Distributed dynamic scheduling method and system for multi-satellite networking under complex link conditions

Abstract

The present invention provides a multi-satellite network distributed dynamic scheduling method and system under complex link state, the steps are: obtain Pod from the queue to be scheduled, obtain the available satellite node queue based on the default pre-selection strategy, further filter the satellite nodes that are not in the geographical partition based on the extended pre-selection strategy of the satellite node geographical partition, score the available satellite nodes based on the distributed selection SpreadSelector and perform comprehensive scoring based on the basic operating status and link status data of the satellite nodes, and determine the optimal node based on the two scores for Pod scheduling and deployment. Using the scheduling method provided by the present invention, the satellite distributed system under the multi-satellite network can provide more efficient and reliable Pod resource scheduling services in a complex link environment.

Description

Technical Field

[0001] The present invention relates to the fields of aerospace technology and distributed systems, and in particular to a Pod resource scheduling method and system in a distributed satellite system in a multi-satellite networking environment under a complex link state. Background Art

[0002] In recent years, satellite Internet technology has developed rapidly. Countries have promoted satellite Internet construction to a national strategy, and my country has also included satellite Internet in the scope of new infrastructure. Generally speaking, satellite Internet has the characteristics of a large number of satellites, dynamic constellation network topology, and complex inter-satellite link environment.

[0003] In order to achieve large-scale satellite management and application, it is a feasible solution to introduce the common distributed system design ideas in ground systems into satellite Internet and to form a distributed satellite system with satellites in a multi-satellite networking environment. In a distributed system, more computing nodes mean more computing resources, and more distributed systems can complete more diversified computing tasks. However, in satellite Internet, due to the limited onboard resources, the cost of building a distributed satellite system is relatively high.

[0004] Compared with traditional virtual machine (VM) technology, container technology can package applications and their operating environments, and achieve isolation between different applications on the basis of sharing an operating system kernel. It has the characteristics of lightweight and good scalability. Container technology represented by Docker has developed rapidly in recent years. It effectively solves the problems of excessive resource consumption, slow scheduling, inconsistent software stack environment and low resource utilization of virtual machines through image technology and image warehouse. To this end, container technology makes it possible to build a distributed satellite system. At the same time, by isolating different application environments, container technology also enables limited satellite resources to meet more diverse mission requirements at the same time.

[0005] Kubernetes is currently a relatively mature distributed system container orchestration management platform in the industry. Pod is the smallest scheduling unit in Kubernetes, which is a collection of one or a group of containers. Kubernetes implements diversified service provision through Pod resource scheduling based on physical nodes. In the implementation, the Kube-scheduler functional component is responsible for Pod resource scheduling, and the Scheduler Extender functional component extends the scheduling function. Due to the stable status of ground network links, traditional Pod resource scheduling generally only considers the node operation status, but not the link status between nodes. However, in the distributed satellite system composed of satellite Internet, in order to improve resource utilization and service quality, the inter-satellite link status is as important as the current operation status of the satellite node. How to perform efficient and reliable Pod resource scheduling on a distributed satellite system in a multi-satellite networking environment under complex link conditions has become an urgent problem to be solved. Summary of the invention

[0006] The technical problem solved by the present invention is: in order to achieve more efficient and reliable Pod resource scheduling, a distributed dynamic scheduling method for multi-satellite networking based on complex link states is provided. After filtering the available satellite nodes based on the default pre-selected scheduling strategy of Kubernetes, in order to improve the scheduling efficiency, the sub-satellite point of each satellite node in the distributed satellite system during operation is mapped to the earth's surface, and divided according to the 0° and 180° longitudes and the equator into four geographical partitions, and the available satellite nodes are secondary filtered according to the partition position where the satellite is located during operation, and then the balance of satellite node resource allocation is comprehensively considered, and the basic operating status of the satellite node and the link status of the satellite node are used to comprehensively evaluate the available satellite nodes filtered twice, so as to achieve distributed dynamic scheduling of multi-satellite networking under complex link states.

