A "cloud-edge-end" collaborative caching method for metaverse scenarios

Abstract

The present invention proposes a "cloud-edge-end" collaborative caching method for a metaverse scene, the method comprising: dividing the metaverse scene into several sub-scenes, and organizing and compressing the resources required for rendering each sub-scene to obtain a sub-scene compression package; using a fast dynamic time warping algorithm to calculate the similarity between different trajectory segments in each sub-scene; based on the similarity between different trajectory segments, using a multi-GPU distributed streaming computing system to match the segment with the smallest similarity in real time and obtain a corresponding number of recommended predicted sub-scene trajectory sequences; distributing a number of recommended predicted sub-scene trajectory sequences to a number of edge servers to calculate the popularity of the sub-scene, and using a cache replacement framework consisting of states, actions and rewards to adjust the edge server cache strategy. The present invention can realize intelligent cache allocation and adjustment according to the sub-scene popularity prediction, improve the cache hit rate, and reduce the delayed response of user requests.

Description

Technical Field

[0001] The present invention relates to the field of metaverse and edge caching, and in particular to a "cloud-edge-end" collaborative caching method for a metaverse scenario. Background Art

[0002] Metaverse, as a digital ecosystem that integrates cutting-edge technologies such as virtual reality, augmented reality, artificial intelligence and blockchain, has attracted widespread attention. Metaverse is not only a virtual online social platform, but also a multi-dimensional space that integrates the real and virtual worlds, where users can socialize, entertain, learn and work. With the advancement of technology and the popularization of the Internet, Metaverse is rapidly rising and becoming an important part of the future digital economy.

[0003] In the construction and operation of the Metaverse, data transmission and caching technologies play a vital role. Efficient data transmission and caching can significantly improve user experience, reduce latency and freezes, and ensure that users enjoy a smooth and realistic experience in the virtual environment. Although the requirements of efficient transmission and rendering have been widely studied in scenarios such as virtual reality, the Metaverse still faces unique challenges, especially in user behavior trajectory prediction and scene cache optimization, which has become a hot topic and difficulty in research.

[0004] However, with the increasing complexity of metaverse scenarios and the diversity of user behaviors, traditional methods have become unable to cope with large-scale, real-time, and personalized needs. Existing research has gradually introduced machine learning and deep learning technologies in cache strategy optimization, using user behavior data for prediction and decision-making, but most of these methods ignore the similarity and dynamics of user behavior trajectories and do not fully utilize trajectory similarity for cache optimization.

[0005] In addition, existing research also has deficiencies in real-time and distributed computing, and it is difficult to efficiently handle data transmission and caching issues in large-scale metaverse scenarios. So far, most metaverses use centralized data processing methods, and rendering work is completed on cloud servers. However, as the complexity of metaverse scenarios increases, this centralized approach faces significant bottlenecks, including latency, bandwidth, and computing resource limitations, making it difficult to meet large-scale, real-time, and personalized requirements. Summary of the invention

[0006] In view of the above situation, the main purpose of the present invention is to propose a "cloud-edge-end" collaborative caching method for a metaverse scenario to solve the above technical problems.

[0007] The present invention proposes a "cloud-edge-end" collaborative caching method for a metaverse scenario, the method comprising the following steps:

[0008] Step 1: Divide the metaverse scene in the horizontal and vertical directions to obtain a preset number of sub-scenes; wherein there is at least regional overlap between adjacent sub-scenes to ensure a smooth transition between the sub-scenes;

[0009] The resources required for rendering each sub-scene are sorted and compressed to obtain a sub-scene compressed package, which is then transmitted and cached through the edge server;

[0010] Step 2: Use the fast dynamic time warping algorithm to calculate the similarity between different trajectory segments in each sub-scene;

[0011] Step 3: Based on the similarity between different trajectory segments, the segment with the smallest similarity is matched in real time through a multi-GPU distributed streaming computing system, and the user's next behavior is predicted based on the similarity between different trajectory segments;

[0012] Step 4: According to the predicted next behavior of the user, obtain a corresponding sequence of recommended prediction sub-scenario trajectories;

[0013] Step 5: Distribute several recommended predicted sub-scenario trajectory sequences to several edge servers to calculate the sub-scenario popularity, and use a cache replacement framework consisting of state, action, and reward to adjust the edge server cache strategy to achieve cache management.

[0014] Compared with the prior art, the present invention has the following beneficial effects:

[0015] 1. The present invention can realize intelligent cache allocation and adjustment according to the sub-scenario popularity prediction, improve the cache hit rate, and reduce the delayed response of user requests;

[0016] 2. The present invention has high real-time performance and accuracy in similarity analysis of user trajectories, thereby effectively optimizing sub-scene resource allocation, significantly reducing transmission delays, and providing users with a smoother virtual experience;

[0017] 3. The present invention is based on a distributed collaborative caching strategy, which can maintain high performance under different cache capacity conditions. It is suitable for diverse and complex metaverse scenarios, and provides a solid foundation and technical support for the future expansion of the generative metaverse.

