Inter-satellite cooperative computing offloading method, system and storage medium based on load balancing

Abstract

The application provides an inter-satellite cooperative computing offloading method and system based on load balancing and a storage medium, the method comprising: a satellite node receiving a service request from a service initiator; and completing balanced distribution of a service corresponding to the service request through the inter-satellite offloading mechanism to determine a satellite node accepting the service request; and completing a processing task of the service through the satellite node accepting the service request. The application uses each satellite node to jointly plan computing offloading of a task, thereby realizing cooperative computing and processing of multiple satellite nodes in a satellite network, solving the problems of inter-satellite computing load imbalance and inter-satellite link bandwidth congestion, and the satellite offloading range of a computing task is not limited by a line-of-sight range, so that more flexible offloading planning can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of inter-satellite computing distribution, and particularly relates to an inter-satellite cooperative computing offloading method and system based on load balancing and a storage medium. BACKGROUND

[0002] In recent years, satellite networks have attracted extensive attention from academia and industry. Compared with traditional ground mobile communication networks, satellite networks have the characteristics of global coverage at all times, little influence from terrain and disasters, etc., and can provide broad economic, social and strategic value in wide-area industrial Internet of Things, wireless broadband access in oceans and remote areas, disaster relief, military, etc. In recent years, with the continuous maturity of rocket launching and satellite technology, the network design capacity of space satellite networks is continuously increasing, and the cost of network deployment is greatly reduced.

[0003] With the emergence of new network services such as AR / VR, 4K high-definition video, and earth remote sensing, some services require the network to have strong computing power, and the wide coverage characteristics of satellites are used to achieve computing access at all times and all places. In previous satellite network research, for user equipment in remote areas and lacking of edge computing infrastructure coverage, satellites are used as relays to forward the computing tasks of user equipment to the ground cloud computing center, and the ground computing center performs computing and returns the results to the user through the satellite. However, the large distance between the satellite and the ground causes large round-trip delay, which has the defect of large computing task feedback delay; in addition, due to the limited uplink and downlink bandwidth between the satellite and the ground, a large amount of raw data without processing will be directly forwarded through the satellite, which will cause uplink and downlink congestion. Therefore, in the satellite network scenario, only using satellites as relays has great defects.

[0004] Due to the different population density of the ground, the satellite network computing task load is unevenly distributed, and at any moment, a large number of satellite resources are in an idle state, which brings feasibility to on-board computing; on-board computing refers to the fact that the satellite not only plays the role of data forwarding as a relay, but also processes the computing task. In the past research on on-board computing, the task is unloaded to a single satellite for computing through the satellite-ground link, but the resources of a single satellite are limited, which still leads to a large amount of data that cannot be processed on the satellite and then forwarded to the ground computing center, so the computing offloading method of a single satellite still cannot effectively alleviate the problems of long feedback delay and uplink and downlink congestion. In the past research on on-board computing, there is also a multi-satellite computing offloading method, that is, a task is divided into multiple parts and sent to multiple satellites for computing through the satellite-ground link. However, the quality of the satellite-ground link is affected by many factors such as weather and obstruction, and the transmission of data through the satellite-ground link is prone to packet loss; and in this method, the user can only unload the task to the satellites within the visible range, which causes a certain degree of limitation. Therefore, the reliability and flexibility of the multi-satellite computing offloading method based on the satellite-ground link are low. SUMMARY

[0005] In view of this, the embodiments of the present application provide a load balancing based inter-satellite cooperative computing offloading method, system and storage medium to eliminate or improve one or more defects in the prior art.

[0006] One aspect of the present application provides a load balancing based inter-satellite cooperative computing offloading method, which comprises:

[0007] receiving a service request from a service initiator by a satellite node in a satellite network; an inter-satellite offloading mechanism for planning and decision-making of the service request is arranged in each satellite node in the satellite network; the current satellite node receiving the service request determines whether the service corresponding to the service request from the service initiator is admitted into the network according to the inter-satellite offloading mechanism for planning and decision-making of the service request; the current satellite node determines whether to accept or forward the service request according to the inter-satellite offloading mechanism for planning and decision-making of the service request admitted into the network, determines the next satellite node to which the service request needs to be forwarded and forwards the service request to the next satellite node, so as to forward the service request hop by hop to the satellite node capable of accepting the service request; the satellite node accepting the service request feeds back the offloading planning result to the corresponding service initiator, so that the service initiator can send the service corresponding to the service request to the satellite node performing the planning and decision-making of the service request in response to the offloading planning result, so that the service is forwarded to the satellite node accepting the service request via the forwarding path of the corresponding service request; and the satellite node accepting the service request completes the processing task of the service.

