An explicit rate allocation paradigm for congestion control in ATM networks.

摘要

Asynchronous Transfer Mode (ATM) has been selected to be the technology supporting the future broadband paradigm. Since its introduction in 1984, ATM has been the subject of intensive research. ATM is expected to carry different types of traffic such as video, audio, data, and any future media that may emerge in the future. A distinct feature of ATM technology is the ability to support different Quality of Service (QoS) for different connections.; Although ATM has been viewed as a very promising technology, the acceptance of ATM is heavily depending on how efficient it can support data. Realizing this fact, the ATM Forum has introduced two service classes to support data, namely Available Bit Rate (ABR) and Unspecified Bit Rate (UBR). Of these two classes, only ABR provides means of congestion control.; In this study, we introduce a new explicit rate allocation scheme for ABR service. We propose a basic algorithm and options that found to improve the performance in different scenarios. Through an extensive simulation study, we show that the proposed scheme is fast, efficient, and fair. We also show that the proposed scheme is superior compared to other proposed schemes in the literature.; In a major contribution, the new scheme and its different options have gone through the complete cycle of the VLSI physical design process using the latest technology available in the world. This has enabled us to study the implementation feasibility of the new scheme and to study its different options from cost/performance stand point.; Although we don't show the detailed extension of the scheme to support multicasting, a mathematical proof has been provided to show that an efficient multicasting scheme should be scalable when mapped to the switch level. This proof implicitly shows that any efficient switch-level rate allocation scheme can be extended feasibly to handle multicasting traffic.; As a case study, we select the TCP/IP, the most dominant Transport layer protocol suite, and study its performance over the new scheme. The study shows that the new scheme was efficient in handling TCP traffic and it is less sensitive to various parameter mistuning.