This mainly includes:

[0038] (a) User mobility analysis: User movement changes the connection between the user and the base station. When a user moves from one base station's coverage area to multiple base stations' coverage areas, the user has more ways to access resources. When a user moves from one base station's coverage area to another, the user still occupies server resources. A user mobility model is established in a multi-base station server scenario to obtain user location information.

[0039] (b) User latency analysis: Local computing tasks generated by user requests can be processed locally, at the edge, or in the cloud. The user latency incurred by local computing tasks only includes processing latency. User latency incurred by edge or cloud computing tasks includes upload latency, processing latency, and return latency. When a user requests cached content or data, the base station provides different uplink and downlink bandwidths to the user, resulting in different network latency.

[0040] (c) User quantity analysis: The user traffic volume in each area of ​​a mobile communication system varies over time and exhibits a tidal effect. The number of user requests at each time point in a day also follows a tidal effect. Changes in the number of users affect the number of dormant base stations.

[0041] (d) System energy consumption analysis: The edge system mainly consists of base stations and edge servers. When users request to obtain edge computing resources, bandwidth resources, and read cached content, corresponding energy consumption will be generated. Among them, the energy consumption of the base station mainly includes transmission energy consumption and basic energy consumption, and the energy consumption generated by the edge server is mainly divided into computing energy consumption and cache energy consumption.

[0042] The user's movement in a continuous time period is converted into movement in a continuous time slice. The latency requirements, duration, and location of each user in each time slice are interconnected. A user's computing latency depends on the method of offloading computing tasks and the attributes of the computing tasks. The user's network latency depends on the content of the edge cache and the network resource requirements. The energy consumption of the base station depends on the number of resource blocks consumed by the user, the computing energy consumption of the edge server depends on the computing resources provided to the user, and the cache energy consumption of the edge server depends on the cache space and the size of the edge content read by the user. The computing tasks generated by the user are divided into local terminal computing tasks and server computing tasks. The network resources requested by the user are divided into channel bandwidth and uplink and downlink rates.

[0043] Step 3: Solve various possible scenarios for base station sleep, and solve the problem of minimizing energy consumption in a given base station sleep scenario through an improved intelligent algorithm, and ultimately select a base station sleep and resource allocation solution that meets user latency requirements; specifically, this includes:

[0044] (a) Base station dormancy analysis: The macro base station is located in the center of the cell and the corresponding edge server caches the most popular content. Therefore, in order to increase the proportion of users obtaining resources at the edge, the macro base station is kept in working state and the small base stations are appropriately dormant;

[0045] (b) System energy saving analysis: Under the condition of a certain base station sleep state, resource allocation is performed with the goal of minimizing the energy consumption of the edge system and meeting the user's latency requirements.

[0046] The implementation scheme of the present invention is further described below with reference to specific embodiments:

[0047] Example 1: Figure 1 、 Figure 2 As shown, the present invention proposes a system energy saving method in a multi-user scenario in a mobile edge computing network, which is mainly divided into two parts: base station sleep and resource allocation. The specific steps include:

[0048] Step S1: Define user requests, edge system parameters, and computation offloading decision variables in a multi-user scenario of base station mobile edge computing. The specific description is as follows:

[0049] S11. The multi-user scenario of base station mobile edge computing includes base stations, edge servers, and mobile users. Base stations are divided into macro base stations and small base stations. A macro base station is deployed in the center of a cell, and multiple small base stations are scattered around it. Each base station has a corresponding edge server. Users move within the cell and can only send one request at a time. Based on the differences in different request parameters, user requests are divided into video, data, and emerging business categories.

[0050] S12. Define user request (N is the total number of user requests), d i Indicates the size of the user's computing task, c i represents the computing resources required to process computing tasks, Indicates the network resources required by the user, Indicates the computational latency requirement, Indicates network delay requirements; defines edge system BS j ={w j ,f j ,B j ,CS j}, j∈{1,2,…,M} (M is the total number of base stations), w j Indicates the channel bandwidth that the base station can provide, f j Indicates the size of computing resources that the server can provide, B j Indicates the network bandwidth that the base station can provide; CS j Indicates the size of the server's cache space. The parameters requested by different users vary greatly. The parameters of different small base stations are the same but different from those of macro base stations.

