有线和无线网络环境下SCTP丢包区分机制的研究

发布时间：2018-06-01 18:08

本文选题：有线和无线网络 + SCTP　；参考：《东北大学》2014年硕士论文

【摘要】：随着互联网和相关技术的飞速发展,网络的应用日益广泛,网络业务量也随之激增。在这样的情况下TCP的固有缺陷表现的愈发明显,SCTP具有下一代互连网多宿多流等相关特性,因此SCTP替代TCP成为下一代互联网的传输协议已被业界达成共识。但是SCTP原本是用于有线网络的传输,使之应用于无线与有线的混合网络环境则会产生许多问题,比如无线移动环境下BER(误码率)明显加大,延迟尖峰出现得更加频繁,出现突发的设备噪声,并且无线信道质量不稳定,因此无线网络要比有线网络的传输环境更恶劣。这些特点将导致频繁产生丢包。在有线网络中,SCTP把所有的丢包简单归结为网络拥塞,路径上一旦出现丢包现象,则该路径的拥塞窗口减半。但是,在无线网络中,丢包现象的发生并不完全是由于网络拥塞所引起的,一旦出现丢包现象就减少拥塞窗口则是不明智的,并且此行为会极大地降低SCTP在无线网络中的性能。为了提高有线和无线网络环境下的利用率,开展下一代丢包区分机制的研究就具有重要的理论意义和应用价值。到目前为止,已经有很多研究人员针对有线和无线网络环境下SCTP的丢包区分机制进行研究,但这些现有的机制或多或少都有一些不甚完善的地方,它们的主要缺陷有：(1)大部分机制都是靠单一参数及其变化趋势进行丢包区分。如基于带宽估计技术的WISE协议只是利用网络的剩余带宽来区分丢包的原因,这相对就比较片面,在不断变化的网络环境下难以取得令人满意的效果。(2)当前网络参数统计、评判的方法是通过仿真获取某一网络模型下的评判标准,这些评判标准大都是确定一定的阈值或区间,从而以当前测量或估算的参数值与此阈值或区间的关系判断丢包原因。这种对模糊问题的简单二值逻辑特性,显然不适合融合网络环境下综合多种概率因素的评判过程。针对上述问题,本文建立了基于模糊综合评判模型(Fuzzy Comprehensive Evaluation Model)的丢包区分机制。模糊综合评判模型是一种十分有效的多因素决策模型,正是针对难以直接用准确的数字进行量化的评价问题提出来的一种很有价值的方法。该方法对原本仅具有模糊和非定量化特征的因素,经过数学处理,使其具有量化的表达形式,从而为决策提供可以进行比较和判别的依据,提高决策的科学性和正确性。其可以客观的评价多个影响因素与决策目标之间的关系,在处理多因素与评判结果之间的不确定性问题上具有一定的优越性。因此在本文的研究中,我们引入模糊综合评判模型来描述网络分组丢失的评判过程,利用模糊逻辑的智能处理能力,在发送端通过综合评判模型来分析处理分组丢失的原因,从而提高拥塞判断的准确性,避免SCTP误启动拥塞控制算法。最后通过仿真实验验证了所设计模型的准确性和合理性。论文最后对课题进行了总结和展望,提出了下一步的研究方向。
[Abstract]:With the rapid development of Internet and related technologies, the application of network is becoming more and more extensive and the volume of network services is increasing. In this case, the inherent defects of TCP are becoming more and more obvious, and SCTP has the related characteristics of the next generation of interconnected multi lodging and multi flow, so the transfer protocol of SCTP instead of TCP to the next generation of Internet has been reached by the industry However, SCTP is originally used for the transmission of wired networks, and the application of it to the wireless and wired hybrid network environment will produce many problems, such as the increase of BER (bit error rate) in wireless mobile environment, the emergence of the delay peak more frequently, the sudden device noise, and the wireless channel quality unstable, so the wireless network is used. It is worse than the transmission environment of the wired network. These features will result in frequent packet loss. In the wired network, SCTP reduces all packets simply as network congestion. Once the packet loss occurs on the path, the congestion window of the path is halved. In the wireless network, the occurrence of the packet loss is not entirely due to the network support. It is unwise to reduce the congestion window once the packet loss occurs, and this behavior will greatly reduce the performance of SCTP in the wireless network. In order to improve the utilization rate in the wired and wireless network environment, the research on the next generation packet loss differentiation mechanism has important theoretical significance and application value. So far, many researchers have studied the packet loss differentiation mechanism of SCTP in wired and wireless network environment, but these existing mechanisms are more or less imperfect. The main defects are as follows: (1) most of the mechanisms are divided by single parameters and their changing trends. For example, based on bandwidth estimation. The WISE protocol of the technology only uses the residual bandwidth of the network to distinguish the cause of the packet loss. This is relatively one-sided, and it is difficult to achieve satisfactory results in the changing network environment. (2) the current network parameter statistics, the evaluation method is through the simulation to obtain the evaluation criteria under a network model, most of these criteria for evaluation. It is to determine a certain threshold or interval, so as to judge the cause of the packet loss by the relation between the values of the measured or estimated parameters and the threshold or the interval. This simple two value logic characteristic of the fuzzy problem is obviously not suitable for the evaluation process of integrating a variety of probability factors under the network environment. The fuzzy comprehensive evaluation model is a very effective multi factor decision model. It is a very valuable method for the evaluation problem which is difficult to quantify directly with accurate numbers. This method is only fuzzy and non quantitative to the original Fuzzy Comprehensive Evaluation Model. The factor of chemical characteristics, through mathematical treatment, makes it have a quantitative expression form, thus providing a basis for comparison and discrimination for decision making, and improving the scientificity and correctness of decision making. It can objectively evaluate the relationship between multiple factors and decision goals, and deal with the uncertainty between the multiple factors and the evaluation results. In this paper, we introduce the fuzzy comprehensive evaluation model to describe the evaluation process of network packet loss, and use the intelligent processing capability of fuzzy logic to analyze the original cause of packet loss through the comprehensive evaluation model at the sending end, so as to improve the accuracy of congestion judgment and avoid SCTP In the end, the accuracy and rationality of the designed model are verified by the simulation experiment. Finally, the thesis is summarized and prospected, and the next research direction is put forward.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.04

【相似文献】