[0007] The technical solution of the present invention is: a multi-satellite network distributed dynamic scheduling method under complex link state, comprising:

[0008] Step 1: Get a Pod from the Pod queue to be scheduled;

[0009] Step 2: Based on the default pre-selection strategy, filter out unavailable satellite nodes and obtain a queue of available satellite nodes;

[0010] Step 3: For the satellite node queue filtered in step 2, use the geographical partition information where the satellite node is running, and further filter the satellite nodes that are not in the specified partition based on the extended pre-selection strategy;

[0011] Step 4: For all satellite nodes in the satellite node queue filtered in step 3, score them based on the SpreadSelector scoring algorithm;

[0012] Step 5: For all satellite nodes in the satellite node queue filtered in step 3, basic operation status data including CPU, memory, and disk are obtained in real time;

[0013] Step 6: For all satellite nodes in the satellite node queue filtered in step 3, real-time acquisition of link status data including latency, packet loss rate, routing hop count, etc.;

[0014] Step 7: Based on the information obtained in Steps 5 and 6, all satellite nodes in the satellite node queue filtered in Step 3 are scored for expansion optimization, that is, the satellite node basic resources and satellite node link resources are scored separately according to various information indicators and weighted summed, and the satellite node score that comprehensively considers the basic operating status of the satellite node and the link status of the satellite node is obtained according to the weighted average method;

[0015] Step 8: Based on the scores of each satellite node in step 4 and step 7, the final score of the satellite node is obtained by weighted average method, which comprehensively considers the balance of satellite node resource allocation, the basic operation status of the satellite node and the link status of the satellite node;

[0016] Step 9: Based on the satellite node scores obtained in step 8, select the satellite node with the highest score for Pod deployment.

[0017] In step 2, the default pre-selection strategy is specified through the distributed cluster configuration file.

[0018] In step 3, if the partition of the satellite node is not considered in the Pod deployment information, this step will be skipped and the satellite node list in step 2 will be directly passed to step 4.

[0019] For the satellite nodes in the satellite node queue filtered in step 3, the scoring is based on the SpreadSelector strategy. The formula is as follows:

[0020]

[0021] Among them, fScore i is the score of the SpreadSelector scheduling strategy for node i. MaxNodeScore is the highest score that can be obtained by this scheduling strategy, and the default value is 100. MaxNodeScore is the maximum number of Pods deployed on all nodes, podCount i is the number of Pods on node i.

[0022] The weighted parameters of each indicator in step six can be designed and adjusted according to the importance of each indicator to resource scheduling.

[0023] The specific process of step seven is as follows:

[0024] 6.1) Using the basic operating data including satellite node CPU, memory, and disk space obtained in step 4, the basic resource score NodeScore of satellite node i is calculated based on the CPU, memory, and disk space of the satellite node separately scored based on the given weights. i ; The weight here indicates the importance of basic resources including CPU, memory, and disk space of satellite nodes to resource scheduling. The sum of the weights of basic resources of each satellite node is 100%; each basic resource will be converted into a standard score with a maximum value of 100*weight ratio of the indicator and a minimum value of 0 according to its operating data;

[0025] 6.2) Using the link status data including the satellite node delay, packet loss rate, and routing hop count obtained in step 5, score the link resources of the satellite node; based on the given weights, the delay, packet loss rate, and routing hop count between satellite nodes are scored separately, and the link resource score LinkScore of satellite node i is calculated. i ; The weight here indicates the importance of link resources including satellite node delay, packet loss rate, and routing hop count to resource scheduling, and the sum of the weights of link resources of each satellite node is 100%; when the specific values ​​of satellite node delay, packet loss rate, and routing hop count link status information are between the default maximum and minimum delay values ​​configured in the system, each link resource will be converted into a standard score with a maximum value of 100*the weight ratio of the indicator and a minimum value of 0 according to its operating data; taking the satellite node delay resource as an example, DELAYSCOREMAX is used to represent the highest score of the delay score, that is, 100*the weight ratio of the delay indicator, then the memory score range of all satellite nodes will be [0, DELAYSCOREMAX];

[0026] 6.3) The basic resource score of the satellite node and the link resource score of the satellite node are weighted and summed according to the given weights, and the total weight is also 100%, and the expansion optimization score of satellite node i is obtained as follows:

[0027]

[0028] sScore i represents the expansion optimization score of satellite node i, represents the basic resource weight of satellite node i, represents the link resource weight of satellite node i; sScore i The larger the value is, the higher the expansion preference score of the satellite node is.