[0018] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description or learned through embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of a "cloud-edge-end" collaborative caching method for a metaverse scenario proposed by the present invention;

[0020] Figure 2A schematic diagram of an edge cache network framework of a "cloud-edge-end" collaborative caching method for a metaverse scenario proposed by the present invention;

[0021] Figure 3 A schematic diagram of a distributed stream computing framework for a "cloud-edge-end" collaborative caching method for a metaverse scenario proposed by the present invention;

[0022] Figure 4 A schematic diagram of a cache optimization solution for a "cloud-edge-end" collaborative caching method for a metaverse scenario proposed by the present invention;

[0023] Figure 5 A schematic diagram of the trajectory prediction execution process of the "cloud-edge-end" collaborative caching method for a metaverse scenario proposed by the present invention that integrates similar trajectory measurements;

[0024] Figure 6 Schematic diagram of the framework structure of the joint deep reinforcement learning and graph convolutional network cache replacement model for the "cloud-edge-end" collaborative caching method for a metaverse scenario proposed in the present invention. DETAILED DESCRIPTION

[0025] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0026] These and other aspects of the embodiments of the present invention will be apparent with reference to the following description and accompanying drawings. In these descriptions and accompanying drawings, some specific implementations of the embodiments of the present invention are specifically disclosed to represent some ways of implementing the principles of the embodiments of the present invention, but it should be understood that the scope of the embodiments of the present invention is not limited thereto.

[0027] See also Figure 1 , an embodiment of the present invention proposes a metaverse scenario "cloud-edge-end" collaborative caching method, the method comprising the following steps:

[0028] Step 1: Divide the metaverse scene in the horizontal and vertical directions to obtain a preset number of sub-scenes; wherein there is at least regional overlap between adjacent sub-scenes to ensure a smooth transition between the sub-scenes;

[0029] The resources required for rendering each sub-scene are sorted and compressed to obtain a sub-scene compression package, which is then transmitted and cached through the edge server.

[0030] Step 2: Use the fast dynamic time warping algorithm to calculate the similarity between different trajectory segments in each sub-scene;

[0031] In step 2, the fast dynamic time warping algorithm is used to calculate the similarity between different trajectory segments in each sub-scene. The corresponding relationship is:

[0032] ;

[0033] in, It is calculated by the fast dynamic time warping algorithm. and All represent trajectory segments, represents a custom distance function, and Both represent low-resolution sequences after downsampling. It is calculated by the dynamic time warping distance algorithm. Indicates The trajectory segment corresponding to the local window, Indicates The trajectory segment corresponding to the local window, Indicates that the distance function is calculated by a custom distance function. Indicates that by and The local window obtained by mapping the alignment path back to the original sequence;

[0034] In the process of calculating the similarity between different trajectory segments using the fast dynamic time warping algorithm, the expression of the custom distance function is:

[0035] ;

[0036] in, Represents the distance between two trajectory segments calculated by a custom function. and All represent trajectory segments, represents the weight of the coordinate distance, represents the weight of the heading distance, represents the weight of the dwell time distance, represents the coordinate distance, Indicates the distance towards Indicates the dwell time distance, and They all represent the coordinates of trajectory points. Represents the coordinates of the heading point, Indicates the coordinates of the stop time point.

[0037] Step 3: Based on the similarity between different trajectory segments, the segment with the smallest similarity is matched in real time through a multi-GPU distributed streaming computing system, and the user's next behavior is predicted based on the similarity between different trajectory segments;

[0038] Please refer to Figure 3 and Figure 5 In step 3, based on the similarity between different trajectory segments, the segment with the smallest similarity is matched in real time through the multi-GPU distributed streaming computing system, and the user's next behavior is predicted based on the similarity between different trajectory segments. The specific steps are as follows:

[0039] Based on the similarity between the historical trajectory segment and the sub-trajectory segment, find and determine the segment with the smallest similarity to the real-time trajectory segment to obtain the best matching segment;

[0040] Based on the best matching segment, obtain a sub-historical trajectory segment with a length of L after the historical trajectory position where the best matching segment is located;

[0041] When the length of the sub-history trajectory segment is less than L, all the best matching segments are obtained and used as the sequence of recommended predicted sub-scene trajectories;

[0042] Furthermore, the main advantage of the distributed streaming computing framework is that it uses multiple GPU servers to distribute data, significantly improving computing efficiency.

