Method and system for dynamic deployment of controllers in software-defined medium and low orbit satellite networks

By calculating the topological change time and predicting load in the medium and low-orbit satellite network, and dynamically adjusting the number and position of the controllers, the problems of limited response capabilities and high control overhead in the existing technology are solved, and efficient control and load prediction are achieved.

CN114374427BActive Publication Date: 2024-11-29SUZHOU QUANSPATIOTEMPORAL INFORMATION TECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202210061886.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-01-19
Publication Date
2024-11-29
Estimated Expiration
2042-01-19

AI Technical Summary

Technical Problem

The prior art cannot realize globally optimized controller dynamic deployment in software-defined medium and low-orbit satellite networks, resulting in limited network response capabilities and high control overhead, and it is impossible to dynamically adjust the number and position of controllers according to real-time network status.

Method used

By calculating the topological change time of the medium and low-orbit satellite network, dividing the topological stability period, predicting the load of satellite and inter-satellite links, building an initial control domain, quantifying the control overhead and efficiency, dynamically adjusting the number, position and control relationship of controllers, and using depth Neural networks and Louvain methods optimize controller deployment strategies, leveraging the regularization migration overhead of relative entropy functions.

Benefits of technology

It realizes dynamic adjustment of the controller adaptively according to the network load and topology, reduces the management and control overhead, improves the management and control efficiency, and ensures efficient transmission and load prediction accuracy of the network.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114374427B_ABST
    Figure CN114374427B_ABST
Patent Text Reader

Abstract

The present invention provides a software-defined method and system for dynamic deployment of controllers in a medium- and low-orbit satellite network, including: calculating the time when the topology change of the medium- and low-orbit satellite network occurs; dividing the satellite network operation cycle into a series of topological stable periods, and calculating the number of data transmission hops between each pair of satellites in the next topological stable period; respectively predicting the loads generated by the ground and the satellite, and predicting the actual load of each satellite and the actual load of each intersatellite link in the next topological stable period; constructing an initial control domain, quantifying the control overhead and control efficiency, and calculating the position, quantity, and control relationship of the satellite controller in the next topological stable period; uploading the calculation results to the satellite, turning on or off the controller according to the calculation results, and the satellite migrating to the corresponding controller according to the control relationship. The present invention combines satellite load and movement to achieve more accurate load prediction accuracy, and can ensure that the total control overhead within the complete network operation cycle is bounded.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of satellite network technology, and in particular to a method and system for dynamically deploying controllers in a software-defined medium- and low-orbit satellite network. Background Art

[0002] In software-defined low- and medium-orbit (NGEO: Non-geostationary) satellite networks, data is transmitted according to the flow table on the node. Since the flow table of each node is requested from the controller, excessive control delays will fundamentally limit the network's ability to respond to emergencies. For example, when the network is congested, if the flow table request time is long, it will lead to a lag in the adjustment of the sending strategy, which will further aggravate the link congestion. Therefore, an optimized controller deployment strategy is the basis for ensuring efficient network transmission, and it needs to be able to dynamically adjust according to the real-time network status during network operation.

[0003] Existing methods cannot achieve global optimization in dynamic software-defined satellite networks. Specifically, the method for controller deployment in ground networks is to place the controller in a fixed position. When the satellite network topology and load distribution change, this deployment strategy will no longer apply. At the same time, existing research on the dynamic deployment of NGEO satellite network controllers has adopted a similar method to that of ground networks, that is, greedily optimizing the control overhead of each time slot in each time slot, so that the satellite has to frequently migrate from one controller to another, introducing a higher migration overhead, which in turn leads to a higher global total overhead. For example, relevant scholars (Sahand Torkamani-Azara, Mohsen Jahanshahib. A newGSO based method for SDN controller placement. Computer Communications, Vol. 163, 2020, pp. 91-108) proposed an efficient ground network controller deployment method, taking into account the heterogeneous computing power of controllers, modeling the ground controller deployment problem as a 0-1 knapsack problem, and proposing a heuristic method to solve the optimal controller deployment location. In addition, Liu Zhiguo et al. (Liu Zhiguo, Lu Meiling, Li Hui, Liu Qingli. Research on multi-controller deployment method of satellite network based on SDN, Computer Simulation, 37(4), Issue 4, 2020, pp. 62-66) proposed a dynamic deployment method of satellite network controllers based on genetic algorithm. First, the satellite motion cycle is divided into time windows, and the controller deployment strategy of each time window is greedily calculated considering the delay and controller load balancing.

[0004] In a Chinese patent document with publication number CN112817605A, a software-defined satellite network controller deployment method, apparatus and related equipment are disclosed. The method constructs a software-defined satellite network architecture, obtains the network state delay of the software-defined satellite network architecture under fault conditions, calculates the optimal solution of the network state delay and the corresponding controller deployment plan based on a simulated annealing algorithm, and finally deploys the controller in the software-defined satellite network architecture based on the controller deployment plan.

[0005] In view of the shortcomings of existing methods, this paper proposes an efficient dynamic deployment method for NGEO satellite network controllers. This method jointly considers satellite motion and load dynamics, adaptively adjusts the number, location and control relationship of satellite controllers, and uses regularization methods to avoid frequent migration of satellites, thereby minimizing the global total control overhead while ensuring control efficiency.

[0006] Patent document CN113595613A (application number: CN202110727381.3) discloses a controller deployment method for a low-orbit software-defined satellite network, the method comprising: selecting multiple preset controller positions within a satellite constellation operation cycle, and deploying SDN controllers at multiple preset controller positions respectively; based on the static deployment method of the SDN controller, the dynamic allocation relationship between the SDN switch and the SDN controller is calculated according to the satellite network topology change, and the SDN switch is allocated to the SDN controller according to the dynamic allocation relationship. However, the invention cannot dynamically and adaptively adjust the number, location, and control relationship of the controller according to the network load and real-time topology. In short, the existing work has brought a large network management overhead in medium and low-orbit satellite networks. Summary of the invention

[0007] In view of the defects in the prior art, the object of the present invention is to provide a method and system for dynamic deployment of controllers in a software-defined medium and low-orbit satellite network.

[0008] A method for dynamically deploying a controller in a software-defined medium- and low-orbit satellite network according to the present invention includes:

[0009] Step S1: Calculate the time when the topology change of the medium and low orbit satellite network occurs according to the orbit parameters of the medium and low orbit satellites;

[0010] Step S2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, and the number of data transmission hops between each pair of satellites in the next topology stable period is calculated;

[0011] Step S3: respectively predict the loads generated by the ground and the satellite, and predict the actual load of each satellite and the actual load of each inter-satellite link in the next topology stability period;

[0012] Step S4: construct an initial control domain, quantify the control overhead and control efficiency according to the satellite load, inter-satellite link load and satellite migration process, and calculate the satellite controller position, quantity, and control relationship in the next topology stability period;

[0013] Step S5: Upload the calculation results to the satellite, turn on or off the controller according to the calculation results, and the satellite migrates to the corresponding controller according to the control relationship.

[0014] Preferably, in step S1:

[0015] According to the orbital parameters of medium and low orbit satellites, the time when the satellite enters and leaves the polar region is calculated. Since the inter-orbit inter-satellite links of the medium and low orbit satellite constellation are disconnected when passing through the polar region, this time is the time when the topology change of the medium and low orbit satellite network occurs, where the topology change refers to the disconnection or reconstruction of the inter-satellite links;

[0016] The topology change time refers to the disconnection of the inter-orbit intersatellite links of the medium and low orbit satellite network when passing through the polar region. The topology change time is calculated by the time it takes for the satellite to enter and exit the polar region;

[0017] In step S2:

[0018] The topology stability period is the longest time window in which the topology does not change. The number of data transmission hops is the distance, and the time interval between two consecutive topology changes is called the topology stability period.

