A digital twin network architecture approach for service placement and caching in mobile edge computing systems

Abstract

This invention discloses a method for a network architecture for service placement and caching in a mobile edge computing system, relating to the fields of digital twins, network architecture, and optimization algorithms. Based on a Merkle Tree (a hash-value-based tree data structure) and an improved genetic algorithm, this method achieves a trade-off between computational complexity and accuracy compared to traditional service placement algorithms. By combining digital twin network technology, wireless communication technology, and specific network communication scenarios in the context of 6G, this method minimizes cloud load while satisfying user constraints, improving the efficiency of user service type requests.

Description

Technical Field

[0001] The present invention relates to the fields of digital twins, network architectures, and optimization algorithms, and in particular to a method for a digital twin network architecture for service placement and caching in a mobile edge computing system. Background Art

[0002] The surge in new mobile services, such as augmented reality, online gaming, and autonomous driving, is driving demand for low-latency access to computing resources that cannot be met by existing centralized cloud systems. MEC, an emerging distributed cloud architecture, pushes significant computing power to the edge of the network, closer to end users, thereby bypassing the fundamental latency limitations of today's prominent centralized cloud systems. This trend is expected to continue and play a significant role in next-generation 6G networks supporting latency-sensitive and compute-intensive services.

[0003] However, hundreds of billions of wireless devices will connect to 6G networks, generating vast amounts of data. This massive amount of data will require significant computing and communication resources from edge servers. Therefore, how edge computing systems efficiently cache and distribute various services is a major challenge in providing high-quality services for emerging and new applications in the 6G environment.

[0004] Combined with the network characteristics of 6G, digital twins have become a promising technology for efficiently placing and requesting services in MEC systems. Digital replicas of physical entities (such as devices, services, and MEC status information) are built on servers based on historical data and real-time operational status. Digital twins enable close monitoring, real-time interaction, and reliable communication between digital spaces and physical systems, thereby optimizing the operation of physical systems. Summary of the Invention

[0005] The purpose of the present invention is to solve the above problems and to design a method for digital twin network architecture for service placement and caching of mobile edge computing systems based on the field of optimization algorithms with the assistance of digital twin networks, so as to minimize the cloud load while satisfying user constraints and improve the efficiency of user service type requests.

[0006] The present invention achieves the above-mentioned purpose through the following technical solutions:

[0007] A method for a digital twin network architecture for service placement and caching in a mobile edge computing system, comprising the following steps:

[0008] S1: Model of mobile edge computing system assisted by digital twin network;

[0009] S2: Initialize the coverage user information of each mobile edge computing server and the status information of each server, and import them into the digital twin network;

[0010] S3: Establish the optimization problem, define the target optimization function and parameter constraints;

[0011] S4: Initialize the types of user service requests and the number of service request users and the types of cloud network service caches;

[0012] S5: Using the actual physical scenario of the digital twin network to perform dimensionality reduction, the optimization problem of the service placement matrix variables and the task migration matrix variables is transformed into an optimization problem only about the service placement matrix variables;

[0013] S6: Combined with the physical scenario, the service request process is abstracted into a hash-based data structure - Merkletree;

[0014] S7: Encode the service placement matrix variables into chromosomes and design the relevant parameters of the genetic algorithm;

[0015] S8: Considering that the actual number of users is much larger than the number of service types placed on the edge server, a chromosome pruning algorithm is involved;

[0016] S9: Combine S7 and S8 to improve and optimize the traditional genetic algorithm;

[0017] S10: setting new species number and crossover mutation parameters, and performing simulation records on the improved genetic algorithm;

[0018] S11: Repeat step S10 multiple times, and continuously update the optimization variables using the optimization iterative algorithm until the algorithm converges to the specified accuracy.

[0019] The beneficial effects of the present invention are: designing a method for digital twin network architecture for service placement and caching in mobile edge computing systems based on discrete data structures, achieving a trade-off between computational complexity and accuracy compared to traditional network architectures, combining digital twin technology, and jointly optimizing the service request process of wireless device users, thereby minimizing the cloud load while satisfying user constraints and improving the efficiency of user service type requests. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of a method for a digital twin network architecture for service placement and caching of a mobile edge computing system according to the present invention;

[0021] Figure 2 This is a scenario model diagram of a method for a digital twin network architecture for service placement and caching in a mobile edge computing system according to the present invention;

[0022] Figure 3 It is a model abstract diagram of the Merkle tree, a tree data structure based on hash values in the present invention;

[0023] Figure 4 It is a schematic diagram of the chromosome encoding of the present invention. DETAILED DESCRIPTION

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the embodiments of the present invention. It should be understood that the described embodiments are only a portion of the embodiments of the present invention, not all of them. Generally, the components of the embodiments of the present invention described and illustrated in the drawings herein may be arranged and designed in a variety of different configurations.

