Analysis Of LEO Satellite Network With Stochastic Network Calculus

With its many advantages such as low propagation delay,low transmitted power and low performance requirement on the terminal hardware,the low-orbit satellite network has become a research hotspot in the fields of communication.As the rapid increase of user number and the variety of service demands,it is a key issue to accurately analyze the network performance and allocate network resource.By using Stochastic Network Calculus(SNC)and martingale theory,the performance of LEO satellite network is analyzed in this paper.The main contributions are listed as follows: 1.To begin with,Stochastic Network Calculus theory is introduced and the method of how to use Stochastic Network Calculus theory to analyze the performance of network is discussed.Then the queuing delay and the backlog of LEO satellite network are analyzed using Stochastic Network Calculus theory.Firstly,the features of arrival process and service process in the LEO satellite network are analyzed.To present the bursty characteristics of traffic,the Markov-Modulated ON-OFF process(MMOO)is adopted to model the arrival process to obtain the suitable stochastic arrival curve.Secondly the service process of satellite slotted Aloha protocol is modelled by a two-state Markov chain to obtain the suitable stochastic service curve.Based on the stochastic arrival curve and stochastic service curve,the delay distribution bounds and the backlog distribution bounds are derived by independent case analysis.Finally,based on the theoretical analysis,the simulation scenario is built with MATLAB to verify whether the theoretical performance bounds are consistent with the actual situation.The simulation results show that though the obtained stochastic bounds are loose,the performance bounds derived from SNC still agree to the practical situations to some extent.2.The reason why the SNC bounds is not tight is analyzed then.In order to fix the weakness of SNC bounds and obtain more accurate performance of LEO satellite network,a more advanced tool,martingale theory,is leveraged in this paper.Based on the Martingale theory and the existing method of building the supermartingale envelope of two-state Markov chain,the arrival supermartingale envelope of arrival process and the service supermartingale envelope of service process in LEO satellite network are obtained.Based on the arrival supermartingale envelope and service supermartingale envelope,the theoretic martingale delay bound and backlog bound are obtained under the FIFO scheduling scheme and SP scheduling scheme.After that,the simulation results are used to verify that the martingale bounds are tighter with the real data sets obtained from simulation results than SNC bounds.The influence of the changes of network node number,traffic arrival rate as well as schedule strategies on system performance is also analyzed.And it is shown by simulation results that the delay and the backlog value in the system would significantly increase with the increase of node number and the traffic arrival rate in the network,and the queue delay can also be influenced by different schedule strategies.The theoretical analysis and simulation results obtained can provide a theoretical guidance for rational resource allocation and QoS guarantee in LEO satellite network.