[0051] S13. When users send requests, computing tasks are generated. When computing tasks are generated on the local terminal, users can choose to execute them on the local terminal, upload them to the edge server, or execute them on the cloud server. When tasks are generated on the server, users can only choose to execute them on the server, such as computing tasks such as video transcoding. The user delays generated by processing the same user computing tasks at different locations are also different. In order to represent the choice of each user to obtain resources, a binary decision variable for executing computing tasks is defined. Binary variables When it is 1, it means that the user's computing task is processed in the local terminal, and the binary variable When it is 1, it means that the user's local computing tasks are offloaded to the edge for processing. If it is 0, it means that the local computing tasks generated by the user are offloaded to the cloud for processing. Decision variable b i (1) 0 is the decision variable b i (2) The prerequisite for execution is judgment b i (2) The prerequisite is b i (1) =0.

[0052] Step S2: Construct a user movement model in a multi-user scenario of base station mobile edge computing, and analyze the impact of user movement on the computing task offloading method and edge system resources. The specific description is as follows:

[0053] S21. The movement of users in continuous time is irregular. In order to analyze the user's mobility status, the continuous time is divided into continuous time slices. The user's movement in each time slice is regarded as having a constant direction and speed. A plane rectangular coordinate system is established with the macro base station in the center of the cell as the origin. Small base stations are distributed around the cell. The coverage area of ​​each base station is approximately regarded as a circle. The original coordinates of the user are defined as (x i ,y i ), the angle between the direction of motion and the x-axis is θ and the speed is v i , the coordinates of the macro base station are (0,0), the coordinates of the small base station BS1 are (A,B), and after a time slice, the coordinates of the mobile user become (x i +v i ·cosθ,y i +v i ·sinθ).

[0054] S22. If the user is only in the coverage area of ​​the macro base station before moving, and after a time slice of movement, the user moves to the overlapping area of ​​the macro base station and the small base station BS1, then the user has more options for offloading computing tasks at the edge; the distance between the user and the base station is related to the upload rate of the local computing task, so it is necessary to calculate the distance between the user and each base station. The distance between the user and the macro base station is The distance between the user and the small base station BS1 is

[0055] S23. If the user's movement causes him to leave the coverage area of ​​the base station, the base station needs to stop providing resources related to the user in a timely manner; in addition, the disordered movement of the user causes the number of user requests in each base station to change. When the number of users in the coverage area of ​​the small base station has been changing within a small range, it is possible to consider dormant the base station to reduce the energy consumption of the edge system. In this case, the users under the corresponding dormant base station can only request related resources from the macro base station or the cloud layer.

[0056] Step S3: Construct an edge cache strategy for a multi-user scenario of mobile edge computing of base stations, combining base station distribution and user request characteristics to improve the edge cache hit rate. The specific description is as follows:

[0057] S31. In the multi-user scenario of base station mobile edge computing, the macro base station is located in the center of the cell and covers the entire cell. The remaining small base stations are distributed throughout the cell. An edge server is deployed near each base station, and the edge cache space of the macro base station server is larger than that of other small base station servers.

[0058] S32. In the multi-user scenario of mobile edge computing of base stations, the goal is to reduce the energy consumption of the edge system, but the user's latency requirements cannot be ignored, because mobile edge computing is introduced into various scenarios to improve user services; by improving the cache hit rate of the edge server, the user's options for obtaining the requested file content can be increased, so that the user's latency requirements can be met.

[0059] S33. Assume that all base stations are in working state, the base stations are connected by optical fiber and their corresponding edge servers share all cached contents; when the base station sleep strategy is executed, the dormant base station and its corresponding edge server no longer provide user resources, and the content of the edge cache cannot be accessed by users. Therefore, in order to improve the cache hit rate of the edge server, an edge cache strategy is designed based on popularity; considering that different user request file contents occupy different cache spaces and have different popularity, in order to avoid the edge server being full of popular video files that occupy a large cache space, multiple edge cache spaces are divided according to user request types to ensure that each edge server has the same number of various file contents; the dormancy of the base station will turn off the cache function of the corresponding server, so the popular content is cached in the macro base station server that will not sleep, and the remaining content is cached evenly in the small base station server; the popularity of the mobile user's request content file follows the Zipf distribution, and its expression is: Among them, p i represents the popularity of each requested content, K represents the total number of cacheable contents, and ξ is the Zipf parameter, which is generally between [0.32, 0.85].

