Resource Management In Satellite Communication Networks Based On Matching Game

Satellite communication network(SCN)is an essential component for achieving informatization,digitization,and modernization in sixth-generation communication technology.It is also recognized as a significant national information network infrastructure to be developed in the "14th Five-Year Plan for National Informatization".SCN is a crucial technology for achieving global coverage,offering global information network services to various users,including ground,marine,and aerial.It consists of low and medium-orbit satellites and geosynchronous orbit satellites in space,satellite ground stations,satellite measurement and control centers,and data processing centers on the ground,as well as many types of users located around the world.Compared with traditional terrestrial networks,SCNs have such outstanding advantages as broader coverage,fast network deployment,extensive applications,and freedom from geographic constraints.It plays an indispensable and pivotal role in global communications,air transportation,ocean voyage,navigation and positioning,emergency rescue,and the expansion of national strategic interests.However,the structure of the SCN is three-dimensional and multilayered,with dynamically changing topologies,heterogeneous network interconnections,and limited network resources.Moreover,the SCN encompasses a wide variety of tasks,each with distinct requirements.Additionally,there are a large number of potential unknown tasks in future networks,which also bring great challenges for resource management.Therefore,it is crucial to design effective and intelligent resource management methods for the SCN to achieve accurate matching between heterogeneous network resources and differentiated task requirements,as well as to realize the efficient utilization of limited resources.This paper investigates the multi-dimensional resource management architecture of SCN and the corresponding resource management method.In particular,considering the requirements of the actual network environment,this paper explores the multi-dimensional resource management method of SCN under incomplete information.This dissertation is organized as follows:1.In order to address challenges of differentiated multi-task simultaneous requests and difficulty of sharing heterogeneous multi-dimensional network resources in SCNs,we proposed a multi-dimensional resource management architecture.It can realizes accurate matching between multiple tasks and multiple resources.Specifically,this architecture realizes network multi-task feature extraction and classification through feature extraction technology;and employs virtualization technology to abstract the differences between underlying infrastructure settings.Subsequently,a multi-dimensional resource management strategy based on matching game is investigated,and feasibility validation is conducted using Earth observation tasks as an example.The research on the resource management architecture in this chapter lays the foundation for the resource management strategy research in Chapters Three and Four.2.In response to the problems of frequent disconnection in the satellite to ground communication links and the difficulty of meeting the demands of complex tasks in SCNs,a multi-dimensional resource allocation approach is proposed.Specifically,a network model is proposed for the cooperation of LEO and GEO satellites in providing observation and data transmission services.Based on the matching game,we constructed a one-to-one matching model that takes into account both observation and transmission resources with the objective of minimizing delay.We design a task-oriented GaleShapley matching algorithm.The two-side matching algorithm based on time-slot is further designed by utilizing the subtle changes of adjacent time slot networks to achieve effective scheduling of resources with lower complexity and overhead.Extensive simulation results validate the effectiveness of the proposed algorithms and enhance user satisfaction.3.Aiming at the problem that the complex and highly dynamic topology of SCNs makes it difficult to obtain the channel state information of all links,a multi-dimensional resource joint allocation method under incomplete network channel state information is proposed.We formulated an optimization problem for maximizing the revenue from access satellites,i.e.,the sum of the priorities of successfully access users.This problem is a mixed-integer nonlinear programming problem.Therefore,we solve the problem using a three-sided matching game within a single time slot.Meanwhile,we employed a model-free reinforcement learning framework to train historical network data to make the incomplete network channel state information trend complete.Finally,resource allocation algorithm for user-optimal and access-satellite optimal are designed and the convergence,stability and complexity are analyzed.A large number of simulation results verify the effectiveness of the proposed algorithms under different network parameters from both single and multiple time slots.