一种非线性RED算法的研究

定量分析了平均队列长度与网络性能的关系，在介绍了尾丢弃和RED算法之后，提出了一种非线性RED算法，并将非线性RED算法与RED算法进行了定量的比较，指出采用非线性RED算法，路由器可以避免在平均队列长度到达最大阈值前过多地丢弃分组，而且可以提高路由器的缓冲区利用率。     

无标度网络上队列资源分配研究

研究了节点队列资源有限的条件下,无标度网络上的信息流动力学过程,发现了网络由自由流通到拥塞的相变现象,提出了一种基于节点度的队列资源分配模型.模型的核心是使节点i的队列长度与kβi成正比(ki为节点i的度,β为分配参数).仿真结果表明,在网络使用最短路径算法进行信息包传送的条件下,β近似等于1.25时队列资源分配最合理,网络容量最大,且该最佳值与队列总资源多少以及网络的规模无关.     

基于元胞蚁群的主动队列管理算法研究

针对网络拥塞现象,基于弃头方式提出了一种新的主动队列管理算法.该算法首先利用元胞蚁群建立了实际网络队长最大值的计算方法,同时通过判断网络队长与阈值的关系,采取从队列头部丢弃N个数据包的方法.最后,以长相关数据进行仿真实验,对比分析了DFCA与RED、DROP-TAIL之间的优劣,结果表明该算法具有较好的适应性.     

基于分形布朗业务模型的差分队列服务建模

差分队列服务(DQS)是一种在有线与无线融合网络中能有效提供服务质量的服务质量模型,文中基于DQS的思想,将具有相同时延要求的到达业务作为一个分形布朗运动(FBM)业务,建立了一个针对多个FBM业务及对应时延要求的差分队列服务模型,并推导出该模型下的丢包率公式.仿真结果表明,文中模型分析的结果与基于数据包粒度的仿真结果是一致的.模型性能分析结果表明,实时业务的突发性对差分队列服务的服务质量保证影响较大.     

An M[X]/G/1 retrial g-queue with single vacation subject to the server breakdown and repair

An M[X]/G/1 retrial G-queue with single vacation and unreliable server is investigated in this paper. Arrivals of positive customers form a compound Poisson process, and positive customers receive service immediately if the server is free upon their arrivals; Otherwise, they may enter a retrial orbit and try their luck after a random time interval. The arrivals of negative customers form a Poisson process. Negative customers not only remove the customer being in service, but also make the server under repair. The server leaves for a single vacation as soon as the system empties. In this paper, we analyze the ergodical condition of this model. By applying the supplementary variables method, we obtain the steady-state solutions for both queueing measures and reliability quantities.     

A Novel Processor Scheduling Schema in Programmable Router Based on Queue Length Thresholds

Programmable router may execute computing tasks except for forwarding IP packets. How to schedule the CPU in the programmable router is an important issue that needs to be solved. In this paper, the authors establish general software architecture of programmable router firstly. Based on this architecture, authors present a novel CPU scheduling algorithm based on queue length thresholds. We model this algorithm using stochastic Petri nets. The analytic results show that this scheduling algorithm can guarantee the requirements of computing of best-effort flows and QoS flows in prograrmmable router at the same time.     

A Self-tuning Fuzzy Queue Management Algorithm for Congestion Control

This letter presents an effective self-tuning fuzzy queue management algorithm for congestion control.With the application of the algorithm, routers in IP network regulate its packet drop probability by a self-tuning fuzzy controller. The main advantage of the algorithm is that, with the parameter self-tuning mechanism, queue length can keep stable in a variety of network environments without the difficulty of parameter configuration. Simulations show that the algorithm is efficient, stable and outperforms the popular RED queue management algorithm significantly.     

基于NS2的优先级队列管理算法设计

网络控制系统的信号以分组的形式在网络中传输,引起了传输时延、丢包等问题;为了更好地研究网络控制系统的性能,利用NS2搭建了网络控制系统的网络传输模型.并以其中的优先级业务分组为研究对象,设计了优先级队列管理算法;在DropTail算法的基础上进行改进,继承了原有算法的优势,并加入了业务优先级识别和弃包选择机制;通过调用两种算法进行实验,从时延、丢包以及吞吐量方面进行分析与对比,验证了PDropTail算法的有效性.     

N次随机丢包的被动队列管理算法

针对主动队列管理参数设置敏感、消耗大量资源等问题,提出了N次随机丢包的被动队列管理算法.当队列满时,根据当前网络中与路由节点链接的发送端数目,决定丢弃数据包N的值,然后再采用改进的随机丢弃策略丢弃N个数据包.新的算法与其他被动随机丢弃策略相比,克服了弃尾队列管理的缺陷,能够更快地响应网络拥塞,提高了网络传输性能,同时也能够在一定程度上保证公平性,NS2仿真表明了算法的有效性.