[0019] Preferably, in step S3:

[0020] The method for predicting satellite load and intersatellite link load includes the following steps:

[0021] Step S3.1: Divide the surface evenly into N lat ×N lon Geographical area, N lat is the number of equal latitude divisions, N lon is the number of equal divisions of longitude;

[0022] Step S3.2: construct a deep neural network for each geographic region, satellite, and each intersatellite link;

[0023] The deep neural network includes two parts, the first part is a long short-term memory network, and the second part is a fully connected neural network. The historical load information is embedded in the fully connected neural network together with other features through the long short-term memory network. The other features include the current time and the current weather conditions.

[0024] Step S3.3: Calculate the total satellite load in the next topological stability period, which is equal to the sum of the loads generated by the ground area corresponding to the satellite plus the load generated by the satellite itself.

[0025] Preferably, in step S4:

[0026] The satellite controller deployment strategy calculation method based on Louvain method includes the following steps:

[0027] Step S4.1: define satellites with computing capabilities and installed preset systems as controller candidates, initialize all controller candidates as satellite controllers, and define the remaining nodes as ordinary satellite nodes;

[0028] Step S4.2: Each controller and all nodes controlled by the controller constitute a control domain. The distance from the node to the control domain is defined as the minimum number of hops from the node to the nodes in the control domain. The ordinary satellite nodes are placed in the control domain of the nearest controller to complete the initial control domain construction. All node attributes are removed and all nodes are defined as ordinary satellite nodes.

[0029] Step S4.3: Calculate the overall network control overhead and control efficiency, and regularize the impact of current decisions on future overhead to ensure the total control overhead within the complete network operation cycle;

[0030] Step S4.4: Calculate the overall network overhead and control efficiency after migrating each node to its adjacent control domain, and calculate the performance improvement brought by the migration;

[0031] Step S4.5: If there is a positive benefit migration, execute the migration that brings the highest performance improvement, and jump to step S4.3;

[0032] Step S4.6: If there is no positive benefit migration, construct a topological aggregation graph, where each point in the graph is a control domain and each edge represents the aggregation of cross-domain links;

[0033] Step S4.7: If the topology aggregation graph is not equal to the original topology graph, jump to step S4.4;

[0034] Step S4.8: If the topology aggregation graph is equal to the original topology graph, calculate the optimal controller position for each control domain, and have the optimal controller control all other satellite nodes in the domain.

[0035] Preferably, in step S4:

[0036] In step S4.2:

[0037] The distance between a node and a control domain is the minimum number of hops from the node to the nodes in the control domain. It is calculated as follows: u,v is the number of hops from node u to node v, then node u to the control domain The distance is

[0038] In step S4.3:

[0039] The control overhead includes satellite migration overhead and routing request overhead, where: the migration overhead of each satellite is equal to the control signaling transmission volume introduced by the satellite during the migration process, and the overall satellite migration overhead is equal to the sum of all satellite migration overheads; the routing request overhead of each satellite is equal to the satellite load multiplied by the control signaling transmission volume introduced by each routing request, and the total routing request overhead is equal to the sum of all satellite routing request overheads; the satellite control efficiency is quantified using the modularity of the network, which is equal to the sum of the load within the domain;

[0040] The regularization method based on the relative entropy function means that in the migration cost, whether the satellite u is migrated to the controller v can be expressed as in is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. is whether satellite u is controlled by controller v in the t-1th topological stability period, () + It is a positive function that returns the input value when the input value is positive, otherwise it outputs 0;

[0041] In order to regularize the impact of current decisions on future migration costs, the relative entropy function is used replace Ensure that the total migration cost within the complete satellite network operation cycle is bounded;

[0042] In step S4.6:

[0043] The construction method of the topological aggregation graph is to define the topological graphs before and after aggregation as G(V, E) and G′(V′, E′), where V and E represent the point set and edge set of the topological graph respectively. The original topological graph G is divided into N non-intersecting control domains, and the i-th control domain is defined as D i , define the jth control domain as D j ,When aggregating, for each control domain D i Build Node For control domain D i and control domain D j , D i Midpoint to D j All links of the middle node are aggregated into links That is, from arrive link, and D j Midpoint to D i All links of the middle node are aggregated into links The link load is the sum of all aggregated link loads.

[0044] A software-defined controller dynamic deployment system in a medium- and low-orbit satellite network provided by the present invention includes:

[0045] Module M1: Calculate the time when the topology change of the medium and low orbit satellite network occurs according to the orbital parameters of the medium and low orbit satellites;

[0046] Module M2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, and the number of data transmission hops between each pair of satellites in the next topology stable period is calculated;

[0047] Module M3: predicts the loads generated by the ground and satellites respectively, and predicts the actual load of each satellite and the actual load of each intersatellite link in the next topology stability period;

[0048] Module M4: Construct the initial control domain, quantify the control overhead and control efficiency according to the satellite load, inter-satellite link load and satellite migration process, and calculate the location, number and control relationship of satellite controllers in the next topology stability period;

[0049] Module M5: uploads the calculation results to the satellite, turns on or off the controller according to the calculation results, and the satellite migrates to the corresponding controller according to the control relationship.

[0050] Preferably, in the module M1:

[0051] According to the orbital parameters of medium and low orbit satellites, the time when the satellite enters and leaves the polar region is calculated. Since the inter-orbit inter-satellite links of the medium and low orbit satellite constellation are disconnected when passing through the polar region, this time is the time when the topology change of the medium and low orbit satellite network occurs, where the topology change refers to the disconnection or reconstruction of the inter-satellite links;

[0052] The topology change time refers to the disconnection of the inter-orbit intersatellite links of the medium and low orbit satellite network when passing through the polar region. The topology change time is calculated by the time it takes for the satellite to enter and exit the polar region;

[0053] In the module M2:

[0054] The topology stability period is the longest time window in which the topology does not change. The number of data transmission hops is the distance, and the time interval between two consecutive topology changes is called the topology stability period.

[0055] Preferably, in the module M3:

[0056] The method for predicting satellite load and intersatellite link load includes the following steps:

[0057] Module M3.1: Divide the surface of the earth evenly according to longitude and latitude lat ×N lon Geographical area, N lat is the number of equal latitude divisions, N lonis the number of equal divisions of longitude;

[0058] Module M3.2: Build a deep neural network for each geographic region, satellite, and each intersatellite link;

[0059] The deep neural network includes two parts, the first part is a long short-term memory network, and the second part is a fully connected neural network. The historical load information is embedded in the fully connected neural network together with other features through the long short-term memory network. The other features include the current time and the current weather conditions.

[0060] Module M3.3: Calculate the total satellite load in the next topological stability period, which is equal to the sum of the loads generated by the ground area corresponding to the satellite plus the load generated by the satellite itself.

[0061] Preferably, in the module M4:

[0062] The satellite controller deployment strategy calculation method based on Louvain method includes the following modules:

[0063] Module M4.1: Define satellites with computing capabilities and installed preset systems as controller candidates, initialize all controller candidates as satellite controllers, and define the remaining nodes as ordinary satellite nodes;

[0064] Module M4.2: Each controller and all nodes controlled by the controller constitute a control domain. The distance from a node to the control domain is defined as the minimum number of hops from the node to the nodes in the control domain. Ordinary satellite nodes are placed in the control domain of the nearest controller to complete the initial control domain construction. All node attributes are removed and all nodes are defined as ordinary satellite nodes.