[0008] In some embodiments of the present application, the base information of the corresponding service recorded in the service request comprises: the task volume of the service corresponding to the service request, the number of machine cycles required for processing the task volume of one bit in the service corresponding to the service request, the decision period of forwarding and unloading of the service request in the satellite node performing the admission plan of the service request, and the deadline of completing the processing task of the service corresponding to the service request.

[0009] In some embodiments of the present application, the method further comprises: feeding back the admission plan result to the service initiator of the service corresponding to the service request by the satellite node performing the admission plan of the service request, so that the service initiator does not send the service corresponding to the service request to the satellite node performing the admission plan of the service request.

[0010] In some embodiments of the present application, each satellite node in the forwarding path of the service request records the label information of the service corresponding to the service request, so that each satellite node determines the service request corresponding to the received service through the label information, and the service is forwarded hop by hop from the satellite node performing the admission plan of the service request to the satellite node accepting the service request via the forwarding path of the corresponding service request.

[0011] In some embodiments of the present application, each satellite node in the satellite network comprises a forwarding queue and a computing queue; the forwarding queue comprises the service request directly sent by the service initiator and the service request forwarded by the previous satellite node; and the computing queue comprises the service request accepted by the satellite node where the computing queue is located.

[0012] In some embodiments of the present application, the step of planning and deciding the service request by the inter-satellite decision mechanism comprises: determining the admission strategy and the unloading strategy of the service request, so that the total system overhead required for completing the processing task of the service corresponding to the service request by each satellite node in the satellite network reaches the minimum.

[0013] In some embodiments of the present application, the step of planning and deciding the service request from the service initiator by the current satellite node receiving the service request according to the inter-satellite unloading mechanism comprises:

[0014] determining the task load threshold of the current satellite node; comparing the total task volume of the services arriving at the current satellite node in the current decision period with the task load threshold of the current satellite node; if the total task volume of the services arriving at the current satellite node in the current decision period is less than the task load threshold of the current satellite node, the service corresponding to the service request from the service initiator received by the current satellite node in the current decision period is admitted into the network, otherwise, it is not admitted into the network.

[0015] In some embodiments of the present application, the step of planning offloading by the current satellite node according to the inter-satellite offloading mechanism for the admitted network service request comprises:

[0016] determining a system overhead threshold of the current satellite node; comparing the system overhead required for task processing of the service corresponding to the admitted network service request in the current decision cycle in the current satellite node with the system overhead threshold of the current satellite node; if the system overhead required for task processing of the service corresponding to the admitted network service request in the current decision cycle in the current satellite node is less than the system overhead threshold of the current satellite node, determining that the current satellite node accepts the network service request for offloading planning in the current decision cycle; otherwise, determining the next satellite node to which the network service request is forwarded; comparing the system overhead required for processing the task of the service corresponding to the network service request by the neighboring satellite node of the current satellite node with the weighted sum of the system backlog of the neighboring satellite node in the current decision cycle; and determining the neighboring satellite node with the minimum weighted sum as the next satellite node to which the network service request is forwarded.

[0017] Another aspect of the present application provides an inter-satellite cooperative computing system based on load balancing, comprising a processor and a memory; the memory stores computer instructions, and the processor is configured to execute the computer instructions stored in the memory, so that the system implements the steps of any one of the above-mentioned inter-satellite cooperative computing methods.

[0018] Another aspect of the present application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the steps of any one of the above-mentioned inter-satellite cooperative computing methods.

[0019] The application discloses an inter-satellite cooperative computing offloading method, system and storage medium based on load balancing. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 A flow chart of inter-satellite cooperative computing offloading in an embodiment of the application;

[0021] Figure 2 A queue model of inter-satellite cooperative computing offloading in an embodiment of the application;

[0022] Figure 3 A flow chart of a load prediction algorithm in an embodiment of the application;

[0023] Figure 4 A structure diagram of a maximum flow minimum cost algorithm in an embodiment of the application. DETAILED DESCRIPTION

[0024] In order to make the objects, technical solutions and advantages of the application clearer, the application will be further described in detail below with reference to the embodiments and drawings. Here, the illustrative embodiments of the application and the descriptions thereof are used to explain the application, but are not used to limit the application.

[0025] Here, it should be noted that, in order to avoid the application being obscured by unnecessary details, only the structures and / or processing steps closely related to the scheme according to the application are shown in the drawings, and other details not closely related to the application are omitted.

[0026] It should be emphasized that the term "comprises / comprising" when used in this specification is taken to specify the presence of stated features, integers, steps or components but does not preclude the presence or addition of one or more other features, integers, steps, components, or groups thereof.

[0027] It should also be noted that, unless otherwise specified, the term "connected" herein can mean directly connected, but can also mean indirectly connected via intervening elements.

