Distributed edge computing network resource arrangement method

Abstract

The invention discloses a distributed edge computing network resource arrangement method, and relates to the technical field of wireless communication and computing. The method comprises the steps that all end nodes of an edge computing network are regarded as intelligent agents, a centralized resource joint arrangement model TAC is initialized, and an edge computing network environment and task flows are initialized; performing centralized training on the TAC model until the model converges; wherein the TAC model is constructed on the basis of a reinforcement learning framework of a single Actor multi-Critic structure and a Transform model; based on a transfer learning method, an original TAC model is converted into a distributed resource arrangement model DTAC, and the DTAC model is used for carrying out local network resource arrangement decisions of users at all user nodes in a distributed mode. Compared with the prior art, the communication and calculation time delay can be reduced.

Description

Technical Field

[0001] The present invention relates to the field of wireless communications and computing technologies, and more specifically, to a distributed edge computing network resource orchestration method. Background Art

[0002] In the field of wireless communications, with the rapid development of IoT devices, it is expected that there will be 50 billion IoT devices in the entire network, making wireless user equipment (UE) evolve into content producers and consumers. These massively interconnected user equipment will generate various data traffic and data requirements, including caching, communication, and computing. Since most UEs in edge networks have only limited computing power, Mobile Edge Computing (MEC) plays a key role in providing computing power. By deploying edge servers in MEC networks, caching services and computing power can be provided to UEs to save computing energy and time costs.

[0003] However, offloading all tasks to edge servers may significantly reduce computing performance due to signaling overhead and queuing delay. To solve this problem, an intelligent computing offloading and cache placement strategy is proposed to solve this problem by selectively offloading and caching appropriate tasks to edge servers. Through efficient offloading and caching strategies, the time and energy consumption of task computing and communication can be minimized. In addition, considering that multiple edge servers in MEC networks share computing and communication resources, efficient joint resource allocation is also crucial to meet the computing needs of large-scale tasks. Therefore, many studies currently consider how to centrally and jointly offload cache and allocate communication computing resources for each user's tasks. However, this centralized joint orchestration approach will also introduce a large amount of computing delay and communication delay caused by uploading task features and distributing orchestration actions, which in turn affects the application of mobile edge computing in delay-sensitive task scenarios, and has therefore become a technical problem that urgently needs to be solved. Summary of the invention

[0004] In order to overcome the inevitable communication and computing delay problems of the centralized resource orchestration in the above-mentioned prior art, the present invention provides a distributed edge computing network resource orchestration method.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] In a first aspect, a distributed edge computing network resource orchestration method includes:

[0007] All end nodes of the edge computing network are regarded as intelligent agents, the centralized resource joint orchestration model TAC is initialized, and the edge computing network environment and task flow are initialized;

[0008] Centrally train the TAC model until the model converges; wherein the TAC model is constructed based on a reinforcement learning framework and a Transformer model of a single-Actor multi-Critic structure;

[0009] Based on the transfer learning method, the original TAC model is converted into a distributed resource orchestration model DTAC, which is used to make local network resource orchestration decisions for users in a distributed manner at each user node.

[0010] In a second aspect, an electronic device includes:

[0011] A memory for storing computer executable instructions or computer programs;

[0012] The processor is used to implement the method described in the first aspect when executing the computer executable instructions or computer program stored in the memory.

[0013] In a third aspect, a computer-readable storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by a processor to implement the method described in the first aspect.

[0014] In a fourth aspect, a computer program product comprises a computer program or computer executable instructions, wherein when the computer program or computer executable instructions are executed by a processor, the method of the first aspect is implemented.

