Satellite-based business load migration method and system based on dynamic region division of heat map

Abstract

The application provides a space-based service load migration method and system based on dynamic region division of heat maps, comprising a heat map generation module, a dynamic region division module, a migration decision module and a fault-tolerant execution module. The dynamic region division module divides the network region according to the service load heat map generated by the heat map generation module, identifies the service hotspot region and quantifies the level; the migration decision module plans a service load migration path according to the service hotspot region; and the fault-tolerant execution module finds a replacement node to generate multiple paths and executes the load migration according to the optimal migration path. The application decomposes the global calculation into multiple orbit plane parallel calculations, realizes the global balance of the space-based satellite networking platform load, and solves the problems of the coexistence of the resource waste and network congestion in some regions caused by the uneven service load in the low-orbit grid-based huge constellation networking.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of space-based network intelligence and satellite edge computing, and particularly relates to a space-based service load migration method and system based on dynamic region division of heat maps, in particular, a space-based service efficient load migration method and device based on dynamic region division of heat maps. BACKGROUND

[0002] Due to the spatial non-uniformity caused by the geographical location distribution of users and the temporal non-uniformity caused by the activity period of users, the user service accesses the space-based platform, causing the non-uniformity of satellite load, and further causing resource waste or network congestion in some areas. In the existing space-based service system, the load migration technology is an important means to ensure service continuity and efficient use of resources.

[0003] However, the traditional load migration method usually performs region division and migration decision based on static or simple dynamic threshold strategy, resulting in significant limitations in actual application. Specifically, the traditional method often relies on fixed geographical or logical boundaries, which is difficult to accurately distinguish real high-load areas, and in the load migration process, a large number of invalid switching problems are caused by repeated migration in high-load areas. Invalid switching will introduce additional signaling interaction and migration delay, increase system overhead, affect real-time requirements of space-based services (such as remote sensing data processing, emergency communication, etc.), and reduce service quality.

[0004] The patent document "Satellite network multi-service QoS routing method based on region division" (CN113067625A) discloses a satellite network multi-service QoS routing method based on region division, which divides time slices through virtual topology, determines service weight, divides heavy and light load areas, and uses ground control center to calculate static routing path offline, and dynamically calculates routing path in heavy load area, reducing on-board computing complexity and routing signaling overhead. However, its is a static routing planning strategy, and the routing path is completely different, which cannot realize adaptive division, migration path optimization and fault tolerance execution.

[0005] The patent document "Polar constellation inter-satellite load balancing routing method and system based on topology switching" (CN117938238A) discloses a polar constellation inter-satellite load balancing routing method based on topology switching, which divides blocks through ground control center, calculates block connectivity and link load strength, generates routing table, and calculates switching time combined with satellite operation law, reducing computing complexity and error. However, it is for polar constellation rather than low-orbit grid-based mega constellation, which has great difference and cannot be migrated, and it does not consider parallel computing and relies on ground offline calculation.

[0006] Therefore, a kind of sky-based service efficient load migration method based on heat map dynamic area division is urgently needed, which can dynamically perceive, adaptively divide and load migration to service hotspot area, realize the global balance of sky-based satellite networking platform load. SUMMARY

[0007] In view of the defects in the prior art, the purpose of the present application is to provide a sky-based service load migration method and system based on heat map dynamic area division.

[0008] According to the sky-based service load migration method based on heat map dynamic area division provided by the present application, the method comprises the following steps:

[0009] Step S1, collect data, and construct service load heat map;

[0010] Step S2, divide network area according to service load heat map, identify service hotspot area and quantify grade;

[0011] Step S3, according to the divided service hotspot area, plan service load migration path;

[0012] Step S4, find multiple paths by replacing nodes, judge and select the optimal migration path to execute load migration.

[0013] Preferably, in the step S1, the data of the whole domain is collected and divided into multiple orbital plane data collection:

[0014]

[0015] Wherein, represents the data of the i-th satellite on the same orbital plane; represents the data of all satellites on the p-th orbital plane; represents the data of all orbital planes; represents the data set of the p-th orbital plane.