[0025] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0026] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0027] In the description of the present invention, it should be understood that the terms "upper", "lower", "inside", "outside", "left", "right", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, or are the orientations or positional relationships in which the inventive product is conventionally placed when in use, or are the orientations or positional relationships conventionally understood by those skilled in the art. These are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0028] In addition, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance. In the description of the present invention, it should also be noted that, unless otherwise clearly stipulated and limited, the terms "setting", "connection", etc. should be understood in a broad sense. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, an indirect connection through an intermediate medium, or it can be a communication between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0029] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0030] The present invention provides a network architecture method that effectively balances computational complexity and computational accuracy, so as to maximize the requests of mobile edge computing servers while minimizing the cloud load under the premise of satisfying user constraints. The method consists of three parts: system model establishment, model transformation, and solution. Figure 1 As shown, the specific steps include:

[0031] S1: A model of a mobile edge computing system based on a digital twin network, specifically:

[0032] As attached Figure 2 As shown in Figure 1, we consider a multi-user MEC system consisting of a set of U wireless device (WD) users, a set of N MEC (Mobile Edge Computing) servers, a digital twin network, and a cloud I. Each WD can request services from a set of different types. The services requested by each wireless device user within the same MEC coverage area must be different. This is because if multiple identical services are requested within the same area, they can be merged into one, while if a wireless device user performs multiple requests, it can be split across multiple wireless device users. Each MEC server can only communicate directly with WDs within the MEC coverage area. The digital twin network is a lightweight network. It does not store the entire service entity, but only the state information of each MEC (such as the current storage capacity, computing resources, and service type) and the route to the MEC where the service is located. The cloud stores all service categories and entities. If a WD requests a service that is not available in all MECs, the WD directly retrieves the service from the cloud.

[0033] S2: Initialize the coverage user information of each mobile edge computing server and the status information of each server, and import them into the digital twin network.

[0034] Establish a communication model and initialize all communication variables in the system. Assume W is the channel bandwidth, u s WD to serve the request. Therefore, we will Meaning for each WD u s The transmission power, σ 2 is defined as the noise power of each WD receiver. The channel power gain of each WD is expressed as

[0035] In MEC and cloud, the time to execute the service requested by WD us can be calculated by the following formula:

[0036]

[0037] Since the amount of data requested and queried from the DTN is small, the communication delay between MEC and DTN can be ignored. Cloud servers are usually deployed far away from users. Therefore, we assume that the time delay for WDs to directly extract services from the cloud is much longer than that of MEC. We use T* to represent the time to obtain services from the cloud. Using Equation (8), WD u s The total time to complete its service request can be calculated using the following formula:

[0038]

[0039] because is a variable determined by x and y, we express equation (9) as We assume that the QoS requirement of each WD is Then, we can get:

[0040]

[0041] S3: Establish an optimization problem, define the target optimization function and parameter constraints:

[0042]

[0043] S4: Initialize the user service request type, the number of service request users, and the cloud network service cache type.

[0044] We use random initialization to initialize the user's requested service type and randomize the service type for service placement on MEC.

[0045] S5: Use the actual physical scenario of the digital twin network to reduce the dimension and transform the optimization problem of the service placement matrix variables and the task migration matrix variables into an optimization problem only about the service placement matrix variables.

[0046] The optimization problem after dimensionality reduction is as follows:

[0047]

[0048] S6: Combined with the physical scenario, the service request process is abstracted into a hash-based data structure - Merkletree.

