A method for dynamically selecting a control node in a low earth orbit constellation edge computing system

Abstract

The present application relates to the field of low-orbit satellite, in particular to a kind of control node dynamic selection method in low-orbit constellation edge computing system.The whole constellation is divided into multiple satellite subareas, and the virtual adjacent node information of each area control node is found and maintained by message interaction;Each area control node uses the message interaction between virtual adjacent node to reach consensus on virtual topology information;Each area control node selects consistent interval control node based on virtual topology information, that is, control node in low-orbit constellation edge computing system.The interval control node selection method used in the present application first relies on the message transmission between each area control node, so that each area control node agrees on the virtual topology of the current dynamic node cluster;Then each area control node uses the same interval control node election algorithm, which can ensure that a unique interval control node is calculated according to the current virtual topology.

Description

Technical Field

[0001] This invention relates to the field of low Earth orbit satellites, and more specifically to a method for dynamically selecting control nodes in a low Earth orbit constellation edge computing system. Background Technology

[0002] In low-Earth orbit (LEO) constellation edge computing systems, each satellite possesses certain computing and storage resources, and they can be interconnected via microwave or laser links. Therefore, there is a possibility of multiple satellites collaboratively providing computing services. However, this collaborative service delivery involves resource allocation, task scheduling, and task migration among the multiple satellites, requiring a management and control method for effective management and control of each satellite.

[0003] The simplest management and control method is to control each satellite by a ground-based centralized controller. However, this method has the following problems: 1) The communication delay between the ground-based centralized controller and some satellites is relatively large, especially when the geographical distance between the geographical location of the ground controller and the geographical location of the satellite's ground point is large; 2) It cannot meet the autonomous operation requirements of the constellation in military application scenarios.

[0004] There are three typical approaches to the autonomous management and control of constellations:

[0005] 1) Centralized: This means that a satellite is pre-determined as the control node of the entire constellation. This satellite makes control decisions and controls other satellites to execute them. The disadvantages are that the communication delay between the constellation control node and some satellites is too large; there is a single point of failure, which does not meet the survivability requirements of military applications; as the number of satellites in the constellation increases, the management and control pressure on the constellation control node becomes too great.

[0006] 2) Distributed: Each satellite makes its own management and control decisions. The disadvantage is that the decisions between different satellites may conflict. For example, satellite A and satellite B may simultaneously assign a certain task to satellite C, causing satellite C to become overloaded.

[0007] 3) Partitioning: The entire constellation is divided into multiple satellite partitions, each containing multiple satellites. The control nodes within each partition manage and control the satellites within that partition. When inter-partition coordination is required, it relies on the inter-partition control nodes for coordination. Summary of the Invention

[0008] The purpose of this invention is to demonstrate how to select interval control nodes within a dynamic node cluster (i.e., a cluster composed of intra-region control nodes). A dynamic control node selection method for a low-Earth orbit constellation edge computing system is proposed. After dynamically partitioning the constellation and forming multiple intra-region control nodes, interval control nodes can be selected from the cluster of these intra-region control nodes, supporting subsequent coordination of each partition by the interval control nodes.

[0009] The technical solution adopted in this invention is as follows:

[0010] A method for dynamically selecting control nodes in a low-Earth orbit constellation edge computing system includes the following steps:

[0011] (1) Divide the entire constellation into multiple satellite partitions. Within each partition, the control nodes discover and maintain their own virtual neighbor node information through message interaction.

[0012] (2) Control nodes in each zone reach a consensus on virtual topology information by using message interaction with virtual neighboring nodes;

[0013] (3) Based on virtual topology information, the control nodes in each zone use the same algorithm to select consistent zone control nodes, namely the control nodes in the low-orbit constellation edge computing system.

[0014] Furthermore, the specific process of step (1) is as follows:

[0015] After the entire constellation is divided into multiple satellite partitions, each partition contains multiple satellites. Within each partition, the intra-partition control node periodically sends DiscoverNeighbor messages to all neighboring satellites. The intra-partition control node receiving the DiscoverNeighbor message returns an AckDiscover message to the sending intra-partition control node. Through this message exchange, virtual neighbor node discovery between intra-partition control nodes is completed. The DiscoverNeighbor message includes the message type, intra-partition control node identifier, message sequence number, and intra-partition control node IP address. The AckDiscover message includes the message type, intra-partition control node identifier, message sequence number, intra-partition control node IP address, and virtual link cost.