[0029] The basic operating data including the satellite node CPU, memory, and disk space obtained in step 4 is used to score the satellite node CPU, memory, and disk space separately based on the given weights to calculate the basic resource score NodeScore of satellite node i i ,include:

[0030] The satellite node memory score calculation formula is as follows:

[0031]

[0032] MemScore i Represents the memory score of satellite node i, and MEMSCOREMAX represents the maximum memory score of satellite nodes; Mem i Represents the memory availability of satellite node i, and maxMem represents the maximum memory availability of all satellite nodes.

[0033] The satellite node CPU score calculation formula is as follows:

[0034]

[0035] CpuScore i represents the CPU score of satellite node i, CPUSCOREMAX represents the maximum CPU score of satellite nodes, and CPU i represents the CPU availability of satellite node i, and maxCpu represents the maximum CPU availability of all satellite nodes;

[0036] The satellite node disk space score calculation formula is as follows:

[0037]

[0038] DiskScore i represents the disk space score of satellite node i, DISKSCOREMAX represents the maximum disk score of satellite nodes, and DISK i represents the disk availability of satellite node i, and maxDisk represents the maximum disk availability of all satellite nodes;

[0039] Sum the above scores to get the basic resource score of satellite node i after filtering in step 3

[0040] NodeScore i =MemScore i +CpuScore i +DiskScore i

[0041] NodeScorei Indicates the basic resource score of satellite node i. The larger the value, the higher the resource availability.

[0042] The link status data including the satellite node delay, packet loss rate, and routing hop count obtained in step 5 are used to score the link resources of the satellite node; based on the given weights, the delay, packet loss rate, and routing hop count between satellite nodes are scored separately, and the link resource score LinkScore of satellite node i is calculated. i ,include:

[0043] The calculation formula for satellite node i delay score is as follows:

[0044]

[0045] delayScore i represents the link delay score of satellite node i, MAXDELAY and MINDELAY represent the default maximum and minimum delay values ​​of the system configuration, respectively. i represents the link delay of satellite node i; DELAYSCOREMAX represents the maximum delay score;

[0046] The calculation formula for the packet loss rate score of satellite node i is as follows:

[0047]

[0048] lossScore i represents the link packet loss rate score of satellite node i, LOSSCOREMAX represents the highest score of packet loss rate, MAXLOSS and MINLOSS represent the default maximum and minimum packet loss rate values ​​of the system configuration, respectively, and loss i represents the link packet loss rate of satellite node i;

[0049] The calculation formula of satellite node i routing hop score is as follows:

[0050]

[0051] hopScore i represents the routing hop score of satellite node i, HOPSCOREMAX represents the highest routing hop score, MAXHOP and MINHOP represent the default maximum and minimum routing hop values ​​of the system configuration, respectively. i represents the number of routing hops of node i;

[0052] Sum the above scores to obtain the link resource score of the satellite node after filtering in step 3;

[0053] LinkScorei =delayScore i +lossScore i +ShopScore i

[0054] LinkScore i Represents the link resource score of satellite node i; the larger the value, the better the link performance.

[0055] The specific process of step eight is as follows:

[0056] The SpreadSelector score in step 4 and the node comprehensive score in step 7 are comprehensively analyzed, that is, the SpreadSelector and the extended selection are weighted and summed according to the given weights, and the total weight is 100%. The final score of satellite node i is:

[0057] Score i =fWeight*fScore i +sWeight*sScore i

[0058] Score i represents the final score of satellite node i, fWeight represents the weight of the spread selection SpreadSelector, and sWeight represents the weight of the expansion selection; Score i The larger the value is, the higher the final score of satellite node i is.

