Space computing power network task migration method and system based on dynamic environment perception

Abstract

The invention discloses a space computing power network task migration method and system based on dynamic environment perception, and relates to spaceflight electronics and data processing technologies, and the method comprises the steps: determining that a user node obtains a computing resource needed by a computing power service from a satellite node in a space computing power network; calculating a task migration moment caused by satellite node movement and an alternative satellite set capable of receiving tasks according to the dynamic environment information; based on a candidate satellite set, an optimal satellite is screened out according to different user expected targets, before the migration moment arrives, a calculation task is migrated to the selected satellite in advance, and seamless continuity of services is guaranteed; and for the high-priority task, carrying out redundancy migration on the alternative scheme without computing resource intersection. According to the method, the problems of low efficiency, service interruption caused by sudden faults, multi-target optimization conflicts, resource shortage scene migration failure and the like of a traditional migration algorithm in a spatial network topology dynamic scene can be effectively solved.

Description

Technical Field

[0001] The present application relates to the fields of aerospace electronics and data processing technologies, and in particular, to a method and system for task migration of a space computing power network based on dynamic environment perception. Background Art

[0002] With the rapid promotion of commercial aerospace, commercial off-the-shelf (COTS) devices are widely used. Some high-performance and low-cost ground devices are simply fortified and then operate in the space environment. For example, the spaceborne supercomputer Spacebourne independently designed and developed by Hewlett-Packard Company was sent to the International Space Station. With the relatively strong computing power on the satellite, communication satellites will gradually expand their capabilities from "communication relay nodes" to "computing service nodes". By "moving" the computing to the satellite, the satellite Internet will gradually transform from a "satellite communication network" to a "space computing power network", thus significantly improving the service response speed, reducing the communication pressure caused by data backhaul, and getting rid of the dependence on ground systems.

[0003] However, limited by power consumption, volume, weight, etc., the types and quantities of resources that a single satellite can carry are limited, resulting in the computing power of a single satellite still being very limited. The space computing power network needs to organize the computing resources scattered on satellites and on the ground through the satellite network in a coordinated manner, build a distributed cloud environment with multi-satellite collaboration and satellite-ground collaboration, realize resource assistance and task collaboration, and obtain a multiplicative effect, which is a technical bottleneck that must be broken through in the future construction of satellite communication networks. Compared with traditional fixed networks, considering that medium and low Earth orbit satellites are in high-speed movement, the topology of the space computing power network has the characteristics of dynamic time-variation. On the one hand, the interconnection relationship between network nodes is in dynamic change, and on the other hand, the relationship between network nodes and user nodes is also in dynamic change. As a result, how to achieve stable and reliable algorithm services in a high-speed dynamic network scenario, task migration is an urgent problem to be solved.

[0004] Different from the static migration algorithms in the scenarios where base stations are fixed in cloud computing or edge computing, the space network represented by medium and low Earth orbit satellites is always in high-speed movement. When performing task migration, it is necessary to consider the dynamicity of the space network topology and the characteristics such as sudden abnormal failures caused by the complex electromagnetic environment affecting space links. When traditional static migration algorithms face dynamic network topologies and breakthrough failures, they will face disadvantages such as frequent task migrations, low migration success rates, and high migration costs.

[0005] In the prior art, task scheduling in satellite edge computing scenarios mostly relies on preset resource allocation strategies or simple dynamic adjustment mechanisms. Patent CN115967433A proposed a satellite edge computing offloading method based on dynamic time-variance, which mainly focuses on the matching degree between the requested resources of tasks and the remaining resources of satellites; CN116347623A is mainly based on the communication duration and execution duration between the current satellite and the candidate satellites. CN117544218A proposed an inter-satellite autonomous migration scheduling method for space-based computing tasks, which is mainly based on a scheduling action network trained on the ground; the satellite-based service migration algorithm proposed in patent CN119155742A optimizes service migration through a particle swarm algorithm. However, the existing task migration methods do not fully consider factors such as the stability of communication links and signal interference, and lack a redundancy mechanism. Once the communication between the target satellite and the task source is interrupted, there is a problem of insufficient reliability. Therefore, in order to achieve continuous and reliable task guarantee in a high-dynamic environment, it is necessary to solve the regular task migration problem triggered by the dual dynamics of high-speed movement of satellite nodes and user nodes, and also solve the sudden task migration problem caused by abnormal failures of satellite nodes or links, so as to greatly improve the reliability of task migration.