[0028] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0029] The present application provides a load-balanced inter-satellite cooperative computing offloading method, and a main inter-satellite cooperative computing offloading process thereof is shown in Figure 1 , including steps S100-S400:

[0030] In step S100, a service request from a service initiator is received by a satellite node in a satellite network; an inter-satellite offloading mechanism for planning and decision-making of the service request is provided in each satellite node in the satellite network;

[0031] In an embodiment, the service initiator adopts a service control method, and sends a service request to the satellite node before sending the actual service to the satellite node, and the service request is provided with basic information of the corresponding service, including: i , h i , and wherein s i represents the task amount of the service i corresponding to the service request, in bit; h i represents the number of machine cycles required to complete the service amount of one bit in the service i corresponding to the service request; represents the decision period of the planning and decision-making of the service request corresponding to the service i in the planning and decision-making of the satellite node executing the service request; represents the deadline for completing the processing task of the service i corresponding to the service request by each satellite node in the satellite network.

[0032] In step S200, the planning and decision-making of each satellite node in the satellite network for the service request is determined according to the inter-satellite offloading mechanism.

[0033] In an embodiment, the queue model for planning and decision-making of the service request in the satellite node is as shown in Figure 2As shown, each satellite node in the satellite network is provided with two virtual queues, including a forwarding queue and a computing queue, the forwarding queue includes service requests directly sent from a service initiator and service requests forwarded from a previous satellite node; the computing queue includes service requests accepted by the satellite node where the computing queue is located. The service requests directly sent from the service initiator and the service requests from the adjacent satellite nodes are sent to the forwarding queue of the current satellite node, an admission policy is made for the service requests directly sent from the service initiator, the service requests directly sent from the service initiator are divided into service requests admitted into the network and service requests not admitted into the network; an offloading policy is made for the service requests admitted into the network and the service requests from the adjacent satellite nodes, the service requests are divided into service requests accepted by the current satellite node and service requests needed to be forwarded to the adjacent satellite node; the service requests needed to be forwarded to the adjacent satellite node are forwarded according to the inter-satellite offloading mechanism in each satellite node in the satellite network until the service request is accepted by a satellite node, and the satellite nodes passed through during the forwarding of the service request are taken as the forwarding path of the service request in the satellite network.

[0034] Each satellite node in the satellite network determines the admission policy and the offloading policy of the service request according to the inter-satellite offloading mechanism, and offloads the to-be-processed service from the satellite node with heavy load to the satellite node with light load according to the forwarding path of the corresponding service request. In order to control the processing task backlog of each satellite node in the satellite network, while reducing the energy consumption of the satellite node required in the process of processing the service and improving the service quality of the satellite node to the service as much as possible, the weighted sum of the system overhead and the system backlog required in the process of processing the service request corresponding to each satellite node in the satellite network is minimized, the system overhead includes system time delay overhead and system energy consumption overhead, and the system backlog includes forwarding queue backlog and computing queue backlog.

[0035] In an embodiment, based on the above queue model and the inter-satellite offloading mechanism, a computing model required for making a planning decision of the service request is set:

[0036] The complete timeline T of the offloading planning of the service request between the satellite nodes in the satellite network is divided into multiple decision periods, The decision period represents a time period required for a satellite node to make a planning decision.

[0037] The forwarding path of the service request corresponding to the service i is set as Wherein, represents a satellite node performing the admission planning of the service request corresponding to the service i, represents a satellite node accepting the service request corresponding to the service i.

[0038] Set the set of service requests in the forwarding queue of each satellite node: denotes the set of service requests in the forwarding queue of satellite node n in decision period t; Set the task amount of the service corresponding to each service request in the forwarding queue of each satellite node: wherein, denotes the sum of the task amounts of the services corresponding to each service request in the forwarding queue of satellite node n in t+1 decision period; If service request i is to be accepted by the current satellite node n, i.e., the service request corresponding to service i is transferred from the forwarding queue of satellite node n to the computing queue, record Otherwise, N n is the set of neighboring satellite nodes of satellite node n, β n (t) denotes the task amount of the service corresponding to the service request transferred out of the current satellite node n in decision period t, β nm (t) denotes the task amount of the service corresponding to the service request transferred from satellite node n to satellite node m in decision period t, if the service request corresponding to service i is transferred from satellite node n to satellite node m, record Otherwise π n (t) denotes the sum of the arrival rates of the services corresponding to all admitted service requests to satellite node n in decision period t, denotes the arrival rate of service i corresponding to the admitted service request to satellite node n in decision period t, I n (t) denotes the set of services i in satellite node n in decision period t.