[0059] A multi-satellite network distributed scheduling system based on a complex link state environment includes the following modules:

[0060] Module 1: Expand the pre-selection module, through which the satellite node list is screened based on the partition;

[0061] Module 2: Satellite node status perception module, which provides the perception of the basic operating status of satellite nodes;

[0062] Module 3: Satellite node link status perception, through this module to obtain satellite node link status data;

[0063] Module 4: Expand the optimization scoring module, through which the basic operation status and link status of satellite nodes obtained in modules 2 and 3 are scored;

[0064] Module 5: Node selection module, through which the best satellite node is selected for Pod scheduling and deployment based on the node scores obtained in module 4.

[0065] The beneficial effects of the present invention are:

[0066] Taking into full consideration the characteristics of the distributed satellite system, such as the large number of satellites, dynamic constellation network topology, and complex inter-satellite link environment, in the process of scheduling Pod resources, the geographical partition location information of the satellite during operation is used to perform secondary filtering on the available satellite nodes, thereby improving the scheduling efficiency. Through the comprehensive design of the decentralized selection algorithm and the extended optimization algorithm, the Pod scheduling is more likely to be scheduled to the satellite nodes with better comprehensive status and better system comprehensive service performance in the distributed satellite system, ensuring that the distributed satellite system can better provide efficient and reliable services. In addition, the entire scheduling process is automatic and has high practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 Flow chart of the distributed dynamic scheduling method for multi-satellite networking under complex link conditions proposed by the present invention;

[0068] Figure 2 It is a process diagram of the SpreadSelector scoring algorithm in a specific implementation manner of the present invention;

[0069] Figure 3 It is a process diagram of the extended optimization scoring algorithm in a specific implementation manner of the present invention. DETAILED DESCRIPTION

[0070] The present invention provides a multi-satellite network distributed dynamic scheduling method under complex link state, which is implemented based on Kubernetes. The present invention uses the Kubernetes scheduling function extension component Scheduler Extender to extend the function of the Kubernetes scheduler component Kube-scheduler. During the operation of Kube-scheduler, the extended scheduling method logic is run by calling the webhook registered by Scheduler Extender to control the decision results of each stage in the scheduling process.

[0071] When service deployment or other operations generate Pod deployment in the distributed satellite system, Kube-scheduler will display a list of Pods to be scheduled, store all Pods to be scheduled, and wait for scheduling through Kube-scheduler. The flowchart of the multi-satellite network distributed dynamic scheduling method under complex link state of the present invention is shown in the attached Figure 1 The basic steps of distributed scheduling of multi-satellite networks based on complex link status environments are as follows:

[0072] Step 1: Get a Pod from the Pod queue to be scheduled;

[0073] Step 2: Based on the default pre-selected scheduling strategy of Kubernetes, Kube-scheduler filters the available satellite nodes through the distributed cluster configuration file, filters the unavailable satellite nodes, and obtains the available satellite nodes;

[0074] Step 3: In order to improve scheduling efficiency, Kube-scheduler calls the newly added extended pre-selection strategy in the present invention through the webhook registered by Scheduler Extender, that is, the partition information in the Pod configuration file is used for partition matching, and the satellite nodes that are not in the specified partition are further filtered. The specific method is to map the sub-satellite point of each satellite node in the distributed satellite system to the surface of the earth when it is running, and divide it into four geographical partitions according to the 0° and 180° longitudes and the equator. The available satellite nodes are secondary filtered according to the geographical partition position where the satellite is running, and only the satellite nodes in the specified geographical partition are retained.

[0075] Step 4: To prevent Pods in the same scheduling queue from being scheduled to a few satellite nodes, which would result in the resources of the remaining satellite nodes not being fully utilized and unbalanced satellite node resource allocation, all satellite nodes obtained after filtering in step 3 are scored based on the SpreadSelector scoring algorithm.

[0076] The SpreadSelector scoring algorithm can distribute Pods to different satellite nodes. The specific process is shown in the attached Figure 2 The node scoring formula is as follows:

[0077]

[0078] Among them, fScore i is the score of the SpreadSelector scheduling strategy for node i. MaxNodeScore is the highest score that can be obtained by this scheduling strategy. The default value is 100. maxPodCount is the maximum number of Pods deployed on all nodes. podCount i is the number of Pods on node i.