[0016] The main star of the orbital plane receives the load index of all satellites on the orbital plane, calculates the corresponding mean and variance, and generates a cluster summary;

[0017] The ground station receives the cluster summary information of all orbital plane main stars, constructs the mapping from orbital plane to plane, and generates service load heat map;

[0018] The received cluster summary information is processed by unsupervised clustering method, and the satellites in high load area are used as the source of migration, and the satellites in low load area are used as the destination candidate of migration.

[0019] Preferably, the main star of the orbital plane preprocesses the load index of each satellite local service access of the orbital plane:

[0020]

[0021] Min-Max normalization processing is performed to generate a standardized feature vector, and clustering summary information is obtained:

[0022]

[0023] wherein, represents the data of the i-th satellite on the same orbital plane;

[0024] represents the data of all satellites on the p-th orbital plane;

[0025] represents the data of all orbital planes;

[0026] represents the k-th load indicator of .

[0027] The data collection period is 30 seconds.

[0028] Preferably, in step S2, the network region is divided according to the number of inter-satellite link hops, the spatio-temporal continuity is detected, the isolated points are merged, the business hotspot region is identified, and the load intensity is quantified.

[0029] The network region is divided by judging the number of inter-satellite link hops. If two satellites communicate through ≤2 hops, they are divided into the same network region. If two satellites communicate through >2 hops, they are divided into different network regions.

[0030] The spatio-temporal continuity detection is to judge whether the points in the same network region are in the same load state for 80% of the time within the continuous time window . If yes, it is determined as a stable hotspot and merged into the same load intensity level. If not, no adjustment is made.

[0031] The load intensity quantification divides the range according to the number of levels set by the task, and the processing strategy is divided into receivable migration, monitoring state and trigger migration.

[0032] Preferably, in step S3, the migration priority of each business hotspot region is calculated:

[0033] Migration priority = regional average load × regional duration

[0034] Through the hierarchical routing planning method, the path with the lowest delay is planned within the region of the same load intensity level.

[0035] When crossing load intensity levels, select the region with the lowest load intensity level and choose the region with the highest migration priority as the migration target region according to the migration priority, and plan a path that meets the specified constraints or requirements of bandwidth and latency between regions.

[0036] In step S4, a fast replacement node is sought in the path within 3 hops using the spatial neighborhood search method. Satellites that share the orbital plane with the main path are excluded, and a new path is generated. Each satellite maintains a local routing table, which includes the real-time latency / load status of 1-hop neighbors and the average latency of 2-hop neighbors.

[0037] The path with the lowest overall cost is selected. If there is only one path with the lowest overall cost, it is taken as the optimal migration path. If there are multiple paths with the same overall cost, it is determined whether they are in the same load intensity level area under the existing constraints.

[0038] If the regions with the same load intensity level are selected, the path with the lowest latency is chosen as the optimal migration path. If the regions with different load intensity levels are selected, the path with the best latency and bandwidth is chosen as the optimal migration path.

[0039] According to the present invention, a space-based service load migration system based on dynamic region division of heat map includes: a heat map generation module, a dynamic region division module, a migration decision module, and a fault-tolerant execution module;

[0040] The dynamic region segmentation module divides network regions based on the service load heat map generated by the heat map generation module, identifies service hotspots, and quantifies their levels.

[0041] The migration decision module plans business load migration paths based on business hotspot areas;

[0042] The fault-tolerant execution module finds replacement nodes, generates multiple paths, judges and selects the optimal migration path to perform load migration.

[0043] Preferably, the heatmap generation module decomposes the global data collection into data collection for multiple orbital planes:

[0044]

[0045] in, This represents the data of the i-th satellite on the same orbital plane; This represents all satellite data for the p-th orbital plane; Data representing all orbital planes; This represents the data set for the p-th orbital plane.

[0046] The primary satellite in the orbital plane acts as a computing node, receiving the load indicators of all satellites in its orbital plane, calculating the corresponding mean and variance, and generating a cluster summary.

[0047] The ground station receives clustering summary information from all primary satellites in the orbital plane, constructs a mapping from the orbital plane to the plane, and generates a service load heatmap.

[0048] An unsupervised clustering method is used on the received clustering summary information to select satellites that form high-load regions as the source of migration and satellites that form low-load regions as the destination candidates for migration.