[0016] Each intra-area control node periodically sends NeighborM messages to all virtual neighbor nodes. Intra-area control nodes that receive NeighborM messages return AckNeighborM messages to the intra-area control node that sent them. Through message exchange, the virtual neighbor node information between intra-area control nodes is maintained. NeighborM messages and AckNeighborM messages both include message type, source intra-area control node identifier, destination intra-area control node identifier, and message sequence number.

[0017] Furthermore, the specific process of step (2) is as follows:

[0018] Control node H within each zone i Upon discovering a new virtual neighbor node H j At that time, to Hj Send a TopoInfoMsg message, which includes the information of the control node H within the area. i The maintained virtual topology information TopoInfo includes the control node H within the area. i All known identifiers of control nodes within the area, virtual adjacency relationships between control nodes within the area, and virtual link costs;

[0019] Control node H within each zone i The UpdtMsg message is periodically sent to all virtual neighbor nodes. The UpdtMsg message includes the control node H within the area. i The virtual neighbor node and virtual link update information Updt is maintained. Updt includes newly added virtual neighbor nodes, changed virtual link costs, and newly deleted virtual neighbor nodes. After each UpdtMsg message is sent, the intra-area control node clears its maintained Updt information. The intra-area control node H, upon receiving the UpdtMsg message... rec Forward the UpdtMsg message to all virtual neighbor nodes except the intra-area control node that sent the UpdtMsg message.

[0020] Furthermore, the specific process of step (3) is as follows:

[0021] The virtual link cost between two control nodes within the calculation area is calculated. If the two control nodes are the same, the virtual link cost is 0.

[0022] For each control node H in the area i Calculate the control nodes H in all other zones. j Cumulative cost between The formula is as follows:

[0023]

[0024] Where, N c PC represents the number of control nodes within the area. ij Representative of the control node H within the representative area i With control node H in the area j The virtual path cost between nodes is calculated using the shortest path algorithm with virtual topology information and virtual link costs within the control node cluster as inputs, and the cumulative cost is calculated. For the control node H in the area i The total cost of the virtual paths between the device and all other control nodes within the same zone;

[0025] The control node with the lowest cumulative cost within the zone is selected as the control node for the interval. If multiple control nodes within the zone have the same cumulative cost, the control node with the lowest or highest index number is selected as the control node for the interval.

[0026] The advantages of this invention compared to the prior art are: it can support the autonomous selection of control nodes in the region when control nodes dynamically join / leave the cluster. Attached Figure Description

[0027] Figure 1 This is a diagram of a constellation without dynamic partitioning.

[0028] Figure 2 This is a schematic diagram after dynamic partitioning.

[0029] Figure 3 This is a diagram illustrating the virtual link and its corresponding actual path. Detailed Implementation

[0030] The invention will be further explained below with reference to the accompanying drawings.

[0031] like Figure 1 The image shown is a partial schematic diagram of a constellation with incomplete dynamic partitioning. Figure 2 The diagram shows the result of dynamic partitioning, which creates three satellite partitions with control nodes for satellites 202, 205, and 403. The objective of this invention is to elect a partition control node from the cluster composed of control nodes 202, 205, and 403.

[0032] Specifically, the following steps are included:

[0033] (1) After dividing the entire constellation into multiple satellite partitions, the control nodes within each partition discover and maintain their own virtual neighbor node information through message exchange; the specific process is as follows:

[0034] To discover virtual adjacent intra-region control nodes, each intra-region control node H... D Each satellite periodically sends DiscoverNeighbor messages to all neighboring satellites. Satellites receiving DiscoverNeighbor messages perform the following actions:

[0035] a) Ordinary satellites that receive DiscoverNeighbor messages (non-regional control nodes, denoted as S) rec The DiscoverNeighbor message is forwarded directly to all neighboring satellites, and the TTL field in the IP protocol is used to limit the number of times the DiscoverNeighbor message is forwarded in the constellation; if ordinary satellite S rec It is from one of its neighboring satellites (denoted as S)send If a DiscoverNeighbor message is received, it will not send a message to satellite S. send Forward this message;

