Cache decision-making and updating method for space-air-ground integrated network

Abstract

The invention discloses a cache decision-making and updating method for a space-air-ground integrated network. The method is applied to a dynamic topology network comprising low earth orbit satellites and high-altitude platform nodes. The method comprises the following steps: a cache node selection step: quantifying node importance based on a node comprehensive importance evaluation model and selecting a key node as a cache node; and a cache content decision-making step: caching or replacing the content on the cache node based on a user closeness index, the index being obtained by fusing the position proximity between the content request user and the cache node, the content dynamic popularity and the recent access frequency through comprehensive calculation. According to the method, through two-stage intelligent decision, the cache hit rate and the resource utilization efficiency are remarkably improved, the access delay is effectively reduced, and the robustness of the network in a dynamic environment and when nodes fail is enhanced.

Description

Technical Field

[0001] This invention relates to the field of communication network technology, and specifically to a method for making caching decisions and updating cached content at caching nodes in a space-air-ground integrated network (SAGIN) composed of a high-altitude platform (HAP), low-Earth orbit (LEO) satellites, and terrestrial networks. Background Technology

[0002] With the growing global demand for ubiquitous and efficient communication, integrated space-air-ground networks have become the core architecture of 6G and future communication systems. This network integrates satellites (space-based), airborne platforms (air-based), and terrestrial networks (ground-based) to provide information services with full coverage, high reliability, and low latency. Content caching technology, as a key means to improve network service efficiency, reduce backbone network load, and lower user access latency, is widely considered one of the core enabling technologies for unlocking the potential of integrated space-air-ground networks.

[0003] Implementing content caching in an integrated air-space-ground network essentially involves pre-storing popular content on edge nodes such as satellites and airborne platforms, bringing data closer to users and thus optimizing the service experience. However, unlike relatively stable terrestrial networks, integrated air-space-ground networks possess unique dynamism, heterogeneity, and resource constraints. This presents significant challenges when directly porting traditional terrestrial network caching and update strategies to applications, specifically in the following three aspects:

[0004] (1) High dynamism of network topology and user service relationships. Non-ground nodes, represented by low-Earth orbit satellites and UAVs, are in high-speed motion, and the ground service areas they cover are constantly and rapidly changing. This results in strong time-varying characteristics in the correlation between nodes and ground users, as well as the network connection status between nodes. For example, a low-Earth orbit satellite may be serving a certain area at one moment and a different area at the next moment. Static caching strategies cannot adapt to the dynamic service relationships caused by mobility, and are prone to mismatch between cached content and user needs in the current service area, resulting in a decrease in cache hit rate and an increase in request latency. Some existing studies have attempted to introduce location-based management strategies, but they still face huge state maintenance overhead and decision lag problems when dealing with global-scale, constantly changing service relationships.

[0005] (2) High heterogeneity of cached resources and difficulties in global coordination. In the integrated air-space-ground network, nodes at different layers, such as satellites and high-altitude platforms, have huge differences in storage capacity, computing power, energy supply, and backhaul link bandwidth. At the same time, user requests have complex locality characteristics in time and space. Many existing caching strategies either optimize only for a single type of node or adopt simplified, centralized global coordination models, which make it difficult to meet diverse business needs while achieving efficient global utilization and coordination of multi-dimensional heterogeneous resources across air, space, and ground.

[0006] (3) Inefficient decision-making for updating cached content in dynamic environments. Content popularity evolves dynamically with time, location, and events. Under the constraints of dynamic topology and limited inter-satellite and satellite-to-ground link resources, deciding when to update, which content to update, and through which path to distribute the updated content is an extremely complex optimization problem. Some existing methods rely on periodic global information collection and centralized decision-making, which introduces unacceptable decision delays and signaling overhead in large-scale satellite networks with significant latency. Other adaptive methods based on machine learning, while able to handle certain uncertainties, often suffer from slow convergence and insufficient policy generalization ability when facing extreme dynamic environments with rapid changes in both network topology and content popularity, making it difficult to achieve efficient and accurate real-time update decisions.

[0007] In summary, current content caching and update technologies in integrated air-space-ground networks face three main challenges: mismatch between cache placement and user needs in highly dynamic network environments; low global collaboration efficiency under heterogeneous resource constraints; and inaccurate and untimely content update decisions under dynamic constraints. The root cause of these problems lies in the fact that existing strategies fail to fundamentally and effectively characterize and address the strong coupling between the inherent dynamics of network topology, the multidimensional heterogeneity of resources, and the spatiotemporal differences in business needs within integrated air-space-ground networks.