[0065] Module M4.3: Calculate the overall network control overhead and efficiency, and regularize the impact of current decisions on future overhead to ensure the total control overhead within the complete network operation cycle;

[0066] Module M4.4: Calculate the overall network overhead and control efficiency after migrating each node to its adjacent control domain, and calculate the performance improvement brought by the migration;

[0067] Module M4.5: If there is a positive benefit migration, execute the migration that brings the highest performance improvement, and jump to run module M4.3;

[0068] Module M4.6: If there is no positive benefit migration, construct a topological aggregation graph, where each point is a control domain and each edge represents the aggregation of cross-domain links;

[0069] Module M4.7: If the topology aggregation graph is not equal to the original topology graph, jump to run module M4.4;

[0070] Module M4.8: If the topology aggregation graph is equal to the original topology graph, calculate the optimal controller position for each control domain, and the optimal controller controls all other satellite nodes in the domain.

[0071] Preferably, in the module M4:

[0072] In the module M4.2:

[0073] The distance between a node and a control domain is the minimum number of hops from the node to the nodes in the control domain. It is calculated as follows: u,v is the number of hops from node u to node v, then node u to the control domain The distance is

[0074] In the module M4.3:

[0075] The control overhead includes satellite migration overhead and routing request overhead, where: the migration overhead of each satellite is equal to the control signaling transmission volume introduced by the satellite during the migration process, and the overall satellite migration overhead is equal to the sum of all satellite migration overheads; the routing request overhead of each satellite is equal to the satellite load multiplied by the control signaling transmission volume introduced by each routing request, and the total routing request overhead is equal to the sum of all satellite routing request overheads; the satellite control efficiency is quantified using the modularity of the network, which is equal to the sum of the load within the domain;

[0076] The regularization method based on the relative entropy function means that in the migration cost, whether the satellite u is migrated to the controller v can be expressed as in is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. is whether satellite u is controlled by controller v in the t-1th topological stability period, () + It is a positive function that returns the input value when the input value is positive, otherwise it outputs 0;

[0077] In order to regularize the impact of current decisions on future migration costs, the relative entropy function is used replace Ensure that the total migration cost within the complete satellite network operation cycle is bounded;

[0078] In the module M4.6:

[0079] The construction method of the topological aggregation graph is to define the topological graphs before and after aggregation as C(V, E) and G′(V′, E′), where V and E represent the point set and edge set of the topological graph respectively. The original topological graph G is divided into N non-intersecting control domains, and the i-th control domain is defined as D i , define the jth control domain as D j,When aggregating, for each control domain D i Build Node For control domain D i and control domain D j , D i Midpoint to D j All links of the middle node are aggregated into links That is, from arrive link, and D j Midpoint to D i All links of the middle node are aggregated into links The link load is the sum of all aggregated link loads.

[0080] Compared with the prior art, the present invention has the following beneficial effects:

[0081] 1. The present invention can dynamically and adaptively adjust the number, location and control relationship of controllers according to network load and real-time topology, greatly reducing the control overhead while ensuring efficient control, and providing guarantee for efficient transmission of satellite networks and integrated space-ground networks;

[0082] 2. The present invention combines historical load and satellite motion to achieve more accurate load prediction accuracy;

[0083] 3. The present invention uses the relative entropy function to regularize the impact of the deployment strategy on the migration overhead, which can ensure that the total control overhead within the complete network operation cycle is bounded. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0085] Figure 1 The present invention provides a controller dynamic deployment method process for software-defined NGEO satellite networks;

[0086] Figure 2 A load prediction model according to an embodiment of the present invention is shown;

[0087] Figure 3 This is a schematic diagram of the control domain adjustment process in an embodiment of the present invention;

[0088] Figure 4 The invention is a schematic diagram of a controller dynamic deployment system for a software-defined NGEO satellite network in an embodiment of the present invention. DETAILED DESCRIPTION

[0089] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0090] Embodiment 1:

[0091] According to a software-defined method for dynamically deploying controllers in a medium- and low-orbit satellite network provided by the present invention, Figure 1-Figure 4 As shown, including:

[0092] Step S1: Calculate the time when the topology change of the medium and low orbit satellite network occurs according to the orbit parameters of the medium and low orbit satellites;

[0093] Step S2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, and the number of data transmission hops between each pair of satellites in the next topology stable period is calculated;

[0094] Step S3: respectively predict the loads generated by the ground and the satellite, and predict the actual load of each satellite and the actual load of each inter-satellite link in the next topology stability period;

[0095] Step S4: construct an initial control domain, quantify the control overhead and control efficiency according to the satellite load, inter-satellite link load and satellite migration process, and calculate the satellite controller position, quantity, and control relationship in the next topology stability period;

[0096] Step S5: Upload the calculation results to the satellite, turn on or off the controller according to the calculation results, and the satellite migrates to the corresponding controller according to the control relationship.

[0097] Specifically, in step S1:

[0098] According to the orbital parameters of medium and low orbit satellites, the time when the satellite enters and leaves the polar region is calculated. Since the inter-orbit inter-satellite links of the medium and low orbit satellite constellation are disconnected when passing through the polar region, this time is the time when the topology change of the medium and low orbit satellite network occurs, where the topology change refers to the disconnection or reconstruction of the inter-satellite links;

[0099] The topology change time refers to the disconnection of the inter-orbit intersatellite links of the medium and low orbit satellite network when passing through the polar region. The topology change time is calculated by the time it takes for the satellite to enter and exit the polar region;

[0100] In step S2:

[0101] The topology stability period is the longest time window in which the topology does not change. The number of data transmission hops is the distance, and the time interval between two consecutive topology changes is called the topology stability period.

[0102] Specifically, in step S3:

[0103] The method for predicting satellite load and intersatellite link load includes the following steps:

[0104] Step S3.1: Divide the surface evenly into N lat ×N lon Geographical area, N lat is the number of equal latitude divisions, N lon is the number of equal divisions of longitude;

[0105] Step S3.2: construct a deep neural network for each geographic region, satellite, and each intersatellite link;

[0106] The deep neural network includes two parts, the first part is a long short-term memory network, and the second part is a fully connected neural network. The historical load information is embedded in the fully connected neural network together with other features through the long short-term memory network. The other features include the current time and the current weather conditions.

[0107] Step S3.3: Calculate the total satellite load in the next topological stability period, which is equal to the sum of the loads generated by the ground area corresponding to the satellite plus the load generated by the satellite itself.

[0108] Specifically, in step S4:

[0109] The satellite controller deployment strategy calculation method based on Louvain method includes the following steps:

[0110] Step S4.1: define satellites with computing capabilities and installed preset systems as controller candidates, initialize all controller candidates as satellite controllers, and define the remaining nodes as ordinary satellite nodes;

[0111] Step S4.2: Each controller and all nodes controlled by the controller constitute a control domain. The distance from the node to the control domain is defined as the minimum number of hops from the node to the nodes in the control domain. The ordinary satellite nodes are placed in the control domain of the nearest controller to complete the initial control domain construction. All node attributes are removed and all nodes are defined as ordinary satellite nodes.

[0112] Step S4.3: Calculate the overall network control overhead and control efficiency, and regularize the impact of current decisions on future overhead to ensure the total control overhead within the complete network operation cycle;

[0113] Step S4.4: Calculate the overall network overhead and control efficiency after migrating each node to its adjacent control domain, and calculate the performance improvement brought by the migration;

[0114] Step S4.5: If there is a positive benefit migration, execute the migration that brings the highest performance improvement, and jump to step S4.3;

[0115] Step S4.6: If there is no positive benefit migration, construct a topological aggregation graph, where each point in the graph is a control domain and each edge represents the aggregation of cross-domain links;

[0116] Step S4.7: If the topology aggregation graph is not equal to the original topology graph, jump to step S4.4;

[0117] Step S4.8: If the topology aggregation graph is equal to the original topology graph, calculate the optimal controller position for each control domain, and have the optimal controller control all other satellite nodes in the domain.

