Proactive cache routing method and system for satellite content distribution networks

Abstract

The application relates to an active cache routing method and system of a satellite content distribution network, the satellite content distribution network comprising at least one satellite node. The method comprises: extracting first semantic information features of distributed content and second semantic information features of historical popular content by using a deep learning model; calculating semantic similarity between the first semantic information features and the second semantic information features; calculating a satellite semantic cache priority value of a satellite node corresponding to the distributed content according to the semantic similarity; and planning an active cache routing path of the distributed content according to the satellite semantic cache priority value of each satellite node. The application can enable the satellite to automatically make a cache decision of the distributed content in the content distribution process, and improve the efficiency of active cache by combining content distribution and cache deployment.

Description

Technical Field

[0001] This application mainly relates to the field of satellite communication and content delivery network technology, and specifically to an active caching routing method and system for satellite content delivery networks. Background Technology

[0002] With the rapid development of low Earth orbit satellite constellations, Satellite Content Delivery Networks (SCDNs) have become a crucial infrastructure for achieving highly available global content services. Traditional terrestrial content delivery networks rely on edge servers to cache frequently accessed content to reduce user access latency. However, in space-based scenarios, satellite nodes exhibit high dynamism, limited cache capacity, and limited inter-satellite link bandwidth, making the caching and scheduling mechanisms of traditional terrestrial content delivery networks unsuitable for direct application.

[0003] For space-based scenarios, proactive caching can be used, which involves predicting content popularity based on historical access records and pre-caching it before a user request occurs. However, in satellite content delivery networks, the high-speed movement of low-Earth orbit satellites causes frequent changes in coverage areas, and the user needs served by the satellites are also constantly changing. This necessitates continuous redeployment of the cache. Furthermore, the proactive cache update process consumes a large amount of inter-satellite link resources, resulting in excessively high overall system traffic load and reduced transmission efficiency.

[0004] Some proactive caching methods cache content data packets through inter-satellite collaboration or periodic updates. However, this approach cannot automatically make caching decisions during content distribution, resulting in low proactive caching efficiency. Summary of the Invention

[0005] The technical problem to be solved by this application is to provide an active caching routing method and system for satellite content delivery networks, which enables satellites to automatically make caching decisions for distributed content during the content distribution process. By combining content distribution with caching deployment, the efficiency of active caching is improved.

[0006] The technical solution adopted in this application to solve the above-mentioned technical problems is an active caching routing method for a satellite content delivery network. The satellite content delivery network includes at least one satellite node. The method includes: extracting first semantic information features of the distributed content and second semantic information features of historically popular content using a deep learning model; calculating the semantic similarity between the first semantic information features and the second semantic information features; calculating the satellite semantic caching priority value of the satellite node corresponding to the distributed content based on the semantic similarity; and planning an active caching routing path for the distributed content based on the satellite semantic caching priority value of each satellite node.

[0007] In one embodiment of this application, calculating the satellite semantic cache priority value of the satellite node corresponding to the distributed content based on the semantic similarity includes: calculating the regional semantic content popularity of the distributed content in the preset region based on the semantic similarity and the request frequency of the historical popular content in the preset region; and calculating the satellite semantic cache priority value based on the regional semantic content popularity.

[0008] In one embodiment of this application, the regional semantic content popularity is calculated using the following formula:

[0009]

[0010] in, This indicates the popularity of semantic content in the region. This refers to the preset area. This refers to the distributed content; This indicates the historical popular content. Indicates the semantic similarity; This indicates the number of historically popular content items; This indicates the frequency of the request.

[0011] In one embodiment of this application, the satellite semantic cache priority value is calculated using the following formula:

[0012]

[0013] in, Indicates the distributed content corresponding satellite nodes The satellite semantic cache priority value; Indicates satellite node The number of future service areas; Indicates the satellite node Future service area The corresponding regional semantic content popularity, superscript Indicates the satellite node The corresponding number A future service area.

[0014] In one embodiment of this application, planning an active caching route for the distributed content based on the satellite semantic cache priority value of each satellite node includes: calculating the path caching benefit of the caching path based on the satellite semantic cache priority value; calculating the caching path benefit penalty value based on the length of the caching path and a preset path length; calculating the final path benefit based on the path caching benefit and the caching path benefit penalty value; and using the caching path corresponding to the largest final path benefit as the active caching route.

[0015] In one embodiment of this application, the path caching benefit is calculated using the following formula:

[0016]

[0017] in, This indicates the caching revenue of the stated path. Indicates the cache path The source satellite node in Indicates the cache path The target satellite node; Indicates the cache path Corresponding relay satellite node set Satellite nodes in; Indicates the distributed content corresponding satellite nodes The satellite semantic cache priority value.

[0018] In one embodiment of this application, the final path benefit is calculated using the following formula:

[0019]

[0020] in, This represents the final path benefit; This indicates the preset maximum path revenue value; This indicates the preset maximum number of additional route hops; This represents the penalty value for the cached path's benefit. This indicates the length of the cache path. This indicates the preset path length; Indicates the first weight. Indicates the second weight. , , .

