A multi-layer service optimization deployment method based on a cloud-edge computing network

Abstract

This invention proposes a multi-layer service optimization deployment method based on cloud-edge computing networks. First, a multi-layer edge computing network scenario is constructed, establishing deployment and communication models. Second, the multi-objective optimization problem of transmission latency and deployment cost is modeled as a single-objective optimization problem. Public services accessible from multiple regions are deployed on the least common parent node of the network tree, and a greedy method is used to progressively create replicas of the services, placing them closer to users. This invention is applicable to multi-layer service optimization deployment in edge computing scenarios, achieving the goal of improving user service quality at low cost by reducing deployment costs and request latency.

Description

Technical Field

[0001] This invention mainly relates to the field of edge computing, and in particular to a method for optimizing the deployment of multi-layer services in a cloud-edge computing network. Background Technology

[0002] In recent years, Mobile Edge Computing (MEC) has emerged as a new computing framework. Simply put, Mobile Edge Computing involves the distributed deployment of network, storage, and computing resources at the network edge to provide high-quality, low-latency request services to processing terminal devices. The advantages of edge computing in terms of rapid request connection, real-time processing, and privacy protection have attracted widespread attention both domestically and internationally. In a 2015 report on MEC service scenarios, the European Telecommunications Standards Institute (ETSI) pointed out that virtual reality service scenarios and video streaming analytics service scenarios have extremely high requirements for bandwidth and storage resources.

[0003] However, compared to data centers in remote clouds with abundant communication and storage resources, the resources of a single edge node are quite limited. Benefiting from the development of virtualization technologies such as virtual machines and containers, users can also send requests to collaborative edge clouds (ECs) via metropolitan area networks (MANs), allowing them to access edge cloud services from outside their one-hop communication range. Multi-layered edge network architectures can comprehensively utilize the upper-layer edge cloud, further increasing the capacity of available edge resources. The current challenge is how to allocate resources on edge servers to better respond to user requests—that is, the service deployment problem on edge nodes.

[0004] Solving service deployment challenges in edge environments is closely related to both users and service providers. From the user's perspective, some services are latency-sensitive. For example, in autonomous driving scenarios, the time from data generation to processing completion by automotive sensors must be less than 50ms, while VR video live streaming services can only tolerate a maximum latency of 70ms. Optimizing service deployment solutions to reduce request latency will help improve the quality of service for users. From the service provider's perspective, they need to coordinate the types and quantities of resources deployed at the network edge. This involves improving the user experience while minimizing deployment costs such as leasing edge resources.

[0005] In conclusion, effectively utilizing remote cloud and edge resources to optimize the deployment of multi-layered services is crucial for both service providers and users, and there is an urgent need to research effective service deployment solutions for different scenarios. Summary of the Invention

[0006] This invention proposes a multi-layer service optimization deployment method based on cloud-edge computing networks, primarily applied to edge computing. Its main advantage is the ability to minimize both service provider deployment costs and user request latency, thereby improving service quality at a lower cost. The specific solution of this invention is as follows:

[0007] 1. Obtain an optimized deployment scheme for multi-layer services in a cloud-edge computing network using a greedy algorithm:

[0008] Step 1: Construct a multi-layer edge computing network scenario. The network contains... Layer edge cloud and a remote cloud , No. The edge of the cloud layer is composed of This indicates that edge clouds at all levels are composed of... This indicates that the user set is composed of This indicates that the type of service requested by the user is determined by... This indicates that the multi-layered edge server network is modeled as a tree structure, in the user... Edge clouds that can be connected within the range of wireless communication For local edge cloud Remote cloud As the root node, the local edge cloud Each edge cloud node serves as a leaf node. Able to interact with parent node and child nodes communication;

[0009] Step 2: Establish a deployment model where user requests are responded to by the edge cloud where the service has been deployed, and time is represented as a discrete variable. ,exist Time users In position Storage and data size are respectively , services Send a request, the request tuple is represented as All requests are recorded as , express 24-hour service Whether deployed on edge cloud Above, the value can be 1 or 0, representing all services. All are deployed on a remote cloud by default. superior, , This indicates the storage capacity of the edge cloud; all services located on the edge cloud must meet this requirement. After a user sends a service request, it is first processed by the local edge cloud. Response, if If the local edge cloud does not store the service, it continues to upload to the next higher-level edge cloud until it is uploaded to the edge cloud that stores the service. ,and , Respond to the service;

[0010] Step 3: Establish a communication model for users. With local edge cloud Information transmission rate is expressed as Edge clouds at different levels and The transmission speed between them is The total transmission latency after the user sends a request. There are three scenarios, users With local edge cloud The transmission delay between them is ,user non-local edge cloud The delay between ,user With remote cloud The delay is expressed as Define the set of path nodes from the user to the remote cloud as follows: Transmission delay ;

[0011] Step 4: Minimize transmission latency and deployment costs by deploying services in edge cloud and remote cloud. Transmission latency is defined as... , will serve The deployment cost is expressed as The total deployment cost of the service provider is expressed as The service deployment problem of minimizing transmission latency and deployment cost is represented as: Based on Pareto optimization, the multi-objective problem is modeled as a single-objective optimization problem, and the definition is... For deployment plan utility function

[0012]

[0013] The goal is to solve Minimum optimal solution ;

[0014] Step 5: Sort the service types according to the number of requests from largest to smallest, and deploy the services sequentially on the common parent node of multiple local edge clouds that access the service. Then, use a greedy method to solve the utility function. The minimum value is obtained, and finally an optimized multi-layer service deployment scheme is obtained.