[0039] Set the set of service requests in the computing queue of each satellite node: denotes the set of service requests in the computing queue of satellite node n in decision period t; Set the task amount of the service corresponding to each service request in the computing queue of each satellite node: wherein, denotes the sum of the task amounts of the services corresponding to each service request in the computing queue of satellite node n in t+1 decision period; denotes the amount of service i completed by satellite node n in decision period t, τ denotes the number of decision periods; wherein, denotes the processing frequency of service i by satellite node n in decision period t, c n denotes the computing capacity of satellite node n, i.e., the number of machine cycles executed per second; denotes the number of service requests in the computing queue of satellite node n executed in decision period t.

[0040] Set the processing delay of each satellite node for service i: wherein, d i，Cdenotes the processing delay of service i in the satellite node.

[0041] Setting the processing energy consumption of the satellite node for service i: where κ is a parameter, usually set to 10 -26 , e i，C denotes the energy consumption of service i in the satellite node.

[0042] Setting the total offloading delay required for service i to make a planning decision in each satellite node: wherein denotes the link bandwidth between the jth and (j+1)th satellite nodes in the offloading path of service request i, denotes the transmission delay between the jth and (j+1)th satellite nodes in the offloading path of service request i, denotes the propagation delay between the jth and (j+1)th satellite nodes in the offloading path of service request i, j represents the serial number of the satellite node on the task forwarding path.

[0043] Setting the total transmission energy consumption required for service i to make a planning decision in each satellite node: wherein denotes the transmission power of the jth satellite node.

[0044] Based on the calculation model set in the above embodiment, the inter-satellite offloading mechanism of each satellite node in the satellite network is determined and to obtain the total system overhead required for each satellite node in the satellite network to complete the processing task of the service corresponding to the service request, wherein N n denotes the set of adjacent satellite nodes of the current satellite node n, denotes the set of current satellite nodes in the satellite network that execute the inter-satellite offloading mechanism. The optimization problem model with the minimum sum of system overhead is:

[0045]

[0046] wherein C(t) denotes the system overhead at decision cycle t, ρ d is the weight coefficient of delay, ρ e is the weight coefficient of energy, denotes the calculation delay of service request i in satellite node n in the system overhead C; denotes the energy consumption of service request i in satellite node n for calculation in the system overhead C; b nm denotes the link bandwidth between satellite node n and satellite node m; denotes the transmission delay of service request i between satellite node n and satellite node m; T nm denotes the propagation delay of service request i between satellite node n and satellite node m; denotes the transmission power between the j-1th satellite node and the jth satellite node in the offloading path of the service request, T denotes the transmission time of the service request, n j denotes the transmission path between the j-1th satellite node and the jth satellite node in the offloading path of the service request.

[0047] When making planning decisions according to the service information in the forwarding queue in each satellite node in the above-mentioned satellite network, the processing delay and energy consumption required for the current satellite node n to perform the processing task of the service i corresponding to the service request are calculated, that is, to determine whether the current satellite node n accepts the service request corresponding to the service i, thereby determining the value of The value of is 0 or 1; the transmission delay and transmission energy consumption required for forwarding the service i from the current satellite node n to the adjacent satellite node m are calculated, that is, to determine the adjacent satellite node m to which the service request needs to be forwarded by the current satellite node n, thereby determining the value of The value of is 0 or 1.

[0048] At this time, in order to realize the above-mentioned optimization model that the sum of the total system overhead required for completing the processing task of the service corresponding to the service request is minimum, the constraint conditions required to be met include:

[0049] 1. The offloading result of the service request is limited to ensure that the offloading result of the admitted service request includes the calculation queue moved into the local satellite node, the forwarding queue forwarded to the adjacent satellite node, and the forwarding queue remaining in the current satellite node:

[0050] 2. The forwarding behavior is limited to ensure that forwarding is only performed to satellite nodes with lower delay:

[0051] 3. The calculation load of the service corresponding to the service request is limited to ensure that the calculation load of the service corresponding to the service request moved into the calculation queue of the local satellite node is lower than the calculation load capacity of the local satellite node: E n denotes the calculation load capacity of the satellite node n;

[0052] 4. The forwarding task amount of the service request in the satellite node is limited to ensure that the forwarding task amount of the service request forwarded to the adjacent node is lower than the forwarding capacity between two satellite nodes: where B nm represents the forwarding capacity of forwarding a service request from a satellite node n to a satellite node m;

[0053] 5. Limiting the system load of each satellite node in the satellite network so that the sum of the forwarding queue and the computing load of the computing queue in each satellite node remains stable: where E represents the computing load.

[0054] Since the solution search space, network topology scale, time range and service quantity required by the optimal solution solving process of the optimization problem model with the minimum sum of system overheads are too large, the calculation complexity of the centralized and offline solving method is too large. Therefore, in this embodiment, the Lyapunov optimization framework is used to convert the long-term offline optimization problem into an online solving problem, and convert the centralized optimization target into a distributed optimization target.