[0006] Problems existing in the prior art include: traditional algorithms do not consider the coupling relationship between the time-variance of satellite network topology and user movement trajectories, and have insufficient dynamic adaptability; there is a lack of a redundant migration guarantee mechanism in scenarios of sudden failures (such as electromagnetic interference and node downtime), resulting in reliability defects; a joint optimization model for multi-dimensional quality of service (delay, stability, resource matching) has not been established, and resource optimization is limited; when the resources of candidate nodes are insufficient, there is a lack of task splitting and reconstruction capabilities, and extreme scenarios are not considered. Summary of the Invention

[0007] The embodiments of the present application provide a task migration method and system for a space computing power network based on dynamic environment perception, which can make full use of prior information such as satellite ephemeris information and user trajectories to perform task migration in advance, and can effectively overcome the problem of low efficiency of traditional migration algorithms in dynamic scenarios of space network topology.

[0008] The embodiments of the present application provide a task migration method for a space computing power network based on dynamic environment perception, which is applied to adaptive task migration in scenarios of satellite node movement, user node movement, and sudden link failures, and includes:

[0009] At the initial moment T0, determine the computing resource C0 occupied by the user node to obtain computing power services from the satellite node S0 in the space computing power network;

[0010] According to the dynamic environment information of satellite movement, calculate the moment T of task migration caused by the movement of satellite nodes 1， The dynamic environment information includes: ephemeris information, user trajectory information, real-time resource status;

[0011] Calculate a set of K alternative satellites that can receive tasks at time T1 according to the dynamic environment information;

[0012] Determine the computing resource situation of all satellites in the alternative satellite set at time T1 to determine a set of K1 candidate satellites whose remaining computing resources are greater than or equal to the computing resource requirement C0 of the task;

[0013] Based on the set of candidate satellites, calculate the difference ΔC between the resources of the i-th candidate satellite and the resources required for task migration i =C i -C0, 1 ≤ i ≤ K1;

[0014] Based on the set of candidate satellites, calculate the transmission duration DTi required to complete the migration of the task from satellite node S0 to the i-th candidate satellite, 1 ≤ i ≤ K1, where the transmission duration includes the transmission duration from satellite node S0 to the candidate satellite and the duration required for the migration task to be completely deployed to the candidate satellite;

[0015] According to the set of candidate satellites, calculate that the i-th candidate satellite receives the task, and calculate the next migration time T 2i to obtain the time interval ΔT from when the i-th candidate satellite receives the task to the next task migration at the next moment i , ΔT i =T 2i -T1, 1 ≤ i ≤ K1;

[0016] Filter the task migration target according to the current priority or demand characteristic information of the user and the task;

[0017] Migrate the computing task to the target satellite for migration.

[0018] The embodiment of the present application also provides a space computing power network task migration system based on dynamic environment perception, including a processor and a memory. A computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the method for migrating tasks in the space computing power network based on dynamic environment perception as described above are implemented.

[0019] The method of the embodiment of the present application makes full use of prior information such as satellite ephemeris information and user trajectories, and performs task migration in advance, which can effectively overcome the problem of low efficiency of traditional migration algorithms in the dynamic scenario of the space network topology.

[0020] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the following specifically illustrates the specific implementation manners of the present application. Description of the Drawings

[0021] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present application. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0022] Figure 1 Schematic diagram of the task migration scenario of the spatial network migration method based on prior information according to an embodiment of the present application;

[0023] Figure 2 Schematic diagram of the basic process of the spatial network migration method based on prior information according to an embodiment of the present application;

[0024] Figure 3 Schematic diagram of the task migration process of the spatial network migration method based on prior information according to an embodiment of the present application;

[0025] Figure 4 Schematic diagram of the migration method of the redundancy mechanism of the spatial network migration method based on prior information according to an embodiment of the present application. Detailed Embodiments

[0026] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be completely conveyed to those skilled in the art.

[0027] Task migration refers to the technology in which a certain node migrates a task to other nodes due to special reasons. At present, there is no research on the task migration problem in the distributed computing application scenario for satellite communication networks. However, in terrestrial distributed computing such as cloud computing and edge computing, task migration is an important research content. In the edge computing scenario, task migration mainly refers to the scenario where computing tasks are migrated from users to mobile edge base stations or even cloud servers; according to the number of users, task migration optimization is mainly divided into two categories: the optimization of task migration strategies in the single-user scenario can be transformed into an optimization problem with task delay and energy consumption as the objectives; the optimization of task migration strategies in the multi-user scenario can usually be modeled as a game problem, and the optimal migration strategy is solved through game theory methods to achieve the optimization of task delay and energy consumption.