[0118] Specifically, in step S4:

[0119] In step S4.2:

[0120] The distance between a node and a control domain is the minimum number of hops from the node to the nodes in the control domain. It is calculated as follows: u,v is the number of hops from node u to node v, then node u to the control domain The distance is

[0121] In step S4.3:

[0122] The control overhead includes satellite migration overhead and routing request overhead, where: the migration overhead of each satellite is equal to the control signaling transmission volume introduced by the satellite during the migration process, and the overall satellite migration overhead is equal to the sum of all satellite migration overheads; the routing request overhead of each satellite is equal to the satellite load multiplied by the control signaling transmission volume introduced by each routing request, and the total routing request overhead is equal to the sum of all satellite routing request overheads; the satellite control efficiency is quantified using the modularity of the network, which is equal to the sum of the load within the domain;

[0123] The regularization method based on the relative entropy function means that in the migration cost, whether the satellite u is migrated to the controller v can be expressed as in is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. is whether satellite u is controlled by controller v in the t-1th topological stability period, () + It is a positive function that returns the input value when the input value is positive, otherwise it outputs 0;

[0124] In order to regularize the impact of current decisions on future migration costs, the relative entropy function is used replace Ensure that the total migration cost within the complete satellite network operation cycle is bounded;

[0125] In step S4.6:

[0126] The construction method of the topological aggregation graph is to define the topological graphs before and after aggregation as C(V, E) and G′(V′, E′), where V and E represent the point set and edge set of the topological graph respectively. The original topological graph G is divided into N non-intersecting control domains, and the i-th control domain is defined as D i , define the jth control domain as D j ,When aggregating, for each control domain D i Build Node For control domain D i and control domain D j , D i Midpoint to D j All links of the middle node are aggregated into links That is, from arrive link, and D j Midpoint to D i All links of the middle node are aggregated into links The link load is the sum of all aggregated link loads.

[0127] Embodiment 2:

[0128] Embodiment 2 is a preferred example of Embodiment 1, and is used to illustrate the present invention in more detail.

[0129] Those skilled in the art may understand the software-defined method for dynamic deployment of controllers in low- and medium-orbit satellite networks provided by the present invention as a specific implementation method of a software-defined system for dynamic deployment of controllers in low- and medium-orbit satellite networks, that is, the software-defined system for dynamic deployment of controllers in low- and medium-orbit satellite networks may be implemented by executing the step flow of the software-defined method for dynamic deployment of controllers in low- and medium-orbit satellite networks.

[0130] A software-defined controller dynamic deployment system in a medium- and low-orbit satellite network provided by the present invention includes:

[0131] Module M1: Calculate the time when the topology change of the medium and low orbit satellite network occurs according to the orbital parameters of the medium and low orbit satellites;

[0132] Module M2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, and the number of data transmission hops between each pair of satellites in the next topology stable period is calculated;

[0133] Module M3: predicts the loads generated by the ground and satellites respectively, and predicts the actual load of each satellite and the actual load of each intersatellite link in the next topology stability period;

[0134] Module M4: Construct the initial control domain, quantify the control overhead and control efficiency according to the satellite load, inter-satellite link load and satellite migration process, and calculate the location, number and control relationship of satellite controllers in the next topology stability period;

[0135] Module M5: uploads the calculation results to the satellite, turns on or off the controller according to the calculation results, and the satellite migrates to the corresponding controller according to the control relationship.

[0136] Specifically, in the module M1:

[0137] According to the orbital parameters of medium and low orbit satellites, the time when the satellite enters and leaves the polar region is calculated. Since the inter-orbit inter-satellite links of the medium and low orbit satellite constellation are disconnected when passing through the polar region, this time is the time when the topology change of the medium and low orbit satellite network occurs, where the topology change refers to the disconnection or reconstruction of the inter-satellite links;

[0138] The topology change time refers to the disconnection of the inter-orbit intersatellite links of the medium and low orbit satellite network when passing through the polar region. The topology change time is calculated by the time it takes for the satellite to enter and exit the polar region;

[0139] In the module M2:

[0140] The topology stability period is the longest time window in which the topology does not change. The number of data transmission hops is the distance, and the time interval between two consecutive topology changes is called the topology stability period.

[0141] Specifically, in the module M3:

[0142] The method for predicting satellite load and intersatellite link load includes the following steps:

[0143] Module M3.1: Divide the surface of the earth evenly according to longitude and latitude lat ×N lon Geographical area, N lat is the number of equal latitude divisions, N lon is the number of equal divisions of longitude;

[0144] Module M3.2: Build a deep neural network for each geographic region, satellite, and each intersatellite link;

[0145] The deep neural network includes two parts, the first part is a long short-term memory network, and the second part is a fully connected neural network. The historical load information is embedded in the fully connected neural network together with other features through the long short-term memory network. The other features include the current time and the current weather conditions.

[0146] Module M3.3: Calculate the total satellite load in the next topological stability period, which is equal to the sum of the loads generated by the ground area corresponding to the satellite plus the load generated by the satellite itself.

[0147] Specifically, in the module M4:

[0148] The satellite controller deployment strategy calculation method based on Louvain method includes the following modules:

[0149] Module M4.1: Define satellites with computing capabilities and installed preset systems as controller candidates, initialize all controller candidates as satellite controllers, and define the remaining nodes as ordinary satellite nodes;

[0150] Module M4.2: Each controller and all nodes controlled by the controller constitute a control domain. The distance from a node to the control domain is defined as the minimum number of hops from the node to the nodes in the control domain. Ordinary satellite nodes are placed in the control domain of the nearest controller to complete the initial control domain construction. All node attributes are removed and all nodes are defined as ordinary satellite nodes.

[0151] Module M4.3: Calculate the overall network control overhead and efficiency, and regularize the impact of current decisions on future overhead to ensure the total control overhead within the complete network operation cycle;

[0152] Module M4.4: Calculate the overall network overhead and control efficiency after migrating each node to its adjacent control domain, and calculate the performance improvement brought by the migration;

[0153] Module M4.5: If there is a positive benefit migration, execute the migration that brings the highest performance improvement, and jump to run module M4.3;

[0154] Module M4.6: If there is no positive benefit migration, construct a topological aggregation graph, where each point is a control domain and each edge represents the aggregation of cross-domain links;

[0155] Module M4.7: If the topology aggregation graph is not equal to the original topology graph, jump to run module M4.4;

[0156] Module M4.8: If the topology aggregation graph is equal to the original topology graph, calculate the optimal controller position for each control domain, and the optimal controller controls all other satellite nodes in the domain.