[0049] Due to the uniqueness of each service type, each service type can be mapped one by one using a unique hash value. To solve this problem, we first introduced a hash-based data structure called Merkle tree, such as Figure 3As shown in Figure 2, we simulate the service placement process and the WD service request process between WD and MEC. DTN, each MEC server, and each WD are abstracted as a node, and each type of service is abstracted as a unique hash value. We assume that the set of WDs associated with each MEC is U i ,i∈N={1,2,...,N}, then U={U1,U2,...,U n}. Each MEC's child node is a set with different hash values, which we can represent as

[0050] We use the set nodes and H = {H1,H2,...,H n},i∈N={1,2,...,N} The hash value set of all nodes is stored in the DTN. Since each service requested by the WD has a unique hash value, before requesting the service, the WD compares the hash value of the requested service with the hash value stored in the local MEC. If the local MEC does not have the corresponding hash value, the local MEC will query the DTN whether there is a hash value corresponding to the service. At the same time, the DTN will also provide routing information (for example, by looking up the hash code in the MEC node) and statement information (such as computing capacity and storage capacity) of the requested service. If the DTN does not store the hash value of the service, the WD will call the service from the cloud, which means greater latency. We introduce the task initialization workflow based on the Merkle tree algorithm as Algorithm 1.

[0051]

[0052] S7: Encode the service placement matrix variables into chromosomes and design the relevant parameters of the genetic algorithm.

[0053] We use a genetic algorithm for candidate pattern pruning to update the service placement matrix x, which can be obtained as follows:

[0054] x={x ns ∈{0,1}|n∈N,s∈S} (16)

[0055] The gene x in the chromosome ns =1 indicates that service s is located in the nth MEC, otherwise x ns = 0. The chromosome coding diagram is as follows Figure 4 shown.

[0056] S8: Considering that the actual number of users is much larger than the number of service types placed on the edge server, a chromosome pruning algorithm is designed.

[0057] We encode the matrix x to represent Figure 3The chromosome shown in , where we denote the entire candidate pattern as M. The chromosome encoding on each MEC node is denoted as pattern M n ,M={M1,...,M N We assume that the hash value set of WD services in each MEC coverage area is represented as When the hash value corresponding to a service deployed on an MEC node is a subset of the hash values of services requested by users in the area covered by the MEC node, the chromosome corresponding to the encoding fragment of the service placed in the MEC node can be pruned. Thus, the chromosome can be continuously shortened in an iterative process to reduce the size of the problem. We introduce the candidate pattern pruning algorithm as Algorithm 2.

[0058]

[0059] S9: Combine S7 and S8 to improve and optimize the traditional genetic algorithm.

[0060] S10: Set new species number and crossover mutation parameters, and simulate and record the improved genetic algorithm.

[0061] S11: Repeat step S10 multiple times, and continuously update the optimization variables using the optimization iterative algorithm until the algorithm converges to the specified accuracy.

[0062] The overall algorithm is described as follows:

[0063]

[0064]

[0065] The technical solution of the present invention is not limited to the above-mentioned specific embodiments. Any technical variations made according to the technical solution of the present invention fall within the protection scope of the present invention.

Claims

1. A method for digital twin network architecture for service placement and caching in mobile edge computing systems, characterized in that: The following steps are involved: S1: A model of a mobile edge computing system assisted by a digital twin network is established, including the following steps: Build a multi-user MEC system consisting of wireless device users, MEC servers, digital twin networks, and the cloud; Define the service placement matrix x and the request routing decision matrix y, and establish storage capacity, computing load, and service quality constraints; Build channel models to calculate uplink data rates and service execution times; S2: Initialize the coverage user information of each mobile edge computing server and the status information of each server, and import them into the digital twin network; S3: Establish the optimization problem, define the target optimization function and parameter constraints; S4: Initialize the types of user service requests and the number of service request users and the types of cloud network service caches; S5: Using the actual physical scenario of the digital twin network to perform dimensionality reduction, the optimization problem of the service placement matrix variables and the task migration matrix variables is transformed into an optimization problem only about the service placement matrix variables; S6: Combined with the physical scenario, the service request process is abstracted into a hash-based data structure - Merkletree; S7: Encode the service placement matrix variables into chromosomes and design the relevant parameters of the genetic algorithm; S8: Considering that the actual number of users is much larger than the number of service types placed on the edge server, a chromosome pruning algorithm is designed; S9: Combine S7 and S8 to improve and optimize the traditional genetic algorithm; S10: setting new species number and crossover mutation parameters, and performing simulation records on the improved genetic algorithm; S11: Repeat step S10 multiple times, and continuously update the optimization variables using the optimization iterative algorithm until the algorithm converges to the specified accuracy.