[0060] Step S4: Construct a user delay and edge system energy consumption model in a multi-user scenario of base station mobile edge computing, and use an algorithm to solve the system energy consumption and the minimum situation under different base station sleep schemes to achieve the purpose of energy saving. The specific description is as follows:

[0061] S41. Assuming that mobile user requests are processed in parallel, the queuing delay of edge computing tasks is not considered. In addition, considering the characteristics of user computing tasks, user computing tasks may be executed locally, at the edge, or in the cloud. If the user generates a computing task at the local terminal, it can be executed locally, at the edge, or in the cloud. If the user generates a computing task within the server, it can only be executed within the server. Mobile terminal devices have certain computing resources. If the computing task is executed locally, the resulting user delay is If the user delay caused by executing the computing task locally is too large, you can choose to offload the computing task to the edge server. The user delay caused in this process includes the upload delay and the task computing delay, that is, If the number of user requests is too large and the computing resources at the edge are insufficient, you can choose to offload the computing tasks to the cloud server, and the resulting user delay is They represent the computing resources provided by mobile terminals, edge servers, and cloud servers to users, respectively. i Indicates the computing resources required to process computing tasks, d i Indicates the size of the computing task generated by the user, r iWith R i They represent the rates of uploading local computing tasks to the MEC server and cloud server, respectively, Represents the channel bandwidth provided by the base station and the cloud layer, p i 、h i , σ 2 represent the terminal transmit power, channel gain, and noise power respectively; the binary computation offloading decision variable represents the execution location of the user computation task, so the user computation delay generated in various situations can be defined as:

[0062] S42. In addition to uploading local computing tasks to the edge server, users will also request cached content and data files from the edge server. In this process, the base station needs to provide network bandwidth. The user's uplink services include video, file and data uploads, and downlink services include video viewing, web browsing and file downloads. The entire process requires the base station to provide uplink and downlink network bandwidth. Assume that the total bandwidth of the edge base station is limited but the ratio of uplink and downlink bandwidth can be adjusted according to actual conditions, while the total bandwidth of the cloud layer is unlimited but can only provide limited bandwidth to each user. When the network resources provided by the base station are less than the user's expectations, there will be a phenomenon of unsmoothness. The user's expected network bandwidth is defined as The actual network bandwidth provided by the base station or cloud layer to the user is The network delay is defined as:

[0063] S43. The size of computing delay and network delay affects the service quality of different user requests. In order to express the maximum tolerance of users for computing delay and network delay generated in the resource allocation process, user delay tolerance is introduced and compared with the actual delay and delay requirement of user requests. That is, the computing delay tolerance is Network delay tolerance The user's overall tolerance for computing delay and network delay is expressed as TT by a normalized method. i =α i ·T i c +β i ·T i n , where α i , β i is a non-negative parameter less than 1 and α i +β i =1, its value is affected by the specific type of user request, indicating the user's preference for computing resources and network resources.

[0064] S44. When users upload local computing tasks, they need to rely on the channel bandwidth provided by the base station. When they send and receive data file contents in the server, they need the base station to provide network bandwidth. In these processes, the system transmission energy consumption and the basic energy consumption of the base station are generated, that is, p0 is the transmission power of each resource block, p1 is the basic equipment power of the base station, η is the energy efficiency coefficient of the power amplifier, n i is the total number of resource blocks provided to users by the edge base station, T is the time size; the number of resource blocks consumed is w0, B0 represent the channel bandwidth and network transmission rate occupied by a resource block respectively. Provides edge base stations with user network bandwidth.