[0015] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0016] By combining the Transformer structure with the reinforcement learning framework in the TAC model, the present invention can make the model adapt to scenarios with different numbers of users in the edge computing network, and can efficiently process high-dimensional state information of multi-user and multi-task characteristics, greatly improving the resource orchestration performance and convergence speed of the model; through the transfer learning method, the original TAC model is converted into a distributed resource orchestration model DTAC, so that the model can gradually no longer need the information of other user nodes and independently perform local resource orchestration, thereby reducing the inevitable communication and computing delays of centralized resource orchestration. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of the cloud-edge-end collaborative edge computing network scenario in Example 1 of the present application;

[0018] Figure 2 This is a flow chart of the distributed edge computing network resource orchestration method in Example 1 of the present application;

[0019] Figure 3 This is a schematic diagram of the round-based multi-user edge computing process in Example 1 of the present application;

[0020] Figure 4 This is a schematic diagram of the training process of the resource orchestration model in Example 1 of the present application;

[0021] Figure 5 This is a schematic diagram of the network structure shared by TAC and DTAC in Example 1 of the present application;

[0022] Figure 6 This is a comparison chart of the task time performance results of each method in the simulation experiment in Example 1 of the present application;

[0023] Figure 7 This is a comparison chart of the energy consumption performance results of each method in the simulation experiment in Example 1 of the present application. DETAILED DESCRIPTION

[0024] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged in appropriate cases, which is only to describe the distinction mode adopted by the objects of the same attributes in the embodiments of the present application when describing. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment containing a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment. The term "determine" widely covers various actions, and may include acquisition, calculation, calculation, processing, derivation, investigation, search (for example, search in a table, database or other data structure), ascertainment, and similar actions, and may also include reception (for example, receiving information), access (for example, accessing data in a memory) and similar actions, and may also include generation, creation, establishment and similar actions, as well as parsing, selection, selection and similar actions, etc. The relevant definitions of other terms will be given in the following description.

[0025] It should be noted that when an element is considered to be "connected" to another element, it can be directly connected to the other element, or connected to the other element through an intermediate element. In addition, the "connection" in the following embodiments should be understood as "electrical connection", "communication connection", etc. if there is transmission of electrical signals or data between the connected objects.

[0026] The drawings are for illustrative purposes only and are not to be construed as limiting the present patent;

[0027] In order to better illustrate the present embodiment, some parts in the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product;

[0028] It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0029] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0030] Example 1

[0031] Consider the edge computing network scenario based on cloud-edge-device collaboration, such as Figure 1 As shown in the figure, there is a single cloud server, multiple edge servers and a large number of edge nodes in the network. Among them, the cloud server is connected to each edge server, and each edge node is wirelessly connected to the edge server. Each end node is only connected to one edge server within the corresponding range. In the edge computing network, each end node will have new computing task requirements at regular intervals. The end node can choose the offloading and computing location of the computing task, including local node computing, computing at the edge server, and computing at the cloud server. When computing locally, the node can decide the CPU frequency for calculation by itself; when computing at the edge or cloud server, the node can decide the channel and transmission power for uplink transmission. The edge and cloud servers will optimize the computing power according to the uploaded user tasks, and return the calculation results to each user.

[0032] For the edge computing network scenario, this embodiment provides a distributed edge computing network resource orchestration method, see Figure 2 ,include:

[0033] S1. Consider all end nodes (i.e. user nodes) of the edge computing network as intelligent agents, initialize the centralized resource joint orchestration model TAC (Transformer-Actor-Critic), and initialize the edge computing network environment and task flow;

[0034] S2. Centrally train the TAC model until the model converges; wherein the TAC model is constructed based on a reinforcement learning framework and a Transformer model of a single-Actor multi-Critic structure;

[0035] S3. Based on the transfer learning method, the original TAC model is converted into a distributed resource orchestration model DTAC (Distributed Transformer-Actor-Critic), which is used to make local network resource orchestration decisions for users in a distributed manner at each user node.

[0036] Those skilled in the art should understand that the network resource scheduling decision includes but is not limited to communication (such as selection of transmission channels, transmission power, etc.), computing resource allocation (such as computing frequency), and task offloading and cache location decisions.

[0037] It should be emphasized that the distributed system performs local communication, computing resource allocation, task offloading, and cache location decisions at the user node without the need for network and task information of other users as in a centralized model, thereby eliminating the communication and computing delays caused by centralized resource orchestration.