2. The method for a digital twin network architecture for service placement and caching in a mobile edge computing system according to claim 1, characterized in that: Included in S1: S101. Consider a multi-user MEC system consisting of a set of U wireless device users, a set of N MEC servers, a digital twin network, and a cloud I. Each wireless device requests the required service from a set of different types of services. The services requested by wireless device users in the same area covered by each MEC server must be different. Each MEC server can only communicate directly with WDs within the MEC coverage area. The digital twin network only stores the status information of each MEC and the route to the MEC where the service is located, and the cloud stores all service categories and entities. If the service requested by WDs does not exist in all MECs, the WD will directly extract the service from the cloud. S102. Define matrix x as service placement and y as request routing decision, and give the following formula: Among them, x ns =1 means service s is placed in the nth MEC, otherwise x ns =0;y nu ∈{0,1} indicates whether the request of WD u is routed to the nth MEC, where y nu =1 means the request is routed to MEC, y nu =0 indicates the routing decision of the cloud; definition As a MEC set that routes directly to WD u, each WD service request needs to be routed to a MEC server or cloud: If s u is the service requested by WD u. In order to route WD u’s request to node n, service s u Place in a node: Use r s Indicates the storage space associated with service S, symbol f s Represents the computing capacity of the service. The total amount of service data placed in the MEC must not exceed its storage capacity: Among them, R n Indicates the storage capacity of each MEC server; The total computational load of user requests routed to a node n must not exceed its computational capacity: Among them F n Represents the computing power used to execute services in each MEC server; S103. Assume W is the channel bandwidth, u s The WD that serves the request will define For each WD u s The transmission power, σ 2 is defined as the acoustic power of each WD receiver; the channel power gain of each WD is expressed as The uplink data rate of each WD to the MEC associated with the WD is given by: Similarly, the uplink data rate from each MEC server to another MEC node is given by the following formula: In MEC and cloud, execute WD u s The time for the requested service is given by the following formula: Assuming that the time delay of WDs extracting services directly from the cloud is much longer than that of MEC, let T* represent the time of obtaining services from the cloud. Using equation (8), WD u s The total time to complete its service request is given by the following formula: because is a variable determined by x and y, and equation (9) can be expressed as Assume that the QoS requirement of each WD is get:

3. The method for a digital twin network architecture for service placement and caching in a mobile edge computing system according to claim 2, characterized in that: In S2, the coverage user information of each mobile edge computing server and the status information of each server are initialized, and the digital twin network and the communication model are imported at the same time, where WD u s The time of service requested is determined by Make confirmation.

4. The method for a digital twin network architecture for service placement and caching in a mobile edge computing system according to claim 3, characterized in that: In S3, establish the optimization problem, define the target optimization function and parameter constraints.

5. The method for a digital twin network architecture for service placement and caching in a mobile edge computing system according to claim 4, characterized in that: In S5, the actual physical scenario of the digital twin network is used for dimensionality reduction, and the optimization problem of the service placement matrix variables and the task migration matrix variables is transformed into an optimization problem only about the service placement matrix variables.

6. The method for a digital twin network architecture for service placement and caching in a mobile edge computing system according to claim 5, characterized in that: In S6, combined with the physical scenario, the service request process is abstracted into a hash-based data structure - Merkle tree, which abstracts the logical process of the user's service request.

7. The method for a digital twin network architecture for service placement and caching in a mobile edge computing system according to claim 6, characterized in that: In S7, the service placement matrix variables are used for chromosome encoding and the relevant parameters of the genetic algorithm are designed.

8. The method for digital twin network architecture for service placement and caching of mobile edge computing systems according to claim 7, characterized in that: In S8, considering that the actual number of users is much larger than the number of service types placed on the edge server, a chromosome pruning algorithm is involved.

9. The method for a digital twin network architecture for service placement and caching in a mobile edge computing system according to claim 8, characterized in that: In S10, new species number and crossover mutation parameters are set, and simulation records are performed on the improved genetic algorithm.

10. The method for digital twin network architecture for service placement and caching of mobile edge computing system according to claim 9, characterized in that: In S11, step S10 is repeated multiple times, and the optimization variables are continuously updated using the optimization iterative algorithm until the algorithm converges to the specified accuracy.

Images