[0036] b) The intra-area control node (denoted as H) that receives the DiscoverNeighbor message R First, check the virtual neighbor node information you maintain (i.e., NeiborInfo(H)). R Does the NeighborMem contain the identifier H of the intra-area control node that issued the DiscoverNeighbor message? D If not, then H D Add your own virtual neighbor node information NeiborInfo(H R In NeighborMem, and the control node H within the area. R With control node H in the area D The virtual link cost between them is assigned to NeiborInfo(H) R The corresponding element in vCost is then sent to the intra-area control node (denoted as H) that issued the DiscoverNeighbor message. D Returns the AckDiscover message. If NeiborInfo(H R The area already has control node identifier H. D If the DiscoverNeighbor message is not received, it will be discarded. There are various ways to calculate the virtual link cost. In this invention, the actual number of hops between intra-area control nodes is used as an example. This invention is also applicable to defining the "virtual link cost" using other similar methods. Intra-area control node H that receives the DiscoverNeighbor message... R The relationship with the control node H within the region can be calculated by reading the TTL. D The actual number of hops between them. For subsequent identical DiscoverNeighbor messages (i.e., satellite H...) D Copies of DiscoverNeighbor messages generated during network transmission are discarded.

[0037] Control node within the area (denoted as H) D A timer T is set when a DiscoverNeighbor message is sent. D If H D In timer T D Before the timeout, other control nodes (denoted as H) were received. R The AckDiscover message sent by ) will then affect the control node H within the area. RThe identifier is added to its own virtual neighbor node information NeiborInfo(H) D In NeighborMem, and the control node H within the area. D With control node H in the area R The virtual link cost between them is assigned to NeiborInfo(H) D The corresponding element in vCost within ).

[0038] Each control node within the zone (denoted as H) S It periodically sends NeighborM messages to all its virtual neighbor nodes, and starts timer T after sending the NeighborM message. WaitAckM The control node within the area that receives the NeighborM message sends a message to H. S Returns the AckNeighborM message; through message interaction, the maintenance of virtual neighbor node information between control nodes within the area is completed.

[0039] If in timer T WaitAckM Before the timeout, the intra-area control node H that sent the NeighborM message... S Received a control node (denoted as H) within a certain area Neighbor If the returned AckNeighborM message is H, then H S H is considered Neighbor It remains its virtual neighbor node; for a control node H within a certain area S If there are multiple consecutive T WaitAckM Timeout occurred because a virtual neighbor node H was not received. Neighbor In the case of the returned AckNeighborM message, then H S H is considered Neighbor It is no longer suitable to be its virtual neighbor node. The intra-area control node H that received the AckNeighborM message... Neighbor The relationship with the control node H within the region can be calculated by reading the TTL. S The actual number of hops between them.

[0040] In this embodiment, virtual adjacency specifically refers to a low actual hop count between two intra-region control nodes. Specifically, if the actual hop count between two intra-region control nodes is less than the upper limit threshold of the hop count, then the two intra-region control nodes are considered to be virtually adjacent.

[0041] The message types involved in this step include DiscoverNeighbor, AckDiscover, NeighborM, and AckNeighborM. The content of each message type is as follows:

[0042] The information that DiscoverNeighbor messages must include includes:

[0043] (a) Message type: a 01 sequence representing "DiscoverNeighbor";

[0044] (b) Control node identifier within the area: that is, the identifier of the control node within the area that sent the DiscoverNeighbor message;

[0045] (c) Message sequence number: used to distinguish DiscoverNeighbor messages sent by control nodes within the same area at different times;

[0046] (d) IP address of the control node in the zone: that is, the IP address of the control node in the zone that sent the DiscoverNeighbor message.

[0047] The AckDiscover message must include the following information:

[0048] (a) Message type: a 01 sequence representing "AckDiscover";

[0049] (b) Control node identifier within the area: that is, the identifier of the control node within the area that issued the AckDiscover message;

[0050] (c) Message sequence number: i.e., the message sequence number of the corresponding DiscoverNeighbor message;

[0051] (d) IP address of the control node in the area: that is, the IP address of the control node in the area that sent the AckDiscover message;

[0052] (e) Virtual link cost: the virtual link cost between the intra-area control node that sends the AckDiscover message and the intra-area control node that sends the corresponding DiscoverNeighbor message.