[0038] It should be noted that, combined with the Transformer structure, the model can adapt to scenarios with different numbers of users in the edge computing network, and can efficiently process high-dimensional state information of multi-user and multi-task characteristics, greatly improving the resource orchestration performance and convergence speed of the model; in addition, based on the transfer learning method, the original TAC model is converted into a distributed resource orchestration model DTAC, so that the model can gradually no longer need information from other user nodes and independently perform local resource orchestration, thereby reducing the inevitable communication and computing delays of centralized resource orchestration.

[0039] It should also be noted that the method described in this embodiment is particularly suitable for round-based task calculation mode, see Figure 3 That is, after the task calculation of each user in the scenario is completed, it will wait for the task calculation of other users to be completed, and the next round of resource scheduling and task calculation will be carried out after the task calculation of all users is completed, so that there are more resources available for better resource scheduling. The duration of each round is the maximum task calculation time of all users in the round, avoiding the situation where some resources are still occupied when allocating communication, computing, and storage resources to users.

[0040] In some preferred embodiments, the TAC model includes an Actor module, a Critic1 module, and a Critic2 module, and a Transformer model is used as a shared model structure of the Actor module, the Critic1 module, and the Critic2 module, and a plurality of different classification layers are added after the Transformer to obtain the outputs of the Actor module, the Critic1 module, and the Critic2 module;

[0041] The Actor module is used to input observations about the user's network and task information. Get resource orchestration action

[0042] The Critic1 module is used to input S for task time consumption t , A t And output the Q value of the corresponding task time

[0043] The Critic2 module targets energy consumption, and uses the input St , A t And output the corresponding energy consumption Q value

[0044] It should be noted that the resource arrangement action of user j Including but not limited to the task offloading location (local, edge server, cloud server), transmission channel, transmission power, local computing power, etc. The network information of user j includes the maximum receiving signal-to-noise ratio from the user to the corresponding edge server, as well as the ideal computing delay and ideal communication delay when the task is offloaded to the edge server or cloud server (i.e., the minimum value that can be achieved without considering interference from other nodes).

[0045] In some specific implementation processes, resource orchestration actions on the edge server and cloud server sides will be quickly obtained on the server side through a convex optimization algorithm instead of being controlled by the end node.

[0046] It should be emphasized that compared with the method of estimating the sum of task time and energy consumption using a single Actor and single Critic network, the method of a single Actor and multiple Critic structure in this embodiment can well adapt to the weight relationship between different task time and energy consumption, and can achieve better estimation effect.

[0047] In some optional embodiments, during the TAC model parameter update process, the Critic1 module and the Critic2 module use the mean square error loss function to perform batch gradient descent, and the loss function of the Actor module is to minimize Q t,c (S t ,A t ) and Q t,e (S t ,A t ) is used as the goal to optimize task time and energy consumption.

[0048] Furthermore, when training the TAC model:

[0049] The loss function of the Critic1 module is:

[0050]

[0051] The loss function of the Critic2 module is:

[0052]

[0053] The loss function of the Actor module is:

[0054]

[0055] In some alternative embodiments, see Figure 4 The transfer learning method is used to convert the original TAC model into a distributed resource orchestration model DTAC, including:

[0056] The attention mechanism in the TAC model that completes centralized training is modified: the expression of the attention mechanism in the original Transformer is:

[0057]

[0058] Where V,Q,K∈R N*h is the attention mechanism parameter, V represents the value mapping of different user input vectors, Q and K represent the queue and key mapping of different user input vectors, respectively, N represents the number of UEs, and h represents the size of the hidden layer;

[0059] Modify the attention mechanism into the following expression:

[0060]

[0061] At this time, the Transformer model only cares about the state of each user node and ignores the state of other nodes. It only obtains the attention weight based on the state information of each user node. The goal is

[0062] Using the TAC model after modifying the attention mechanism as the DTAC model;

[0063] The local information of each user node is used as the input of the Actor module in the DTAC model to decide the distributed action of each user node, and the Q value of the distributed action is obtained through the Critic1 module and the Critic2 module in the DTAC model. The loss function of the Actor module in the DTAC model is calculated according to the Q value and the parameters are updated, so that the DTAC model gradually reduces the demand for other node information while keeping the action decision output similar.