[0079] Step 5: Configure the open source node monitoring service Prometheus in the distributed satellite system, and use the Prometheus service to obtain the basic operating status data of each satellite node, such as CPU, memory, and disk space, filtered in step 3, in real time;

[0080] Step 6: Configure an intersatellite network link simulator in the distributed satellite system, set the link information such as delay, packet loss rate and routing hop count between satellite nodes, and use this information to describe the link status information such as delay, packet loss rate and routing hop count between each satellite node filtered in step 3; for a specified satellite node, use the average value of the link delay, packet loss rate and routing hop count between the node and all other satellite nodes in the available satellite node list to reflect the node link status;

[0081] Step 7: Based on the basic operation data of satellite nodes and the link status information of satellite nodes obtained in Steps 5 and 6, the satellite nodes filtered in Step 3 are scored for expansion optimization. The scoring process is shown in the attached Figure 3 .

[0082] First, using the basic operating data of the satellite node CPU, memory, disk space, etc. obtained in step 4, the basic resource score of the satellite node is calculated based on the given weights to score the satellite node's CPU, memory, disk space, etc. separately. Here, the weight represents the importance of the basic resources such as the satellite node's CPU, memory, disk space, etc. to resource scheduling, and the sum of the weights of the basic resources of each satellite node is 100%. Each basic resource will be converted into a standard score with a maximum value of (100*the weight ratio of the indicator) and a minimum value of 0 according to its operating data. Taking memory resources as an example, MEMSCOREMAX is used to represent the highest score of the memory score, that is, 100*the weight ratio of the memory indicator, then the memory score range of all satellite nodes will be [0,MEMSCOREMAX].

[0083] In the present invention, the satellite node memory score calculation formula is as follows:

[0084]

[0085] MemScore i Represents the memory score of satellite node i, Mem i Represents the memory availability of satellite node i, and maxMem represents the maximum memory availability of all satellite nodes.

[0086] The satellite node CPU score calculation formula is as follows:

[0087]

[0088] CpuScore i represents the CPU score of satellite node i, CPUSCOREMAX represents the maximum CPU score of satellite nodes, and CPU i It represents the CPU availability of satellite node i, and maxCpu represents the maximum CPU availability of all satellite nodes.

[0089] The satellite node disk space score calculation formula is as follows:

[0090]

[0091] DiskScore i represents the disk space score of satellite node i, DISKSCOREMAX represents the maximum disk score of satellite nodes, and DISK i represents the disk availability of satellite node i, and maxDisk represents the maximum disk availability of all satellite nodes.

[0092] The above scores are summed up to obtain the score of the basic resources of satellite node i after filtering in step three.

[0093] NodeScore i =MemScore i +CpuScore i +DiskScore i

[0094] NodeScore i Indicates the basic resource score of satellite node i. The larger the value, the higher the resource availability.

[0095] Secondly, the link resources of the satellite nodes are scored using the link status data such as the delay, packet loss rate, and routing hop count of the satellite nodes obtained in step 5. The link resource scores of the satellite nodes are calculated based on the given weights for scoring the delay, packet loss rate, and routing hop count between the satellite nodes. Here, the weight represents the importance of the link resources such as the delay, packet loss rate, and routing hop count of the satellite nodes to resource scheduling, and the sum of the weights of the link resources of each satellite node is 100%. When the specific values ​​of the link status information such as the delay, packet loss rate, and routing hop count of the satellite nodes are between the default maximum and minimum delay values ​​configured in the system, each link resource will be converted into a standard score with a maximum value of (100*the weight ratio of the indicator) and a minimum value of 0 according to its operating data. Taking the satellite node delay resource as an example, DELAYSCOREMAX is used to represent the highest score of the delay score, that is, 100*the weight ratio of the delay indicator, and the memory score range of all satellite nodes will be [0, DELAYSCOREMAX].

[0096] The calculation formula for satellite node i delay score is as follows:

[0097]

[0098] delayScore irepresents the link delay score of satellite node i, MAXDELAY and MINDELAY represent the default maximum and minimum delay values ​​of the system configuration, respectively. i represents the link delay of satellite node i.