[0008] Therefore, there is an urgent need for a new content caching and updating method that can deeply perceive and adapt to the dynamic topology and global state of the integrated air-space-ground network, and make intelligent collaborative cache placement and updating decisions under complex constraints, thereby optimizing the overall network performance. Summary of the Invention

[0009] To address the needs of existing technologies, this invention provides a caching decision and update method for integrated air-space-ground networks, solving the caching and update challenges faced by existing technologies in integrated air-space-ground networks.

[0010] A cache decision and update method for integrated air-space-ground networks, applied to communication networks with dynamic topologies, the method comprising:

[0011] Cache node selection steps: Based on the topological characteristics and location attributes of nodes, the importance of each node in the network is quantified through a node comprehensive importance evaluation model, and a preset number of nodes are selected as cache nodes based on the quantification results.

[0012] Cache content decision steps: On the cache node, based on the user proximity index, a decision is made to cache or replace the arriving content; wherein, the user proximity index is a quantitative evaluation value calculated for each piece of content to be cached, based on the location proximity between the user who made the content request and the cache node, the dynamic popularity of the content, and the recent access frequency of the content.

[0013] Furthermore, the node comprehensive importance assessment model is constructed by fusing the node's importance transmission value with its KSGC importance.

[0014] Furthermore, the importance transmission value is calculated based on the transmission capability between nodes and the degree centrality of the node itself; the KSGC importance is calculated based on the gravity model and incorporates the k-shell value of the node.

[0015] Furthermore, the location proximity is calculated based on the reciprocal of the average distance from the cache node to all other nodes in the network;

[0016] The dynamic popularity is dynamically calculated based on the number of times each user requests the content within a historical time period;

[0017] The recent access frequency is calculated based on the reciprocal of the time interval between the user's two most recent consecutive requests for the content.

[0018] Furthermore, the cached content decision-making step includes:

[0019] If the storage space of the cache node is not full, the newly arrived content will be cached;

[0020] If the storage space of the cache node is full, calculate the user proximity of the new content and compare it with the user proximity of each cached content, and replace the content with the lowest user proximity.

[0021] Furthermore, the method is executed within the information center network architecture and works in conjunction with the routing and forwarding mechanism of that architecture.

[0022] Furthermore, the collaboration specifically includes: recording the content request status through an enhanced PIT table; carrying content popularity and recent frequency information for calculating the user proximity in the interest packet; and the cache node executing the cached content decision-making steps based on the information in the received interest packet.

[0023] The beneficial effects of this invention are: improved cache hit rate and reduced latency: by constructing a two-level caching architecture and intelligent caching strategy, user requests can be responded to by a more stable high-altitude platform or satellite, which shortens the content retrieval distance from the network topology perspective; simulation verification shows that compared with traditional random or connectivity-based node selection methods, this method can improve the overall network cache hit rate by about 20% to 30% and effectively reduce the end-to-end average access latency;

[0024] Optimize the utilization of heterogeneous cache resources: A multi-dimensional node comprehensive importance evaluation model is adopted for cache node selection. By quantitatively analyzing multi-dimensional parameters such as node topology characteristics, transmission capacity and spatial location, key nodes that connect upstream and downstream in the network are accurately identified, thereby deploying limited and expensive cache resources in the most efficient positions and saving valuable inter-satellite and satellite-to-ground bandwidth resources.

[0025] Achieving dynamic and accurate content updates: The cache replacement strategy based on user proximity creatively integrates three dimensions—locational proximity, temporal locality, and global popularity—to dynamically evaluate content value. Compared with traditional LRU or strategies based solely on popularity, this strategy can improve content caching efficiency by approximately 15% to 25% in dynamic scenarios and can more sensitively adapt to user request patterns that change over time and space.

[0026] Enhance overall network robustness: By collaborating with the information center network architecture and utilizing enhanced PIT tables and interest packets carrying content popularity information, the network is given the ability to perceive content distribution and request status. This enables the network to maintain basic services through alternative paths and intelligently distributed cached content even when some nodes fail or links are interrupted, effectively reducing system service interruption time and significantly improving reliability in scenarios such as emergency communication. Attached Figure Description

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

[0028] Figure 2 This is a schematic diagram of the integrated air-space-ground network architecture used in the embodiments of the present invention. Detailed Implementation

[0029] The present invention will now be described in detail with reference to the accompanying drawings. 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 denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The directional terms such as left, center, right, top, and bottom in the embodiments of the present invention are only relative concepts or referenced to the normal use state of the product, and should not be considered restrictive.