[0015] 2. Furthermore, The user request at any given time is known information, including the type of service requested, the storage space required for the service, the size of the service, and the location of the request.

[0016] 3. Furthermore, during the calculation of the multi-layer service deployment scheme, the storage constraints of the edge cloud should always be met.

[0017] 4. Further, the services are initially deployed sequentially according to the number of service requests. Then, a greedy method is used to create copies of the services to further reduce transmission latency. Finally, the services are deployed in containers on various edge clouds according to the service deployment plan, thereby minimizing transmission latency and deployment costs.

[0018] 5. Furthermore, a greedy multi-tiered service deployment scheme includes at least the following steps:

[0019] 1) Statistics All requests to the service are recorded. All local edge clouds that issued the request The position is Each edge cloud Corresponding to a geographical location ;

[0020] 2) Place each position Regarding services The number of requests is denoted as Then each location provides services The average number of requests is denoted as ,definition It is the impact factor. ,like ,but ;

[0021] 3) Each service Deploy to By finding the least common parent node of all elements in the network tree, a preliminary deployment plan is obtained with minimal deployment cost. };

[0022] 4) Existing services The deployment scheme is represented as For service After creating a copy, it is deployed on the next level's common parent node, resulting in the deployment scheme as follows: Utility gain is denoted as ;

[0023] 5) For all services calculate Ultimately, the current optimal option is selected. ,according to Update deployment plan ;

[0024] 6) Repeat steps 4) and 5) until the algorithm ends, yielding the final service deployment plan. .

[0025] Compared with existing technologies, the advantages of this method are:

[0026] A multi-layer service deployment network model, computation model, deployment model, and communication model based on cloud-edge computing networks were established. Single-layer or dual-layer MEC servers suffer from limited storage capacity, making them unable to handle data-intensive tasks. Secondly, this invention extracts public services and deploys them to the upper layer of the network based on statistical analysis of the similarities and differences in service requests from different geographical locations, reducing service deployment costs. This invention also optimizes the objective function, creating replicas for services and performing a greedy optimal selection process. Finally, a multi-objective optimized multi-layer service deployment scheme in a cloud-edge computing network is obtained. Attached Figure Description

[0027] Figure 1 This is a flowchart of the present invention.

[0028] Figure 2 This is a deployment flowchart for the present invention.

[0029] Figure 3 This diagram illustrates the deployment of services for this invention on a common parent node and the creation of copies of the services. Detailed Implementation

[0030] The following is in conjunction with the appendix Figure 3 The present invention will be described in further detail below.

[0031] Step 1: Construct a multi-layer edge computing network scenario, by Figure 3 (a) Example, existing in the network Layer edge cloud and a remote cloud , ,in , , The service type requested by the user is ;

[0032] Step 2: Establish a deployment model, assuming the service's storage and data size are both [value missing]. That is, set Please represent the tuple as follows: , , , , , , , , , , , Storage resources of the edge cloud located in the first, second, and third layers are represented as follows: , , , ;

[0033] Step 3: Establish a communication model for users. With local edge cloud Information transmission rate is expressed as The edge clouds between the first and second layers and The transmission speed between them is The edge clouds of the second and third layers and The transmission speed between them is The third layer and the edge cloud of remote clouds and The transmission speed between them is ;

[0034] Step 4: By deploying services on edge clouds and remote clouds, transmission latency and deployment costs are minimized, thus extending service availability. The deployment cost is expressed as The total deployment cost for the service provider is expressed as The service deployment problem of minimizing transmission latency and deployment cost is represented as: The multi-objective problem is modeled as a single-objective optimization problem, and defined as follows: For deployment plan utility function , ,in For deployment plan Total delay, and These are the minimum and maximum transmission delays, respectively. For deployment plan Total deployment cost at the time, maximum deployment cost Lowest deployment cost ;

[0035] Step 5: Sort the service types according to the number of requests from largest to smallest, and deploy the services sequentially on the common parent nodes of multiple local edge clouds that access the service. Then, use a greedy method to solve the utility function. The minimum value is obtained, and the optimized multi-layer service deployment scheme is finally obtained. This step includes at least the following steps:

[0036] 1) Statistics All requests to the service are recorded. All local edge clouds that issued the request Location, , , , , ;

[0037] 2) Service at each location The average number of requests is denoted as ,definition =0.4, therefore it does not exist. The situation;

[0038] 3) By Figure 3 (b) Example, each service Deploy to A preliminary deployment plan was obtained by finding the least common parent node of all elements in the network tree. , , ;

[0039] 4) Existing services The deployment scheme is represented as For service After creating a copy, it is deployed on the next level's common parent node, resulting in the deployment scheme as follows: Utility gain is denoted as ;

[0040] 5) For all services calculate Ultimately, the current optimal option is selected. At the end of the first round, the result was... hour, Maximum, update deployment plan , , ;

[0041] 6) Repeat steps 4) and 5), by Figure 3 (c) Example, until The algorithm concludes, yielding the final service deployment solution. .