[0157] Specifically, in the module M4:

[0158] In the module M4.2:

[0159] The distance between a node and a control domain is the minimum number of hops from the node to the nodes in the control domain. It is calculated as follows: u,vis the number of hops from node u to node v, then node u to the control domain The distance is

[0160] In the module M4.3:

[0161] The control overhead includes satellite migration overhead and routing request overhead, where: the migration overhead of each satellite is equal to the control signaling transmission volume introduced by the satellite during the migration process, and the overall satellite migration overhead is equal to the sum of all satellite migration overheads; the routing request overhead of each satellite is equal to the satellite load multiplied by the control signaling transmission volume introduced by each routing request, and the total routing request overhead is equal to the sum of all satellite routing request overheads; the satellite control efficiency is quantified using the modularity of the network, which is equal to the sum of the load within the domain;

[0162] The regularization method based on the relative entropy function means that in the migration cost, whether the satellite u is migrated to the controller v can be expressed as in is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. is whether satellite u is controlled by controller v in the t-1th topological stability period, () + It is a positive function that returns the input value when the input value is positive, otherwise it outputs 0;

[0163] In order to regularize the impact of current decisions on future migration costs, the relative entropy function is used replace Ensure that the total migration cost within the complete satellite network operation cycle is bounded;

[0164] In the module M4.6:

[0165] The construction method of the topological aggregation graph is to define the topological graphs before and after aggregation as G(V, E) and G′(V′, E′), where V and E represent the point set and edge set of the topological graph respectively. The original topological graph G is divided into N non-intersecting control domains, and the i-th control domain is defined as D i , define the jth control domain as D j ,When aggregating, for each control domain D i Build Node For control domain D i and control domain D j , D i Midpoint to D j All links of the middle node are aggregated into links That is, from arrive link, and D j Midpoint to D iAll links of the middle node are aggregated into links The link load is the sum of all aggregated link loads.

[0166] Embodiment 3:

[0167] Embodiment 3 is a preferred example of Embodiment 1, and is used to illustrate the present invention in more detail.

[0168] A method for dynamically deploying a controller in a software-defined NGEO satellite network comprises the following steps:

[0169] Step S1: Calculate the time when the satellite enters / leaves the polar region according to the NGEO satellite orbit parameters. Since the inter-orbit inter-satellite links of the NGEO satellite constellation are disconnected when passing through the polar region, this time is the time when the topology change of the NGEO satellite network occurs, where the topology change refers to the disconnection or reconstruction of the inter-satellite links.

[0170] Step S2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, that is, the longest time window in which the topology does not change, and the number of data transmission hops (i.e., distance) between each pair of satellites in the next topology stable period is calculated, where the time interval between two consecutive topology changes is called the topology stable period;

[0171] Step S3: The ground management layer decouples the effects of satellite motion and ground activities on satellite loads, and predicts the loads generated by the ground and satellites based on the long short-term memory network. Then, combined with the satellite motion law, it predicts the actual load of each satellite and the actual load of each inter-satellite link in the next topological stability period.

[0172] Step S4: The ground management layer constructs an initial control domain according to satellite capabilities, jointly considers satellite load, inter-satellite link load and satellite migration process, quantifies control overhead and control efficiency, and calculates the satellite controller position, quantity and control relationship in the next topology stability period based on the Louvain algorithm with the goal of ensuring control efficiency while minimizing control overhead;

[0173] In step S5, the ground management layer uploads the calculation results to each satellite, each controller is turned on or off according to the calculation results, and each satellite is migrated to the corresponding controller according to the control relationship.

[0174] The method for predicting satellite load and intersatellite link load in step S3 comprises the following steps:

[0175] Step S3.1: Divide the surface evenly into N lat ×N lon Geographical area, N lat is the number of equal latitude divisions, N lon is the number of equal divisions of longitude;

[0176] Step S3.2: construct a deep neural network for each geographic region, satellite, and each intersatellite link;

[0177] Step S3.3: Calculate the total satellite load in the next topological stability period, which is equal to the sum of the loads generated by the ground area corresponding to the satellite plus the load generated by the satellite itself.

[0178] The deep neural network in step S3.2 includes two parts, the first part is a long short-term memory network, and the second part is a fully connected neural network. The historical load information is embedded in the fully connected neural network together with other features through the long short-term memory network. The other features include but are not limited to the current time, current weather conditions, etc.;

[0179] The satellite controller deployment strategy calculation method based on the Louvain method in step S4 comprises the following steps:

[0180] Step S4.1: define satellites with computing capabilities and installed with the necessary systems as controllers as candidate controllers, initialize all candidate controllers as satellite controllers, and the remaining nodes are called ordinary satellite nodes;

[0181] Step S4.2: Each controller and all its managed nodes constitute a management domain. The distance from a node to a management domain is defined as the minimum number of hops from a node to a node in the management domain. Ordinary satellite nodes are placed in the nearest controller management domain to complete the initial management domain construction. All node attributes are removed, that is, all nodes are regarded as ordinary satellite nodes.

[0182] Step S4.3: Calculate the overall network control overhead and control efficiency, and use the relative entropy function to regularize the impact of current decisions on future overhead to ensure the total control overhead within the complete network operation cycle;

[0183] Step S4.4: Calculate the overall network overhead and control efficiency after migrating each node to its adjacent control domain, and calculate the performance improvement brought by the migration;

[0184] Step S4.5: If there is a positive benefit migration, execute the migration that brings the highest performance improvement, and return to step S4.3;

[0185] Step S4.6: construct a topological aggregation graph, where each point in the graph is a control domain and each edge represents the aggregation of cross-domain links;

[0186] Step S4.7: If the topology aggregation graph is not equal to the original topology graph, return to step S4.4;

[0187] Step S4.8: Calculate the optimal controller position for each control domain, and have the controller control all other satellite nodes in the domain.

[0188] The control overhead in step S4.3 includes but is not limited to satellite migration overhead and routing request overhead, wherein: each satellite migration overhead is equal to the control signaling transmission amount introduced by the satellite during the migration process, and the overall satellite migration overhead is equal to the sum of all satellite migration overheads; each satellite routing request overhead is equal to the satellite load multiplied by the control signaling transmission amount introduced by each routing request, and the total routing request overhead is equal to the sum of all satellite routing request overheads; the satellite control efficiency is quantified using the modularity of the network, which is equal to the sum of the loads within the domain;

[0189] The regularization method based on relative entropy function in step S4.3 means that in the migration cost, whether satellite u is migrated to controller v can be expressed as in is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. is whether satellite u is controlled by controller v in the t-1th topological stability period, () + It is a positive function, that is, it returns the input value when the input value is positive, otherwise it outputs 0;

[0190] In order to regularize the impact of current decisions on future migration costs, the relative entropy function is used replace This method can ensure that the total migration cost within the complete satellite network operation cycle is bounded;

[0191] The method for constructing the topological aggregation graph in step S4.6 is to define the topological graphs before and after aggregation as G(V, E) and G′(V′, E′), respectively, where V and E represent the point set and edge set of the topological graph, respectively, and the original topological graph G is divided into N non-intersecting control domains, and the i-th control domain is defined as D i ,When aggregating, for each control domain D i Build Node For control domain D i and control domain D j , D i Midpoint to D j All links of the middle node are aggregated into links That is, from arrive link, and D j Midpoint to D i All links of the middle node are aggregated into links The link load is the sum of all aggregated link loads;

[0192] A controller dynamic deployment system in a software-defined medium and low orbit satellite network includes the following modules:

[0193] Module M1: Calculate the time when the topology change of the medium and low orbit satellite network occurs according to the orbital parameters of the medium and low orbit satellites;

[0194] Module M2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, and the number of data transmission hops between each pair of satellites in the next topology stable period is calculated;

[0195] Module M3: predicts the loads generated by the ground and satellites respectively, and predicts the actual load of each satellite and the actual load of each intersatellite link in the next topology stability period;

[0196] Module M4: Construct the initial control domain, quantify the control overhead and control efficiency according to the satellite load, inter-satellite link load and satellite migration process, and calculate the location, number and control relationship of satellite controllers in the next topology stability period;

[0197] Module M5: uploads the calculation results to the satellite, turns on or off the controller according to the calculation results, and the satellite migrates to the corresponding controller according to the control relationship.