[0028] For the task migration scenario triggered by the high-speed movement of satellite network nodes or user nodes, the periodic law of satellite network topology changes can be fully utilized, combined with the ephemeris information of the satellite constellation and the movement trajectory of the user, etc., to predict the trigger time of task migration in advance. Among them, how to select the appropriate migrating satellite and determine the advance migration time t is the key. This project aims to minimize the cost of task migration delay, select the appropriate task migration satellite, complete the task migration a certain time in advance, and achieve a migration that is imperceptible to users, such as Figure 1 shown. For non-deterministic users who cannot provide movement trajectories, the next position prediction can be made based on the user's historical location information and behavior, and optimization can be carried out with the goal of reducing task interruption caused by task migration.

[0029] The embodiment of the present application provides a spatial network migration method based on prior information, which is applied to adaptive task migration in multiple scenarios such as satellite node movement, user node movement, and sudden link failures, such as Figure 1 shown. For the task migration scenario triggered by satellite node movement, at time T1, the on-board computing unit A is responsible for observing and processing the target. At time T2 - Δt1, the on-board computing unit A starts to migrate task data to B, and just completes the task migration after Δt1 time. At time T2, the on-board computing unit B will just relay to observe the target ship; for the migration scenario triggered by user movement, at time T3, the on-board computing unit C is responsible for observing the target. At time T4 - Δt2, C starts to migrate the relevant data of the observation task to D. At time T4, the target enters the visible range of D and the task data migration is completed, and D will relay to observe the target aircraft. Such as Figure 2 shown, the dynamic environment-aware spatial computing power network task migration method of the embodiment of the present application includes:

[0030] Initial service state. In step S101, at the initial moment T0, determine the computing resource C0 occupied by the user node to obtain computing power services from the satellite node S0 in the spatial computing power network.

[0031] Migration time prediction. In step S102, according to the dynamic environment information of satellite movement, calculate the time T of task migration triggered by satellite node movement 1， The dynamic environment information includes: ephemeris information, user trajectory information, real-time resource status.

[0032] Candidate satellite set construction. In step S103, according to the dynamic environment information, calculate the set of K alternative satellites that can receive tasks at time T1.

[0033] Candidate satellite resource screening. In step S104, determine the remaining (idle) computing resources Ci of all satellites in the alternative satellite set at time T1, 1 ≤ i ≤ K, to determine a set of K1 candidate satellites with remaining computing resources Ci (1 ≤ i ≤ K) greater than or equal to the computing resource requirement C0 of the task.

[0034] Resource redundancy parameter calculation. In step S105, based on the candidate satellite set, calculate the difference ΔC between the resources of the i-th candidate satellite and the resources required for task migration i =C i -C 0， 1 ≤ i ≤ K 1；

[0035] Transmission duration parameter calculation. In step S106, based on the candidate satellite set, calculate the transmission duration DTi required to complete the migration of the task from S0 to the i-th candidate satellite, 1 ≤ i ≤ K1. The transmission duration includes the transmission duration from S0 to the candidate satellite and the duration required for the migrated task to be completely deployed on the candidate satellite.

[0036] Next migration parameter calculation. According to the candidate satellite set, calculate the time T when the i-th candidate satellite receives the task, and calculate the next migration time T 2i to obtain the time interval ΔT from when the i-th candidate satellite receives the task to the next task migration at the next moment i ,ΔT i =T 2i -T1, 1 ≤ i ≤ K1.

[0037] In step S107, according to the current priority or demand characteristic information of the user and the task, screen the task migration target.

[0038] In step S108, migrate the computing task to the target satellite for migration.

[0039] In some embodiments, screening the task migration target according to the current priority or demand characteristic information of the user and the task includes:

[0040] For the shortest time that the user expects to migrate, with S next =Min{DTi, 1 ≤ i ≤ K1} as the optimization goal, select the migration satellite;

[0041] For the least number of migrations that the user expects, with S next =Max{ΔTi=T 2i -T1, 1 ≤ i ≤ K1} as the optimization goal, select the migration satellite;

[0042] For the best match of the new satellite resources that the user expects to migrate to, with S next= Max{ΔC i = C i - C0 (1 ≤ i ≤ K1)}, select the migrating satellite;

[0043] Aiming at the balance of migration time, migration times, and idle computing resources for users, establish a multi-objective joint optimization problem S next = Max{1 / DT i , ΔT i, ΔC i , 1 ≤ i ≤ K1}, select the migrating satellite.