[0065] S45. After the user uploads the local computing task to the edge server, the edge server starts to process the user's computing task, thereby generating server computing energy consumption and some basic computing energy consumption, namely Users can read the file data cached in the edge server through the network bandwidth provided by the base station, and in this process, corresponding cache energy consumption will be generated, that is, Where p2 represents the basic computing power of the edge server, p m represents the maximum computing power of the edge server, f i Indicates the size of computing resources obtained by the user, f m represents the maximum computing resources that can be allocated to the server, p3 represents the power generated by the edge server caching content, and e c It represents the energy consumption of each bit of content read by the user, d i e Indicates the size of content read by users from the edge server.

[0066] S46. In a multi-user scenario of mobile edge computing using base stations, the macro base station located in the center of the cell will always be in operation. When the number of user requests is small, the energy consumption of the edge system can be reduced by dormant small base stations in the cell. The resource allocation problem for each determined base station in dormancy is aimed at minimizing the energy consumption of the edge system. This problem is modeled as an optimization problem in a single time slice by combining edge caching results, user movement conditions, user delay model, and edge system energy consumption model, namely:

[0067]

[0068]

[0069]

[0070]

[0071] C4:f i e <f i max

[0072]

[0073]

[0074]

[0075] Among them, N involved in the optimization objective represents the total number of user requests in the current time slice, and each user request may generate a certain amount of system energy consumption; the two binary decision variables in C1 represent the execution location of the user computing task, which affects the size of the user computing delay; the three variables in C2 represent the channel bandwidth, computing resources and network bandwidth obtained by the user from the edge system. When the user uses local computing resources or cloud resources, the edge system will not generate energy consumption; C3 and C4 represent the maximum value of the edge channel bandwidth and computing resources obtained by the user at one time; C5, C6, and C7 represent the maximum value of the channel bandwidth, computing resources and network resources available to each base station and edge server, respectively, where w m It is the maximum channel resource bandwidth that the base station signal can provide, and the unit is: Hertz, B m It represents the maximum rate that the base station channel can provide to users, measured in bits per second. Macro and small base station systems share the same available resources, except for edge buffer space.

[0076] S47. From S46, we know that this optimization problem is a mixed integer nonlinear programming problem. It is difficult for general algorithms to obtain its optimal solution. Usually, intelligent algorithms can be improved to solve the suboptimal solution. For the solution of base station dormancy, a backtracking algorithm can be used. Therefore, the backtracking algorithm and intelligent algorithm are improved and combined to solve the base station dormancy and resource allocation problem in the multi-user scenario of base station mobile edge computing:

[0077] In the first step, the number of user requests and small base stations is set to N and M respectively, where the initial value of the number of base station sleep ss is 0.

[0078] In the second step, a recursive backtracking method is used to solve all situations when the number of base station sleep is ss. Pruning operations are added in the solution process to determine in turn whether each base station sleep situation is a further sleep situation under the unqualified base station sleep scheme. If so, the subsequent operations are abandoned and the first step is returned to increase the number of base station sleep by one.

[0079] The third step is to solve the resource allocation problem under the condition of base station dormancy by improving the intelligent algorithm, that is, the optimization problem in S46. First, randomly generate an array S = [s1, s2, s3, ..., s N ] T , the elements in the array Indicates the situation where users obtain edge resources. The randomly generated array S is composed into a cell matrix P of size M, that is, P = {[S1], [S2], ..., [S M ]}; Then the objective function of the optimization problem is used as the fitness value to judge the quality of the solution, that is, fitness = E 1 +E 2 +E 3 ; Then, the arrays in the cell matrix are selected by the roulette wheel method to form a new matrix. The probability of each array being selected is Randomly select two arrays in the new matrix, use multi-point crossover to generate a new array, and then randomly change the number of elements in an array according to the mutation probability. and value, thereby changing the corresponding resource acquisition method and size. After a certain number of iterations, the solution to the optimization problem can be obtained.

[0080] The fourth step is to obtain the average user delay tolerance under the corresponding situation based on the solution of the optimization problem. If it is greater than 90% of the user's expected value, the corresponding base station sleep and resource allocation plan will be retained; otherwise, it will be regarded as an unqualified base station sleep plan and discarded.