[0198] Embodiment 4:

[0199] Embodiment 4 is a preferred example of Embodiment 1, and is used to illustrate the present invention in more detail.

[0200] The present invention provides a method and system for dynamic deployment of controllers in a software-defined medium and low orbit (NGEO) satellite network, including: NEGO satellite network topology calculation, that is, according to the NGEO satellite orbit parameters, the NGEO satellite network topology change time point and the NEGO satellite network topology stable period state are calculated; NGEO satellite network load prediction, that is, the ground management layer predicts the satellite load status in the next topology stable period based on historical information and satellite positions; NGEO satellite network controller optimal strategy calculation, that is, according to the topology information and load information, the ground management layer calculates the number, position and control relationship of satellite controllers in the next topology stable period with the goal of minimizing control overhead while ensuring management efficiency; satellite controller dynamic deployment, that is, the ground management layer uploads the calculation results to each satellite, each controller is turned on or off according to the controller deployment strategy, and each satellite migrates to the corresponding controller. The present invention jointly considers the dynamics of the satellite network time-varying topology and load, adaptively adjusts the number, position and control relationship of controllers, greatly reduces the control overhead while ensuring management efficiency, and provides guarantee for the efficient transmission of data in the NGEO satellite network and the integrated network of the earth and the sky.

[0201] According to a software-defined NGEO satellite network controller dynamic deployment method provided by the present invention, refer to Figure 1 , including the following steps:

[0202] Step S1: Calculate the time when the satellite enters / leaves the polar region based on the NGEO satellite orbit parameters, and on this basis calculate the time when the NGEO satellite network topology change occurs, where the topology change refers to the disconnection or reconstruction of the inter-satellite link;

[0203] Step S2: Based on the NGEO satellite orbit parameters, the number of data transmission hops between each pair of satellites in the next topology stability period is calculated, where the time interval between two consecutive topology changes is called the topology stability period;

[0204] Step S3: The ground management layer decouples the effects of satellite motion and ground activities on satellite loads, and predicts the loads generated by the ground and satellites based on the long short-term memory network. Then, combined with the satellite motion law, it predicts the actual load of each satellite and the actual load of each intersatellite link in the next topological stability period. Compared with the traditional direct prediction method, this prediction method can achieve higher prediction accuracy.

[0205] Step S4: The ground management layer constructs an initial control domain according to satellite capabilities, jointly considers routing requests and satellite migration processes to quantify control overhead and control efficiency, and calculates the satellite controller location, quantity, and control relationship in the next topology stability period based on the Louvain method with the goal of ensuring control efficiency while minimizing control overhead;

[0206] In step S5, the ground management layer uploads the calculation results to each satellite, each controller is turned on or off according to the controller deployment strategy, and each satellite migrates to the corresponding controller according to the control relationship.

[0207] Specifically, the topology change time in step S1 refers to the time when the inter-orbit intersatellite links of the NGEO satellite network are disconnected when passing through the polar region. Therefore, the topology change time can be calculated by the time when the satellite enters / exits the polar region. Specifically, define T m T is the time interval between a satellite entering the polar region and the last time a satellite leaves the polar region, and T is defined as c is the time interval between two consecutive satellites entering the polar region, then the set of time points when the topology changes is {k·T c ,k·T c +T m |k∈[0,+∞)};

[0208] Specifically, the method for predicting satellite load and intersatellite link load in step S3 comprises the following steps:

[0209] Step S3.1: Divide the surface evenly into N lat ×N lon geographical area;

[0210] Step S3.2: construct a deep neural network for each geographic region, satellite, and each intersatellite link;

[0211] Step S3.3: Calculate the total satellite load in the next topology stable period, which is equal to the sum of the loads generated by the satellite coverage area plus the load generated by the satellite itself.

[0212] Specifically, in step S3.1, the ground area division may divide the ground into 24×12 areas in units of 5°×5°;

[0213] Specifically, the deep neural network in step S3.2 includes the following: Figure 2 The two parts shown are: the first part is a long short-term memory network, and the second part is a fully connected neural network. The historical load information is embedded in the fully connected neural network together with other features through the long short-term memory network. The other features include but are not limited to the current time, current weather conditions, etc.;

[0214] Specifically, the satellite controller deployment strategy calculation method based on the Louvain method in step S4 includes the following: Figure 3 The following steps are shown:

[0215] Step S4.1: define satellites with computing capabilities and installed with the necessary systems as controllers as candidate controllers, initialize all candidate controllers as satellite controllers, and the remaining nodes are called ordinary satellite nodes;

[0216] Step S4.2: Each controller and all its managed nodes constitute a management domain. The distance between a node and a management domain is defined as the minimum number of hops from a node to a node in the management domain. Ordinary satellite nodes are placed in the nearest controller management domain to complete the initial management domain construction. All node attributes are removed, that is, all nodes are regarded as ordinary satellite nodes.

[0217] Step S4.3: Calculate the overall network control overhead and control efficiency, and use the relative entropy function to regularize the impact of current decisions on future overhead to ensure the total control overhead within the complete network operation cycle;

[0218] Step S4.4: Calculate the overall network overhead and control efficiency after migrating each node to its adjacent control domain, and calculate the performance improvement brought by the migration;

[0219] Step S4.5: If there is a positive benefit migration, execute the migration that brings the highest performance improvement, and return to step S4.3, such as Figure 3 As shown, moving satellite u to control domain D1 brings the highest performance improvement;

[0220] Step S4.6: Construct a topological aggregation graph, where each point is a control domain and each edge represents the aggregation of cross-domain links. Figure 3As shown, the original topology contains three control domains, which are aggregated into three points;

[0221] Step S4.7: If the topology aggregation graph is not equal to the original topology graph, return to step S4.4;

[0222] Step S4.8: Calculate the optimal controller position for each control domain, and have the controller control all other satellite nodes in the control domain;

[0223] Specifically, the distance between the node and the control domain in step S4.2 is the minimum number of hops from the node to the nodes in the control domain, which is calculated as follows: u,v is the number of hops from node u to node v, then node u to the control domain The distance is

[0224] Specifically, the control overhead in step S4.3 includes but is not limited to satellite migration overhead and routing request overhead, wherein: each satellite migration overhead is equal to the control signaling transmission amount introduced by the satellite during the migration process, and the overall satellite migration overhead is equal to the sum of all satellite migration overheads; each satellite routing request overhead is equal to the satellite load multiplied by the control signaling transmission amount introduced by each routing request, and the total routing request overhead is equal to the sum of all satellite routing request overheads; the satellite control efficiency is quantified using the modularity of the network, which is equal to the sum of the traffic within the domain;

[0225] Specifically, the migration cost from the t-1th topology stable period to the tth topology stable period is calculated as follows: is the cost of migrating satellite u to controller v, is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. Indicates whether satellite u migrates to controller v from the t-1th topology stable period to the yth topology stable period. The migration cost from the t-1th topology stable period to the tth topology stable period is equal to

[0226]

[0227] Specifically, the routing request cost of the tth topology stable period is calculated as follows: t is the length of the tth topological stable period, is the load prediction value of satellite u in t topology stable periods, k is the unit routing request overhead, is the number of hops between satellite u and satellite v in the tth topology stable period, then the routing request overhead in the tth topology stable period is equal to

[0228]

[0229] Specifically, the control efficiency of the tth topology stability period is calculated as follows, where e(m,n) is the link from node m to node n, and is defined as is the load of link e(m,n) in the yth topology stable period, and the control efficiency in the tth topology stable period is equal to

[0230]