[0043] Step 4: According to the predicted next behavior of the user, obtain a corresponding sequence of recommended prediction sub-scenario trajectories;

[0044] Step 5: Distribute several recommended predicted sub-scenario trajectory sequences to several edge servers to calculate the sub-scenario popularity, and use a cache replacement framework consisting of state, action, and reward to adjust the edge server cache strategy to achieve cache management;

[0045] In step 5, several recommended predicted sub-scene trajectory sequences are distributed to several edge servers to calculate the sub-scene popularity. The corresponding process has the following relation:

[0046] ;

[0047] in, represents the popularity of the sub-scene, represents a sub-scene sequence, represents the sequence of recommended predicted sub-scene trajectories, Indicates The first A sequence of sub-scenes, Indicates processing by the indicator function, Represents the number of all predicted trajectory segments of the current edge server at the optimal time interval.

[0048] Please refer to Figure 4 and Figure 6 In step 5, a cache replacement framework consisting of state, action and reward is used to adjust the edge server cache strategy, where the state is obtained by combining the cache state and popularity. The corresponding process has the following relationship:

[0049] ;

[0050] in, Indicates the status, Indicates the cache status. Indicates popularity;

[0051] Further is through In time The sub-scenarios are sorted by the number of requests.

[0052] The reward calculation process specifically includes:

[0053] The real-time sub-scenario mixed storage hit rate is calculated through the content cached by the edge server:

[0054] ;

[0055] in, Indicates the real-time sub-scene cache hit rate, Indicates Request content, Indicates the content of the edge server cache. Indicates the number of requested contents;

[0056] Based on the change value of the sub-scene cache hit rate, the cache replacement penalty is calculated, and the corresponding process has the following relationship:

[0057] ;

[0058] in, represents the penalty coefficient, Indicates the change value of the sub-scene cache hit rate. Indicates the sub-scene cache hit rate at the previous moment, represents cache replacement penalty;

[0059] Based on the cache replacement penalty and the real-time sub-scenario cache hit rate, the reward is calculated. The corresponding relationship is:

[0060] ;

[0061] in, Indicates reward.

[0062] Further, action Deciding whether to perform a cache replacement operation is a binary decision. Specifically, the action Determines whether the edge server performs cache replacement. If This means that based on the current popularity, the first few word scenes are intercepted for cache replacement. , no replacement is performed at the current moment.

[0063] For further information, please refer to Figure 2 The present invention optimizes the processing and distribution of scene data to achieve efficient processing of complex scenes. The various levels of the framework work together to provide efficient data management and scene rendering support.

[0064] Furthermore, in step 5, GCN is combined with the Actor-Critic architecture to make full use of the advantages of GCN in processing graph structure data, enhance the expressiveness of the model, and optimize the cache replacement strategy through the Actor-Critic algorithm. GCN is used to extract the complex relationships and similarities between cache scenarios. Through the graph convolution layer, node features are aggregated layer by layer to generate higher-level feature representations, thereby capturing the global and local information of the cache scenario. In the Actor-Critic architecture, the Actor network generates a cache replacement strategy and decides whether to perform cache replacement through the strategy output layer; the Critic network evaluates the value of the current state and provides a state value estimate through the value output layer. The Actor network and the Critic network work together to improve the effect of the cache replacement strategy and reduce resource consumption through strategy optimization and value evaluation. By combining GCN with Actor-Critic, the model can better understand the relationship between cache scenarios, dynamically adjust the cache strategy, improve the cache hit rate, and reduce resource consumption of edge servers.

[0065] Specifically, the cloud layer: as the network hub, responsible for comprehensive management, providing functions such as application center, long-term storage, and real-time data synchronization through storage servers and synchronization servers;

[0066] Edge layer: The edge server collaborates with the neighboring GPU server to perform similarity calculations, recommends the optimal sub-scene sequence as a prediction, and calculates the sub-scene popularity based on the prediction data. The GCNAC scheme is used to determine the sub-scene resource compression packages that should be cached first to meet the real-time needs of users;

[0067] Terminal layer: In the request-intensive area, GPU servers are configured to be responsible for data collection, request encryption, and rendering sub-scenes, and form a distributed real-time trajectory recommendation system with other GPU servers under the edge server to predict the user's sub-scene access sequence. A dynamic resource allocation mechanism is adopted to dynamically adjust the number and computing power of terminal layer GPU servers according to actual needs. For users in non-intensive request areas, the edge server requests the cloud to obtain a compressed package for rendering on the client. The terminal device (head-mounted display or computer + head-mounted display) interacts directly with the user and is responsible for scene perception and scene display.