[0055] The Lyapunov function is:

[0056] where Θ(t) = [Q F (t), Q B (t)] represents the set of queue vectors of each satellite node in the satellite network, where represents the task set of the service corresponding to the service request in the forwarding queue of all satellite nodes, represents the task set of the service corresponding to the service request in the computing queue of all satellite nodes, indicates a threshold set artificially to provide a reference for planning decisions, and in an ideal state, ||Q F (t)-θ|| should be close to 0 in all decision-making periods of the offloading decision;

[0057] The Lyapunov drift is Δ(Θ(t)) = E{L(Θ(t+1))-L(Θ(t)|Θ(t))}, and by limiting the Lyapunov drift Δ(Θ(t)) to remain stable during the planning decision process of the service request in each satellite node in the satellite network, the limitation of the system load of each satellite node in the satellite network can be met.

[0058] After the optimization of the above Lyapunov optimization framework, the online solving problem model obtained is min π(t)，s(t) Δ(Θ(t))+V·E{C(t)|Θ(t)}, where V represents an artificial setting parameter for balancing the system load and the system overhead, and E indicates the operator for taking the mean.

[0059] The online problem solving model is combined with the calculation model set in the above inter-satellite offloading mechanism; after scaling change, an optimization target of a single satellite node planning and decision-making for a service request is formed, so as to find a satellite node with the minimum weighted sum of calculation time delay, calculation energy consumption, transmission time delay, transmission energy consumption, propagation time delay, forwarding queue and calculation queue backlog of the service corresponding to the service request in adjacent satellite nodes, as the next satellite node for forwarding the service request. The optimization target of a single satellite node planning and decision-making for a service request is:

[0060]

[0061] wherein, represents the task amount of the service corresponding to the admitted service request in the forwarding queue of the satellite node n in the decision-making period t, θ n indicates a threshold set artificially, if x≤0, then otherwise represents a set of satellite nodes in the satellite network, n represents a satellite node in the satellite network, I n (t) represents a set of service initiators accessing the satellite node n in the decision-making period t, represents the arrival rate of the service request i to the satellite node n in the decision-making period t; represents the task amount of the service request from the service initiator in the decision-making period t; Θ(t) represents the queue vector of the forwarding queue and the calculation queue in the decision-making period t.

[0062] At this time, in order to achieve the above optimization target of a single satellite node planning and decision-making for a service request, the following limiting conditions need to be met:

[0063] 1. The offloading result of the service request is limited to ensure that the offloading result of the admitted service request includes moving into the calculation queue of the local satellite node, forwarding to the forwarding queue of the adjacent satellite node and continuing to stay in the forwarding queue of the current satellite node:

[0064] 2. The forwarding behavior is limited to ensure that forwarding is only performed to a satellite node with lower time delay:

[0065] 3. The calculation load of the service corresponding to the service request is limited to ensure that the calculation load of the service corresponding to the service request moved into the calculation queue of the local satellite node is lower than the task load capacity of the local satellite node: E n represents the task load capacity of the satellite node n;

[0066] 4. Limiting the forwarding task amount of the service request in the satellite node to ensure that the forwarding task amount of the service request forwarded to the adjacent node is lower than the forwarding capacity between the two satellite nodes: where B nm represents the forwarding capacity of the service request from the satellite node n to the satellite node m;

[0067] In the solving process of the optimization objective of the above-mentioned single satellite node for planning and decision-making of the service request, the values of and are also related to the load of the current satellite node in the future decision-making period. Therefore, in this embodiment, the delay online learning is used to predict the load of the current satellite node n in the future decision-making period, so as to determine the values of and . The load of the satellite node n in each decision-making period after the decision-making period t is predicted, and the algorithm flow of the delay online learning of and is shown in Figure 3 , which includes determining the prediction step of the satellite node and the actual calculation load of the current decision-making period t. Predicting the load of the calculation queue of the satellite node n in the future decision-making period: where nt is the gradient at ; the prediction result of the load of the calculation queue of the satellite node n in the future decision-making period is informed to the adjacent satellite node, and the load prediction result of the adjacent satellite node is received; for each task corresponding to the service request in , the time delay and energy consumed by the task in the calculation of the node n and any one node m adjacent to the node n are predicted, and the specific process is as follows: setting a variable s' for representing the remaining task amount which has not been calculated, a variable t' for representing the decision-making period, and a variable e' for representing the energy consumed in the calculation of a task; initializing the above three variables into s i , t, 0 respectively; when s' > 0, calculating the processing frequency of the task i in the decision-making period t' in the satellite node m as calculating the service amount of the task i completed by the satellite node m in the decision-making period t' as calculating the energy e' consumed by the satellite node m in the calculation of the task i as resetting the value of s' as s' - e' * f resetting the value of t' as t' + 1; judging whether s' > 0, if yes, repeating the above steps, and if no, obtaining is t-t', e is e', and t' is t-t'. is e'; repeating the above steps for the next node m until the node n and the neighboring nodes of n complete the above steps; repeating the above steps for the next task in until all tasks in and complete the above steps. Figure 3 The algorithm aims to minimize the time delay and energy consumption of the tasks calculated on different nodes by the algorithm. as part of the input of the optimization goal, determine which node in different tasks are offloaded to for calculation, thereby completing the planning and decision of the inter-satellite offloading mechanism in this embodiment.