[0049] Preferably, the load index for local service access of each satellite in the orbital plane preprocessing orbital plane is as follows:

[0050]

[0051] Min-Max normalization is performed to generate standardized feature vectors, and cluster summary information is obtained.

[0052]

[0053] in, This represents the data of the i-th satellite on the same orbital plane;

[0054] This represents all satellite data for the p-th orbital plane;

[0055] Data representing all orbital planes;

[0056] express The kth load metric.

[0057] The data collection period is 30 seconds.

[0058] Preferably, the dynamic region division module divides the network region according to the number of inter-satellite link hops, performs spatiotemporal continuity detection, merges isolated points, identifies service hotspot areas, and quantifies the load intensity by level.

[0059] The division of network regions is determined by the number of hops in the inter-satellite links. If two satellites communicate with each other through ≤2 hops, they are divided into the same network region. If two satellites communicate with each other through >2 hops, they are divided into different network regions.

[0060] The spatiotemporal continuity detection determines whether points within the same divided network region are within a continuous time window. If an element is under the same load condition 80% of the time, it is considered a stable hotspot and merged into the same load intensity level; otherwise, no adjustment is made.

[0061] The load intensity grading quantification is evenly divided into ranges based on the number of levels set for the task, and the processing strategy is divided into acceptable migration, monitoring status, and triggered migration.

[0062] Preferably, the migration decision module calculates the migration priority for each business hotspot area:

[0063] Migration priority = Area average load × Area duration

[0064] By using a hierarchical routing planning method, the path with the lowest latency is planned within an area of ​​the same load intensity level;

[0065] When crossing load intensity levels, select the region with the lowest load intensity level and choose the region with the highest migration priority as the migration target region according to the migration priority, and plan a path that meets the specified constraints or requirements of bandwidth and latency between regions.

[0066] The fault-tolerant execution module seeks fast replacement nodes in paths within 3 hops using a spatial neighborhood search method, excludes satellites that share the orbital plane with the main path, generates new paths, and each satellite maintains a local routing table, which includes the real-time latency / load status of 1-hop neighbors and the average latency of 2-hop neighbors.

[0067] Select the path with the lowest overall cost. If there is only one path with the lowest overall cost, then take it as the optimal migration path. If there are multiple paths with the same overall cost, then determine whether they are in the same load intensity level area under the existing constraints.

[0068] If the regions with the same load intensity level are selected, the path with the lowest latency is chosen as the optimal migration path. If the regions with different load intensity levels are selected, the path with the best latency and bandwidth is chosen as the optimal migration path.

[0069] Compared with the prior art, the present invention has the following beneficial effects:

[0070] 1. This invention achieves global load balancing of the space-based satellite networking platform by dynamically sensing, adaptively dividing, and migrating service hotspot areas.

[0071] 2. This invention solves the problem of resource waste and network congestion in some areas caused by uneven service load in low-orbit gridded mega-constellation networking by decomposing global computing into parallel computing on multiple orbital planes.

[0072] 3. This invention implements a dynamic routing scheme through migration priority calculation and fault tolerance mechanism, which improves migration efficiency and reduces losses while meeting bandwidth and latency requirements. Attached Figure Description

[0073] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0074] Figure 1 This is a schematic diagram of the space-based service load migration method based on dynamic region division of heatmaps.

[0075] Figure 2 A schematic diagram illustrating the process of constructing a business load heatmap based on an unsupervised clustering method;

[0076] Figure 3 A flowchart illustrating the method for adaptive business hotspot area identification and load level classification;

[0077] Figure 4 A schematic diagram of the global business load migration decision-making process for dynamic routing planning;

[0078] Figure 5 This is a schematic diagram illustrating a high-reliability load migration process based on a fault-tolerant mechanism.

[0079] Figure 6 This is a schematic diagram of a space-based service load migration system based on dynamic region division using a heatmap. Detailed Implementation

[0080] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0081] This invention provides a space-based service load migration method based on dynamic region division using heatmaps. By dynamically sensing, adaptively dividing, and migrating service hotspot areas, it achieves global load balancing on the space-based satellite networking platform, solving the problem of resource waste and network congestion in some areas caused by uneven service loads in low-Earth orbit gridded mega-constellation networks. Figure 1 For example, the specific steps include:

[0082] Step S1: Construct a business load heatmap based on an unsupervised method, and generate the business load heatmap through spatial hierarchical management clustering.