[0021] In one embodiment of this application, the distributed content is a distributed content data packet constructed according to a preset routing protocol; after the step of planning the active caching routing path of the distributed content according to the satellite semantic caching priority value of each satellite node, the method further includes: programming a programmable routing header of the distributed content data packet according to the active caching routing path, wherein the programmable routing header is used to enable each satellite node to actively cache the distributed content data packet during the content distribution process.

[0022] In one embodiment of this application, programming the programmable routing header of the distributed content data packet according to the active caching routing path includes: in response to the satellite semantic cache priority value corresponding to the active caching routing path being greater than a preset priority threshold, marking the satellite node corresponding to the satellite semantic cache priority value as an intermediate cache satellite node, wherein the intermediate cache satellite node is used to instruct the satellite node to copy and cache the distributed content data packet before forwarding it to the next-hop satellite node when receiving the distributed content data packet; or in response to the satellite semantic cache priority value corresponding to the active caching routing path being less than the preset priority threshold, marking the satellite node corresponding to the satellite semantic cache priority value as an intermediate bypass satellite node, wherein the intermediate bypass satellite node is used to instruct the satellite node to directly forward the distributed content data packet to the next-hop satellite node when receiving the distributed content data packet.

[0023] To address the aforementioned technical problems, this application also proposes an active caching routing system for a satellite content delivery network, suitable for executing the active caching routing method described above. The active caching routing system includes: a semantic information extraction and feature encoding module configured to extract first semantic information features of the distributed content and second semantic information features of historically popular content using a deep learning model; a semantic similarity calculation module configured to calculate the semantic similarity between the first semantic information features and the second semantic information features; a regional semantic content popularity calculation and satellite semantic cache priority value prediction module configured to calculate the satellite semantic cache priority value of the satellite node corresponding to the distributed content based on the semantic similarity; an active caching routing mechanism module configured to plan the active caching routing path of the distributed content based on the satellite semantic cache priority value of each satellite node; and a programmable routing protocol module configured to program the programmable routing header of the distributed content data packets based on the active caching routing path.

[0024] The technical solution of this application extracts semantic information from the content and combines it with historical popular content data for content similarity analysis and regional demand prediction. This intelligently predicts the satellite semantic cache priority value for each satellite node, improving cache hit rate. By dynamically planning the optimal proactive cache routing path, it reduces latency and load on the satellite system. This application essentially constructs a semantically driven proactive cache routing mechanism for satellite content distribution networks, enabling satellites to automatically make caching decisions during content distribution. This achieves coordinated optimization of caching and routing, improving the efficiency of content distribution caching, reducing transmission latency, and optimizing network resource utilization. Attached Figure Description

[0025] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings, wherein:

[0026] Figure 1 This is a flowchart illustrating the collaboration between modules of an active caching routing system for a satellite content delivery network according to an embodiment of this application.

[0027] Figure 2 This is a flowchart of an active caching routing method for a satellite content delivery network according to an embodiment of this application;

[0028] Figure 3 This is a flowchart illustrating how a satellite node actively caches and distributes content data packets based on a programmable routing header during content distribution, according to one embodiment of this application.

[0029] Figure 4 This is a system block diagram of an active caching routing system for a satellite content delivery network according to another embodiment of this application. Detailed Implementation

[0030] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0031] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein, and therefore this application is not limited to the specific embodiments disclosed below.

[0032] As illustrated in this application, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0033] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously. Furthermore, other operations may be added to these processes, or one or more steps may be removed from these processes.

[0034] This paper first introduces the active caching routing system for the satellite content delivery network of this application to facilitate understanding of the active caching routing method described later. In practical applications, the satellite content delivery network includes at least one satellite node.

[0035] Figure 1 This is a flowchart illustrating the collaboration between modules of an active caching routing system for a satellite content delivery network according to an embodiment of this application. (Reference) Figure 1 As shown, this application proposes an active caching routing system 100 for a satellite content delivery network, suitable for executing the active caching routing method described below. The active caching routing system 100 includes: a semantic information extraction and feature encoding module 101, a semantic similarity calculation module 102, a regional semantic content popularity calculation and satellite semantic cache priority value prediction module 103, an active caching routing mechanism module 104, and a programmable routing protocol module 105. This application constructs an efficient and low-load space-based communication system through the modules in the active caching routing system 100, enabling active caching deployment of content during content distribution.

[0036] For example, the aforementioned modules in the active caching routing system 100 of the satellite content delivery network of this application can be implemented by a computer hardware device consisting of a processor and a memory. Specifically, the aforementioned modules are stored as program units in the memory, and the processor executes the program units stored in the memory to achieve thread control. This active caching routing system 100 can be set in the onboard computer of each satellite node.

