CN-116366135-B - Distributed multi-user computing force routing strategy formulation method for space-based computing force network

CN116366135BCN 116366135 BCN116366135 BCN 116366135BCN-116366135-B

Abstract

The invention discloses a method for formulating a distributed multi-user computing force routing strategy for a space-based computing force network, which comprises the steps of firstly constructing a satellite element network composed of satellites and functional modules, then modeling user services as a DAG model, realizing the determination of a computing strategy by mapping the DAG to corresponding functions in the space-based computing force network, then modeling satellite computing force resource allocation problems of different users when different access satellites initiate services as a distributed potential game process, realizing the formulation of the multi-user computing force routing strategy under the condition of minimum user loss, and finally solving the mapping strategy of Nash equilibrium points through a distributed self-learning Xi Diedai algorithm, wherein the strategy at the moment is used as the optimal computing force routing strategy in the all day long-based computing force network. The invention can connect the satellite computing power in series, realize the scheduling of satellite computing power resources and solve the problem of computing power resource preemption conflict of multi-user multitasking access to the satellite computing power network.

Inventors

REN BAOQUAN
LI HONGJUN
GONG XIANGWU
ZHONG XUDONG
DONG NANNAN
LV JIAZHENG
HE JI
HU TIANZHU

Assignees

军事科学院系统工程研究院系统总体研究所

Dates

Publication Date: 20260505
Application Date: 20230407

Claims (6)

