City Research Online

The design of a packet-switched local area network

Morling, R. C. S. The design of a packet-switched local area network. (Unpublished Doctoral thesis, The City University)

Abstract

MININET Is a local area network designed for instrumentation and other real-time applications. It is a true store-and-forward network using small fixed-length packets. A very high degree of transparency is required, such that the Network Service can be totally invisible to the user devices. The design of the network is based around a hierarchical architectural model which is similar to. but not identical with, the ISO OSI Reference Model. Its small 32—bit packet size was a logical consequence of the transparency and real-time service requirements. This size was found to be optimum for average user message lengths of around 13 bits, as far as the buffer utilization efficiency was concerned. A simple, but robust, full-duplex data link protocol, which avoids sequence errors and uses only a single sequence number field to interlock the packet stream, was developed. Network congestion is avoided by means of a flow control algorithm, which uses active backpressure vectors and a separate buffer allocation for each destination node. to guarantee freedom from store-and-forward deadlock. A highly reliable, half-duplex, end-to-end protocol providing a Transport Service for the network management entities, has been developed. The network compatible DIM intermediate interface has been specified, together with DIM-CPC. its basic user protocol providing flow control, initialization and error recovery procedures. A routing protocol, that maintains packet sequency even in the event of node or link failure, has been developed. This distributed algorithm constructs a separate tree rooted at each destination node in the network. It uses short messages transferred only between adjacent nodes. A quad-phasic update cycle is used to guarantee loop freedom at all times and to flush old pathways before routing changes are made. Thus, packet sequency is maintained without any packets being dropped. A high-speed implementation of the network Station has been designed and constructed. This uses a two-dimensional polling technique in order to maintain fairness, whilst ensuring that no blockages occur within the node. A design technique has been pioneered for the construction of PROM-based system controllers which are exceptionally agile.

Publication Type: Thesis (Doctoral)
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Q Science > QA Mathematics > QA76 Computer software
Departments: School of Science & Technology
School of Science & Technology > School of Science & Technology Doctoral Theses
Doctoral Theses
[thumbnail of Morling thesis 1988_Redacted PDF-A.pdf]
Preview
Text - Accepted Version
Download (128MB) | Preview

Export

Add to AnyAdd to TwitterAdd to FacebookAdd to LinkedinAdd to PinterestAdd to Email

Downloads

Downloads per month over past year

View more statistics

Actions (login required)

Admin Login Admin Login