[0037] Continue to refer to Figure 1 As shown, the semantic information extraction and feature encoding module 101 is configured to: extract first semantic information features of the distributed content and second semantic information features of historically popular content using a deep learning model. For example, this application performs semantic analysis on content (such as multimodal data like text, images, audio, and video), extracts multimodal semantic features based on a deep learning model, and generates a unified semantic vector representation. The semantic information extraction and feature encoding module 101 aims to mine deep feature representations that reflect the inherent semantics of the content from the original content. The semantic vectors generated in this module are used to support semantic similarity calculation and description generation.

[0038] The semantic similarity calculation module 102 is configured to calculate the semantic similarity between the first semantic information feature and the second semantic information feature. For example, the semantic similarity calculation module 102 combines the content semantic vector representation output by the semantic information extraction and feature encoding module 101 to calculate the semantic similarity between different content semantic vector representations, thereby representing the degree of similarity between content. These semantic vectors are formed in a unified high-dimensional semantic space, reflecting the semantic core and potential related features of each content. This semantic similarity measurement provides a foundation for subsequent content regional popularity prediction and satellite cache priority prediction. Through semantic-level measurement, the satellite system can effectively identify semantically similar but differently expressed content, thereby achieving a more intelligent and semantically driven network resource management strategy.

[0039] The regional semantic content popularity calculation and satellite semantic cache priority value prediction module 103 is configured to calculate the satellite semantic cache priority value of the satellite node corresponding to the distributed content based on semantic similarity. For example, the regional semantic content popularity calculation and satellite semantic cache priority value prediction module 103 can be divided into a regional semantic popularity calculation submodule and a satellite semantic cache priority value prediction submodule.

[0040] For the regional semantic popularity calculation submodule, it predicts the potential popularity of content in the target region based on the historical requests and semantic feature similarity of users in different regions, enabling the dynamic generation of semantic content heatmaps. By introducing semantic similarity weighting in the regional content popularity prediction process, even if content is not frequently accessed within the service area, its potential "semantic neighborhood" content (i.e., semantically similar popular content) can positively influence its predicted popularity. This design effectively alleviates the problem of inaccurate prediction of "cold start content" using traditional statistical frequency-based methods, achieving more granular and generalizable content popularity estimation.

[0041] For the satellite semantic cache priority prediction submodule, it combines satellite orbit information, visibility window, and future demand in the service area to calculate the cache priority of content on different satellite nodes, providing a basis for proactive caching decisions. By introducing a dual weighting mechanism of semantic similarity and request frequency in regional popularity prediction, and performing multi-region fusion in satellite-level cache decision-making, it achieves semantic-driven fine-grained content caching optimization. This can capture the semantic migration characteristics of user interests, alleviate the content cold start problem, and improve satellite cache hit rate and link utilization efficiency.

[0042] The proactive caching routing mechanism module 104 is configured to plan proactive caching routing paths for content distribution based on the satellite semantic caching priority value of each satellite node. For example, to implement proactive caching routing, both the hop count and the benefit within the proactive caching routing path need to be considered simultaneously. To maximize the benefit of the proactive caching routing path, as many high-demand satellite nodes as possible should be selected from the available routes; simultaneously, to avoid excessive latency in a single content distribution, the hop count of the proactive caching routing path needs to be guaranteed as much as possible. To address both benefit issues simultaneously, this application transforms the planning of the proactive caching routing path into a multi-objective optimization problem and employs a deep reinforcement learning algorithm to solve it.

[0043] The programmable routing protocol module 105 is configured to programmatically distribute content data packets using a programmable routing header based on an active caching routing path. This programmable routing header enables each satellite node to actively cache and distribute content data packets during content distribution. For example, the programmable routing protocol of this application employs IPv6-based Segment Routing over IPv6 (SRv6), programming the packet header through an optimal active caching routing path to achieve active caching routing functionality. The SRv6 routing protocol has a programmable mechanism; the Function field provided in the Segment Routing Header (SRH) of the SRv6 header provides programmable instruction functionality. This application specifies different Function fields in the SRH header, allowing each Function field to include a caching instruction field to instruct satellite nodes to perform cache replication or direct forwarding operations, thereby integrating content distribution and caching deployment within the content distribution path.

[0044] The active caching routing system 100 of the satellite content delivery network of this application implements a semantic-driven intelligent content distribution and caching routing mechanism in space-based satellite communication networks. Compared with traditional methods, this application introduces for the first time a content similarity prediction method based on semantic features. Simultaneously, based on programmable technology using programmable routing protocols (such as SRv6), it achieves active caching deployment of multiple satellite nodes in a single content distribution process, thereby realizing active caching routing. This application integrates content distribution and caching deployment at the system level, maintaining low request latency and high cache hit rate while reducing satellite network traffic load, improving the overall performance and resource utilization of the space-based communication system, and possessing high intelligence and scalability.

[0045] The active caching routing method for satellite content delivery networks described in this application will be introduced later.

[0046] This application proposes an active caching routing method for satellite content delivery networks, applicable to scenarios where active caching deployment of each satellite node is completed during the content distribution process in low-Earth orbit satellite content delivery networks. The active caching routing method of this application can run within the satellite, for example, within the controller of the satellite's onboard computer.