[0083] Specifically, with Figure 2 For example, the collection of data across the entire domain can be broken down into the collection of data from multiple orbital planes:

[0084]

[0085] in, For data from the i-th satellite on the same orbital plane, For all satellite data in the p-th orbital plane, This represents data for all orbital planes. Global computation is decomposed into parallel computations across multiple orbital planes, with each plane processing only its own satellite data.

[0086] The primary satellite in the orbital plane acts as a computing node, receiving the normalized statistical values ​​of the payloads of all satellites in the plane, and calculating the mean and variance of each corresponding value.

[0087] Specifically, the load metrics for local service access of each satellite in the preprocessing orbital plane (processor utilization, memory usage, current ISL connection status, and orbital position) are as follows:

[0088]

[0089] Furthermore, heterogeneous multi-indicator data are normalized to eliminate the influence of dimensions, generate standardized feature vectors, and reduce transmission volume. The Min-Max normalization method is used, which is expressed as:

[0090]

[0091] Meanwhile, the data collection cycle was set to 30 seconds to accommodate the rapid movement of the constellations.

[0092] The ground station receives cluster summary information from the primary satellites on each orbital plane, constructs a mapping from the orbital plane to the ground plane to generate a global view, i.e., a service load heatmap. An unsupervised clustering method (such as K-means) is used on the received data to identify which satellites form high-load areas, serving as sources for migration; and which satellites form low-load areas, serving as candidate destinations for migration.

[0093] Step S2: Adaptive business hotspot area identification and load level classification. Through spatiotemporal continuity detection and load level quantification methods, business hotspot areas are identified and their levels are quantified, providing a basis for migration direction.

[0094] Specifically, with Figure 3 For example, based on the inter-satellite link hop count, if two satellites can communicate with each other through ≤2 hops, they belong to the same network region. Considering that a single continuous high-load area may be misjudged as multiple isolated hotspots, spatiotemporal continuity detection is performed, and isolated points are merged into regions.

[0095] Spatial continuity refers to merging points with a hop count of ≤2 hops; temporal continuity refers to the continuity within a time window. Determining persistent hotspots within a spatial region r. For example, when a continuous time window... If a region is under the same load condition 80% of the time, it is considered a stable hotspot and merged into the same load region.

[0096] Specifically, the spatiotemporal continuity is determined using a simple method, which involves considering points within the same region and determining whether they fall within a continuous time window. Within. The same area refers to a business hotspot area identified based on the business load heatmap, within a continuous time window. The time interval should be set relatively, but not too long; a second-level interval is preferred. This ensures spatial consistency during region division and reduces instability and fluctuations in the time interval.

[0097] The load intensity is graded and quantified into 8 levels (0-7), and the following provisions are made:

[0098]

[0099] In more preferred embodiments, the load strength level classification is manually set according to specific needs, and the range is generally evenly divided according to the required number of levels.

[0100] Step S3: Global service load migration decision based on dynamic routing planning. Through hierarchical routing planning method, the path with the lowest latency is planned within the same load level area, and the path between areas that meets the stability conditions of inter-satellite link ISL is planned across load areas.

[0101] Specifically, with Figure 4 For example, the starting point is selected within the neighborhood of high-load cluster centers, and the ending point is selected within the neighborhood of low-load cluster centers. The migration priority for each region is calculated using the following formula:

[0102] Migration priority = Area average load × Area duration

[0103] A hierarchical routing planning strategy is adopted. Within the same load area, the path with the lowest latency is selected first. When migrating across regions, the low-load area is selected and the area with the highest migration priority is selected as the migration target area according to the migration priority. The stability of the inter-satellite link must be met under the constraints of bandwidth and latency between regions.

[0104] The inter-satellite link stability condition refers to a stable link that can guarantee the required bandwidth and latency constraints or requirements; a link that can meet the bandwidth and latency requirements is considered stable.

[0105] Step S4: Perform a highly reliable load migration based on a fault-tolerant mechanism.

[0106] Specifically, with Figure 5For example, based on the existing routing plan, considering the situation where a node suddenly becomes unreachable, the spatial neighborhood search method is used to consider replacing nodes in paths within 3 hops and select multiple paths.