[0081] In the fifth step, if the number of base station sleep is less than M, the number of base station sleep is increased by one and the process returns to the first step. Otherwise, all available base station sleep solutions are compared and the one with the lowest edge system energy consumption is selected as the final solution.

[0082] Therefore, the present invention proposes a system energy-saving method in a multi-user scenario in a mobile edge computing network. The overall solution is divided into two parts: base station sleep and resource allocation. The impact of user type, base station distribution, and edge cache on base station sleep is considered respectively. In the process of resource allocation, the impact of local, edge, and cloud resources on user latency and edge system energy consumption is considered. Combined with user mobility, user latency, base station energy consumption, and edge server energy consumption models, base station sleep and resource allocation algorithms are designed to solve the edge system energy-saving solution.

[0083] Example 2: This example provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method of the present invention are implemented, which will not be repeated here.

[0084] Example 3: This embodiment also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. It should be noted that when the processor executes the computer program, it corresponds to the specific steps of the method provided in the embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method. For technical details not fully described in this embodiment, please refer to the method provided in the embodiment of the present invention.

[0085] The program code for implementing the methods of the present application can be written in any combination of one or more programming languages. Such program code can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the program code is executed by the processor or controller, the functions / operations specified in the flow charts and / or block diagrams are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0086] In the context of the present application, a machine-readable medium can be a tangible medium that can contain or store a program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0087] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0088] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A system energy saving method in a multi-user scenario in a mobile edge computing network, characterized in that: The specific steps are as follows: Step 1: Design an edge caching strategy based on the distribution of base station edge servers, server cache space, and the types of user requests in the local database. Each edge server caches different content based on the popularity of the user requested content and the type and number of files in the local database. Step 2: Analyze the impact of user mobility, user latency requirements, and the number of user requests on edge system energy consumption. Combined with the edge system energy consumption model, the energy consumption minimization problem in a given base station sleep scenario is transformed into a mixed integer nonlinear programming problem. Step 3: A backtracking pruning algorithm is used to solve various possible scenarios for base station sleep. An intelligent algorithm is then used to solve the problem of minimizing energy consumption under a given base station sleep scenario. Ultimately, a base station sleep and resource allocation solution that meets user latency requirements is selected. In step 2, we analyze the impact of user mobility, user latency requirements, and the number of user requests on edge system energy consumption. Specifically: S201. Calculate user latency in different scenarios: (1) If the computing task is executed locally, the resulting user delay is (2) If the user delay caused by executing the computing task locally is too large, the computing task will be offloaded to the edge server. The user delay caused by this process includes the upload delay and the task computing delay, that is, (3) If the number of user requests is too large and the computing resources at the edge are insufficient, the computing tasks are offloaded to the cloud server, resulting in a user delay of in, They represent the computing resources provided by mobile terminals, edge servers, and cloud servers to users, respectively. i Indicates the computing resources required to process computing tasks, d i Indicates the size of the computing task generated by the user, r i With R i Represents the rate of uploading local computing tasks to MEC servers and cloud servers respectively; Represents the channel bandwidth provided by the base station and the cloud layer, p i 、h i , σ 2 Represent the terminal transmit power, channel gain, and noise power respectively; The user computing delay generated in various situations is defined as: Binary variables When it is 1, it means that the computing task generated by the user is processed in the local terminal, and the binary variable When it is 1, it means that the local computing tasks generated by the user are offloaded to the edge for processing. If it is 0, it means that the local computing tasks generated by the user are offloaded to the cloud for processing. The decision variable b i (1) When all 0s are present, the decision variable b i (2) Only then can its significance be realized; S202: Define the user's expected network bandwidth as The actual network bandwidth provided by the base station or cloud layer to the user is The network delay is defined as: S203, calculate the delay tolerance Network delay tolerance The user's overall tolerance for computing delay and network delay is expressed as TT by a normalized method. i =α i ·T i c +β i ·T i n , where α i , β i is a non-negative parameter less than 1 and α i +β i =1; S204, calculate the system transmission energy consumption and