Bidirectional token flow system

  • Authors:
  • M. E. Ulug;G. M. White;W. J. Adams

  • Affiliations:
  • -;-;-

  • Venue:
  • SIGCOMM '81 Proceedings of the seventh symposium on Data communications
  • Year:
  • 1981

Quantified Score

Hi-index 0.00

Visualization

Abstract

Over the last few years there has been an avalanche of technical literature on techniques for allowing nodes to access and share one of the most valuable resources of a computer communication network, i.e., the bandwidth of the communication media.1 This report describes the design of a new type of Local Area Computer Network called BID that uses a bidirectional token flow as a form of token passing.2 The design of BID not only allows it to process messages having various properties but also makes it possible to operate simultaneously in a packet and circuit-switching mode. This latter mode, using minipackets, enables BID to be used for those applications having real-time response requirements. In describing BID this report defines the basic concepts of the implicit token, token carrier, token generator, and token wait time. It also describes a new preemption technique which reduces the walk time normally associated with token and polling systems. A GPSS computer simulation was made of the BID system to determine the average delay times experienced by the various priority classes of messages. An analytical model using a single priority class of messages is presented, and the results of the computer simulation are compared with this. The simulation results also indicate the uniformity of the average delays experienced by the Bus Interface Units across the bus resulting from the use of bidirectional flow of tokens.