[0030] A cache decision and update method for integrated air-space-ground networks, such as Figure 1 and Figure 2 As shown, it includes the following:

[0031] Step 1: Construct an integrated air-space-ground network model;

[0032] To achieve seamless global coverage and on-demand access communication services in 6G scenarios, this invention proposes an innovative architecture that integrates terrestrial and satellite networks; such as Figure 1 As shown, this architecture constructs a SAGIN network system with space-air-ground caching resources, mainly composed of three core elements: low-Earth orbit (LEO) satellites, high-altitude platforms (HAP), and ground users (GT). In this system, popular requests are intelligently cached on LEO satellites or HAP platforms, allowing ground users' requests to be directly served by these cache nodes without needing to forward requests to the core network via backhaul links. The application scenarios of this invention are mainly for areas where ground base stations do not achieve full coverage, such as remote areas, disaster areas, and other special environments.

[0033] In this architecture, LEO satellites periodically provide services to HAP platforms within their coverage area, while the HAP platforms directly connect with ground users, thereby building an efficient and reliable integrated air-ground communication network. The HAP acts as a buffering middle layer and access point, improving the link performance between satellite and ground terminals.

[0034] Step 2: Cache node selection based on comprehensive node importance assessment;

[0035] Step 2.1: Define the inter-node transmission capabilities;

[0036] A computational model for inter-node transmission capacity is constructed by integrating three dimensions: optimal path length, optimal path quantity, and information propagation rate. In an unweighted network, the optimal path is the shortest path; nodes To the node Transmission capability is defined as:

[0037] (1)

[0038] in, Represents a node To the node Transmission capability; Represents a node and nodes The length of the shortest path between them; This represents the number of optimal paths between these two nodes; The information dissemination rate is represented by the dissemination threshold setting in infectious disease models, and the average network degree is used. The reciprocal of is approximated as the information propagation rate;

[0039] Step 2.2: Construct the importance transfer matrix;

[0040] Based on the propagation effect of the optimal path on second- and third-order neighbor nodes, an importance transmission matrix is ​​constructed. In the network topology, due to the strong reachability between nodes, the influence of a node is not limited to its directly connected first-order neighbors, but is also transmitted to second- and even third-order neighbor nodes through the optimal path. Considering that the importance of a node depends not only on its own characteristics, but also on the combined effect of the influence of its neighbor nodes, this invention combines the two key indicators of node degree centrality and transmission capability to construct an importance transmission matrix, as shown in formula (2):

[0041] (2)

[0042] in, Assign parameters to transmission capacity when a node Located at node When within the range of third-order neighbors The value is 1 if it is set to 1, and 0 otherwise. The characteristics of a node are represented by its degree centrality, which is used to calculate the degree centrality of the node. Represents a node For nodes The importance transmission value is determined by both the node's degree value and its transmission capacity. Therefore, the importance transmission value of a node is positively correlated with its influence on nodes within its third-order neighbor range. That is, the higher the importance transmission value, the more significant the influence of the node on its third-order neighbor nodes.

[0043] Step 2.3: Calculate KSGC importance; The location information of a node is also one of the key attributes for measuring the importance of a network node. Based on the classic gravity model, this invention introduces node location information and combines it with the global features of the node to design the KSGC importance of the node, as shown in formula (3).

[0044] (3)

[0045] in, As an attraction coefficient, it is used to adjust the attractiveness of the central node to other nodes in the network, and can be calculated by formula (4):

[0046] (4)

[0047] in, and These are nodes and nodes The k-shell value; and These represent the maximum and minimum k-shell values ​​of nodes in the current network, respectively.

[0048] Step 2.4: Calculate the overall importance of nodes and select cache nodes: The importance transfer value obtained based on the importance transfer matrix can effectively reflect the influence of a node in its local range, while the KSGC importance of a node can accurately characterize the node's positional attributes and its global characteristics; In order to fully consider the positional characteristics and local influence of nodes, this invention proposes the concept of node overall importance (Combined Importance), and its calculation formula is shown in Equation (5);

[0049] (5)

[0050] in, For nodes Overall importance;

[0051] Based on the node comprehensive importance evaluation model, a certain proportion of the most critical nodes in the network are intelligently selected as cache nodes to achieve optimal allocation of cache resources.