[0044] In some embodiments, aiming at the balance of migration time, migration times, and idle computing resources for users, use the deep learning method to solve the established multi-objective joint optimization problem.

[0045] In some embodiments, as Figure 3 shown, according to the calculated computing duration required to migrate the task to the i-th candidate satellite task, migrating the computing task to the migration target satellite includes:

[0046] At T2 - DTi, based on the calculated migration target S next , migrate the computing task to the migration target satellite; and,

[0047] The satellite node S0 continues to provide services. At T2, the task migration is completed, and the user node starts to obtain services from S next .

[0048] In the case of burst task migration triggered by satellite node or link abnormal faults, it also includes:

[0049] For high-priority tasks, based on the fault-tolerant migration algorithm of redundant backup of non-intersecting computing resources, select the optimal migration plan from the migration plans with the shortest migration time, the fewest migration times, the best resource matching or balanced. For tasks with other priorities, select alternative migration plans. There is no intersecting computing resource between the alternative migration plan and the optimal migration plan. For example, in the case of burst task migration scenarios triggered by satellite node or link abnormal faults, considering that electromagnetic interference usually only occurs in local areas, in order to avoid service interruption caused by the failure of this part of the on-board computing unit, for high-priority tasks, based on the fault-tolerant migration algorithm of redundant backup of non-intersecting computing resources, select the optimal task migration according to the above task migration method; for tasks with other priorities, select the second or even the third set of task migration plans from other alternative plans, and ensure that there is no intersecting computing resource between the two sets of task migration plans, that is, the non-overlapping resource constraint of the primary and backup migration paths, which can effectively solve the service guarantee problem in abnormal situations.

[0050] In some embodiments, the satellite communication network is in an electromagnetic open space and is vulnerable to electromagnetic interference or attacks. For the scenario of burst task migration triggered by abnormal failures of satellite nodes or links, the characteristics of wide-area coverage of the satellite network are fully utilized. Considering that electromagnetic interference, etc., usually only occurs in local areas, in order to avoid service interruption caused by the failure of on-board computing units in this part, the embodiments of the present application adopt a highly reliable migration algorithm with remote data redundancy backup. When arranging and allocating resources according to tasks, the optimal resource arrangement plan is selected; then, with the physical space distance greater than the typical value D as a condition, as Figure 4 shown, the second set of resource allocation plans is selected from other alternative plans, and it is ensured that the two plans do not have intersecting computing resources, which can effectively solve the service guarantee problem in abnormal situations.

[0051] In some embodiments, in the case of K1 = 0, the tasks undertaken by the satellite node S0 are disassembled into at least two subtasks, so as to calculate the computing duration required for migration based on the disassembled subtasks, and the computing tasks are migrated to the target satellite for migration. That is, if K1 = 0, the computing tasks undertaken by S0 can be disassembled into two or more detailed computing subtasks Task01, Task02, Task03, and so on iteratively until a candidate satellite can be found for each task size, and then the migration is carried out according to the above steps. In a specific example, the method of the present application also designs an adaptive task splitting mechanism. In the scenario where the candidate satellite set is empty, an adaptive task decomposition and distributed migration strategy based on resource constraints is adopted, that is, K1 = 0, and the computing task Task undertaken by S0 is disassembled into two or even more detailed computing tasks Task01, Task02, Task03, and so on iteratively until a candidate satellite can be found for each task size, and then the migration is carried out according to the above steps.

[0052] The migration algorithm based on prior information in the embodiments of the present application can fully utilize the periodic law of the change of the satellite network topology for the task migration scenario caused by the dual dynamics of high-speed movement of network nodes and movement of user nodes, combined with the ephemeris information of the satellite constellation and the movement trajectory of users and other information, predict the trigger time of task migration in advance, and select a suitable task migration satellite with the goal of minimizing the task migration delay as the cost, and complete the task migration a certain time in advance to achieve a migration without the user noticing.

[0053] The embodiments of the present application also propose a spatial network migration system based on prior information, including a processor and a memory. A computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the spatial network migration method based on prior information as described above are implemented.

[0054] It should be noted that in the embodiments of the present application, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "including one..." does not exclude the existence of additional identical elements in the process, method, article or device including such element.

[0055] The serial numbers of the embodiments of the present application above are only for description and do not represent the superiority or inferiority of the embodiments.

[0056] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the embodiments of the present application.