[0047] Figure 2 This is a flowchart of an active caching routing method for a satellite content delivery network according to an embodiment of this application. (Reference) Figure 2 As shown, the active caching routing method of the satellite content delivery network in this embodiment includes the following steps:

[0048] Step S110: Use a deep learning model to extract the first semantic information features of the distributed content and the second semantic information features of historically popular content.

[0049] Step S120: Calculate the semantic similarity between the first semantic information feature and the second semantic information feature.

[0050] Step S130: Calculate the satellite semantic cache priority value of the satellite node corresponding to the distributed content based on semantic similarity.

[0051] Step S140: Plan the active caching route for the distributed content based on the satellite semantic cache priority value of each satellite node.

[0052] For example, in some embodiments, the distributed content is a distributed content packet constructed according to a preset routing protocol (such as the SRv6 routing protocol). Step S140 is followed by step S150.

[0053] Step S150: Program the programmable routing header for distributing content data packets according to the actively cached routing path. The programmable routing header is used to enable each satellite node to actively cache the content data packets during content distribution. This application realizes intelligent caching operations and transparent data transmission of intermediate satellite nodes through the network programming mechanism supported by the SRv6 routing protocol.

[0054] The following details steps S110 to S150 described above:

[0055] In step S110, a deep learning model is used to extract the first semantic information features of the distributed content and the second semantic information features of historically popular content.

[0056] For example, regarding content This application utilizes a deep learning model that integrates multiple algorithms. Deep semantic feature extraction is performed on it, and a unified semantic vector representation is generated. Specifically, textual content can be extracted using pre-trained language models, such as Bidirectional Encoder Representations from Transformers (BERT), to capture semantic relationships and potential topic information between sentences. Image content can be extracted using visual encoders, such as Vision Transformer (ViT) or Residual Network with 50 layers (ResNet-50). For multimodal content containing both text and images, joint encoding can be performed using cross-modal fusion models, such as Bootstrapping Language-Image Pre-training (BLIP) or Object-Semantics Aligned Pre-training (OSCAR), enabling the model to semantically understand the correspondence between image regions and text descriptions.

[0057] This application uses the following formula (1) to construct a unified semantic vector representation. .

[0058] (1)

[0059] As the unique representation of the content in the semantic space, it is used for subsequent tasks such as similarity calculation, content retrieval, and cache priority prediction. For example, the above formula (1) can be used to obtain the semantic vector of the first semantic information feature of the distributed content and the second semantic information feature of the historical popular content.

[0060] In step S120, the semantic similarity between the first semantic information feature and the second semantic information feature is calculated.

[0061] For example, for two pieces of content (such as a certain distribution content) With a certain popular historical content This application extracts its deep semantic features and generates semantic vectors of the first semantic information features. semantic vectors with second semantic information features Then, using the following formula (2) based on the semantic similarity calculation model, the similarity distance between the semantic vectors of the two contents is calculated. , used to characterize and The degree of semantic similarity between them.

[0062] (2)

[0063] in, This represents a function for calculating semantic similarity distance. Through similarity calculation, a distance between [aspects] and [aspects] can be obtained. Similarity score between The closer the value is to 1, the more similar the semantics; the closer the value is to 0, the greater the semantic difference. Therefore, we can use... This is used to characterize the semantic similarity between two pieces of content.

[0064] In step S130, the satellite semantic cache priority value of the satellite node corresponding to the distributed content is calculated based on semantic similarity. In some embodiments, step S130 includes:

[0065] Step S1301: Calculate the regional semantic content popularity of the distributed content in the preset area based on semantic similarity and the request frequency of historically popular content in the preset area (such as the satellite service area determined by the predictability of satellite orbits).

[0066] Step S1302: Calculate the satellite semantic cache priority value based on the regional semantic content popularity.

[0067] For example, this application can accurately predict the content demand in the future service area of ​​the satellite through step S130, thereby improving the hit rate of popular content.

[0068] In practical applications, after obtaining the semantic similarity calculation results, this application incorporates them into region-level semantic content popularity prediction. A specific region is defined. (i.e., the preset area), the top [number] users in its historical request data. The set of frequently requested content is as follows The request frequency for each piece of content is These frequencies reflect the region The actual access popularity of each piece of content by users within a certain time window.

[0069] For any content to be predicted (i.e., content distribution), which is located in the region The popularity of regional semantic content can be predicted using the following formula (3). Therefore, in some embodiments, the popularity of regional semantic content in step S1301 is calculated using the following formula (3):

[0070] (3)

[0071] in, Indicates the popularity of semantic content in a region. Indicates the preset area. Indicates the content to be distributed; This indicates popular historical content. Indicates the content to be distributed Popular historical content Semantic similarity between them; This indicates the number of historically popular content items; Indicates the request frequency. Reflects the region For content The actual request situation.

[0072] For example, this application combines semantic similarity with historical request frequency through formula (3), which can quantify the popularity index of regional semantic content and accurately reflect the real demand distribution of regional users.

