| Although the terrestrial network has entered the 5G era,there are still network coverage blind spots such as mountains,deserts,and seas.As a powerful supplement and extension of the terrestrial network,satellite network has become the focus of international attention.To improve the communication capacity of satellite networks,beam design and its scheduling method become the key.Beam hopping technology has the characteristics of flexible resource allocation and has become one of the new trends in the development of beam technology.This paper mainly studies the beam hopping resource scheduling method.The main innovations and contributions are as follows:First,a high-throughput beam resource scheduling method for differentiated services is studied.To simultaneously guarantee the differentiated service quality requirements of multi-type services and the system service requirements of high throughput of the satellite network,a beam resource scheduling model oriented to delay tolerance of differentiated services is proposed.The deep reinforcement learning method is used to optimize the matching relationship between beam scheduling actions and business characteristics by using the interactive feedback between beam service coverage,throughput,and beam scheduling actions,to obtain the dual guarantee of network throughput and differentiated service quality.The simulation results show that,while ensuring the network throughput,the proposed method improves service coverage by 6%,7%,11%,and 16%,respectively,compared with the on-demand polling,longest queue first,sequential,and random allocation methods.Second,a lightweight beam-hopping scheduling method is studied.With the increase of the number of satellite beam and wave-bit cells,the search space for global beam hopping increases sharply,and the algorithm complexity is high,which brings severe challenges to the on-board computing resources.A beam hopping scheduling model based on load balancing clustering is established,which converts a complex global scheduling task into multiple local scheduling tasks.The lightweight beam scheduling is realized by combining artificial immune algorithm with deep reinforcement learning.The simulation results show that this method accelerates the convergence rate of beam scheduling by two times,and achieves the throughput and service coverage rate performance similar to the global scheduling.With the accelerating pace of LEO satellite constellation construction,the demand for efficient beam scheduling technology will become increasingly urgent.The proposed beam hopping resource scheduling method will provide a feasible solution for LEO constellation and its beam scheduling system design. |