[0052] Step 3: Update cached content based on user proximity. Compared to terrestrial networks, HAP and LEO nodes move at high speeds, resulting in significant changes in network topology. The location information of cache nodes changes constantly, and cache resources are limited. For requesting users, the greater the popularity of the content and the higher the frequency of its recent appearance, the closer the content is to the requesting user. The closer the cache node is to the requesting user, the greater the proximity of the cache node to the requesting user. User proximity is defined using location proximity, content popularity, and content recent appearance frequency, and then efficient updates of cached content are achieved based on user proximity.

[0053] Step 3.1: Calculate the proximity of locations;

[0054] In satellite networks, the closer a cache node is to the requesting user, the faster the user can retrieve content via inter-satellite or satellite-to-ground links, which facilitates information transmission. Identifying key influential nodes in the SAGIN network that are close to the requesting user helps improve network transmission performance. In the SAGIN network, a high proximity of a cache node to the requesting user means the node is closer to the user's request, and the faster the user retrieves data through that node. Timely data retrieval is crucial for the requesting user. For the low-Earth orbit satellite network topology diagram G= , For the number of satellite nodes, For inter-satellite links, where satellite nodes Average distance to other satellite nodes The calculation formula is shown in equation (6):

[0055] (6)

[0056] in, Represents a node arrive distance, The smaller the value, the more cache nodes... The smaller the average distance to other nodes, the better the invention will achieve its goals. Centralized processing to better represent the metric, that is... The reciprocal of is defined as the proximity of the cache node, as shown in formula (7);

[0057] (7)

[0058] Step 3.2: Calculate dynamic content popularity, content popularity;

[0059] Cache nodes with high proximity can deliver data to requesting users faster, but at the same time, more data will pass through the node. At this time, satellite nodes will have data redundancy, and the content in the cache space will be frequently changed, affecting network performance. Therefore, it is considered to prioritize storing content with high popularity in cache nodes with high proximity. There are many ways to represent content popularity, such as the number of times the content is accessed within a certain period. The more times the content is requested within a certain period, the greater the content popularity. This method ignores the differences between requesting users. Different requesting users have different needs for the content. This invention dynamically calculates the content popularity from the user's perspective, considering that different requesting users have different needs for the content in the network. It records the number of times each user requests each content and dynamically calculates the content popularity. The content popularity based on the user's perspective is represented by formula (8):

[0060] (8)

[0061] in This indicates that each user requests content The number of requests, Indicates the amount of content requested by the user;

[0062] Step 3.3: Calculate the recent access frequency of the content; For the requesting user, popular content will be requested frequently. When performing cache content updates, keeping popular content in cache nodes with high proximity may result in the problem of previously popular content occupying cache space for a long time. If a user frequently requests a certain data in a short period of time, it indicates that the requesting user has a high level of attention to the data recently. This invention uses two-dimensional Euclidean distance to represent the recent frequency of the content. In particular, when the requesting node has only been requested once, it is represented by the difference between the two times. The more frequently the user requests the content recently, the smaller this indicator is, as shown in formula (9):

[0063] (9)

[0064] In the formula The time of this request for the content. The time of the last request for the content. Given the time of the previous request, the recent frequency of the content can be expressed by formula (10):

[0065] (10)

[0066] The higher the value, the more frequently the user has requested the content recently; the more recent the content is, the greater the user's recent level of attention to the content.

[0067] Step 3.4: Calculate user proximity and make a decision;

[0068] User proximity is represented by the following three indicators: First, the location of the cache node and the requesting user. The closer the cache node is to the requesting user, the faster the user can obtain the content. Second, the popularity of the content from the user's perspective. The higher the frequency of the user's request for the content, the greater the popularity of the content. Third, the recent frequency of the content. The higher this indicator is, the more frequently the requesting user has recently requested the content. This invention integrates the three indicators into user proximity and prioritizes the retention of content with high user proximity on satellite nodes that are close to the requesting user, as shown in formula (11).

[0069] (11)

[0070] in, , and The coefficients for the three indicators are all taken as 1 / 3 in this invention, and all three indicators have been normalized.

[0071] Step 4: Cooperative routing and forwarding strategy (in ICN architecture). To integrate the above decisions into network operations, this invention proposes a routing and cache update mechanism that is cooperative with the Information Center Network (ICN) architecture.

