The routing and flow control techniques developed for wide-area, local-area, and metropolitan-area networks are surveyed. A classification that shows the characteristics that are desirable for high-speed wide-area networks is developed. On the basis of the classification, techniques that should and should not be considered for future high-speed networks are identified. In particular, an attempt is made to determine the effect of increasing transmission rates on the routing and flow control algorithms that are used in packet-switched networks. >

Routing and flow control in high-speed wide-area networks

Advances in communication technology have made the design of sophisticated computer networks feasible. A principal advantage of a computer network is that it permits the sharing and exchange of information and computational resources by a large number of users. An evolutionary extension to the concept of data sharing is now possible: geographically distributed and physically dissimilar databases may be treated as though they comprise a single, logically uniform database.

Modeling and Design of Distributed Information Systems

This paper describes a general point of view on the subject of store-and-forward queueing networks. A systematic approach to a store-and-forward network as a whole is applied. This allows network problems which involve dynamic optimization of congestion control and routing to be treated in a common framework. The dynamic model applied is non-linear and can be handled straightforwardly by optimal control system theory. It is shown that under normal operational conditions the optimal flow pattern is given by the steady-state solution of the costate equations. Properties of the optimal flow pattern are examined in detail. Then, schemes of the operation of the network under different operating conditions are discussed at length. It is shown that the optimal control strategies depend on the total volume of traffic forwarded to the network. Algorithms are developed for different traffic demand patterns, and a numerical example is also given.

Optimal control of store‐and‐forward networks

Expressions are given for the first and second moments of the queue-length distribution of the single-server queue with constant service time and general arrival distribution. A general relationship between the moments of the queue-length and message-delay distributions holds in the particular case of a first-come-first-served queue discipline. The results for a single queue enable mean message delay to be estimated for a network of queues, with both deterministic and adaptive routing doctrines. Criteria for the rational choice of a deterministic doctrine for a given network and traffic matrix follow directly. A simulation study of a practical network supports the theoretical discussion, and reveals the message-delay distribution in greater detail. Adaptive simulation emerges as a useful tool for designing a network to carry a given traffic loading most efficiently.

https://ieeexplore.ieee.org/iel5/5247218/5249341/05249345.pdf

Queueing model for message-switching networks with constant-length messages

An inherent capability of packet-switched networks is the speed at which they can be reconfigured; various dynamic or adaptive routing techniques have been conceived to exploit this capability. In this study, existing techniques are described and an "ultra-dynamic" technique, delta routing, is invented. Several promising techniques are then selected for comparison with one another and with a network's ultimab carrying capacity. The goal is to shed light on the questions if and when one should use which kind of adaptive routing. First, a taxonomy of routing strategies is presented. In addition to delta routing, random, proportional, shortest path, and shortest path fixed-for-session duration are selected for comparison and their mechanisms described. The delay and efficiency performance of the five techniques are then compared with one another and with ideal behavior via simulations. These have been carried out concentrating on four very small networks, each with very different characteristics with the intention of gaining in- sight into the strengths and weaknesses of the various techniques. A ten-node network has also been simulated. The results favor delta routing which is most effective in highly interconnected network enviroments.

On Routing and "Delta Routing": A Taxonomy and Performance Comparison of Techniques for

Three practical constraints on the choice of a deterministic routing doctrine for a connected network are defined, leading to a class of doctrines for which a weighted graph of the network exists so that the specified routes are given uniquely by minimum-cost routes through the graph. A technique for finding such a weighted graph is given. An adaptive routing doctrine is defined, which considers the current delays along links of the network. The implementation of such a doctrine is discussed, and a ‘distributed’ solution described which is efficient, decentralised, and resilient under fault conditions. More general adaptive routing doctrines are defined, considering two parameters associated with each link, where the routing of a message may depend on its history as well as its present position and destination. A device called a dynamic route selector is described, which takes routing decisions according to the most general doctrine discussed. It is faster than a digital computer. Finally a method for first introducing deterministic, and then adaptive, routing on a computer-based network is outlined.

D.J. Silk

Papers

Routing doctrines and their implementation in message-switching networks