基于离散滑模控制的通信网络拥塞控制(英文)

Congestion control in communication networks using discrete sliding mode control

最近30年来,离散滑模控制(DSMC)被加以研究.在现有离散滑模控制的文献中,不但有完善的理论,并且有应用的实例.计算机通信网络(涵盖有线和无线情形)的发展给控制专家提出了具有挑战性的问题.此类系统的响应很快,因此要求控制器必须能快速反应.同时,此类系统中的数据包和路由采用固有的切换方式.因此,在该网络系统中采用与其固有切换类似的滑动模态控制成为一个有趣的问题.本文应用离散滑模控制方法来解决通信网络的拥塞控制问题.在过去几年中,许多反馈控制方法被用来解决通信网络的许多问题,如流量控制,拥塞控制和功率控制.本文提出了一种新的应用滑动模态控制来实现通信网络拥塞控制的方法.由于计算机通信网络处理的是数字数据且在本质上是离散的,因此其是一类离散时间系统,因而可应用离散滑模控制.在该系统中,比特率保持恒定不变,通过调整缓冲区的长度来控制拥塞.出于模拟与分析的目的,缓冲区的最大长度被设为250bit.本文展示了离散滑模控制能够实现通讯网络不同特性的控制.

Discrete sliding mode control （DSMC） has been investigated for last three decades. It has reached to a point wherein a well established theory as well as implementation reports are available in the literature. Advances in computer communication networks with wired and wireless configurations have imposed a challenging problem for the control community. The response of such system is very fast and fast acting controller needs to be designed. At the same time, such system uses inherent switching of data packets and routes. Hence, it has been found interesting to use inherent switching like control, a sliding mode control for such networks. This paper explores the utilization of the discrete sliding mode control approach for congestion control in communication networks. For the past several years many feedback control methods have been used for the control of various parameters of communication networks like Traffic Control, Congestion Control and Power Control. The work presented here proposes a new application of sliding mode control for congestion control in communication networks. As computer communication networks deal with digital data and are inherently discrete in nature, discrete sliding mode control was applied to discrete time systems. The bit rate has been kept constant and buffer length has been controlled for control of congestion. The buffer length has been kept at maximum of 250 bits for simulation and analysis purposes. Thus, it has been shown that discrete sliding mode control can be used to control various parameters of communication networks.