[0068] In step S210, the satellite node plans admission for the service request accessed by the satellite node according to the inter-satellite offloading mechanism, that is, determines whether the service corresponding to the service request accessed by the satellite node is admitted into the network.

[0069] The step of planning admission for the service request from the service initiator by the current satellite node receiving the service request according to the inter-satellite offloading mechanism includes: determining the task load threshold of the current satellite node; comparing the total task amount of the services arriving at the current satellite node in the current decision period with the task load threshold of the current satellite node; if the total task amount of the services arriving at the current satellite node in the current decision period is less than the task load threshold of the current satellite node, the service corresponding to the service request from the service initiator received by the current satellite node in the current decision period is admitted into the network, otherwise, it is not admitted into the network.

[0070] According to the optimization goal obtained from the analysis conclusion of the planning and offloading decision of the service request according to the inter-satellite offloading mechanism in the above embodiment, in this embodiment, the decision basis for the admission planning of the service request is set as: wherein E n represents the task load threshold of the current satellite node n.

[0071] In step S220, the current satellite node plans offloading for the service request admitted into the network according to the inter-satellite offloading mechanism, determines whether to accept or forward the service request, determines the next satellite node to which the service request needs to be forwarded and forwards the service request to the next satellite node, so as to forward the service request hop by hop to the satellite node capable of accepting the service request.

[0072] In the embodiment, based on the planning decision obtained according to the inter-satellite offloading mechanism, from the satellite node performing the access planning of the service request to the satellite node accepting the service request, all the satellite nodes through which the service request is forwarded in the process of hop-by-hop forwarding of the service request according to the offloading planning in the satellite network form a forwarding path of the service request according to the forwarding order of the service request; and each satellite node in the forwarding path of the service request records the label information of the service corresponding to the service request, so that each satellite node determines the service request corresponding to the received service through the label information, so that each satellite node can hop-by-hop forward the subsequently received service from the satellite node performing the access planning of the service request to the satellite node accepting the service request via the forwarding path of the corresponding service request.

[0073] The step of offloading the service request admitted into the network by the current satellite node according to the inter-satellite offloading mechanism includes:

[0074] determining the system overhead threshold of the current satellite node; comparing the system overhead required for task processing of the service corresponding to the service request admitted into the current satellite node in the current decision period with the system overhead threshold of the current satellite node; if the system overhead required for task processing of the service corresponding to the service request admitted into the current satellite node in the current decision period is less than the system overhead threshold of the current satellite node, determining that the service request for which offloading planning is performed in the current decision period is accepted by the current satellite node; otherwise, determining the next satellite node to which the service request is forwarded. According to the analysis result of the optimization problem model that the sum of the total system overhead required for each satellite node in the satellite network to complete the processing task of the service corresponding to the service request is minimum in the above embodiment, in the embodiment, the decision basis for accepting the service request by the current satellite node is: C n represents the system overhead threshold of the current satellite node n.

[0075] The system overhead required for processing the task of the service corresponding to the service request by the neighboring satellite node of the current satellite node and the weighted sum of the system backlog of the neighboring satellite node in the current decision period are compared; the neighboring satellite node with the minimum weighted sum is determined as the next satellite node to which the service request is forwarded. By calculating the system overhead required for processing the task of the service corresponding to the service request in the forwarding queue of the current satellite node n, the satellite node with the minimum weighted sum of the processing delay, processing energy consumption, transmission delay, transmission energy consumption, propagation delay, forwarding queue and calculation queue backlog required for processing the service corresponding to the service request is selected as the next satellite node in the forwarding path of the service request. According to the optimization target obtained from the analysis conclusion in the above embodiment for planning and offloading decisions of service requests according to the inter-satellite offloading mechanism, in the embodiment, the decision basis for the offloading path of the service request in a single decision period of a single satellite node is:

[0076] wherein s n (t) represents the planning decision of the service request in the current satellite node to determine the next satellite node n to which the service request is forwarded, represents delay nm represents T nm .