[0073] Furthermore, this application considers predicting the distribution content. In satellite Satellite semantic cache priority value Assuming a satellite Future service area collection , This is equivalent to determining the satellite service area based on the predictability of satellite orbits. Each satellite Future service area Corresponding to a set of regional semantic content popularity In order to integrate satellites The distribution of user interests within the future service area can be predicted using the following formula (4) to forecast the content to be distributed. In satellite The satellite semantic cache priority value. Therefore, in some embodiments, the satellite semantic cache priority value in step S1302 is calculated using the following formula (4):

[0074] (4)

[0075] in, Indicates the content to be distributed corresponding satellite nodes Satellite semantic cache priority value; Indicates satellite node The number of future service areas will depend on the fact that satellites are constantly orbiting the Earth, therefore the number of satellite nodes... The number of service areas in the future can be 0, 1, or more; Indicates satellite node Future service area The corresponding regional semantic content popularity, superscript Indicates satellite node The corresponding number A future service area.

[0076] For example, the satellite semantic cache priority value reflects the distributed content. In satellite The average potential popularity of the content within the covered area. The higher the value, the more likely the content is to be requested by users within the satellite service range, and therefore it should receive a higher storage priority in the allocation of satellite cache resources. This application uses formula (4) to calculate the average content popularity of the future service area of ​​the satellite node to generate an index of satellite semantic cache priority, which can take into account the coverage characteristics of satellite movement in a balanced way.

[0077] In step S140, an active caching route for the distributed content is planned based on the satellite semantic cache priority value of each satellite node. In some embodiments, step S140 includes:

[0078] Step S1401: Calculate the path caching benefit of the cache path based on the satellite semantic cache priority value.

[0079] Step S1402: Calculate the cache path benefit penalty value based on the length of the cache path and the preset path length.

[0080] Step S1403: Calculate the final path revenue based on the path caching revenue and the cached path revenue penalty value.

[0081] Step S1404: Select the cached path corresponding to the largest final path benefit as the active cached route path.

[0082] For example, steps S1401 to S1404 of this application, through the concept of multi-objective optimization, not only prioritize the maximization of caching benefits for high-priority content, but also suppress long-distance transmission losses through path length penalties, thus planning the optimal active caching route.

[0083] This application designs an active caching routing mechanism. Specifically, in a satellite content delivery network, let the source satellite node be... The target satellite node is The content to be distributed is Planned active caching path The following two types of objectives must be met simultaneously:

[0084] (1) Maximize caching benefits: Store the content along the relay satellite nodes on the caching path. The overall benefits that follow should be as large as possible.

[0085] (2) Minimize hop count overhead: The cache path length (hop count) should not be too long to avoid excessive consumption of transmission resources and increased latency.

[0086] This application enables active caching of paths. The set of relay nodes is The path caching benefit can be defined using the following formula (5). Therefore, in some embodiments, the path caching benefit is calculated using the following formula (5):

[0087] (5)

[0088] in, Indicates the benefits of path caching. Indicates the cache path The source satellite node in Indicates the cache path The target satellite node; Indicates the cache path Corresponding relay satellite node set Satellite nodes in; Indicates the content to be distributed corresponding satellite nodes The satellite semantic cache priority value, i.e. It also indicates the distribution of content. In the relay node set satellite nodes in The semantic cache priority value of the satellites on the cache path. For example, this application can quantify the overall value of the path by accumulating the semantic cache priority of all satellite nodes on the cache path using formula (5).

[0089] This application specifies that the cache path length is [length missing]. To control the number of hops added relative to a baseline path (e.g., the shortest path), a preset path length is introduced. (This can be set to the shortest path length or the desired maximum length), and the quantity to be minimized is constructed as follows: Therefore, this application will optimize the active caching path. The planning is defined as a multi-objective optimization problem, as shown in the objective equations (6) and (7) below:

[0090] (6)

[0091] (7)

[0092] in, Indicates source satellite node To the target satellite node The application considers the path set of all candidate satellite inter-satellite links. To solve this multi-objective optimization problem, this application combines the benefits and penalties into a single objective using a weighted approach, where a larger objective value is better.

[0093] In some embodiments, the final path gain in step S1403 is calculated using the following formula (8):

[0094] (8)

[0095] in, Indicates the final path benefit; This represents the preset maximum path revenue value. It can be set as the theoretical maximum path benefit value; This indicates the preset maximum number of additional route hops. It can be set to the theoretical maximum number of additional route hops; This represents the benefit / penalty value for cached paths. Indicates the length of the cache path. Indicates the preset path length. It can be set to the theoretical shortest path length; Indicates the first weight. Indicates the second weight. , , , and This is used to control the relative importance of revenue and hop count penalty. This application, through formula (8), can simultaneously consider cache value and transmission efficiency in the evaluation of path revenue. The goal of this application's proactive cached routing is to select an optimal path. make maximum.

[0096] Furthermore, to address path search in large-scale topologies (characterized by a large action space and high real-time requirements), this application employs a deep reinforcement learning algorithm to search for near-optimal paths within the satellite topology embedding space and provide cache or forwarding instructions. For example, this application constructs a state space, action space, and reward function.