[0068] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0069] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0070] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A "cloud-edge-end" collaborative caching method for a metaverse scenario, characterized in that: The method comprises the following steps: Step 1: Divide the metaverse scene in the horizontal and vertical directions to obtain a preset number of sub-scenes; wherein there is at least regional overlap between adjacent sub-scenes to ensure a smooth transition between sub-scenes; The resources required for rendering each sub-scene are sorted and compressed to obtain a sub-scene compressed package, which is then transmitted and cached through the edge server; Step 2: Use the fast dynamic time warping algorithm to calculate the similarity between different trajectory segments in each sub-scene; Step 3: Based on the similarity between different trajectory segments, the segment with the smallest similarity is matched in real time through a multi-GPU distributed streaming computing system, and the user's next behavior is predicted based on the similarity between different trajectory segments; Step 4: According to the predicted next behavior of the user, obtain a corresponding sequence of recommended prediction sub-scenario trajectories; Step 5: Distribute several recommended predicted sub-scenario trajectory sequences to several edge servers to calculate the sub-scenario popularity, and use a cache replacement framework consisting of state, action, and reward to adjust the edge server cache strategy to achieve cache management; In step 5, several recommended predicted sub-scene trajectory sequences are distributed to several edge servers to calculate the sub-scene popularity. The corresponding process has the following relation: ； in, represents the popularity of the sub-scene, represents a sub-scene sequence, represents the sequence of recommended predicted sub-scene trajectories, Indicates The first A sequence of sub-scenes, Indicates processing by the indicator function, represents the number of all predicted trajectory segments of the current edge server at the optimal time interval; The cache replacement framework consisting of state, action and reward is used to adjust the edge server cache strategy, where the state is obtained by combining the cache state and popularity. The corresponding process has the following relationship: ； in, Indicates the status, Indicates the cache status. Indicates popularity.

2. According to claim 1, a metaverse scenario "cloud-edge-end" collaborative caching method is characterized in that: In step 2, the fast dynamic time warping algorithm is used to calculate the similarity between different trajectory segments in each sub-scene, and the relationship between the corresponding process is: ； in, It is calculated by the fast dynamic time warping algorithm. and All represent trajectory segments, represents a custom distance function, and Both represent low-resolution sequences after downsampling. It is calculated by the dynamic time warping distance algorithm. Indicates The trajectory segment corresponding to the local window, Indicates The trajectory segment corresponding to the local window, Indicates that the distance function is used for calculation. Indicates that by and The local window obtained after mapping the alignment path back to the original sequence.

3. According to claim 2, a metaverse scenario "cloud-edge-end" collaborative caching method is characterized in that: In the process of calculating the similarity between different trajectory segments using the fast dynamic time warping algorithm, the expression of the custom distance function is: ； in, Represents the distance between two trajectory segments calculated by a custom function. and All represent trajectory segments, represents the weight of the coordinate distance, represents the weight of the heading distance, represents the weight of the dwell time distance, represents the coordinate distance, Indicates the distance towards Indicates the dwell time distance, and They all represent the coordinates of trajectory points. Represents the coordinates of the heading point, Represents the coordinates of the stop time point.

4. According to claim 3, a metaverse scenario "cloud-edge-end" collaborative caching method is characterized in that: In step 3, based on the similarity between different trajectory segments, the segment with the smallest similarity is matched in real time through a multi-GPU distributed streaming computing system, and the user's next behavior is predicted based on the similarity between different trajectory segments. The specific steps are as follows: Based on the similarity between the historical trajectory segment and the sub-trajectory segment, find and determine the segment with the smallest similarity to the real-time trajectory segment to obtain the best matching segment; Based on the best matching segment, obtain a sub-historical trajectory segment with a length of L after the historical trajectory position where the best matching segment is located; When the length of the sub-history trajectory segment is less than L, all the best matching segments are obtained and used as the sequence of recommended predicted sub-scene trajectories.

5. According to claim 1, a metaverse scenario "cloud-edge-end" collaborative caching method is characterized in that: In step 5, a cache replacement framework consisting of state, action and reward is used to adjust the edge server cache strategy, wherein the reward calculation process specifically includes: The real-time sub-scenario mixed storage hit rate is calculated through the content cached by the edge server: ； in, Indicates the real-time sub-scene cache hit rate, Indicates Request content, Indicates the content of the edge server cache. Indicates the number of request contents.

6. According to claim 5, a metaverse scenario "cloud-edge-end" collaborative caching method is characterized in that: In the step of adjusting the edge server cache strategy by using the cache replacement framework consisting of state, action and reward, the reward calculation process also includes: Based on the change value of the sub-scene cache hit rate, the cache replacement penalty is calculated, and the corresponding process has the following relationship: ； in, represents the penalty coefficient, Indicates the change value of the sub-scene cache hit rate. Indicates the sub-scene cache hit rate at the previous moment, represents cache replacement penalty; Based on the cache replacement penalty and the real-time sub-scenario cache hit rate, the reward is calculated. The corresponding process has the following relationship: ； in, Indicates reward.

Images