Research On Satellite Beam Management Algorithm For Non-Terrestrial Network

After a long development,the mobile communication system has come to 5G(the fifth generation of mobile communication systems).Compared with the current cellular mobile network,5G can provide effective support in the three scenarios,including enhanced mobile broadband,ultra reliable and low latency communication and massive machine type of communication.To realize such a technological revolution,it is necessary to change the core network and its access mode thoroughly.In the face of this demand,non terrestrial networks(NTN)begins to come into people's sight.NTN network is a kind of network which uses the air or space carrier as a part of communication.It has the characteristics of wide coverage,strong ability to deal with disasters,and not vulnerable to physical attacks on the ground.3GPP R15 and R16 have included NTN network in the discussion of 5G standard.Low orbit satellite has the characteristics of low orbit height and short electromagnetic wave round-trip time,which can effectively solve the delay problem in the process of satellite communication,and the emergence of high-throughput multi-beam satellite also provides a strong support for higher throughput data services.However,the rapid movement of LEO satellite will cause frequent switching of user calls,which challenges the beam management technology of satellite beam.First of all,this paper discusses and analyzes the NTN network architecture and the beam management technology of LEO satellite.by analyzing and studying the characteristics of the current typical LEO satellite constellations,divide the LEO satellite constellations into two categories: fixed beam system and staring beam system according to whether the beam direction angle can be adjusted,and pays attention to their main research issues as well as the research in this direction at home and abroad,which provides a theoretical basis for the later study of satellite beam management algorithm.Then,for the fixed beam satellite system,this paper focuses on the channel allocation in the process of beam switching.By studying the motion characteristics of fixed beam satellite,a simulation model with periodic motion characteristics is established.Aiming at the two concerns of channel allocation and channel reservation in beam switching,the dynamic channel allocation algorithm and dynamic channel reservation algorithm are introduced respectively.The simulation results show that the combination of dynamic channel allocation and reserved beam switching algorithm can be better than the traditional fixed channel allocation or reserved channel.Finally,aiming at the beam scheduling problem of staring beam system,this paper studies the modeling from two aspects.One is to divide the satellite and hot spot area into two sets,and model the system optimization problem as the optimal matching problem of bipartite graph,and solve it based on Kuhn munkres algorithm.On the other hand,the process of beam scheduling is regarded as the process of interaction between agent and environment,which is solved by multi-agent deep Q network algorithm.Finally,the comparison between the two algorithms shows that the former is better than the latter in coverage index,and the latter is better than the former in handover rate by adjusting the parameters of reinforcement learning network.