[0107] Seek quick replacement of nodes, exclude satellites sharing the orbital plane with the main path, and generate a new path to avoid in-plane cascading failures and ensure the reliability of the overall migration path.

[0108] Each satellite maintains a local routing table, containing: the real-time status (latency / load) of its 1-hop neighbors and summary information (average latency) of its 2-hop neighbors. The system selects the path with the lowest overall cost. If multiple paths exist, the path with the lowest latency and highest bandwidth, under existing constraints, is chosen; this is a dynamic routing scheme. Within the same area, the system considers the lowest latency; across areas, it considers both latency and bandwidth constraints.

[0109] The present invention also provides a space-based service load migration system based on dynamic region division of heatmaps. The space-based service load migration system based on dynamic region division of heatmaps can be implemented by executing the process steps of the space-based service load migration method based on dynamic region division of heatmaps. That is, those skilled in the art can understand the space-based service load migration method based on dynamic region division of heatmaps as a preferred embodiment of the space-based service load migration system based on dynamic region division of heatmaps.

[0110] According to the present invention, a space-based service load migration system based on dynamic region division using heatmaps is provided. Figure 6 For example, including:

[0111] Heatmap generation module: used for heatmap construction and generation;

[0112] Dynamic region partitioning module: used for dynamic partitioning of load regions;

[0113] Migration decision module: used for migration direction selection and dynamic route planning;

[0114] Fault-tolerant execution module: used for standby node switching and rapid route replanning.

[0115] The dynamic region segmentation module divides network regions based on the service load heat map generated by the heat map generation module, identifies service hotspots, and quantifies their levels.

[0116] The migration decision module plans business load migration paths based on business hotspot areas;

[0117] The fault-tolerant execution module finds replacement nodes, generates multiple paths, judges and selects the optimal migration path to perform load migration.

[0118] In more preferred embodiments, the heatmap generation module decomposes the global data collection into data collection for multiple orbital planes:

[0119]

[0120] in, This represents the data of the i-th satellite on the same orbital plane; This represents all satellite data for the p-th orbital plane; Data representing all orbital planes; This represents the data set for the p-th orbital plane.

[0121] The primary satellite in the orbital plane acts as a computing node, receiving the load indicators of all satellites in its orbital plane, calculating the corresponding mean and variance, and generating a cluster summary.

[0122] The ground station receives clustering summary information from all primary satellites in the orbital plane, constructs a mapping from the orbital plane to the plane, and generates a service load heatmap.

[0123] An unsupervised clustering method is used on the received clustering summary information to select satellites that form high-load regions as the source of migration and satellites that form low-load regions as the destination candidates for migration.

[0124] In more preferred embodiments, the load index for local service access of each satellite in the orbital plane preprocessing orbital plane is as follows:

[0125]

[0126] Min-Max normalization is performed to generate standardized feature vectors, and cluster summary information is obtained.

[0127]

[0128] in, This represents the data of the i-th satellite on the same orbital plane;

[0129] This represents all satellite data for the p-th orbital plane;

[0130] Data representing all orbital planes;

[0131] express The kth load metric.

[0132] The data collection period is 30 seconds.

[0133] In more preferred embodiments, the dynamic region division module divides network regions based on the number of inter-satellite link hops, performs spatiotemporal continuity detection, merges isolated points, identifies service hotspot areas, and quantifies load intensity by level.

[0134] The division of network regions is determined by the number of hops in the inter-satellite links. If two satellites communicate with each other through ≤2 hops, they are divided into the same network region. If two satellites communicate with each other through >2 hops, they are divided into different network regions.

[0135] The spatiotemporal continuity detection determines whether points within the same divided network region are within a continuous time window. If an element is under the same load condition 80% of the time, it is considered a stable hotspot and merged into the same load intensity level; otherwise, no adjustment is made.

[0136] The load intensity grading quantification is evenly divided into ranges based on the number of levels set for the task, and the processing strategy is divided into acceptable migration, monitoring status, and triggered migration.

[0137] In more preferred embodiments, the migration decision module calculates the migration priority for each business hotspot area:

[0138] Migration priority = Area average load × Area duration

[0139] By using a hierarchical routing planning method, the path with the lowest latency is planned within an area of ​​the same load intensity level;

[0140] When crossing load intensity levels, select the region with the lowest load intensity level and choose the region with the highest migration priority as the migration target region according to the migration priority, and plan a path that meets the specified constraints or requirements of bandwidth and latency between regions.