[0064] It should be understood that in this embodiment, the structures of the DTAC model and the TAC model are similar, the difference being that the attention mechanism in the Transformer is different.

[0065] Figure 5 This is the neural network model structure shared by the TAC and DTAC models in this embodiment, which mainly adopts the Transformer framework. The resource orchestration model is implemented through a multi-head attention mechanism to adapt to scenarios with different numbers of users and to achieve efficient extraction and processing of high-dimensional state features of multiple users and multiple tasks, so as to achieve better orchestration effects and convergence performance.

[0066] It should be noted that in the original Transformer (i.e., the Transformer in TAC), by comparing the Q mapping and K mapping of each user, W = QK T , we can get the attention weight matrix between each user, which is also the main operation of the Transformer that uses the information between nodes; in order to remove this part of the operation from the original Transformer, so that the model can only use the local information of the node and the original centralized model for distributed decision-making, this embodiment uses decentralized transfer training to improve the attention mechanism, so that the Transformer model only cares about the state of each node itself and ignores the state of other nodes, and can be obtained only based on the state information of each node itself. For the sake of distinction, the Transformer after decentralized transfer learning is defined as DistributedTransformer in this embodiment.

[0067] In the above optional embodiment, the centralized trained TAC model guides the distributed DTAC model to perform efficient training, so that the Transformer in the original TAC can maintain similar action decision outputs during training while gradually reducing the demand for other node information, and gradually transform the TAC model into a DTAC model. After the final training is completed, the DTAC model can make the best orchestration action close to the centralized orchestration model TAC only through the user's local without the help of other node information.

[0068] In some optional embodiments, a multi-dimensional state input form of "number of users×local state length" is adopted to obtain an action output of an adaptive number of users of "number of users×action length" based on the Transformer.

[0069] It should be noted that in order to adapt to scenarios with different numbers of users, this embodiment improves the original reinforcement learning method of concatenating different user states into a single-dimensional state input, and instead adopts a multi-dimensional state input form of "number of users × local state length", so that Transformer can be used to obtain an action output of "number of users × action length" that is adaptive to the number of users.

[0070] In some preferred embodiments, in step S2, the centralized training of the TAC model includes:

[0071] In any time slot t:

[0072] Let each user j receive their own task τ j And get the corresponding observation o j, the cloud server aggregates the observations of each user As the input of the TAC model, the corresponding resource orchestration action is obtained It is then sent back to each user, including the task offloading location, the selected transmission channel and transmission power, and the calculated frequency allocation. The corresponding optimal caching strategy is obtained by the heuristic algorithm.

[0073] Each user j schedules action a according to resources j Get the corresponding task time and energy consumption And the corresponding rewards and The observation S t , choreography action A r And the corresponding reward R t,c ,R t,e Stored in the experience replay pool as training data;

[0074] After every T time slots: reset the edge computing network environment and replay the training data in the pool based on the experience Centralized training is performed through the PPO algorithm, and model parameters are updated until the TAC model converges.

[0075] It should be noted that for the network environment, each reset will be randomly initialized, and the number of end nodes in the network, the task type of the end nodes, and the location of the end nodes will change.

[0076] In a simulation experiment, the radius of the random initial position of the end node is set to 200m, the distance radius from the edge server to the cloud server is 300m, the total number of edge servers is 5, the number of end nodes covered by each edge server is 4 to 6, and the number of task types is 200. The distributed DTAC method proposed in this embodiment, as well as the centralized TAC, the traditional transfer learning method (Transfer Leaning), and the multi-intelligence reinforcement learning method HAPPO are observed for task time consumption and energy consumption performance. The observation results are shown in Figure 6 , Figure 7 It can be seen that the distributed DTAC method proposed in this embodiment can achieve task time and energy consumption performance close to that of the centralized TAC method (Centralized Transformer), and is significantly better than the traditional transfer learning method and the HAPPO method in the field of multi-agent reinforcement learning.

