Analisis And Optimization On Capacity, Energy,Delay Of Satellite Communication System Based On Lyapunov Method

Satellite communication system has been widely used in various scenes of both civil and national defense domain, and has broad prospects for development. In recent years, as number of users increases and demands for service varies, how to allocate network resource and how to use relationship among capacity, energy and delay have become important issues. However, due to the characteristics of satellite communication system itself, technology and conclusions of ground-based communication system cannot be directly applied to satellite communications system. Hence, it is of necessity to research satellite communication systems specificly. By using Lyapunov method in access networks and satellite networks, research on the analytical relationships are researched, and optimization algorithms are developed. Main contributions are listed as follows:(1) Research on hybrid Adhoc networks with satellites has been carried out by using cell-partitioned model. First, theorems about analytical expressions of network capacity and upper bound of minimum energy function are proposed and proved. Algorithm aiming at maximizing capacity and minimizing energy cost are presented and analyzed by Lyapunov drift method. Second, these two theorems are applied to several types of Ad Hoc networks. Expressions of network capacity and minimum energy function are obtained. Third, capacity property of a type of hybrid Ad Hoc networks is analyzed in detail. Relationship among limitation of capacity, node density and coverage of base stations are investigated. Numerical analysis and simulation are carried out.(2) Research on degree-limited satellite networks has been carried out. The problem of flow control together with power allocation to antennas on satellite with arbitrary link states has been considered, so as to maximize the utility function while stabilizing the network. Inspired by Lyapunov optimization method, a Degree-Limited Scheduling Algorithm (DLSA) is proposed with a control parameter V, which requires no stochastic knowledge of link state. Discussion about implementation is carried out about the complexity of DLSA and several approximation methods to reduce complexity. Analyze shows DLSA stabilizes the network and the gap between utility function under DLSA and the optimal value is arbitrarily close to zero on the order of O(1/V). Simulation results verify DLSA on a simple network.(3) Research on satellite networks with large propogation has been carried out. A model of satellite networks with propagation delay is given first. Then a Quadratic Lyapunov function based Algorithm specific for satellite networks with propagation delay is proposed to greedily minimizing an expression containing Lyapunov drift. We prove satellite networks with propagation delay under QLA is stable and power constraints are met no matter how large propagation delay is. Theoretical analysis establishes upper bounds of queue backlogs and a lower bound of utility function, and also demonstrates how propagation delay jeopardizes utility performance. Simulation of QLA is carried out in a scenario consists of six nodes, and simulation results are in accordance with theoretical conclusions.(4) The performance of using Lyapunov drift and perturbation in QLAs has been investigated. By analyzing attraction points and utility performance of four variants of OQLA (Original QLA), we examine the rationality of OQLA for using the first-order part of an upper bound of Lyapunov drift of a function L1. It is proved that either using the real Lyapunov function (L2) of networks under QLA or using the entire expression of Lyapunov drift doesn’t improve backlogutility performance. The linear relationship between the attraction point of backlog and perturbation in the queue is found and thus average backlog can be controlled using perturbation. As an example of manipulation of the backlog by perturbation, an algorithm named QLA-VPDQ (QLA with Variable Perturbation of Data Queue) is proposed to serve as a substitution of OQLA when estimating the attraction point in the Phase One of FQLA-G (Fast-QLA-General) consuming less buffer, which is useful for networks with less buffer. Simulations verify the results above.