[0141] The fault-tolerant execution module seeks fast replacement nodes in paths within 3 hops using a spatial neighborhood search method, excludes satellites that share the orbital plane with the main path, generates new paths, and each satellite maintains a local routing table, which includes the real-time latency / load status of 1-hop neighbors and the average latency of 2-hop neighbors.

[0142] Select the path with the lowest overall cost. If there is only one path with the lowest overall cost, then take it as the optimal migration path. If there are multiple paths with the same overall cost, then determine whether they are in the same load intensity level area under the existing constraints.

[0143] If the regions with the same load intensity level are selected, the path with the lowest latency is chosen as the optimal migration path. If the regions with different load intensity levels are selected, the path with the best latency and bandwidth is chosen as the optimal migration path.

[0144] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0145] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method for migrating space-based service load based on dynamic region division using heatmaps, characterized in that, include: Step S1: Collect data and construct a business load heatmap; Step S2: Divide the network into regions based on the service load heatmap, identify service hotspots, and quantify their levels; Step S3: Based on the identified business hotspot areas, plan the business load migration path; Step S4: Find replacement nodes to generate multiple paths, determine and select the optimal migration path to perform load migration; In step S3, the migration priority of each business hotspot area is calculated: Migration priority = Area average load × Area duration By using a hierarchical routing planning method, the path with the lowest latency is planned within an area of ​​the same load intensity level; When crossing load intensity levels, select the region with the lower load intensity level and choose the region with the higher migration priority as the migration target region according to the migration priority, and plan the path that meets the specified constraints or requirements of bandwidth and latency between regions. In step S4, a fast replacement node is sought in the path within 3 hops by using the spatial neighborhood search method, satellites that share the orbital plane with the main path are excluded, and a new path is generated. Each satellite maintains a local routing table, which includes the real-time latency / load status of 1-hop neighbors and the average latency of 2-hop neighbors. Select the path with the lowest overall cost. If there is only one path with the lowest overall cost, then take it as the optimal migration path. If there are multiple paths with the same overall cost, then determine whether they are in the same load intensity level area under the existing constraints. If the regions with the same load intensity level are selected, the path with the lowest latency is chosen as the optimal migration path. If the regions with different load intensity levels are selected, the path with the best latency and bandwidth is chosen as the optimal migration path.

2. The space-based service load migration method based on dynamic region division of heatmaps according to claim 1, characterized in that, In step S1, the data collection of the entire domain is decomposed into data collection of multiple orbital planes: in, This represents the data of the i-th satellite on the same orbital plane; This represents all satellite data for the p-th orbital plane; Data representing all orbital planes; This represents the data set for the p-th orbital plane. The primary satellite in the orbital plane acts as a computing node, receiving the load indicators of all satellites in its orbital plane, calculating the corresponding mean and variance, and generating a cluster summary. The ground station receives clustering summary information from all primary satellites in the orbital plane, constructs a mapping from the orbital plane to the plane, and generates a service load heatmap. An unsupervised clustering method is used on the received clustering summary information to select satellites that form high-load regions as the source of migration and satellites that form low-load regions as the destination candidates for migration.

3. The space-based service load migration method based on dynamic region division of heatmaps according to claim 2, characterized in that, The load index for local service access of each satellite in the orbital plane preprocessing orbital plane is as follows: Min-Max normalization is performed to generate standardized feature vectors, and cluster summary information is obtained. in, This represents the data of the i-th satellite on the same orbital plane; This represents all satellite data for the p-th orbital plane; Data representing all orbital planes; express The kth load metric; The data collection period is 30 seconds.

4. The space-based service load migration method based on dynamic region division of heatmaps according to claim 1, characterized in that, In step S2, network regions are divided according to the number of hops in the inter-satellite links, spatiotemporal continuity is detected, isolated points are merged, service hotspot areas are identified, and load intensity is classified and quantified. The division of network regions is determined by the number of hops in the inter-satellite links. If two satellites communicate with each other through ≤2 hops, they are divided into the same network region. If two satellites communicate with each other through >2 hops, they are divided into different network regions. The spatiotemporal continuity detection determines whether points within the same segmented network region are within a continuous time window. If a hotspot is under the same load condition 80% of the time, it is considered a stable hotspot and merged into the same load intensity level; otherwise, no adjustment is made. The load intensity grading quantification is evenly divided into ranges based on the number of levels set for the task, and the processing strategy is divided into acceptable migration, monitoring status, and triggered migration.