[0053] The information that a NeighborM message must include includes:

[0054] (a) Message type: a 01 sequence representing "NeighborM";

[0055] (b) Source area control node identifier: that is, the identifier of the control node in the area that sent the NeighborM message;

[0056] (c) Destination area control node identifier: that is, the identifier of the control node in the area to which the NeighborM message is sent;

[0057] (d) Message sequence number: used to distinguish NeighborM messages sent by control nodes within the same area at different times.

[0058] The AckNeighborM message must include the following information:

[0059] (a) Message type: a 01 sequence representing "AckNeighborM";

[0060] (b) Source area control node identifier: that is, the identifier of the control node in the area that sent the AckNeighborM message;

[0061] (c) Destination area control node identifier: that is, the identifier of the control node in the area to which the AckNeighborM message is sent;

[0062] (d) Message sequence number: that is, the message sequence number of the corresponding NeighborM message.

[0063] The information types maintained by the control nodes within the zone involved in this step include NeiborInfo and logicalClk. The content of each type of information is as follows:

[0064] NeiborInfo(H i The information includes the control node (denoted as H) within a certain area. i The virtual neighbor node information of ) can be represented as<infoClk,NeighborMem,vCost> The meanings of its parts are as follows:

[0065] infoClk: Represents this NeiborInfo(H i The timeliness of the information is determined by the value generated by the NeiborInfo(H) function. i The logicalClk value (logical clock) maintained by the control node within the information zone;

[0066] NeighborMem: Represents the control node H within the zone. i The set of virtual neighboring nodes;

[0067] vCost: Represents the control node H within the area. i The set of virtual link costs (corresponding one-to-one with the virtual neighbor nodes in NeighborMem).

[0068] NeiborInfo(H i The information in the document comes from the process of discovering and maintaining virtual neighboring nodes.

[0069] The `logicalClk` information is a value maintained locally by each control node within a zone, essentially a logical clock. Logical time measures the timeliness of information; a higher logical time indicates newer information. When the zone control node selection program starts, the value of `logicalClk` is 0 for each control node. Each control node increments its maintained `logicalClk` by 1 under the following conditions:

[0070] A) Discover new control nodes within virtual adjacent zones;

[0071] B) It was found that an existing control node in a virtual neighboring zone is no longer suitable to continue serving as a control node in a virtual neighboring zone;

[0072] C) The cost of the virtual link between the control node and the existing virtual neighboring area changes.

[0073] (2) Control nodes within each zone reach a consensus on virtual topology information through message exchange with virtual neighboring nodes; the specific process is as follows:

[0074] Control node H within each zone send Upon discovering a new virtual neighbor node H rec At that time, to H rec Send a TopoInfoMsg message containing the control node H within the area. send Maintained TopoInfo information;

[0075] For the control node within the area that receives the TopoInfoMsg message (denoted as H) rec The TopoInfoMsg message is sent to the control node H within the area. rec The system compares the TopoInfo it maintains with the system's own data and updates the TopoInfo according to the following rules:

[0076] (a) For entries included in the TopoInfoMsg message but not included in the TopoInfo information, add the entry to the TopoInfo information;

[0077] (b) For entries included in both the TopoInfoMsg message and the TopoInfo information, if the infoClk value of the entry in the TopoInfoMsg message is larger, it indicates that the entry in the TopoInfoMsg message is more timely. Therefore, the entry in the TopoInfo information should be replaced with the entry in the TopoInfoMsg message.

[0078] (c) Other entries in the TopoInfo information remain unchanged.

[0079] In addition, each zone's control node (denoted as H) srcEach node periodically sends an UpdtMsg message to all virtual adjacent intra-region control nodes, which carries the intra-region control node H... src Updt information maintained. The intra-area control node (denoted as H) that receives the UpdtMsg message. rec Forward this message to all virtual adjacent intra-area control nodes. If intra-area control node H rec From a control node within a certain area (denoted as H) send If this message is received, it will not be sent to the control node H within the area. send Forward this message.