[0099] The calculation formula for the packet loss rate score of satellite node i is as follows:

[0100]

[0101] lossScore i represents the link packet loss rate score of satellite node i, LOSSCOREMAX represents the highest score of packet loss rate, MAXLOSS and MINLOSS represent the default maximum and minimum packet loss rate values ​​of the system configuration, respectively, and loss i represents the link packet loss rate of satellite node i.

[0102] The calculation formula of satellite node i routing hop score is as follows:

[0103]

[0104] hopScore i represents the routing hop score of satellite node i, HOPSCOREMAX represents the highest routing hop score, MAXHOP and MINHOP represent the default maximum and minimum routing hop values ​​of the system configuration, respectively. i Indicates the number of routing hops of node i.

[0105] The above scores are summed to obtain the link resource score of satellite node i after filtering in step three.

[0106] LinkScore i =delayScore i +lossScore i +ShopScore i

[0107] LinkScore i Represents the link resource score of satellite node i. The larger the value, the better the link performance. The basic resource score of the satellite node and the link resource score of the satellite node are weighted and summed according to the given weights. The total weight is also 100%, and the extended optimization score of satellite node i is obtained as follows:

[0108]

[0109] sScore i represents the expansion optimization score of satellite node i, represents the basic resource weight of satellite node i, Represents the link resource weight of satellite node i. i The larger the value is, the higher the expansion preference score of the satellite node is.

[0110] Step 8: Comprehensively analyze the SpreadSelector score in step 4 and the node comprehensive score in step 7, that is, weighted sum of the SpreadSelector and the extended selection according to the given weights, and the total weight is also 100%, and the final score of satellite node i is:

[0111] Score i =fWeight*fScore i +sWeight*sScore i

[0112] Score i represents the final score of satellite node i, fWeight represents the weight of the spread selection SpreadSelector, and sWeight represents the weight of the expansion selection. i The larger the value is, the higher the final score of satellite node i is.

[0113] Step 9: Sort the nodes by their total scores in step 8 and select the node with the highest score for Pod scheduling and deployment.

[0114] A multi-satellite network distributed scheduling system based on a complex link state environment in the present invention implements the scheduling of a distributed satellite system through multiple functional modules. The functional modules of a multi-satellite network distributed dynamic scheduling system under a complex link state are as follows:

[0115] Module 1: Expand the pre-selection module. Through this module, the system performs secondary screening of satellite nodes based on the satellite node partitions pre-specified during Pod deployment. If the partitions of satellite nodes are not considered during Pod deployment, this module will not work, and the satellite node list will be directly passed to subsequent modules.

[0116] Module 2: Satellite node status perception module. Through this module, the system reads the basic operating status of the specified satellite node from Prometheus, a status monitoring service based on distributed cluster deployment, including CPU usage, memory usage, disk space usage, etc.

[0117] Module 3: Satellite node link status perception module. Through this module, the system reads the link status data such as link delay, packet loss rate, and routing hop count between the specified satellite node and all other satellite nodes in the available satellite node list from the inter-satellite link data table pre-calculated based on satellite orbit parameters, and takes the average value to reflect the current node link status;

[0118] Module 4: Expand the optimization scoring module. Through this module, the system scores the satellite nodes based on the basic operation status and link status of the satellite nodes obtained in modules 2 and 3;

[0119] Module 5: Node selection module. Through this module, the system selects the best satellite node for Pod scheduling and deployment based on the satellite node scores obtained in module 4.