5. A space-based service load migration system based on dynamic region division using heatmaps, characterized in that, include: The module includes a heatmap generation module, a dynamic region division module, a migration decision module, and a fault-tolerant execution module. The dynamic region segmentation module divides network regions based on the service load heat map generated by the heat map generation module, identifies service hotspots, and quantifies their levels. The migration decision module plans business load migration paths based on business hotspot areas; The fault-tolerant execution module finds replacement nodes, generates multiple paths, determines and selects the optimal migration path, and executes load migration. The migration decision module calculates the migration priority for each business hotspot area: Migration priority = Area average load × Area duration By using a hierarchical routing planning method, the path with the lowest latency is planned within an area of ​​the same load intensity level; When crossing load intensity levels, select the region with the lower load intensity level and choose the region with the higher migration priority as the migration target region according to the migration priority, and plan the path that meets the specified constraints or requirements of bandwidth and latency between regions. The fault-tolerant execution module seeks fast replacement nodes in paths within 3 hops using a spatial neighborhood search method, excludes satellites that share the orbital plane with the main path, generates a new path, and each satellite maintains a local routing table, which includes the real-time latency / load status of 1-hop neighbors and the average latency of 2-hop neighbors. Select the path with the lowest overall cost. If there is only one path with the lowest overall cost, then take it as the optimal migration path. If there are multiple paths with the same overall cost, then determine whether they are in the same load intensity level area under the existing constraints. If the regions with the same load intensity level are selected, the path with the lowest latency is chosen as the optimal migration path. If the regions with different load intensity levels are selected, the path with the best latency and bandwidth is chosen as the optimal migration path.

6. The space-based service load migration system based on dynamic region division of heatmaps according to claim 5, characterized in that, The heatmap generation module decomposes the data collection from the entire region into data collection from multiple orbital planes: in, This represents the data of the i-th satellite on the same orbital plane; This represents all satellite data for the p-th orbital plane; Data representing all orbital planes; This represents the data set for the p-th orbital plane. The primary satellite in the orbital plane acts as a computing node, receiving the load indicators of all satellites in its orbital plane, calculating the corresponding mean and variance, and generating a cluster summary. The ground station receives clustering summary information from all primary satellites in the orbital plane, constructs a mapping from the orbital plane to the plane, and generates a service load heatmap. An unsupervised clustering method is used on the received clustering summary information to select satellites that form high-load regions as the source of migration and satellites that form low-load regions as the destination candidates for migration.

7. The space-based service load migration system based on dynamic region division of heatmaps according to claim 6, characterized in that, The load index for local service access of each satellite in the orbital plane preprocessing orbital plane is as follows: Min-Max normalization is performed to generate standardized feature vectors, and cluster summary information is obtained. in, This represents the data of the i-th satellite on the same orbital plane; This represents all satellite data for the p-th orbital plane; Data representing all orbital planes; express The kth load metric; The data collection period is 30 seconds.

8. The space-based service load migration system based on dynamic region division of heatmaps according to claim 5, characterized in that, The dynamic region division module divides the network region based on the number of inter-satellite link hops, performs spatiotemporal continuity detection, merges isolated points, identifies service hotspot areas, and quantifies and classifies load intensity. The division of network regions is determined by the number of hops in the inter-satellite links. If two satellites communicate with each other through ≤2 hops, they are divided into the same network region. If two satellites communicate with each other through >2 hops, they are divided into different network regions. The spatiotemporal continuity detection determines whether points within the same segmented network region are within a continuous time window. If a hotspot is under the same load condition 80% of the time, it is considered a stable hotspot and merged into the same load intensity level; otherwise, no adjustment is made. The load intensity grading quantification is evenly divided into ranges based on the number of levels set for the task, and the processing strategy is divided into acceptable migration, monitoring status, and triggered migration.

Images