DYNAMIC BEHAVIOR AND CONTROL OF COMMUNICATIONS NETWORKS.

Abstract

An arbitrary store-and-forward network where messages without priorities can enter the network at any node and be destined for any other node is considered. The problem of routing messages is considered to be a feedback control problem in which information about the number of bits waiting at various places in the network is fed back to points where decisions are made. Mathematical models which represent system behavior are developed. The model of a single channel is a new Markov process queuing model with random arrivals and deterministic service times. Simple models which predict future queue lengths are developed. A way of reducing a set of series-parallel channels to a single equivalent channel is presented. A method of determining message routing decision rules which are based on what is known at a decision point is developed. A method of determining whether and how often to obtain information about another part of the network is developed. Many of the methods apply only to series-parallel networks. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1965
Accession Number
AD0617723

Entities

People

  • John U. Beusch

Organizations

  • Massachusetts Institute of Technology

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Feedback
  • Markov Processes
  • Mathematical Models
  • Models

Fields of Study

  • Computer science

Readers

  • Computer Networking
  • Control Systems Engineering.
  • Graph Algorithms and Convex Optimization.