[0097] Specifically, the state space consists of different satellite node types (source satellite, destination satellite, buffer satellite, bypass satellite) and their corresponding... The matrix is ​​composed of the action space, which is the set of optional inter-satellite links (ISLs) for each satellite node. At this point, the action is to choose the next jump point (i.e., the next jump). This is equivalent to selecting which ISL to use from the current node, and the optional caching decision action (such as caching or not caching).

[0098] Reward function: The reward is designed to allow the agent to simultaneously pursue high cache gains and low hop count. When the search reaches the target satellite node... At that time, a finish line reward formula (9) is obtained.

[0099] (9)

[0100] This application trains the estimation network and target network in a deep reinforcement learning agent, enabling the agent to eventually learn the optimal active caching path.

[0101] In step S150, the programmable routing header (such as the SRH header of the SRv6 programmable routing protocol) of the content data packets is programmed and distributed according to the active caching routing path. For example, this application plans the optimal active caching routing path based on the active caching routing mechanism. ,for The set of all satellite nodes passed through Satellite nodes Based on the predicted satellite semantic cache priority value A priority threshold can be set. And based on the satellite semantic cache priority value With priority threshold The comparison results are used to determine whether satellite nodes need to cache distributed content data packets.

[0102] In some embodiments, step S150 is based on the satellite semantic cache priority value. With priority threshold The comparison results will determine whether to proceed to step S150A or step S150B.

[0103] Step S150A: In response to the satellite semantic cache priority value corresponding to the active cache route path Greater than (or greater than or equal to) the preset priority threshold The satellite nodes corresponding to the satellite semantic cache priority values. Marked as an intermediate cache satellite node, the intermediate cache satellite node is used to instruct the satellite node to copy and cache the distribution content data packet when receiving the distribution content data packet before forwarding it to the next hop satellite node.

[0104] Step S150B: In response to the satellite semantic cache priority value corresponding to the active cache route path Less than (or less than or equal to) the preset priority threshold The satellite nodes corresponding to the satellite semantic cache priority values. Marked as an intermediate bypass satellite node, an intermediate bypass satellite node is used to instruct a satellite node to directly forward content data packets to the next-hop satellite node.

[0105] For example, this application utilizes the SRv6 encapsulation mechanism to optimize the active caching route path. The intermediate buffer satellite nodes and intermediate bypass satellite nodes are identified. When programming the SRH header of SRv6, the Function section of the corresponding SRH header is programmed according to the identifier of the corresponding satellite node. For example, for intermediate buffer satellite nodes, the Function section is set to 1; for intermediate bypass satellite nodes, the Function section is set to 1. The purpose of Function is to implement content caching at intermediate satellite nodes while distributing content. Function indicates whether a satellite node should perform caching operations. It can be specified that when Function=1, the satellite node will copy and cache the received data packet before forwarding the original data packet to the next-hop satellite node; when Function=0, no processing is required, and the original data packet is directly forwarded to the next-hop satellite node. This application achieves integrated content distribution and caching deployment through the programmable network routing function provided by the SRv6 routing protocol.

[0106] For example, this application enables satellite nodes to synchronously complete intelligent caching deployment during the forwarding of data packets by directly embedding the decision results of active caching into the programmable routing header of the distributed content data packets.

[0107] The following specific example illustrates the process by which satellite nodes actively cache and distribute content data packets based on programmable routing headers during content distribution.

[0108] Figure 3 This is a flowchart illustrating how a satellite node actively caches and distributes content data packets based on a programmable routing header during content distribution, according to one embodiment of this application. For example, see [link to relevant documentation]. Figure 3As shown, in step S310, the source satellite node sends the distribution content data packet; in step S320, the SRv6 packet header of the distribution content data packet is programmed according to the optimal active caching route path; in step S330, the distribution content data packet is transmitted to the intermediate satellite node on the active caching route path; in step S340, the intermediate satellite node detects the Function part of the packet header of the distribution content data packet. If the Function part of the packet header is 1, then step S350 is executed first to generate a copy of the distribution content data packet and cache it, and then step S360 is executed to forward the distribution content data packet to the target satellite node; if the Function part of the packet header is 0, then step S360 is executed directly to forward the distribution content data packet to the target satellite node.

[0109] This application is equivalent to constructing an active caching and routing mechanism for a semantically driven satellite content delivery network, enabling satellites to automatically make caching decisions for distributed content during the content distribution process. This achieves coordinated optimization of caching and routing, improves the efficiency of content distribution caching, reduces transmission latency, and optimizes network resource utilization.

[0110] This application also includes an active caching routing system for a satellite content delivery network, comprising a memory and a processor. The memory stores instructions executable by the processor; the processor executes these instructions to implement the active caching routing method for the satellite content delivery network described above.