[0077] In the embodiment, the maximum flow minimum cost algorithm is used to make decisions on the offloading path of the service request in a single decision period of a single satellite node; based on the forwarding queue of the satellite node n in the decision period t, as shown in Figure 4 , a maximum flow minimum cost graph is constructed, the virtual source node S in the graph is the service initiator, the task layer includes the service corresponding to each service request signaling, the node layer includes the neighboring satellite nodes m1, m2,... of the current satellite node n and the current satellite node n, and the virtual target node is the satellite node that completes the processing task of the service. The calculation rule of the system overhead of the edge in the graph is: for each vertex in the task layer, if the above-mentioned restriction on the forwarding behavior is met, a directed edge to the corresponding neighbor node is established, the capacity limit of the edge is 1, and the overhead is Each vertex in the task layer establishes a directed edge to the vertex n in the node layer, the capacity limit of the edge is 1, and the overhead is Each vertex in the node layer establishes a directed edge to the virtual target node, the capacity limit of the edge from the vertices m1, m2,... to the virtual target node is B nm , which meets the above-mentioned restriction on the calculation load of the service corresponding to the service request, and the overhead is 0, the capacity limit of the edge from the vertex n to the virtual target node is E nsatisfy the limitation on the forwarding task amount of the service request in the satellite node. In combination with the capacity limitation of all edges, the limitation on the offloading result of the service request can be satisfied; thus, the decision on the offloading path of the service request by the maximum flow algorithm with the minimum cost as the target can satisfy the optimization target of the offloading path of each service request.

[0078] In step S300, the satellite node accepting the service request feeds back the offloading planning result to the corresponding service initiator, so that the service initiator can send the service corresponding to the service request to the satellite node performing the admission planning of the service request in response to the offloading planning result.

[0079] In step S300, the satellite node performing the admission planning of the service request also feeds back the admission planning result to the service initiator corresponding to the service request not admitted, so that the service initiator does not send the service corresponding to the service request to the satellite node performing the admission planning of the service request.

[0080] In an embodiment, the feedback step of the offloading planning result includes: for the service request not admitted to the network, the satellite node performing the admission planning of the service request feeds back the admission planning result of the service request not admitted to the network to the corresponding service initiator, and sends the service request to the corresponding service initiator, so that the service initiator does not send the corresponding service to the satellite node performing the admission planning of the service request; for the accepted service request, the satellite node accepting the service request feeds back the offloading planning result of the service request to the corresponding service initiator, and sends the service request to the corresponding service initiator, so that the service initiator sends the corresponding service to the satellite node performing the admission planning of the service request.

[0081] In step 400, the satellite node accepting the service request completes the processing task of the service.

[0082] In an embodiment, the satellite node receiving the service corresponding to the service request initiated by the service initiator according to the access plan of the service request, determines the service request corresponding to the received service according to the recorded label information of the corresponding service, and determines whether the service needs to complete the corresponding processing task at the current satellite node or needs to be forwarded to the next satellite node according to the unloading plan of the current satellite node; the service is forwarded hop by hop in each satellite node in the satellite network according to the forwarding path of the service request, from the satellite node executing the access plan of the service request to the satellite node accepting the service request, and the processing task of the service is completed by the satellite node accepting the service request, and the forwarding unloading of the service is completed. The processing task of the service is completed by the satellite node accepting the service request, and the processing result of the service is fed back to the corresponding service initiator.

[0083] Corresponding to the above method, the application also provides an inter-satellite cooperative computing system based on load balancing, comprising a processor and a memory; the memory stores computer instructions, and the processor is used to execute the computer instructions stored in the memory; when the computer instructions are executed by the processor, the system implements the steps of the above-mentioned inter-satellite cooperative computing method based on load balancing.

[0084] The embodiment of the application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the above-mentioned inter-satellite cooperative computing method based on load balancing. The computer readable storage medium can be a tangible storage medium, such as a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable storage disk, a CD-ROM, or any other form of storage medium known in the technical field.

[0085] Those of ordinary skill in the art should understand that the example components, systems and methods described in connection with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether the implementation is in hardware or software depends on the specific application and design constraints imposed on the overall system. Those of skill can use various approaches to implement the described functionality depending on the specific application and design constraints. Skilled artisans should understand that the scope of the application is not limited to a particular implementation approach. When implemented in hardware, the hardware can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), a suitable firmware, a plug-in, a functional card, etc. When implemented in software, the elements of the application are the program or code segments to perform the necessary tasks. The program or code segments can be stored in a machine readable medium, or transmitted by a data signal carried in a carrier wave over a transmission medium or communication link.

[0086] It is to be expressly understood that the invention is not limited to the specific configurations and process described above and illustrated in the accompanying drawings. For the sake of clarity, detailed descriptions of known methods are omitted. In the above-described embodiments, several specific steps are described and illustrated as examples. However, the method processes of the present invention are not limited to the specific steps described and illustrated, and various changes, modifications and additions can be made thereto by one of ordinary skill in the art without departing from the spirit of the present invention, and the order of the steps can be changed.