1. A method for formulating a distributed multi-user computing routing strategy for a space-based computing network is characterized by comprising the following steps: step 1, constructing a satellite element network consisting of satellites and functional modules; step 2, modeling the user service as a DAG model, and realizing the determination of a calculation strategy by mapping the DAG to a corresponding function in a space-based computing power network; Step 3, modeling satellite computing power resource allocation problems of different users when different access satellites initiate services as a distributed potential game process, and realizing formulation of a multi-user computing power routing strategy under the condition of minimum user loss, wherein the method comprises the following steps of: step 3.1, modeling time delay and energy consumption; Step 3.2, establishing a utility function; step 3.2, constructing a game model and proving; step 4, solving a mapping strategy of Nash equilibrium points through a distributed self-learning Xi Diedai algorithm, and taking the strategy at the moment as an optimal force routing strategy in a all day long-most basic force network; modeling the time delay and the energy consumption in the step 3.1, specifically as follows: For nodes Task of access The computation delay is expressed as ; Wherein the method comprises the steps of Is that The shortest path between adjacent matrices through satellite elements Obtaining; Is that To the point of Is used to determine the cumulative time delay of the (c), For subtasks Is a task amount of (1); And for the node Energy consumption of (1) and (2) order Is a node The set of functions that participate in the computation, expressed as ; Wherein the method comprises the steps of Is a node Is used for the idle energy consumption of the (a), Is a node Is not limited by the energy consumption of the work; The utility function is established in the step 3.2, which is specifically as follows: Setting node The loss function of (2) is composed of time delay and energy consumption, and the time delay and energy consumption are converted into no unit number by the loss function Is that ; Wherein the method comprises the steps of And (3) with In order to convert the factor(s), Related to energy consumption, mapping results at known nodes In the case of (a) , Is a node Is considered to be a set of functions, Is a node Other node function occupancy cases, where for a node The time delay generated by the task calculation of the access is expressed as ; Energy consumption Represented as ; Wherein, the For node idle energy consumption Dynamic energy consumption for node operation, node Is expressed as a loss function of ; Wherein the method comprises the steps of And For normalizing loss unit price for a given node The aim is to be as small as possible I.e. ; The game model and the demonstration are constructed in the step 3.3, which is specifically as follows: modeling mapping process of tasks accessed by different nodes as a game process ; The three elements are a user set, a strategy set and a loss set respectively; Definition 1 Nash equalization for We call we For the Nash equilibrium point, if and only if there are no satellite nodes changing their strategy again to achieve lower losses, i.e ; Definition 2, potential Game, during the game In which there is a potential function The following conditions are satisfied , ; Function of making potential ; Proof first of all ; ; Because policy changes on each node do not affect the mapping of other tasks, therefore ; To fully utilize the computing resources on the satellite nodes, let I.e. the starting function on the node will fully occupy the computing resource on the node, while the non-starting function will not consume energy, the above formula is rewritten as ; I.e. This is true.
2. The method for making a distributed multi-user computing power routing strategy for a space-based computing power network according to claim 1, wherein the construction of the satellite element network consisting of satellites and functional modules in step 1 is specifically as follows: Step 1.1, determining the network topology of a low-orbit satellite by analyzing the communication relation among satellites; And 1.2, determining the network topology of the satellite elements by analyzing the distribution of the functional modules on the satellite.
3. The method for making a distributed multi-user power-computing routing strategy for a space-based power network according to claim 2, wherein the determining of the low orbit satellite network topology by analyzing the communication relationship between satellites in step 1.1 is specifically as follows: calculating the distance between satellites according to the time-varying coordinates of each satellite, and calculating the signal to noise ratio of channels between two satellites according to the free propagation loss of the inter-satellite links, wherein the channel capacity between any two satellites is obtained through a shannon formula, so that the on-off of the inter-satellite links is judged: establishing a three-dimensional coordinate system by taking the earth center as an origin, wherein the coordinates of any satellite can be obtained by using Kepler six parameters To determine, wherein Is a half-long axis of the machine, In order to achieve the eccentricity ratio, In order to be the inclination angle, In order to raise the right-hand meridian of the intersection point, Is the amplitude angle of the near-near place, For true near-earth angle, for satellite nodes in space-based power network Is the position of (2) Can be obtained by three-dimensional coordinates; According to the free space loss and shannon formula, The inter-satellite link capacity between is expressed as: ; Wherein the method comprises the steps of For the bandwidth of the inter-satellite link, Is the signal-to-noise ratio over time, The value of (2) is related to the distance between satellites, and the minimum capacity of the communication between satellites is set as When (when) In the time-course of which the first and second contact surfaces, The inter-satellite links are in a connected state, and the link capacity is Otherwise, the link is in a disconnected state; because satellite network topology tends to change in a short period of time, a fixed point in time is employed Representing a network topology over a period of time using a 01 adjacency matrix To represent, then, the elements in the matrix Expressed as: ; ; Wherein the method comprises the steps of The capacity thresholds for inter-satellite link connectivity, respectively.
4. The method for making a distributed multi-user computing power routing strategy for a space-based computing power network according to claim 2, wherein the step 1.2 is to determine the topology of the satellite element network by analyzing the distribution of functional modules on the satellite, and specifically comprises the following steps: The node of the element space-time expansion graph is divided into a satellite and a functional module, the functional module is abstracted into a node directly connected to the satellite, and the functional module equipped with the satellite node does not change with time, so that the time slot change is not required to be considered when the element is discussed Seed functional module, all The functional module types of the species are represented as groups Wherein Representative satellite First of the equipment A functional module of the type described above, Represent the first Satellites occupied by functional modules of the type Proportion of computational power resources, therefore, all functional module nodes in the satellite network are used A representation; Constructing element satellite network adjacency matrix based on the definition of the functional module ; Wherein the method comprises the steps of , Delay for transmitting a unit of task quantity, wherein , In order to calculate the time delay per unit amount of tasks, At node for function x The ratio of the occupancy in the resource is calculated, Is a node Is used for the calculation of the calculation capacity of (a); where 0 represents no time delay from function to satellite.
5. The method for formulating the distributed multi-user computing power routing strategy for the space-based computing power network according to claim 1, wherein the modeling of the user service in step 2 is a DAG model, and the determination of the computing strategy is realized by mapping the DAG to the corresponding function in the space-based computing power network, specifically as follows: Modeling user service as a DAG model, and determining a calculation strategy by mapping the DAG to a corresponding function in a space-based power network to form a path for connecting power resources in series, namely a power calculation route; Modeling user traffic as a DAG model Wherein To represent nodes in the DAG representing each subtask, The edge of DAG represents the logic relation between different subtasks, and only one source satellite and one destination satellite of the path in the power network are calculated according to the day, so the adopted DAG model is a source node And a destination node ; Order the The task amount is as follows , The subtask amount of (2) is Order-making machine Representation of Thus (2) Task amount to be calculated Represented by the formula: ; Wherein the method comprises the steps of Is a data scaling factor, takes the value as ; Order the Representation of To the point of Mapping of (c), and Default mapping to task originated satellite functionality I.e. Mapping other tasks onto other nodes of the transmission path, and enabling shortest paths among different functional nodes to pass through And (5) calculating to obtain the product.
6. The method for formulating the distributed multi-user computing power routing strategy for the space-based computing power network according to claim 1, wherein the mapping strategy of the Nash equilibrium point is solved by the distributed self-learning Xi Diedai algorithm in the step 4, and the strategy at the moment is taken as the optimal computing power routing strategy in the all day long-based computing power network, and the method is specifically as follows: the Nash equilibrium solution is solved by adopting a self-learning-based distributed mapping algorithm, and the specific algorithm comprises the following steps: (1) Initializing a null decision matrix , Adjacent matrix to satellite element ; (2) For iteration number ; (3) For each user i; (4) According to Calculation of ; (5)If ; (6) Holding Unchanged; (7)Else; (8) Updating Updating ; (9) Sending RTU information; (10)Endif (11)Endfor (12) If no RTU information can be sent (13)Break; (14)Endif (15)Endfor (16) Outputting the decision matrix at the moment Calculating the calculation time delay of each user task Each game participation node in the algorithm pursues lower loss as much as possible When a user appears a button response, it sends request update information, i.e. RTU information, which includes the user ID and the latest decisions.