[0080] Control node H within the area rec Based on the comparison between the TopoInfo information maintained and the UpdtMsg message, determine the further processing method for the UpdtMsg message and the method for updating the TopoInfo information. Specifically, assume that the id value in the Updt information of the UpdtMsg message is H. updt If both of the following conditions are met:

[0081] A) Control node H within the area rec The TopoInfo contains <H updt ,NeiborInfo(H updt )> Entries;

[0082] B) In the UpdtMsg message, the oldClk of the Updt information is equal to NeiborInfo(H updt infoClk in )

[0083] Then, the UpdtMsg message will not be placed in PendingUpdt, and NeiborInfo in TopoInfo will be updated as follows: updt ):

[0084] 1) infoClk: Assigns the value newClk of the Updt information in the UpdtMsg message;

[0085] 2) NeighborMem: Add the intra-area control node identifier set corresponding to the newSet of the Updt information in the UpdtMsg message, and delete the intra-area control node identifier set corresponding to the rmvSet of the Updt information in the UpdtMsg message;

[0086] 3) vCost: Update according to the vCostSet of the Updt information in the UpdtMsg message.

[0087] If both of the following conditions are met:

[0088] A) Control node H within the arearec The TopoInfo contains <H updt ,NeiborInfo(H updt )> Entries;

[0089] B) The newClk in the UpdtMsg message is less than or equal to NeiborInfo(H updt infoClk in )

[0090] Then the UpdtMsg message will not be placed in PendingUpdt, nor will TopoInfo be updated.

[0091] In other cases, the UpdtMsg message is placed in PendingUpdt, and TopoInfo is not updated.

[0092] This embodiment calculates the virtual link cost based on the actual hop count between control nodes within the area. Figure 3 As shown in the example, the virtual link cost between intra-area control node 1 and intra-area control node 2 is the number of hops of the actual path between them, which is 4.

[0093] The message types involved in this step include TopoInfoMsg and UpdtMsg messages, and the content of each message is as follows:

[0094] The TopoInfoMsg message must include the following information:

[0095] (a) Message type: a 01 sequence representing “TopoInfoMsg”;

[0096] (b) Source area control node identifier: that is, the identifier of the control node in the area that sent the TopoInfoMsg message;

[0097] (c) Virtual topology information: that is, the TopoInfo information maintained by the control node in the area that sends the TopoInfoMsg message.

[0098] The UpdtMsg message must include the following information:

[0099] (a) Message type: a 01 sequence representing "UpdtMsg";

[0100] (b) Updt information: This is a record of updates to virtual neighboring nodes and virtual links.

[0101] The information types maintained locally by the control nodes within the region in this step include TopoInfo, NeiborInfo, logicalClk, Updt, and PendingUpdt. The content of each type of information is as follows:

[0102] The TopoInfo information includes the NeiborInfo information for each control node within a given area, maintained by the control node within that area. Each entry in the TopoInfo list can be represented as...<id,NeiborInfo(id)> , where id is the identifier of the intra-area control node corresponding to the entry, and NeiborInfo(id) includes the virtual neighbor node information of the intra-area control node identified by id.

[0103] For a certain control node within a region (denoted as H) i When a new control node (denoted as H) is discovered in a neighboring area j When ), if the control node H in the area i The TopoInfo list does not contain information about the control node H within the zone. j For entries, the control node H within the area i To control node H in the area j Send a message containing the TopoInfo information it maintains (i.e., TopoInfoMsg) to enable the control node H within the area. j Quickly obtain control node H within the area j The virtual topology being maintained.

[0104] The NeiborInfo and logicalClk information have already been described in the previous step and will not be repeated here.

[0105] Updt information is a record about updates to virtual neighbor nodes and virtual links. The content of Updt can be represented as...<id,newSet,vCostSet,rmvSet,oldClk,newClk> Here, id represents the identifier of the intra-area control node that generated this update record, newSet represents the set of newly added virtual neighbor nodes of the intra-area control node identified by id, vCostSet represents the set of changed virtual link costs of the intra-area control node identified by id (including the virtual link costs corresponding to newly added virtual neighbor nodes and the changed existing virtual link costs), rmvSet represents the set of intra-area control node identifiers that are no longer "virtually adjacent" to the intra-area control node identified by id, oldClk is the logicalClk (logical clock) before updating this Updt record, and newClk is the logicalClk after updating this Updt record.