[0077] Example 2

[0078] This embodiment provides a computer-readable storage medium, on which is stored at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor, so that the processor executes part or all of the steps of the method provided in Example 1 of the present application.

[0079] It is understood that the storage medium may be transient or non-transient. Exemplarily, the storage medium includes, but is not limited to, a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0080] Exemplarily, the processor may be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP), or a field programmable gate array (FPGA).

[0081] Exemplarily, the read-only memory includes but is not limited to MASK ROM, PROM, EPROM, EEPROM, Flash, etc.

[0082] Exemplarily, the random access memory includes but is not limited to DRAM, SRAM, SDRAM, DDR SDRAM, etc.

[0083] In some examples, a computer program product is provided, which can be implemented in hardware, software or a combination thereof. As a non-limiting example, the computer program product can be embodied as the storage medium, or as a software product, such as an SDK (Software Development Kit).

[0084] As a non-limiting example, a computer program product is provided, the computer program product includes a computer program or a computer executable instruction, the computer program or the computer executable instruction is stored in a computer-readable storage medium. A processor of an electronic device reads the computer program or the computer executable instruction from the computer-readable storage medium, and the processor executes the computer executable instruction, so that the electronic device performs some or all steps of the method described in the embodiment of the present application.

[0085] In some examples, a computer program is provided, comprising a computer-readable code, and when the computer-readable code is run in a computer device, a processor in the computer device executes a part or all of the steps in the method.

[0086] This embodiment also proposes an electronic device, including a memory and a processor, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and when the processor executes the at least one instruction, at least one program, a code set or an instruction set, it implements part or all of the steps of the method described in Example 1.

[0087] In some examples, a hardware entity of the electronic device is provided, including: a processor, a memory and a communication interface; wherein the processor generally controls the overall operation of the electronic device; the communication interface is used to enable the electronic device to communicate with other terminals or servers through a network; the memory is configured to store instructions and applications executable by the processor, and can also cache data to be processed or processed by the processor and various modules in the electronic device (including but not limited to image data, audio data, voice communication data and video communication data), which can be implemented by flash memory (FLASH), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM) or random access memory (RAM).

[0088] The processor may include one or more processing elements. Thus, the processor may include one or more integrated circuits (ICs) configured to perform the functions of the processor. In addition, each integrated circuit may include a circuit (e.g., a first circuit, a second circuit, and other circuits, etc.) configured to perform the functions of the processor.

[0089] Furthermore, data may be transmitted between the processor, the communication interface and the memory via a bus, which may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together.

[0090] It can be understood that the options in the above-mentioned embodiment 1 are also applicable to this embodiment, so they will not be described again here.

[0091] The same or similar reference numerals correspond to the same or similar components;

[0092] The terms used to describe the positional relationship in the drawings are only used for illustrative purposes and should not be construed as limiting the present application;

[0093] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application may be combined with each other.

[0094] In different specific implementations, the method or system described in the present application can be implemented in software, hardware or a combination thereof. In addition, the order of the steps of the method can be changed, and various elements can be added, reordered, combined, omitted, modified, etc.

[0095] Obviously, the above-mentioned embodiments of the present application are merely examples for clearly illustrating the present application, and are not intended to limit the implementation methods of the present application, and are not intended to limit the present application. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. Each discrete structure / function module or unit can be integrated together to form an independent part, or each module can exist separately, or two or more modules can be integrated to form an independent part, and the structure and function of the discrete components can be implemented as a combined structure or component. It is not necessary and impossible to enumerate all the implementation methods here. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the claims of the present application.

Claims

1. A distributed edge computing network resource orchestration method, It is characterized in that include: All end nodes of the edge computing network are regarded as intelligent agents, the centralized resource joint orchestration model TAC is initialized, and the edge computing network environment and task flow are initialized; Centrally train the TAC model until the model converges; wherein the TAC model is constructed based on a reinforcement learning framework and a Transformer model of a single-Actor multi-Critic structure; Based on the transfer learning method, the original TAC model is converted into a distributed resource orchestration model DTAC, which is used to make local network resource orchestration decisions for users in a distributed manner at each user node.