[0231] Specifically, the overall control performance in step S4.3 refers to the consideration that the goal of controller deployment is to minimize control overhead while maximizing control efficiency. First, maximizing network efficiency (i.e., modularity) is converted into minimizing inter-domain traffic. Then the overall performance is the weighted sum of control overhead and inter-domain traffic, and λ is defined as the weight. The overall performance is

[0232]

[0233] Specifically, the calculation method of the inter-domain traffic in the tth topology stable period is:

[0234]

[0235] Specifically, the regularization method based on relative entropy function in step S4.3 means that in the migration cost, whether satellite u migrates to controller v can be expressed as in is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. In order to regularize the impact of the current decision on the future migration cost, the relative entropy function is used. replace This method can ensure that the total migration cost within the complete satellite network operation cycle is bounded;

[0236] Dynamic deployment method of controller for software-defined NGEO satellite network: The construction method of the topological aggregation graph in step S4.6 is to define the topological graphs before and after aggregation as G(V, E) and G′(V′, E′), respectively, where V and E represent the point set and edge set of the topological graph, respectively, and the original topological graph G is divided into N non-overlapping control domains, and the i-th control domain is defined as D i ,When aggregating, for each control domain D i Build Node For control domain D i and control domain D j , D i Midpoint to D j All links of the middle node are aggregated into links And D j Midpoint to D i All links of the middle node are aggregated into links The link load is the sum of all aggregated link loads;

[0237] Specifically, in step S4.7, the aggregated topology map is the same as the original topology map if V=V′ and E=E′;

[0238] Specifically, the optimal controller deployment position in step S4.8 refers to the controller deployment position that minimizes the control overhead. Specifically, satellite v is selected as the control domain D i The controller is if and only if in The satellite v is selected as the control domain D in the tth topological stability period i The control overhead of the controller;

[0239] The transport layer adaptive congestion control system for heterogeneous ground-ground integrated networks implemented according to the above-mentioned invention technology is referred to Figure 4 The system includes the following modules:

[0240] Module M1: Calculate the time when the topology change of the medium and low orbit satellite network occurs according to the orbital parameters of the medium and low orbit satellites;

[0241] Module M2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, and the number of data transmission hops between each pair of satellites in the next topology stable period is calculated;

[0242] Module M3: predicts the loads generated by the ground and satellites respectively, and predicts the actual load of each satellite and the actual load of each intersatellite link in the next topology stability period;

[0243] Module M4: Construct the initial control domain, quantify the control overhead and control efficiency according to the satellite load, inter-satellite link load and satellite migration process, and calculate the location, number and control relationship of satellite controllers in the next topology stability period;

[0244] Module M5: uploads the calculation results to the satellite, turns on or off the controller according to the calculation results, and the satellite migrates to the corresponding controller according to the control relationship.

[0245] Those skilled in the art know that, in addition to implementing the system, device and its various modules provided by the present invention in a purely computer-readable program code, it is entirely possible to implement the same program in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. Therefore, the system, device and its various modules provided by the present invention can be considered as a hardware component, and the modules included therein for implementing various programs can also be considered as structures within the hardware component; the modules for implementing various functions can also be considered as both software programs for implementing the method and structures within the hardware component.

[0246] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for dynamic deployment of controllers in a software-defined medium and low orbit satellite network, characterized in that: include: Step S1: Calculate the time when the topology change of the medium and low orbit satellite network occurs according to the orbit parameters of the medium and low orbit satellites; Step S2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, and the number of data transmission hops between each pair of satellites in the next topology stable period is calculated; Step S3: respectively predict the loads generated by the ground and the satellite, and predict the actual load of each satellite and the actual load of each inter-satellite link in the next topology stability period; Step S4: construct an initial control domain, quantify the control overhead and control efficiency according to the satellite load, inter-satellite link load and satellite migration process, and calculate the satellite controller position, quantity, and control relationship in the next topology stability period; Step S5: Upload the calculation results to the satellite, turn on or off the controller according to the calculation results, and the satellite migrates to the corresponding controller according to the control relationship; In step S4: The satellite controller deployment strategy calculation method based on Louvain method includes the following steps: Step S4.1: define satellites with computing capabilities and installed preset systems as controller candidates, initialize all controller candidates as satellite controllers, and define the remaining nodes as ordinary satellite nodes; Step S4.2: Each controller and all nodes controlled by the controller constitute a control domain. The distance from the node to the control domain is defined as the minimum number of hops from the node to the nodes in the control domain. The ordinary satellite nodes are placed in the control domain of the nearest controller to complete the initial control domain construction. All node attributes are removed and all nodes are defined as ordinary satellite nodes. Step S4.3: Calculate the overall network control overhead and control efficiency, and regularize the impact of current decisions on future overhead to ensure the total control overhead within the complete network operation cycle; Step S4.4: Calculate the overall network overhead and control efficiency after migrating each node to its adjacent control domain, and calculate the performance improvement brought by the migration; Step S4.5: If there is a positive benefit migration, execute the migration that brings the highest performance improvement, and jump to step S4.3; Step S4.6: If there is no positive benefit migration, construct a topological aggregation graph, where each point in the graph is a control domain and each edge represents the aggregation of cross-domain links; Step S4.7: If the topology aggregation graph is not equal to the original topology graph, jump to step S4.4; Step S4.8: If the topology aggregation graph is equal to the original topology graph, calculate the optimal controller position for each control domain, and have the optimal controller control all other satellite nodes in the domain.

2. The method for dynamic deployment of controllers in a software-defined medium and low orbit satellite network according to claim 1, characterized in that: In step S1: According to the orbital parameters of medium and low orbit satellites, the time when the satellite enters and leaves the polar region is calculated. Since the inter-orbit inter-satellite links of the medium and low orbit satellite constellation are disconnected when passing through the polar region, this time is the time when the topology change of the medium and low orbit satellite network occurs, where the topology change refers to the disconnection or reconstruction of the inter-satellite links; The topology change time refers to the disconnection of the inter-orbit intersatellite links of the medium and low orbit satellite network when passing through the polar region. The topology change time is calculated by the time it takes for the satellite to enter and exit the polar region; In step S2: The topology stability period is the longest time window in which the topology does not change. The number of data transmission hops is the distance, and the time interval between two consecutive topology changes is called the topology stability period.

3. The method for dynamic deployment of controllers in a software-defined medium and low orbit satellite network according to claim 1, characterized in that: In step S3: The method for predicting satellite load and intersatellite link load includes the following steps: Step S3.1: Divide the surface evenly into N lat ×N lon Geographical area, N lat is the number of equal latitude divisions, N lon is the number of equal divisions of longitude; Step S3.2: construct a deep neural network for each geographic region, satellite, and each intersatellite link; The deep neural network includes two parts, the first part is a long short-term memory network, and the second part is a fully connected neural network. The historical load information is embedded in the fully connected neural network together with other features through the long short-term memory network. The other features include the current time and the current weather conditions. Step S3.3: Calculate the total satellite load in the next topological stability period, which is equal to the sum of the loads generated by the ground area corresponding to the satellite plus the load generated by the satellite itself.