[0120] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solution of the present invention by using the technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. A distributed dynamic scheduling method for multi-satellite networking under complex link conditions, characterized in that: include: Step 1: Get a Pod from the Pod queue to be scheduled; Step 2: Based on the default pre-selection strategy, filter out unavailable satellite nodes and obtain a queue of available satellite nodes; Step 3: For the satellite node queue filtered in step 2, use the geographical partition information where the satellite node is running, and further filter the satellite nodes that are not in the specified partition based on the extended pre-selection strategy; Step 4: For all satellite nodes in the satellite node queue filtered in step 3, score them based on the SpreadSelector scoring algorithm; Step 5: For all satellite nodes in the satellite node queue filtered in step 3, basic operation status data including CPU, memory, and disk are obtained in real time; Step 6: For all satellite nodes in the satellite node queue filtered in step 3, real-time acquisition of link status data including latency, packet loss rate, and routing hop count; Step 7: Based on the information obtained in Steps 5 and 6, all satellite nodes in the satellite node queue filtered in Step 3 are scored for expansion optimization, that is, the satellite node basic resources and satellite node link resources are scored separately according to various information indicators and weighted summed, and the satellite node score that comprehensively considers the basic operating status of the satellite node and the link status of the satellite node is obtained according to the weighted average method; Step 8: Based on the scores of each satellite node in step 4 and step 7, the final score of the satellite node is obtained by weighted average method, which comprehensively considers the balance of satellite node resource allocation, the basic operation status of the satellite node and the link status of the satellite node; Step 9: Based on the satellite node scores obtained in step 8, select the satellite node with the highest score for Pod deployment.

2. The method for distributed dynamic scheduling of multi-satellite networks under complex link conditions according to claim 1, characterized in that: In step 2, the default pre-selection strategy is specified through the distributed cluster configuration file.

3. The distributed dynamic scheduling method for multi-satellite networking under complex link conditions according to claim 1, characterized in that: In step 3, if the partition of the satellite node is not considered in the Pod deployment information, this step will be skipped and the satellite node list in step 2 will be directly passed to step 4.

4. The distributed dynamic scheduling method for multi-satellite networking under complex link conditions according to claim 1, characterized in that: For the satellite nodes in the satellite node queue filtered in step 3, the scoring is based on the SpreadSelector strategy. The formula is as follows: Among them, fScore i is the score of the SpreadSelector scheduling strategy for node i. MaxNodeScore is the highest score that can be obtained by this scheduling strategy, and the default value is 100. MaxNodeScore is the maximum number of Pods deployed on all nodes, podCount i is the number of Pods on node i.

5. The multi-satellite network distributed dynamic scheduling method under complex link conditions according to claim 1, characterized in that: The weighted parameters of each indicator in step six can be designed and adjusted according to the importance of each indicator to resource scheduling.

6. The multi-satellite network distributed dynamic scheduling method under complex link conditions according to claim 4, characterized in that: The specific process of step seven is as follows: 6.1) Using the basic operating data including satellite node CPU, memory, and disk space obtained in step 4, the basic resource score NodeScore of satellite node i is calculated based on the CPU, memory, and disk space of the satellite node separately scored based on the given weights. i ; The weight here indicates the importance of basic resources including CPU, memory, and disk space of satellite nodes to resource scheduling. The sum of the weights of basic resources of each satellite node is 100%; each basic resource will be converted into a standard score with a maximum value of 100*weight ratio of the indicator and a minimum value of 0 according to its operating data; 6.2) Using the link status data including the satellite node's delay, packet loss rate, and routing hop count obtained in step 5, score the satellite node's link resources; Based on the given weights, the delay, packet loss rate, and routing hop count between satellite nodes are scored separately, and the link resource score LinkScore of satellite node i is calculated. i ; The weight here indicates the importance of link resources including satellite node delay, packet loss rate, and routing hop count to resource scheduling, and the sum of the weights of link resources of each satellite node is 100%; when the specific values ​​of link status information of satellite node delay, packet loss rate, and routing hop count are between the default maximum and minimum delay values ​​of the system configuration, each link resource will be converted into a standard score with a maximum value of 100*the weight ratio of the indicator and a minimum value of 0 according to its operating data; taking the satellite node delay resource as an example, DELAYSCOREMAX is used to represent the highest score of the delay score, that is, 100*the weight ratio of the delay indicator, then the memory score range of all satellite nodes will be [0, DELAYSCOREMAX]; 6.3) The basic resource score of the satellite node and the link resource score of the satellite node are weighted and summed according to the given weights, and the total weight is also 100%, and the expansion optimization score of satellite node i is obtained as follows: sScore i represents the expansion optimization score of satellite node i, represents the basic resource weight of satellite node i, represents the link resource weight of satellite node i; sScore i The larger the value is, the higher the expansion preference score of the satellite node is.