[0111] Figure 4 This is a system block diagram of an active caching routing system for a satellite content delivery network according to another embodiment of this application. (Reference) Figure 4As shown, the active caching routing system 400 of the satellite content delivery network may include an internal communication bus 401, a processor 402, a read-only memory (ROM) 403, a random access memory (RAM) 404, and a communication port 405. The active caching routing system 400 may also include a hard disk 406. The internal communication bus 401 enables data communication between the components of the active caching routing system 400. The processor 402 can perform judgments and issue prompts. In some embodiments, the processor 402 may consist of one or more processors. The communication port 405 enables data communication between the active caching routing system 400 and external systems. In some embodiments, the active caching routing system 400 can send and receive information and data from the network through the communication port 405. The active caching routing system 400 of the satellite content delivery network may also include different types of program storage units and data storage units, such as hard disk 406, read-only memory (ROM) 403, and random access memory (RAM) 404, capable of storing various data files used for computer processing and / or communication, as well as possible program instructions executed by processor 402. The processor executes these instructions to implement the main part of the method. The results of processor processing are transmitted to ground user equipment via communication ports and displayed on the ground user interface.

[0112] The above-described active caching routing method for satellite content delivery networks can be implemented as a computer program, stored in hard disk 406, and loaded into processor 402 for execution to implement the active caching routing method for satellite content delivery networks of this application.

[0113] This application also includes a computer-readable medium storing computer program code that, when executed by a processor, implements the active caching routing method for the satellite content delivery network described above.

[0114] When the active caching routing method of a satellite content delivery network is implemented as a computer program, it can also be stored as an article of art in a computer-readable storage medium. For example, computer-readable storage media can include, but are not limited to, magnetic storage devices (e.g., hard disks, floppy disks, magnetic stripes), optical discs (e.g., compact discs (CDs), digital multifunction discs (DVDs)), smart cards, and flash memory devices (e.g., electrically erasable programmable read-only memory (EPROM), cards, sticks, key drives). Furthermore, the various storage media described herein can represent one or more devices and / or other machine-readable media used for storing information. The term "machine-readable medium" can include, but is not limited to, wireless channels and various other media (and / or storage media) capable of storing, containing, and / or carrying code and / or instructions and / or data.

[0115] It should be understood that the embodiments described above are merely illustrative. The embodiments described herein may be implemented in hardware, software, firmware, middleware, microcode, or any combination thereof. For hardware implementation, the processor may be implemented within one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, and / or other electronic units designed to perform the functions described herein, or combinations thereof.

[0116] Some aspects of this application can be executed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The aforementioned hardware or software may be referred to as a "data block," "module," "engine," "unit," "component," or "system." The processor may be one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DAPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, or combinations thereof. Furthermore, aspects of this application may manifest as computer products residing in one or more computer-readable media, including computer-readable program code. For example, computer-readable media may include, but are not limited to, magnetic storage devices (e.g., hard disks, floppy disks, magnetic tapes, etc.), optical discs (e.g., compressed CDs, digital multifunction DVDs, etc.), smart cards, and flash memory devices (e.g., cards, sticks, key drives, etc.).

[0117] A computer-readable medium may contain a propagated data signal containing computer program code, for example, on baseband or as part of a carrier wave. This propagated signal may take various forms, including electromagnetic, optical, and so on, or suitable combinations thereof. A computer-readable medium can be any computer-readable medium other than a computer-readable storage medium, which can be connected to an instruction execution system, apparatus, or device to enable communication, propagation, or transmission of a program for use. The program code located on the computer-readable medium can be propagated through any suitable medium, including radio, cable, fiber optic cable, radio frequency signals, or similar media, or any combination of the above media.

[0118] The basic concepts have been described above. Obviously, for those skilled in the art, the above disclosure is merely illustrative and does not constitute a limitation of this application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this application. Such modifications, improvements, and corrections are suggested in this application, and therefore remain within the spirit and scope of the exemplary embodiments of this application.

[0119] Furthermore, this application uses specific terms to describe embodiments of the application. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic related to at least one embodiment of the application. Therefore, it should be emphasized and noted that "an embodiment," "one embodiment," or "an alternative embodiment" mentioned twice or more in different locations in this specification do not necessarily refer to the same embodiment. In addition, certain features, structures, or characteristics in one or more embodiments of the application can be appropriately combined.

[0120] In some embodiments, numbers describing the quantity of components and attributes are used. It should be understood that such numbers used to describe embodiments are sometimes modified by the terms "approximately," "approximately," or "generally." Unless otherwise stated, "approximately," "approximately," or "generally" indicates that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in this application are approximate values, which may be changed depending on the characteristics required by individual embodiments. In some embodiments, numerical parameters should take into account specified significant digits and employ a general method of digit reservation. Although the numerical ranges and parameters used to confirm their breadth of range in some embodiments of this application are approximate values, in specific embodiments, such values ​​are set as precisely as feasible.

