Study Of Rate-Based Flow Control Mechanisms In ATM Network

With the coming of information society in 21th century, both users and ISPs hope to establish a single network, which can transfer either low or high-speed data and adapt not only the miscode of voice but also the time delay of picture. B-ISDN based on ATM can meet all the services in one network that have different needs in bandwidth, real-time and transport quality. ATM technology aims at support a wide variety of services and applications. ATM networks have four service categories to allow audio, video and data to share the same network. Of these, ABR and UBR service categories are designed to efficiently support data traffic.A primary role of traffic management in ATM networks is to protect the network and the end-system from congestion in order to achieve network performance objectives. The additional role is to promote the efficient use of network resources, such as link bandwidth, switch buffer and computational resources. Lots of traffic mechanisms are defined, which the network may utilize to meet the QoS objectives. The flow control mechanisms for ABR service category may be used to adaptively share the available bandwidth among several competition users.First the shortage of existed flow control mechanisms is pointed out and the need of further research is convinced.Then the performance of several existed rate-based flow control mechanisms in the face of ABR, TCP and self-similar service is studied with the experiments carried on the LINUX platform by NIST ATM network simulation software. The mechanisms' advantages and disadvantages are pointed out. Our work here provides the experiment basis about how to select an appropriate scheme according to the type of transport service. Furthermore, two improved mechanisms based on control theory aiming at EFCI and ER mechanisms are brought out. Their improved performance is proved by simulation.In summary, in this dissertation, on one side fundamental study on performance and improvement of the ATM rate-based flow control mechanisms are addressed. On the other side the performance of several typical rate-based flow control mechanisms in the face of some types of service is also studied.This research is supported by SEU Natural Science pre-studied project Foundation.