[0106] The method for generating Updt records is as follows:

[0107] With control node H in the area iBy way of example, when the previous Updt record is sent by using an UpdtMsg message, a new Updt record is generated, and the Updt record is filled in the following manner:

[0108] id: H i

[0109] newSet: empty;

[0110] vCostSet: empty;

[0111] rmvSet: empty;

[0112] oldClk: the current intra-area control node H i 's logicalClk;

[0113] newClk: empty;

[0114] The method for updating an Updt record is as follows:

[0115] A) For a certain intra-area control node (denoted as H i ), when a new adjacent intra-area control node (denoted as H j ) is found, the Updt record is updated in the following manner:

[0116] id: unchanged

[0117] newSet: add H j ;

[0118] vCostSet: add the virtual link cost between the intra-area control node H i and H j ;

[0119] rmvSet: if H j is in this set, delete H j ;

[0120] oldClk: unchanged;

[0121] newClk: the logicalClk of the intra-area control node H i after the new adjacent intra-area control node is found.

[0122] B) For a certain intra-area control node (denoted as H i ), when it is detected that the virtual link cost corresponding to an adjacent intra-area control node (denoted as H j ) changes, the Updt record is updated in the following manner:

[0123] id: unchanged

[0124] newSet: unchanged;

[0125] vCostSet: if there is already an entry corresponding to H j , update the cost value of the virtual link therein; otherwise, add a new entry corresponding to H j and assign the cost value of the virtual link therein to the new virtual link cost;

[0126] rmvSet: unchanged;

[0127] oldClk: unchanged;

[0128] newClk: the logicalClk of intra-region control node H i after the virtual link cost is changed.

[0129] C) For a certain intra-region control node (denoted as H i ), when it is found that an adjacent intra-region control node (denoted as H j ) no longer serves as its adjacent intra-region control node, the Updt record is updated in the following manner:

[0130] id: unchanged

[0131] newSet: if H j is present in the set, delete H j ;

[0132] vCostSet: if exists, remove the virtual link cost between intra-region control node H i and H j ;

[0133] rmvSet: add H j ;

[0134] oldClk: unchanged;

[0135] newClk: the logicalClk of intra-region control node H i after the virtual link cost is changed.

[0136] PendingUpdt information stores a series of UpdtMsg messages that cannot be temporarily used to update TopoInfo information. Whenever TopoInfo changes, each UpdtMsg message in PendingUpdt needs to be traversed, and it is determined in the following manner whether to move the UpdtMsg message out of PendingUpdt and whether to update TopoInfo.

[0137] For a certain UpdtMsg message in PendingUpdt, assume that the id value of the Updt information in the UpdtMsg message is Hupdt If both of the following conditions are met:

[0138] A) The TopoInfo of the control node within the area contains... <H updt ,NeiborInfo(H updt )> Entries;

[0139] B) In the UpdtMsg message, the oldClk of the Updt information is equal to NeiborInfo(H updt infoClk in )

[0140] Then remove the UpdtMsg message from PendingUpdt and update NeiborInfo in TopoInfo as follows: updt ):

[0141] 1) infoClk: Assigns the value newClk of the Updt information in the UpdtMsg message;

[0142] 2) NeighborMem: Add the intra-area control node identifier set corresponding to newSet in the UpdtMsg message, and delete the intra-area control node identifier set corresponding to rmvSet in the Updt information in the UpdtMsg message;

[0143] 3) vCost: Update the vCostSet according to the Updt information in the UpdtMsg message.

[0144] (3) Based on virtual topology information, the control nodes within each zone use the same algorithm to select consistent zone control nodes, i.e., control nodes in the low-Earth orbit constellation edge computing system; the specific process is as follows:

[0145] The virtual link cost between two control nodes within the calculation area is calculated. If the two control nodes are the same, the virtual link cost is 0.