7. The method for distributed dynamic scheduling of multi-satellite networks under complex link conditions according to claim 6, characterized in that: The basic operating data including the satellite node CPU, memory, and disk space obtained in step 4 is used to score the satellite node CPU, memory, and disk space separately based on the given weights to calculate the basic resource score NodeScore of satellite node i i ,include: The satellite node memory score calculation formula is as follows: MemScore i Represents the memory score of satellite node i, and MEMSCOREMAX represents the maximum memory score of satellite nodes; Mem i represents the memory availability of satellite node i, maxMem represents the maximum memory availability of all satellite nodes, The satellite node CPU score calculation formula is as follows: CpuScore i represents the CPU score of satellite node i, CPUSCOREMAX represents the maximum CPU score of satellite nodes, and CPU i represents the CPU availability of satellite node i, and maxCpu represents the maximum CPU availability of all satellite nodes; The satellite node disk space score calculation formula is as follows: DiskScore i represents the disk space score of satellite node i, DISKCOREMAX represents the maximum disk score of satellite nodes, and DISK i represents the disk availability of satellite node i, and maxDisk represents the maximum disk availability of all satellite nodes; Sum the above scores to get the basic resource score of satellite node i after filtering in step 3 NodeScore i =MemScore i +CpuScore i +DiskScore i NodeScore i Indicates the basic resource score of satellite node i. The larger the value, the higher the resource availability.

8. The multi-satellite network distributed dynamic scheduling method under complex link conditions according to claim 7, characterized in that: Using the link status data including the satellite node's delay, packet loss rate, and routing hop count obtained in step 5, the link resources of the satellite node are scored; Based on the given weights, the delay, packet loss rate, and routing hop count between satellite nodes are scored separately, and the link resource score LinkScore of satellite node i is calculated. i ,include: The calculation formula for satellite node i delay score is as follows: delayScore i represents the link delay score of satellite node i, MAXDELAY and MINDELAY represent the default maximum and minimum delay values ​​of the system configuration, respectively. i represents the link delay of satellite node i; DELAYSCOREMAX represents the maximum delay score; The calculation formula for the packet loss rate score of satellite node i is as follows: lossScore i represents the link packet loss rate score of satellite node i, LOSSCOREMAX represents the highest score of packet loss rate, MAXLOSS and MINLOSS represent the default maximum and minimum packet loss rate values ​​of the system configuration, respectively, and loss i represents the link packet loss rate of satellite node i; The calculation formula of satellite node i routing hop score is as follows: hopScore i represents the routing hop score of satellite node i, HOPSCOREMAX represents the highest routing hop score, MAXHOP and MINHOP represent the default maximum and minimum routing hop values ​​of the system configuration, respectively. i represents the number of routing hops of node i; Sum the above scores to obtain the link resource score of the satellite node after filtering in step 3; LinkScore i =delayScore i +lossScore i +hpoScore i LinkScore i represents the link resource score of satellite node i; The larger the value, the better the link performance.

9. The multi-satellite network distributed dynamic scheduling method under complex link conditions according to claim 1, characterized in that: The specific process of step eight is as follows: The SpreadSelector score in step 4 and the node comprehensive score in step 7 are comprehensively analyzed, that is, the SpreadSelector and the extended selection are weighted and summed according to the given weights, and the total weight is 100%. The final score of satellite node i is: Score i =fWeight*fScore i +sWeight*sScore i Score i represents the final score of satellite node i, fWeight represents the weight of the spread selection SpreadSelector, and sWeight represents the weight of the expansion selection; Score i The larger the value is, the higher the final score of satellite node i is.

10. A multi-satellite network distributed scheduling system based on a complex link state environment, used to implement the method described in claim 1, characterized in that: Includes the following modules: Module 1: Expand the pre-selection module, through which the satellite node list is screened based on the partition; Module 2: Satellite node status perception module, which provides the perception of the basic operating status of satellite nodes; Module 3: Satellite node link status perception, through this module to obtain satellite node link status data; Module 4: Expand the optimization scoring module, through which the basic operation status and link status of satellite nodes obtained in modules 2 and 3 are scored; Module 5: Node selection module, through which the best satellite node is selected for Pod scheduling and deployment based on the node scores obtained in module 4.

Images