Claims

1. An active caching routing method for a satellite content delivery network, characterized in that, The satellite content delivery network includes at least one satellite node, and the method includes: A deep learning model is used to extract the first semantic information features of the distributed content and the second semantic information features of historically popular content. Calculate the semantic similarity between the first semantic information feature and the second semantic information feature; Calculating the satellite semantic cache priority value of the satellite node corresponding to the distributed content based on the semantic similarity includes: calculating the regional semantic content popularity of the distributed content in the preset region based on the semantic similarity and the request frequency of the historically popular content within the preset region; calculating the satellite semantic cache priority value based on the regional semantic content popularity; wherein, the regional semantic content popularity is calculated based on the preset region, the distributed content, the historically popular content, the semantic similarity, the number of historically popular content items, and the request frequency; the satellite semantic cache priority value is calculated based on the number of future service regions of the satellite node and the regional semantic content popularity corresponding to the future service regions of the satellite node; and The active caching routing path for the distributed content is planned based on the satellite semantic cache priority value of each satellite node.

2. The active caching routing method for a satellite content delivery network as described in claim 1, characterized in that, The popularity of the semantic content of the region is calculated using the following formula: in, This indicates the popularity of semantic content in the region. This refers to the preset area. This refers to the distributed content; This indicates the historical popular content. Indicates the semantic similarity; This indicates the number of historically popular content items; This indicates the frequency of the request.

3. The active caching routing method for a satellite content delivery network as described in claim 1 or 2, characterized in that, The satellite semantic cache priority value is calculated using the following formula: in, Indicates the distributed content corresponding satellite nodes The satellite semantic cache priority value; Indicates satellite node The number of future service areas; Indicates the satellite node Future service area The corresponding regional semantic content popularity, superscript Indicates the satellite node The corresponding number A future service area.

4. The active caching routing method for a satellite content delivery network as described in claim 1, characterized in that, The active caching routing path for the distributed content is planned based on the satellite semantic cache priority value of each satellite node, including: The path caching benefit is calculated based on the satellite semantic cache priority value. Calculate the cache path benefit penalty value based on the length of the cache path and the preset path length; Calculate the final path revenue based on the path caching revenue and the cached path revenue penalty value; The cached path corresponding to the largest final path benefit is taken as the active cached routing path.

5. The active caching routing method for a satellite content delivery network as described in claim 4, characterized in that, The path caching benefit is calculated using the following formula: in, This indicates the caching revenue of the stated path. Indicates the cache path The source satellite node in Indicates the cache path The target satellite node; Indicates the cache path Corresponding relay satellite node set Satellite nodes in; Indicates the distributed content corresponding satellite nodes The satellite semantic cache priority value.

6. The active caching routing method for a satellite content delivery network as described in claim 5, characterized in that, The final path revenue is calculated using the following formula: in, This represents the final path benefit; This indicates the preset maximum path revenue value; This indicates the preset maximum number of additional route hops; This represents the penalty value for the cached path's benefit. This indicates the length of the cache path. This indicates the preset path length; Indicates the first weight. Indicates the second weight. , , .

7. The active caching routing method for a satellite content delivery network as described in claim 1, characterized in that, The distributed content is a data packet constructed according to a preset routing protocol; Following the step of planning the proactive caching route path for the distributed content based on the satellite semantic cache priority value of each satellite node, the method further includes: The programmable routing header of the distributed content data packet is programmed according to the active caching routing path, and the programmable routing header is used to enable each satellite node to actively cache the distributed content data packet during the content distribution process.

8. The active caching routing method for a satellite content delivery network as described in claim 7, characterized in that, Programming the programmable routing header of the distributed content data packet according to the actively cached routing path includes: In response to a satellite semantic cache priority value corresponding to the active cache routing path being greater than a preset priority threshold, the satellite node corresponding to the satellite semantic cache priority value is marked as an intermediate cache satellite node. The intermediate cache satellite node is used to instruct the satellite node to copy and cache the distribution content data packet before forwarding it to the next-hop satellite node when receiving the distribution content data packet; or In response to the satellite semantic cache priority value corresponding to the active cache routing path being less than the preset priority threshold, the satellite node corresponding to the satellite semantic cache priority value is marked as an intermediate bypass satellite node. The intermediate bypass satellite node is used to instruct the satellite node to directly forward the distributed content data packet to the next-hop satellite node.

9. An active caching routing system for a satellite content delivery network, characterized in that, Suitable for performing the active caching routing method as described in any one of claims 1-8, the active caching routing system comprising: The semantic information extraction and feature encoding module is configured to use a deep learning model to extract the first semantic information features of the distributed content and the second semantic information features of historical popular content. The semantic similarity calculation module is configured to calculate the semantic similarity between the first semantic information feature and the second semantic information feature; The regional semantic content popularity calculation and satellite semantic cache priority value prediction module is configured to calculate the satellite semantic cache priority value of the satellite node corresponding to the distributed content based on the semantic similarity. The proactive caching routing mechanism module is configured to plan the proactive caching routing path of the distributed content based on the satellite semantic caching priority value of each satellite node; A programmable routing protocol module is configured to programmatically distribute programmable routing headers for content data packets based on the actively cached routing path.

Images