[0072] Step 4.1: Enhanced PIT table and interest packet generation. During the content caching and update process, each requesting user maintains a modified PIT table, which records the content tag, forwarding port, number of requests, and request time. When a user requests content, it first checks its own PIT table. If the entry already exists, it updates the number of requests, records the new request time, and calculates the current content popularity and recent frequency based on formulas (8) and (10). Then, it generates an interest packet forwarding with content tag, content popularity, and recent frequency, and records the forwarding port in the PIT table. If the entry does not exist, it creates the corresponding entry, records the number of requests and request time, calculates the popularity and recent frequency of the content, and generates an interest packet with content tag, content popularity, and recent frequency. In the PIT tables of other cache nodes, the content tag and forwarding port are recorded. When a cache node has already received the corresponding interest packet, the forwarding port is added to the corresponding entry, and the interest packet is discarded. Otherwise, a new entry is created.

[0073] Step 4.2: Processing and forwarding interest packets

[0074] When a user requests a query for an interest packet forwarded by the FIB, the forwarding port is recorded in its own PIT table entry. When the interest packet arrives at the next cache node, the cache node queries the CS. If the corresponding content is already cached, the data packet is modified according to the content popularity and recent frequency recorded in the interest packet, and then the data packet is returned at the forwarding port. If the corresponding data packet is not cached in the cache space, the cache node will query the PIT table to see if the corresponding entry is recorded. If it exists, the forwarding port is recorded and forwarding is performed. If it does not exist, the corresponding entry is created and the forwarding port is recorded.

[0075] Step 4.3: Packet processing and caching decisions;

[0076] When an interest packet receives a response at the CS of a cache node, the satellite node queries the popularity and recent frequency of the content recorded in the interest packet, modifies the data recorded in the corresponding data packet, and forwards it according to the port. When the data packet arrives at the next cache node, the cache node first queries its own CS. If the cache space is not full and the corresponding data packet is not cached, the cache node executes the caching decision strategy. If the cache space is full and the corresponding data packet is not cached, the cache node will query the popularity and recent frequency of the content in the data packet, obtain the proximity of the satellite node at this time, calculate the user proximity of the data packet according to formula (11), evict the content with the lowest user proximity in the cache space, retain the newly arrived data packet, and forward the data packet. If the corresponding data packet already exists in the cache space of the cache node, the PIT table is queried and the data packet is forwarded. When the interest packet does not receive a response at any cache node, the interest packet will be forwarded to the source server node, which queries the popularity and recent frequency of the content recorded in the interest packet, generates the corresponding data packet, and forwards it to the next cache node.

[0077] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for cache decision and update for space-air-ground integrated network, characterized in that, Applied to communication networks containing dynamic topologies, the method includes: Cache node selection steps: Based on the topological characteristics and location attributes of nodes, the importance of each node in the network is quantified through a node comprehensive importance evaluation model, and a preset number of nodes are selected as cache nodes based on the quantification results. Cache content decision steps: On the cache node, based on the user proximity index, a decision is made to cache or replace the arriving content; wherein, the user proximity index is a quantitative evaluation value calculated for each piece of content to be cached, based on the location proximity between the user who made the content request and the cache node, the dynamic popularity of the content, and the recent access frequency of the content.

2. The method of claim 1, wherein, The node comprehensive importance assessment model is constructed by fusing the node's importance transmission value with its KSGC importance.

3. The cache decision and update method for integrated air-space-ground networks according to claim 2, characterized in that, The importance transmission value is calculated based on the transmission capability between nodes and the degree centrality of the node itself; the KSGC importance is based on the gravity model and incorporates the k-shell value of the node for calculation.

4. The cache decision and update method for integrated air-space-ground networks according to claim 1, characterized in that: The location proximity is calculated based on the reciprocal of the average distance from the cache node to all other nodes in the network; The dynamic popularity is dynamically calculated based on the number of times each user requests the content within a historical time period; The recent access frequency is calculated based on the reciprocal of the time interval between the user's two most recent consecutive requests for the content.

5. The cache decision and update method for integrated air-space-ground networks according to claim 1, characterized in that, The cached content decision-making steps include: If the storage space of the cache node is not full, the newly arrived content will be cached; If the storage space of the cache node is full, calculate the user proximity of the new content and compare it with the user proximity of each cached content, and replace the content with the lowest user proximity.

6. The cache decision and update method for integrated air-space-ground networks according to claim 1, characterized in that, The method is executed within an information-centric network architecture and works in conjunction with the architecture's routing and forwarding mechanisms.

7. The cache decision and update method for integrated air-space-ground networks according to claim 6, characterized in that, The collaboration specifically includes: recording the content request status through an enhanced PIT table; carrying content popularity and recent frequency information for calculating the user proximity in interest packets; and the cache node executing the cached content decision-making steps based on the information in the received interest packets.

Images