4. The method for dynamic deployment of controllers in a software-defined medium and low orbit satellite network according to claim 1, characterized in that: In step S4: In step S4.2: The distance between a node and a control domain is the minimum number of hops from the node to the nodes in the control domain. It is calculated as follows: u,v is the number of hops from node u to node v, then node u to the control domain The distance is In step S4.3: The control overhead includes satellite migration overhead and routing request overhead, where: the migration overhead of each satellite is equal to the control signaling transmission volume introduced by the satellite during the migration process, and the overall satellite migration overhead is equal to the sum of all satellite migration overheads; the routing request overhead of each satellite is equal to the satellite load multiplied by the control signaling transmission volume introduced by each routing request, and the total routing request overhead is equal to the sum of all satellite routing request overheads; the satellite control efficiency is quantified using the modularity of the network, which is equal to the sum of the load within the domain; The regularization method based on the relative entropy function means that in the migration cost, whether the satellite u is migrated to the controller v can be expressed as in is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. is whether satellite u is controlled by controller v in the t-1th topological stability period, () + It is a positive function that returns the input value when the input value is positive, otherwise it outputs 0; In order to regularize the impact of current decisions on future migration costs, the relative entropy function is used replace Ensure that the total migration cost within the complete satellite network operation cycle is bounded; In step S4.6: The construction method of the topological aggregation graph is to define the topological graphs before and after aggregation as G(V, E) and G′(V′, E′), where V and E represent the point set and edge set of the topological graph respectively. The original topological graph G is divided into N non-intersecting control domains, and the i-th control domain is defined as D i , define the jth control domain as D j ,When aggregating, for each control domain D i Build node v Di , for control domain D i and control domain D j , D i Midpoint to D j All links of the middle node are aggregated into links That is, from arrive link, and D j Midpoint to D i All links of the middle node are aggregated into links The link load is the sum of all aggregated link loads.

5. A software-defined controller dynamic deployment system in a medium and low orbit satellite network, characterized in that: include: Module M1: Calculate the time when the topology change of the medium and low orbit satellite network occurs according to the orbital parameters of the medium and low orbit satellites; Module M2: Based on the topology change time, the satellite network operation cycle is divided into a series of topology stable periods, and the number of data transmission hops between each pair of satellites in the next topology stable period is calculated; Module M3: predicts the loads generated by the ground and satellites respectively, and predicts the actual load of each satellite and the actual load of each intersatellite link in the next topology stability period; Module M4: Construct the initial control domain, quantify the control overhead and control efficiency according to the satellite load, inter-satellite link load and satellite migration process, and calculate the location, number and control relationship of satellite controllers in the next topology stability period; Module M5: upload the calculation results to the satellite, turn on or off the controller according to the calculation results, and the satellite migrates to the corresponding controller according to the control relationship; In the module M4: The satellite controller deployment strategy calculation method based on Louvain method includes the following modules: Module M4.1: Define satellites with computing capabilities and installed preset systems as controller candidates, initialize all controller candidates as satellite controllers, and define the remaining nodes as ordinary satellite nodes; Module M4.2: Each controller and all nodes controlled by the controller constitute a control domain. The distance from a node to the control domain is defined as the minimum number of hops from the node to the nodes in the control domain. Ordinary satellite nodes are placed in the control domain of the nearest controller to complete the initial control domain construction. All node attributes are removed and all nodes are defined as ordinary satellite nodes. Module M4.3: Calculate the overall network control overhead and efficiency, and regularize the impact of current decisions on future overhead to ensure the total control overhead within the complete network operation cycle; Module M4.4: Calculate the overall network overhead and control efficiency after migrating each node to its adjacent control domain, and calculate the performance improvement brought by the migration; Module M4.5: If there is a positive benefit migration, execute the migration that brings the highest performance improvement, and jump to run module M4.3; Module M4.6: If there is no positive benefit migration, construct a topological aggregation graph, where each point is a control domain and each edge represents the aggregation of cross-domain links; Module M4.7: If the topology aggregation graph is not equal to the original topology graph, jump to run module M4.4; Module M4.8: If the topology aggregation graph is equal to the original topology graph, calculate the optimal controller position for each control domain, and the optimal controller controls all other satellite nodes in the domain.

6. The software-defined system for dynamic deployment of controllers in a medium and low-orbit satellite network according to claim 5, characterized in that: In the module M1: According to the orbital parameters of medium and low orbit satellites, the time when the satellite enters and leaves the polar region is calculated. Since the inter-orbit inter-satellite links of the medium and low orbit satellite constellation are disconnected when passing through the polar region, this time is the time when the topology change of the medium and low orbit satellite network occurs, where the topology change refers to the disconnection or reconstruction of the inter-satellite links; The topology change time refers to the disconnection of the inter-orbit intersatellite links of the medium and low orbit satellite network when passing through the polar region. The topology change time is calculated by the time it takes for the satellite to enter and exit the polar region; In the module M2: The topology stability period is the longest time window in which the topology does not change. The number of data transmission hops is the distance, and the time interval between two consecutive topology changes is called the topology stability period.

7. The software-defined controller dynamic deployment system in a medium and low orbit satellite network according to claim 5, characterized in that: In the module M3: The method for predicting satellite load and intersatellite link load includes the following steps: Module M3.1: Divide the surface of the earth evenly according to longitude and latitude lat ×N lon Geographical area, N lat is the number of equal latitude divisions, N lon is the number of equal divisions of longitude; Module M3.2: Build a deep neural network for each geographic region, satellite, and each intersatellite link; The deep neural network includes two parts, the first part is a long short-term memory network, and the second part is a fully connected neural network. The historical load information is embedded in the fully connected neural network together with other features through the long short-term memory network. The other features include the current time and the current weather conditions. Module M3.3: Calculate the total satellite load in the next topological stability period, which is equal to the sum of the loads generated by the ground area corresponding to the satellite plus the load generated by the satellite itself.

8. The software-defined system for dynamic deployment of controllers in a medium- and low-orbit satellite network according to claim 5, characterized in that: In the module M4: In the module M4.2: The distance between a node and a control domain is the minimum number of hops from the node to the nodes in the control domain. It is calculated as follows: u,v is the number of hops from node u to node v, then node u to the control domain The distance is In the module M4.3: The control overhead includes satellite migration overhead and routing request overhead, where: the migration overhead of each satellite is equal to the control signaling transmission volume introduced by the satellite during the migration process, and the overall satellite migration overhead is equal to the sum of all satellite migration overheads; the routing request overhead of each satellite is equal to the satellite load multiplied by the control signaling transmission volume introduced by each routing request, and the total routing request overhead is equal to the sum of all satellite routing request overheads; the satellite control efficiency is quantified using the modularity of the network, which is equal to the sum of the load within the domain; The regularization method based on the relative entropy function means that in the migration cost, whether the satellite u is migrated to the controller v can be expressed as in is a binary decision variable, representing whether satellite u is controlled by controller v in the tth topological stability period. is whether satellite u is controlled by controller v in the t-1th topological stability period, ( ) + It is a positive function that returns the input value when the input value is positive, otherwise it outputs 0; In order to regularize the impact of current decisions on future migration costs, the relative entropy function is used replace Ensure that the total migration cost within the complete satellite network operation cycle is bounded; In the module M4.6: The construction method of the topological aggregation graph is to define the topological graphs before and after aggregation as G(V, E) and G′(V′, E′), where V and E represent the point set and edge set of the topological graph respectively. The original topological graph G is divided into N non-intersecting control domains, and the i-th control domain is defined as D i , define the jth control domain as D j ,When aggregating, for each control domain D i Build Node For control domain D i and control domain D j , D i Midpoint to D j All links of the middle node are aggregated into links That is, from arrive link, and D j Midpoint to D i All links of the middle node are aggregated into links The link load is the sum of all aggregated link loads.

Citation Information

Patent Citations

  • Software defined satellite network controller deployment method and device and related equipment

    CN112817605A

  • Controller deployment method for low-orbit software-defined satellite network

    CN113595613A

  • Controller Deployment Methods for Low-Earth Orbit Software-Defined Satellite Networks

    CN113595613B

  • Control message snapshot routing method of software-defined satellite network based on LEO / MEO

    CN108768504A

  • SDN-based space-air-ground controller deployment architecture and control method

    CN111342993A