2. A distributed edge computing network resource orchestration method according to claim 1, It is characterized in that The TAC model includes an Actor module, a Critic1 module and a Critic2 module, and uses a Transformer model as a shared model structure of the Actor module, the Critic1 module and the Critic2 module, and adds multiple different classification layers after the Transformer to obtain the outputs of the Actor module, the Critic1 module and the Critic2 module; The Actor module is used to input observations about the user's network and task information. Get resource orchestration action The Critic1 module is used to input S t , A t And output the Q value of the corresponding task time The Critic2 module is used to input S t , A t And output the corresponding energy consumption Q value 3. A distributed edge computing network resource orchestration method according to claim 3, It is characterized in that During the TAC model parameter update process, the Critic1 module and the Critic2 module use the mean square error loss function to perform batch gradient descent, and the loss function of the Actor module is to minimize Q t,c (S t ,A t ) and Q t,e (S t ,A t ) is used as the goal to optimize task time and energy consumption.

4. A distributed edge computing network resource orchestration method according to claim 4, It is characterized in that The loss function of the Critic1 module is: The loss function of the Critic2 module is: The loss function of the Actor module is:

5. A distributed edge computing network resource orchestration method according to claim 3, It is characterized in that The transfer learning method is used to convert the original TAC model into a distributed resource orchestration model DTAC, including: The attention mechanism in the TAC model that completes centralized training is modified: the expression of the attention mechanism in the original Transformer is: Where V,Q,K∈R N*h is the attention mechanism parameter, V represents the value mapping of different user input vectors, Q and K represent the queue and key mapping of different user input vectors, respectively, N represents the number of UEs, and h represents the size of the hidden layer; Modify the attention mechanism into the following expression: At this time, the Transformer model only cares about the state of each user node and ignores the state of other nodes. It only obtains the attention weight based on the state information of each user node. The goal is Using the TAC model after modifying the attention mechanism as the DTAC model; The local information of each user node is used as the input of the Actor module in the DTAC model to decide the distributed action of each user node, and the Q value of the distributed action is obtained through the Critic1 module and the Critic2 module in the DTAC model. The loss function of the Actor module in the DTAC model is calculated according to the Q value and the parameters are updated, so that the DTAC model gradually reduces the demand for other node information while keeping the action decision output similar.

6. A distributed edge computing network resource orchestration method according to claim 3, It is characterized in that A multi-dimensional state input form of "number of users × local state length" is adopted to obtain an action output of adaptive number of users of "number of users × action length" based on Transformer.

7. A distributed edge computing network resource orchestration method according to any one of claims 1 to 6, It is characterized in that The centralized training of the TAC model includes: In any time slot t: Let each user j receive their own task τ j And get the corresponding observation s j , the cloud server aggregates the observations of each user As the input of the TAC model, the corresponding resource orchestration action is obtained It is then sent back to each user, including the task offloading location, the selected transmission channel and transmission power, and the calculated frequency allocation. The corresponding optimal caching strategy is obtained by the heuristic algorithm. Each user j schedules action a according to resources j Get the corresponding task time and energy consumption And the corresponding rewards and The observation S t , choreography action A t And the corresponding reward R t,c ,R t,e Stored in the experience replay pool as training data; After every T time slots: reset the edge computing network environment and replay the training data in the pool based on the experience Centralized training is performed through the PPO algorithm, and model parameters are updated until the TAC model converges.

8. An electronic device, It is characterized in that include: A memory for storing computer executable instructions or computer programs; A processor, used to implement the method according to any one of claims 1 to 7 when executing computer executable instructions or computer programs stored in the memory.

9. A computer-readable storage medium, It is characterized in that The storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by a processor to implement the method as described in any one of claims 1-7.

10. A computer program product comprising a computer program or computer executable instructions, It is characterized in that When the computer program or computer executable instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images