Browse Prior Art Database

Minimizing Average Service Time on a Synchronous Loop Communications System

IP.com Disclosure Number: IPCOM000079387D
Original Publication Date: 1973-Jun-01
Included in the Prior Art Database: 2005-Feb-26
Document File: 4 page(s) / 57K

Publishing Venue

IBM

Related People

Spragins, JD: AUTHOR

Abstract

It is desired to minimize the average service waiting time and queue length for messages entered at each highly buffered terminal, connected in a synchronous loop communication system. Usually, messages are transmitted on the loop via fixed length and formatted time slots. The method contemplates the steps of having a central station generating unassigned (no designated address) time slots available for downstream terminal seizure, as an inverse function of the average number of messages per time slot as a parameter. If the average is less then 0.5, then the central station alternates sending an addressed and an unassigned time slot in succession. If the average is less than 0.1, then the central station inserts an unassigned time slot after ten addressed time slots have been sent.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 35% of the total text.

Page 1 of 4

Minimizing Average Service Time on a Synchronous Loop Communications System

It is desired to minimize the average service waiting time and queue length for messages entered at each highly buffered terminal, connected in a synchronous loop communication system. Usually, messages are transmitted on the loop via fixed length and formatted time slots. The method contemplates the steps of having a central station generating unassigned (no designated address) time slots available for downstream terminal seizure, as an inverse function of the average number of messages per time slot as a parameter. If the average is less then 0.5, then the central station alternates sending an addressed and an unassigned time slot in succession. If the average is less than 0.1, then the central station inserts an unassigned time slot after ten addressed time slots have been sent.

Such a method may be considered as a queueing system having an input process, a service mechanism, a queue discipline, and a channel. In this regard, the input process consists of the Poisson arrival of messages being queued in a buffer of infinite size at each terminal. The service mechanism includes the generating of unassigned slots by the central station of fixed length, which slots are randomly seized by the terminals. The queue discipline is first in-first out, with the channel being characterized as a single channel with a single server.

The techniques described here are developed for a loop-control technique involving centralized control, fixed-slot duration and random-slot seizure. Time multiplexed slots carry information (see Fig. 1). Each slot is of the same duration, T, and may typically contain data plus any control or addressing information necessary. Random-slot seizure implies that any terminal which is ready to transmit can seize the first free slot it sees and place its data in this slot. The control discipline gives priority to terminals near the central (just downstream from it), since they see slots first. The central has the highest priority of all, since it generates the slots. No release of slots by terminals is considered here.

For the loop control stated above, it can be shown that the system average waiting time W(ay) (the average, over all terminals, of the time a terminal waits for a free slot) is (1) W(av) = (1 + C/2/(S)) T over 2 [1-(lambda o+N lambda)T] with T the duration of a slot, lambda T the percentage of the slots used for data distribution, and lambda T the percentage of the slots required to service each of N terminals attached. The term C is given by (2) C/2/(S) = lambda o[(1- lambda oT) B (C/2/(B) + 1) - lambda oT] with B and C(B) the mean and coefficient of variation of the duration of data distribution bursts (use of successive slots for data distribution). With lambda o and T normally fixed by the hardware and application, these equations indicate that the optimum system (in terms of minimizing W(w)) minimizes the product of B and C/2...