Description

Distributed multi-user computing force routing strategy formulation method for space-based computing force network Technical Field The invention relates to the technical field of distributed computing of communication technology, in particular to a distributed multi-user computing routing strategy formulation method for a space-based computing network. Background Along with the continuous development of information technology, the demand range of people for the service of the Internet of things is continuously expanded, and the service is gradually expanded from a traditional urban scene to a rural scene and further gradually expanded to an area which cannot be covered by a traditional communication mode. This means that even in the ocean, desert, gobi, outer space, etc., the user wants to be able to get real-time, seamless, efficient internet of things services. In order to meet the increasing business demands of users, satellite communication has received great attention in recent years as a communication technology capable of providing efficient coverage, and is regarded as one of the cores of 6G communication technology. However, in the traditional satellite communication scene, the satellite is only used for data transparent forwarding, so that the service of the Internet of things of the user needs to go through a plurality of steps of uploading the satellite, forwarding the ground cloud center, returning the result to the satellite, sending the result back to the user by the satellite and the like, the service delay of the service of the user is greatly increased, and the service requirement of the user is difficult to meet. In order to solve the problem, a concept of a space computer is proposed, and an attempt is made to directly provide computing services for users by using satellites, so that service processing time delay can be greatly reduced. However, because satellites generally have limited computing power and storage power, are susceptible to errors caused by the influence of space environment, and the computing and storage power of a single satellite are weak, the concept of a space-based computing power network is further proposed, and a plurality of satellites in a satellite constellation are tried to be combined to realize collaborative computing. The power computing routing technology is the core of the power computing network, and can connect satellite power computing in series to realize the scheduling of satellite power computing resources. However, since the service initiating satellites are not unique, the number of satellites is large, the coverage area is wide, and a control center is difficult to find to formulate a calculation routing strategy, the realization of the calculation routing in the space-based calculation network faces a great challenge. The conventional routing strategy and the satellite cooperative computing strategy cannot adapt to satellite scenes of non-central multiple users. Disclosure of Invention The invention aims to provide a distributed multi-user computing power routing strategy formulation method for a space-based computing power network, which can connect satellite computing power in series and realize the scheduling of satellite computing power resources. The technical scheme for realizing the purpose of the invention is that the method for formulating the distributed multi-user computing routing strategy for the space-based computing network comprises the following steps: step 1, constructing a satellite element network consisting of satellites and functional modules; step 2, modeling the user service as a DAG model, and realizing the determination of a calculation strategy by mapping the DAG to a corresponding function in a space-based computing power network; step 3, modeling satellite computing power resource allocation problems of different users when different access satellites initiate services as a distributed potential game process, and making a multi-user computing power routing strategy under the condition of minimum user loss; and 4, solving a mapping strategy of Nash equilibrium points through a distributed self-learning Xi Diedai algorithm, and taking the strategy at the moment as an optimal force routing strategy in the all day long-most basic force network. Compared with the prior art, the invention has the remarkable advantages that (1) in order to accurately achieve specific functions on the satellite, a satellite element network with abstract function modules is provided so as to achieve abstract representation of satellite computing power resources, a mat is provided for subsequent research of user service mapping, (2) a game theory method is adopted to realize distributed resource allocation among a plurality of game participants, the problem of conflict of computing power resource preemption of a multi-user multitask access satellite computing power network is solved, and (3) in order to realize distributed solving of a game mo