the basic energy consumption of the base station during operation, that is, p0 is the transmission power of each resource block, p1 is the basic equipment power of the base station, η is the energy efficiency coefficient of the power amplifier, n i is the total number of resource blocks provided to users by edge base stations, T is the time size; the number of resource blocks consumed w0, B0 represent the channel bandwidth and network transmission rate occupied by a resource block respectively. Provide edge base stations with user network bandwidth; S205, calculate the energy consumption of the server and the energy consumption of the basic calculation, that is, The user reads the file data cached in the edge server through the network bandwidth provided by the base station, and generates corresponding cache energy consumption in the process, that is, Where p2 represents the basic computing power of the edge server, p m represents the maximum computing power of the edge server, f i Indicates the size of computing resources obtained by the user, f m represents the maximum computing resources that can be allocated to the server, p3 represents the power generated by the edge server caching content, and e c Indicates the energy consumption generated by the user reading each bit of content, Indicates the size of the content read by the user from the edge server; In step 2, the energy consumption minimization problem under a given base station sleep scenario is transformed into a mixed integer nonlinear programming problem, specifically: The resource allocation problem for each determined base station in sleep mode aims to minimize the energy consumption of the edge system. Combining the edge cache results, user movement, user delay model, and edge system energy consumption model, it is modeled as an optimization problem in a single time slice, namely: Among them, N involved in the optimization goal represents the total number of user requests in the current time slice. Each user request may generate a certain amount of system energy consumption, where w m It is the maximum channel resource bandwidth that the base station signal can provide, and the unit is: Hertz, B m It indicates the maximum rate that the base station channel can provide to users, in bits per second. Step 3 combines the improved backtracking algorithm and intelligent algorithm to solve the base station dormancy and resource allocation problems in the multi-user scenario of base station mobile edge computing. The details are as follows: S301, set the number of user requests and small base stations to N and M respectively, where the initial value of the number of dormant base stations ss is 0; S302, using a recursive backtracking method to solve all situations when the number of base station sleep is ss, adding a pruning operation in the solution process to sequentially determine whether each base station sleep situation is a further sleep situation under the unqualified base station sleep solution, and if so, abandon the subsequent operation and return to S301, and increase the number of base station sleep by one; S303: solving the resource allocation problem in the case of base station dormancy by improving the intelligent algorithm: (1) Randomly generate an array S = [s1, s2, s3, ..., s N ] T , the elements in the array Indicates the situation where users obtain edge resources. The randomly generated array S is composed into a cell matrix P of size M, that is, P = {[S1], [S2], ..., [S M ]}; (2) The objective function of the optimization problem is used as the fitness value to judge the quality of the solution, that is, fitness = E 1 +E 2 +E 3 ; (3) The roulette wheel method is used to select arrays in the cell matrix to form a new matrix. The probability of each array being selected is Randomly select two arrays in the new matrix, use multi-point crossover to generate a new array, and then randomly change the number of elements in an array according to the mutation probability. and value, thereby changing the corresponding resource acquisition method and size. After a certain number of iterations, the solution to the optimization problem can be obtained; (4) Obtain the average user delay tolerance under the corresponding situation based on the solution of the optimization problem If it is greater than 90% of the user's expected value, the corresponding base station sleep and resource allocation plan will be retained; otherwise, it will be regarded as an unqualified base station sleep plan and discarded; (5) If the number of base station sleeps is less than M, then the number of base station sleeps is increased by one and the process returns to the first step. Otherwise, all available base station sleep schemes are compared and the one with the lowest edge system energy consumption is selected as the final scheme.

2. A system energy saving method in a multi-user scenario in a mobile edge computing network according to claim 1, characterized in that: Step 1 is as follows: S101. Classify user requests into video, data, and emerging service categories based on latency differences. S102. Divide multiple edge cache spaces according to user request types to ensure that each edge server has the same amount of various file contents; cache relatively popular content in macro base station servers that will not sleep, and cache the remaining content evenly in small base station servers.

3. A system energy saving method in a multi-user scenario in a mobile edge computing network according to claim 2, characterized in that: The popularity of the file content requested by mobile users follows the Zipf distribution, which is expressed as: Among them, p i represents the popularity of each requested content, K represents the total number of cacheable contents, and ξ is the Zipf parameter, which ranges from [0.32 to 0.85].

4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 3 are implemented.

Images