[0146] For each control node H in the area i Calculate the control nodes H in all other zones. j Cumulative cost between The formula is as follows:

[0147]

[0148] Where, N c PC represents the number of control nodes within the area. ij Representative of the control node H within the representative area i With control node H in the area jThe virtual path cost between nodes is calculated using the shortest path algorithm with virtual topology information and virtual link costs within the control node cluster as inputs, and the cumulative cost is calculated. For the control node H in the area i The total cost of the virtual paths between the device and all other control nodes within the same zone;

[0149] The control node with the lowest cumulative cost within the zone is selected as the control node for the interval. If multiple control nodes within the zone have the same cumulative cost, the control node with the lowest or highest index number is selected as the control node for the interval.

Claims

1. A method for dynamically selecting control nodes in a low-Earth orbit constellation edge computing system, characterized in that, Includes the following steps: (1) Divide the entire constellation into multiple satellite partitions. Within each partition, the control nodes discover and maintain their own virtual neighbor node information through message interaction. (2) Control nodes in each zone reach a consensus on virtual topology information by using message interaction with virtual neighboring nodes; (3) Based on virtual topology information, the control nodes in each zone use the same algorithm to select consistent zone control nodes, i.e., control nodes in the low-orbit constellation edge computing system; The specific process of step (3) is as follows: The virtual link cost between two control nodes within the calculation area is calculated. If the two control nodes are the same, the virtual link cost is 0. For each control node within the zone Calculate the control nodes in all other zones Cumulative cost between The formula is as follows: Where, N c The number of control nodes within the area. Representatives of control nodes within the representative area With control nodes within the area The virtual path cost between nodes is calculated using the shortest path algorithm with virtual topology information and virtual link costs within the control node cluster as inputs, and the cumulative cost is calculated. Control nodes within the area The total cost of the virtual paths between the device and all other control nodes within the same zone; The control node with the lowest cumulative cost within the zone is selected as the control node for the interval. If multiple control nodes within the zone have the same cumulative cost, the control node with the lowest or highest index number is selected as the control node for the interval.

2. The method for dynamically selecting control nodes in a low-Earth orbit constellation edge computing system according to claim 1, characterized in that, The specific process of step (1) is as follows: After the entire constellation is divided into multiple satellite partitions, each partition contains multiple satellites. Within each partition, the intra-partition control node periodically sends DiscoverNeighbor messages to all neighboring satellites. The intra-partition control node receiving the DiscoverNeighbor message returns an AckDiscover message to the sending intra-partition control node. Through this message exchange, virtual neighbor node discovery between intra-partition control nodes is completed. The DiscoverNeighbor message includes the message type, intra-partition control node identifier, message sequence number, and intra-partition control node IP address. The AckDiscover message includes the message type, intra-partition control node identifier, message sequence number, intra-partition control node IP address, and virtual link cost. Each intra-area control node periodically sends NeighborM messages to all virtual neighbor nodes. Intra-area control nodes that receive NeighborM messages return AckNeighborM messages to the intra-area control node that sent them. Through message exchange, the virtual neighbor node information between intra-area control nodes is maintained. NeighborM messages and AckNeighborM messages both include message type, source intra-area control node identifier, destination intra-area control node identifier, and message sequence number.

3. The method for dynamically selecting control nodes in a low-Earth orbit constellation edge computing system according to claim 1, characterized in that, The specific process of step (2) is as follows: Control nodes within each zone When a new virtual neighbor node is discovered, a TopoInfoMsg message is sent to the virtual neighbor node, which includes the intra-area control node. The maintained virtual topology information, TopoInfo, includes the control nodes within the area. All known identifiers of control nodes within the area, virtual adjacency relationships between control nodes within the area, and virtual link costs; Control nodes within each zone UpdtMsg messages are periodically sent to all virtual neighbor nodes. The UpdtMsg messages include the control node within the area. The virtual neighbor node and virtual link update information (Updt) is maintained. The Updt includes newly added virtual neighbor nodes, changed virtual link costs, and newly deleted virtual neighbor nodes. After each UpdtMsg message is sent, the intra-area control node clears its maintained Updt information. The intra-area control node that receives the UpdtMsg message... Forward the UpdtMsg message to all virtual neighbor nodes except the intra-area control node that sent the UpdtMsg message.

Images