[0057] The embodiments of the present application have been described above in conjunction with the accompanying drawings. However, the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present application, those of ordinary skill in the art can also make many forms without departing from the purpose of the present application and the scope protected by the claims. All of these are within the protection scope of the present application.

Claims

1. A spatial computing network task migration method based on dynamic environment perception, characterized in that: Adaptive task migration applied to scenarios of satellite node movement, user node movement, and burst link failures, including: At the initial moment T0, determine the computing resources C0 occupied by the user node to obtain computing services from the satellite node S0 in the space computing network; Calculate the moment T of task migration caused by the movement of satellite nodes according to the dynamic environment information of satellite movement 1， The dynamic environment information includes: ephemeris information, user trajectory information, real-time resource status; According to the dynamic environment information, calculate a set of K alternative satellites that can receive tasks at time T1; Determine the available computing resource situation of all satellites in the alternative satellite set at time T1 to determine a set of K1 candidate satellites whose remaining computing resources are greater than or equal to the computing resource requirement C0 of the task; Based on the set of candidate satellites, calculate the difference ΔC between the resources of the i-th candidate satellite and the resources required for mission migration i =C i -C 0， 1≤i≤K1; Based on the candidate satellite set, calculate the transmission duration DTi required to complete the migration of the task from the satellite node S0 to the i-th candidate satellite, 1 ≤ i ≤ K1, where the transmission duration includes the transmission duration between the satellite node S0 and the candidate satellite and the duration required for the migrated task to be completely deployed on the candidate satellite; According to the candidate satellite set, calculate the receiving task of the $i$-th candidate satellite, and calculate the next migration time $T$ through satellite ephemeris information and user orbit information 2i , so as to obtain the time interval $\Delta T$ from after the $i$-th candidate satellite receives the task to the next task migration i , $\Delta T$ i $= T$ 2i $- T_1$, $1\leq i\leq K_1$; Filter the task migration target according to the current priority or demand characteristic information of the user and the task; Migrate the computing task to the target satellite for migration.

2. The method for spatial network migration based on prior information according to claim 1, wherein Filtering the task migration target according to the current priority or demand characteristic information of the user and the task includes: The user expects the migration time to be the shortest, with S next =Min{DTi,1≤i≤K1} is the optimization target, and the migration satellite is selected; The number of user-expected migrations is the least, with S next =Max{ΔTi=T 2i -T1,1≤i≤K1} is the optimization target, and the migration satellite is selected; For the optimal matching of user expectations to migrate to new satellite resources, let S next = Max{ΔC i = C i - C0 (1 ≤ i ≤ K1)}, select the migrating satellite; Aiming at the goals of user's migration time, migration times, and balanced idle computing resources, a multi-objective joint optimization problem S is established next = Max{1 / DT i , ΔT i, ΔC i , 1 ≤ i ≤ K1}, select the migrating satellite.

3. The spatial network migration method based on prior information according to claim 2, wherein Aiming at the balance of the user's optimal migration time, migration times, and satellite resource matching, a deep learning method is used to solve the established multi-objective joint optimization problem.

4. The method for spatial network migration based on prior information according to claim 2, wherein Migrating the computing task to the target satellite for migration includes: At T2-DT i moment, based on the calculated migration target S next , migrate the computing task to the migration target satellite; and, The satellite node S0 continues to provide services. At time T2, the task migration is completed, and the user node starts to obtain services from S next ​ 5. The method for spatial network migration based on prior information according to claim 2, wherein In the case of burst task migration triggered by satellite node or link abnormal failures, it also includes: For high-priority tasks, based on a fault-tolerant migration algorithm with redundant backup of non-intersecting computing resources, select the optimal migration plan from the migration plans with the shortest migration time, the fewest migration times, the optimal or balanced resource matching. For tasks with other priorities, select alternative migration plans, and there is no intersecting computing resource between the alternative migration plan and the optimal migration plan.

6. The method for spatial network migration based on prior information according to claim 5, wherein In the case of K1 = 0, disassemble the task undertaken by the satellite node S0 into at least two subtasks, calculate the transmission duration required for migration based on the disassembled subtasks, and migrate the computing task to the target satellite for migration.

7. A spatial computing network task migration system based on dynamic environment perception, characterized in that: It includes a processor and a memory, and a computer program is stored on the memory. When the computer program is executed by the processor, it realizes the steps of the method for task migration in a space computing network based on dynamic environment perception as described in any one of claims 1 to 6.

Images