Designing a robust active queue management(RAQM)is mandatory to avoid congestion in networks with wireless access links,because transmission control protocol(TCP)can detect con-gestion after its occurrence in a commun...Designing a robust active queue management(RAQM)is mandatory to avoid congestion in networks with wireless access links,because transmission control protocol(TCP)can detect con-gestion after its occurrence in a communication network and wireless links suffer from bottleneck capacity variations caused by fading and packet error rate(PER)in the acknowledgement pack-ets.Furthermore,the average window size cannot be measured explicitly from the output signal and input and state delay are imposed to a network,which complicate theRAQMdesign problem in nonlinear models.So,the main contribution of this study is to design a robust observer based control procedure based on integral sliding mode protocol to estimate the average window size,to control congestion in a TCP/RAQM network and to compensate input and state delay.Sim-ulation results via professional simulator NS-2 and SIMULINK confirm that the procedure can effectively estimate the window size and can robustly avoid congestion.展开更多
主动队列管理(active queue management,简称AQM)是网络拥塞控制的研究热点之一,其中的关键问题是如何设计反馈控制策略.提出一种新的基于D稳定域和时间乘以误差绝对值乘积积分(integral of time-weighted absolute error,简称ITAE)性...主动队列管理(active queue management,简称AQM)是网络拥塞控制的研究热点之一,其中的关键问题是如何设计反馈控制策略.提出一种新的基于D稳定域和时间乘以误差绝对值乘积积分(integral of time-weighted absolute error,简称ITAE)性能准则的比例-积分-微分(proportional-integral-differential,简称PID)优化设计方法(简称DITAE-PID),并用于AQM控制器的设计,控制闭环系统的理想动态性能.首先在复平面上设定一组理想的D稳定域,然后以ITAE为目标函数,通过数值优化算法求出控制器的参数,使得闭环系统的所有特征根都在D稳定域内,以降低排队延时,提高有效吞吐量.对比仿真实验结果表明,该算法能够预先探测和控制拥塞,有较好的鲁棒性,链路利用率更高,丢包率更小,平均队列长度更趋于期望值,同时,趋于期望队列长度的时间更短,其综合性能明显优于典型的随机早期探测(random early detection,简称RED)和比例-积分(proportional-integral,简称PI)算法.展开更多
文摘Designing a robust active queue management(RAQM)is mandatory to avoid congestion in networks with wireless access links,because transmission control protocol(TCP)can detect con-gestion after its occurrence in a communication network and wireless links suffer from bottleneck capacity variations caused by fading and packet error rate(PER)in the acknowledgement pack-ets.Furthermore,the average window size cannot be measured explicitly from the output signal and input and state delay are imposed to a network,which complicate theRAQMdesign problem in nonlinear models.So,the main contribution of this study is to design a robust observer based control procedure based on integral sliding mode protocol to estimate the average window size,to control congestion in a TCP/RAQM network and to compensate input and state delay.Sim-ulation results via professional simulator NS-2 and SIMULINK confirm that the procedure can effectively estimate the window size and can robustly avoid congestion.
基金Supposed by the National Natural Science Foundation of China under Grant No.60373053(国家自然科学基金)the National High-Tech Research and Development Plan of China under Grant Nos.2004AA1Z2100+3 种基金2005AA113140(国家高技术研究发展计划(863))the State Education Ministry Scientific Research Foundation for the Returned Overseas Chinese Scholars under Grant No.[2003]406(国家教育部留学回国人员科研启动基金)the One-Hundred-Talent Program of the Chinese Academy of Sciences(中国科学院"百人计划")the Chinese Academy of Sciences and Royal Society of United Kingdom for the Joint Research Project under Grant No.20030389(中国科学院与英国皇家学会国际合作项目)
文摘主动队列管理(active queue management,简称AQM)是网络拥塞控制的研究热点之一,其中的关键问题是如何设计反馈控制策略.提出一种新的基于D稳定域和时间乘以误差绝对值乘积积分(integral of time-weighted absolute error,简称ITAE)性能准则的比例-积分-微分(proportional-integral-differential,简称PID)优化设计方法(简称DITAE-PID),并用于AQM控制器的设计,控制闭环系统的理想动态性能.首先在复平面上设定一组理想的D稳定域,然后以ITAE为目标函数,通过数值优化算法求出控制器的参数,使得闭环系统的所有特征根都在D稳定域内,以降低排队延时,提高有效吞吐量.对比仿真实验结果表明,该算法能够预先探测和控制拥塞,有较好的鲁棒性,链路利用率更高,丢包率更小,平均队列长度更趋于期望值,同时,趋于期望队列长度的时间更短,其综合性能明显优于典型的随机早期探测(random early detection,简称RED)和比例-积分(proportional-integral,简称PI)算法.
文摘针对基于流体流理论提出的网络模型在推导过程中的一些近似使得该模型对网络行为描述的不精确问题,提出了网络流量的改进模型,并且基于该模型把一种新的类Proportional IntegralDifferential(PID)设计方法用于主动队列管理(AQM)控制器的设计,利用带约束的数值优化方法寻找控制器参数.理论分析和仿真实验表明,该控制算法的综合性能优于已有的Random Early Detection(RED)、Proportional Integral(PI)等算法.表现为平均队列长度更趋于期望值;调节时间更短;队列长度的抖动更小;抗突发业务流干扰能力更强;自适应能力更强.
