卫星网络的拥塞控制策略研究

发布时间：2018-01-13 15:01

本文关键词：卫星网络的拥塞控制策略研究　出处：《北京邮电大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着人们不断增长的多媒体通信需求和日益增大的带宽需要,下一代全球信息网络必将向新一代Internet,新一代移动通信网以及下一代卫星通信网的异构结合的方向发展,而卫星网络也将在全球互联网中扮演更加重要的角色。带宽的发展对卫星网络的服务质量(Quality of Service, QoS)提出了新的要求,保证信息在卫星网络中传输的高效性、稳定性、可靠性成为重中之重。因此,进行卫星网络拥塞控制研究具有十分重要的意义。论文在研究拥塞控制原理的基础上,主要从卫星网络的主要特点出发,对卫星网络中的拥塞控制策略进行了深入的研究,论文主要的研究工作为： (1)在研究传统TCP拥塞控制算法在卫星网络上的不适用性和分析卫星网络中经典TCP拥塞控制算法的优劣的基础上,提出了针对卫星网络的基于带宽估计的窗口恢复改进TCP拥塞控制算法TCP-BRE。算法通过可用带宽估计提高了窗口增大速度,通过引入窗口加速因子加快丢包后的窗口恢复速度,同时增加了丢包检测机制来区分丢包原因,减小误码对TCP性能的影响,仿真研究了算法的吞吐量性能。研究结果表明,TCP-BRE在卫星网络传输条件下与Reno相比,能够更快的传输小数据量短流,并有效提高了稳态吞吐量,在误码条件下吞吐量仍能保持在很高的水平。TCP-BRE算法有效提高了卫星网络的资源利用效率。 (2)在分析经典主动队列管理的算法并指出这些算法在卫星网络中的不适用性的基础上,提出了一种基于链路误码率的自适应随机指数标记算法EAREM。该算法在星载路由系统中,基于REM算法,结合卫星链路误码率高的特点,在价格中增加了误码率项并改进了REM参数自适应的方式。算法通过价格机制降低了误码率较高的输入流的占用带宽,从而降低了整个网络因误码造成的丢包,仿真研究了算法的队列长度稳定性和丢包率等性能。研究结果表明：EAREM在卫星高误码率网络中比AREM、REM算法有更低的丢包率,同时保证了队列长度的稳定性,提高了响应速度。EAREM算法还能够根据不同输入流所经链路的误码率情况,完成带宽的合理分配。在误码率突变情况下,算法也有较好的适应性。 (3)在分析针对缓存门限优化的随机流模型(SFM)原理基础上,提出了一种基于随机实验法的SFM动态缓存门限优化算法。算法首先提出基于输入反馈的星载路由器随机流模型,在此模型上构建平衡丢包、时延性能的代价函数,并考虑星载路由处理能力采用随机试验法来实现缓存门限值的动态优化。仿真研究了算法的缓存门限值收敛性能。研究结果表明所提的基于随机试验法的SFM动态缓存门限优化算法收敛性能较好,在不同门限初值情况下能达到较快的收敛,且算法复杂度低,易于在星载路由器上部署。论文还深入探究了平衡系数、反馈因子、业务流速比例等关键参数对收敛曲线的影响。
[Abstract]:With the increasing demand for multimedia communication and increasing bandwidth, the next generation of global information network will become a new generation of Internet. The new generation mobile communication network and the next generation satellite communication network are developing in the direction of heterogeneous combination. Satellite network will also play a more important role in the global Internet. The development of bandwidth will improve the quality of service of satellite network. QoS) puts forward new requirements to ensure the high efficiency, stability and reliability of the transmission of information in satellite networks. Therefore, it is of great significance to study the congestion control of satellite networks. Based on the research of congestion control principle and the main characteristics of satellite network, this paper makes a deep research on the congestion control strategy in satellite network. The main research work of this paper is as follows: 1) on the basis of studying the inapplicability of traditional TCP congestion control algorithm in satellite network and analyzing the advantages and disadvantages of classical TCP congestion control algorithm in satellite network. An improved TCP congestion control algorithm, TCP-BRE, for satellite networks based on bandwidth estimation is proposed. The algorithm improves the speed of window growth through the estimation of available bandwidth. The window acceleration factor is introduced to accelerate the window recovery speed after packet loss and the packet loss detection mechanism is added to distinguish the cause of packet loss and to reduce the impact of error codes on TCP performance. The simulation results show that TCP-BRE can transmit small data short stream faster than Reno in satellite network. The steady-state throughput is improved effectively, and the throughput can be maintained at a high level. TCP-BRE algorithm can effectively improve the resource utilization efficiency of satellite networks. On the basis of analyzing the classical active queue management algorithms and pointing out the inapplicability of these algorithms in satellite networks. An adaptive random exponential marking algorithm based on link error rate (LBER) is proposed, which is based on REM algorithm and high BER in satellite link routing system. The BER item is added to the price and the adaptive mode of REM parameters is improved. The algorithm reduces the bandwidth of the input stream with high BER through the price mechanism. In order to reduce the packet loss caused by error code, the performance of queue length stability and packet loss rate of the algorithm is studied by simulation. The results show that the performance of AREM in the satellite high bit error rate network is better than that of the AREM. The REM algorithm has lower packet loss rate, at the same time ensures the stability of queue length, and improves the response speed. EAREM algorithm can also be based on different input streams through the link error rate. In the case of BER mutation, the algorithm also has good adaptability. 3) based on the analysis of the stochastic flow model (SFM) for buffer threshold optimization. This paper presents a dynamic buffer threshold optimization algorithm for SFM based on random experiment. Firstly, a random flow model of spaceborne router based on input feedback is proposed and balanced packet loss is constructed on the model. The cost function of delay performance. Considering the spaceborne routing processing ability, the random test method is used to realize the dynamic optimization of the buffer threshold. The convergence performance of the buffer threshold of the algorithm is studied by simulation. The results show that the proposed SFM based on the stochastic test method is based on the proposed algorithm. Dynamic buffer threshold optimization algorithm has good convergence performance. In the case of different threshold initial value can achieve faster convergence, and the algorithm complexity is low, easy to deploy on the spaceborne router. The paper also deeply explore the balance coefficient, feedback factor. The influence of the key parameters, such as the ratio of service velocity, on the convergence curve.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN927.2

【参考文献】