[0042] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. For those skilled in the art, improvements and modifications obtained without departing from the inventive concept should also be considered within the scope of protection of the present invention.

Claims

1. A method for optimizing the deployment of multi-layer services in a cloud-edge computing network, comprising the following steps: Step 1: Construct a multi-layer edge computing network scenario. The network contains... Layer edge cloud and a remote cloud , No. The edge of the cloud layer is composed of This indicates that edge clouds at all levels are composed of... This indicates that the user set consists of This indicates that the type of service requested by the user is determined by... This indicates that the multi-layered edge server network is modeled as a tree structure, in the user... Edge clouds that can be connected within the range of wireless communication For local edge cloud Remote cloud As the root node, the local edge cloud Each edge cloud node serves as a leaf node. Able to interact with parent node and child nodes communication; Step 2: Establish a deployment model where user requests are responded to by the edge cloud where the service has been deployed, and time is represented as a discrete variable. ,exist Time users In position Storage and data size are respectively , services Send a request, the request tuple is represented as All requests are recorded as , express 24-hour service Whether deployed on edge cloud Above, the value can be 1 or 0, representing all services. All are deployed on a remote cloud by default. superior, , This indicates the storage capacity of the edge cloud; all services located on the edge cloud must meet this requirement. After a user sends a service request, it is first processed by the local edge cloud. Response, if If the local edge cloud does not store the service, it continues to upload to the next higher-level edge cloud until it is uploaded to the edge cloud that stores the service. ,and , Respond to the service; Step 3: Establish a communication model for users. With local edge cloud Information transmission rate is expressed as Edge clouds at different levels and The transmission speed between them is The total transmission latency after the user sends a request. There are three scenarios, users With local edge cloud The transmission delay between them is ,user non-local edge cloud The delay between ,user With remote cloud The delay is expressed as Define the set of path nodes from the user to the remote cloud as follows: Transmission delay ; Step 4: Minimize transmission latency and deployment costs by deploying services in edge cloud and remote cloud. Transmission latency is defined as... , will serve The deployment cost is expressed as The total deployment cost of the service provider is expressed as The service deployment problem of minimizing transmission latency and deployment cost is represented as: Based on Pareto optimization, the multi-objective problem is modeled as a single-objective optimization problem, and the definition is... For deployment plan Utility function: The goal is to solve Minimum optimal solution ,in For deployment plan Total delay, and These are the minimum and maximum transmission delays, respectively. For deployment plan Total deployment cost at that time and These represent the minimum and maximum deployment costs, respectively. Step 5: Sort the service types according to the number of requests from largest to smallest, and deploy the services sequentially on the common parent nodes of multiple local edge clouds that access the service. Then, use a greedy method to solve the utility function. The minimum value is obtained, and the optimized multi-layer service deployment scheme is finally obtained. This step includes at least the following steps: 1) Statistics All requests to the service are recorded. All local edge clouds that made the request The position is Each edge cloud Corresponding to a geographical location All local edge cloud locations in the network constitute a set. and ; 2) Place each position Regarding services The number of requests is denoted as Then each location provides services The average number of requests is denoted as ,definition It is the impact factor. ,like ,but ; 3) Each service Deploy to By finding the least common parent node of all elements in the network tree, a preliminary deployment plan is obtained with minimal deployment cost. }; 4) Existing services The deployment scheme is represented as For service After creating a copy, it is deployed on the next level's common parent node, resulting in the deployment scheme as follows: Utility gain is denoted as ; 5) For all services calculate Ultimately, the current optimal option is selected. ,according to Update deployment plan ; 6) Repeat steps 4) and 5) until... The algorithm concludes, yielding the final service deployment solution. .

2. The method for optimizing the deployment of multi-layer services in a cloud-edge computing network according to claim 1, characterized in that... The aforementioned The user request at any given time contains known information, including the type of service requested, the storage space required for the service, the size of the service, and the location of the request.

3. The method for optimizing the deployment of multi-layer services in a cloud-edge computing network according to claim 1, characterized in that... During the calculation of multi-layer service deployment schemes, the storage constraints of edge cloud should always be met.

4. The method for optimizing the deployment of multi-layer services in a cloud-edge computing network according to claim 1, characterized in that... The services are initially deployed sequentially based on the number of service requests. Then, a greedy method is used to create replicas of the services to further reduce transmission latency. Finally, the services are deployed as containers on various edge clouds according to the service deployment plan, thereby minimizing transmission latency and deployment costs.

Images