Intelligent Transmission Of Satellite Network

"Satellite Internet" based on low earth orbit satellite constellations is an important component of high-speed,ubiquitous,integrated,and secure digital infrastructure.As a representative of the new generation of information infrastructure,it has become a national strategic project.Satellite internet solves the problems of limited coverage,poor disaster resistance,and difficulty in coping with special terrain and emergencies of traditional ground networks,effectively combining ground,marine,and airborne users,and is a key solution for the interconnection of all things.Among them,the Low Earth Orbit Satellite Network(LEO-SN),with its unique advantages of low latency and low cost,have become an important infrastructure for building future space-airground integrated networks.Due to significant differences between satellite networks and ground networks in network topology,underlying links,and available resources,current research on ground networks cannot fully adapt to the development needs of future satellite network transmissions.The high dynamic topology of satellite networks accelerates the expiration of routing,making it difficult to obtain stable transmission paths.Open wireless channels are susceptible to interference from space noise,making it difficult for traffic to be reliably transmitted.Differentiated inter-satellite links make traditional scheduling methods inefficient and unable to adapt to complex and changing satellite network conditions.These characteristics pose new challenges to network routing,transmission,and scheduling,which require further research and exploration.In view of the aforementioned challenges in satellite network transmission,this dissertation focuses on three aspects:multipath routing,deterministic and reliable transmission,and intelligent traffic scheduling.We designs the system architecture and transmission model respectively,and utilizes technologies such as Ant Colony Optimization(ACO)algorithm,Network Coding(NC),and Deep Reinforcement Learning(DRL)to achieve adaptive multipath routing,deterministic satellite network transmission,and intelligent traffic scheduling,thereby achieving an intelligent satellite network transmission mechanism.The main research content of this dissertation can be summarized as follows.1.Multipath routing discovery method based on intelligent perception.Aiming at solving path switching caused by the dynamic movement of satellites in the ubiquitous connected network,this dissertation proposes a link state-aware mechanism and a path quality evaluation mechanism based on the SDN(Software Defined Network,SDN)architecture.First,the topology prediction and time-slot partitioning techniques based on ephemeris information are used to characterize all the upcoming satellite communication opportunities.Then,by constructing a resource dynamic update mechanism,a resource topology graph is generated to represent the characteristics of network resources in space and time dimensions,and a spatio-temporal network topology graph with a dynamic resource state description is built.An adaptive multipath optimization method based on ACO algorithm is proposed,which uses ant packets to continuously explore paths and collect state information,and obtains real-time path information to assist multipath planning,realizing multipath routing discovery under the guidance of intelligent perception information.2.Reliable transmission mechanism of satellite networks based on network coding.This dissertation proposes a deterministic network coding algorithm to solve the problem of unreliable traffic caused by lossy and dynamically heterogeneous wireless transmission conditions.Firstly,by considering a transmission scenario without feedback links,a deterministic mission transmission model is established with constraints on delay and delivery rate,and a coding vector matrix with decoding guarantees and adaptive coding redundancy schemes are designed to achieve intelligent adaptation of batch size and redundancy to the transmission environment.Furthermore,a coding vector method that does not require transmission and a deterministic mission execution scheme based on resource reservation are proposed to support the deployment of deterministic missions in satellite networks.Finally,a heuristic multipath coding packet allocation scheme based on synchronized arrival is proposed to shorten the joint transmission delay and guarantee the transmission performance of deterministic missions.3.Intelligent multipath traffic scheduling method for satellite networks.To address the problem of reduced observation accuracy and scheduling performance caused by the complex and changing states of satellite networks,this dissertation first designs an information state collection method based on ACO.Then,by defining pheromone state,traffic scheduling action,and immediate transmission reward,an intelligent multipath traffic scheduling optimization model based on pheromone state is established.Next,the Deep Deterministic Policy Gradient(DDPG)method is introduced to solve the challenge of high-dimensional continuous state-action space and achieve intelligent multipath traffic scheduling for satellite networks.The above three topics respectively investigate three key network transmission technologies:multipath routing,reliable transmission,and traffic scheduling.They have overcome a series of scientific problems in satellite network end-to-end transmission path construction,transmission reliability,and traffic scheduling quality assurance,and have realized an intelligent transmission solution for satellite networks.This effectively improves the transmission performance of satellite networks and explores new solutions for China’s future satellite mobile communication network.