[0087] In the present invention, features described and / or illustrated with respect to one embodiment can be used in the same or a similar way in one or more other embodiments, and / or in combination with or instead of features of other embodiments.

[0088] The above description is merely illustrative of the application, and is not intended to limit the scope of the application. Various modifications and changes can be made by one of ordinary skill in the art without departing from the spirit and scope of the application. Any modification, equivalent replacement, improvement, and the like made within the spirit and principle of the application should be included in the scope of the application.

Claims

1. A load balancing based inter-satellite cooperative computing offloading method, characterized in that, The method comprises: receiving, by a satellite node in a satellite network, a service request from a service initiator; each satellite node in the satellite network is provided with an inter-satellite offloading mechanism for planning and decision of the service request; planning and decision of the service request according to the inter-satellite offloading mechanism, comprising: determining an admission policy and an offloading policy of the service request, so that the total system overhead required for each satellite node in the satellite network to complete the processing task of the service corresponding to the service request is minimized; admission planning of the service request from the service initiator according to the admission policy by a current satellite node receiving the service request, to determine whether the service corresponding to the service request from the service initiator is admitted into the network; offloading planning of the service request admitted into the network according to the offloading policy by the current satellite node, to determine whether the current satellite node accepts the service request or forwards the service request, and to determine the next satellite node to which the service request needs to be forwarded and forward the service request to the next satellite node, so as to forward the service request hop by hop to a satellite node capable of accepting the service request; feedback of the offloading planning result to the corresponding service initiator by the satellite node accepting the service request, so that the service initiator can send the service corresponding to the service request to the satellite node performing the admission planning of the service request in response to the offloading planning result, so that the service is forwarded to the satellite node accepting the service request via the forwarding path of the corresponding service request; completion of the processing task of the service by the satellite node accepting the service request.

2. The method of claim 1, wherein, The basic information of the corresponding service is recorded in the service request, including: the task amount of the service corresponding to the service request, the number of machine cycles required for processing the task amount of one bit in the service corresponding to the service request, the decision period of forwarding offloading of the service request in the satellite node performing the admission planning of the service request, and the deadline for completing the processing task of the service corresponding to the service request.

3. The method of claim 1, wherein, The method further comprises: feedback of the admission planning result to the service initiator of the service request not admitted by the satellite node performing the admission planning of the service request, so that the service initiator does not send the service corresponding to the service request to the satellite node performing the admission planning of the service request.

4. The method of claim 1, wherein, The method further comprises: each satellite node in the forwarding path of the service request records the tag information of the service corresponding to the service request, so that each satellite node determines the service request corresponding to the received service through the tag information, so that the service is forwarded hop by hop from the satellite node performing the admission planning of the service request to the satellite node accepting the service request via the forwarding path of the corresponding service request.

5. The method of claim 1, wherein, Each satellite node in the satellite network comprises a forwarding queue and a computing queue; the forwarding queue comprises service requests directly sent from a service initiator and service requests forwarded from a previous satellite node; the computing queue comprises service requests accepted by the satellite node where the computing queue is located.

6. The method of claim 1, wherein, The step of planning admission of the service request from the service initiator according to the inter-satellite offloading mechanism by the current satellite node receiving the service request comprises: determining a task load threshold of the current satellite node; comparing a total task amount of the services arriving at the current satellite node in the current decision period with the task load threshold of the current satellite node; if the total task amount of the services arriving at the current satellite node in the current decision period is less than the task load threshold of the current satellite node, the services corresponding to the service requests from the service initiator received by the current satellite node in the current decision period are admitted into the network, otherwise not admitted into the network.

7. The method of claim 1, wherein, The step of planning offloading of the admitted service request according to the inter-satellite offloading mechanism by the current satellite node comprises: determining a system overhead threshold of the current satellite node; comparing a system overhead required for processing the tasks of the services corresponding to the admitted service requests in the current satellite node in the current decision period with the system overhead threshold of the current satellite node; if the system overhead required for processing the tasks of the services corresponding to the admitted service requests in the current satellite node in the current decision period is less than the system overhead threshold of the current satellite node, determining that the current satellite node accepts the service request for offloading planning in the current decision period; otherwise determining a next satellite node to which the service request is forwarded; comparing a system overhead required for processing the tasks of the services corresponding to the service request by the neighboring satellite node of the current satellite node with a weighted sum of the system overheads of the neighboring satellite nodes in the current decision period; determining the neighboring satellite node with the minimum weighted sum as the next satellite node to which the service request is forwarded.

8. A load-balancing based inter-satellite cooperative computing system, characterized in that, comprising a processor and a memory; The memory stores computer instructions, and the processor is configured to execute the computer instructions stored in the memory, so that the system implements the steps of the method according to any one of claims 1 to 7.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps of the method according to any one of claims 1 to 7.

Images