The parallel and distributed simulation of network systems

Hosseinzaman, A., 1995. The parallel and distributed simulation of network systems. PhD, Nottingham Trent University.

Preview

Text 10182994.pdf - Published version Download (27MB) | Preview

Abstract

Successful computer simulation of water distribution system prompted the industry to consider development of on_line decision support systems which would incorporate existing telemetry systems and simulation software. Unfortunately, early attempts to accomplish this undertaking proved to be unsuccessful, largely due to the rapid increase of computational requirements of the simulation algorithms as a result of the increase in the physical network size.

In spite of using sparsity exploiting techniques, the nonlinear network solving algorithm demonstrates quadratic numerical complexity therefore is unable to cater for network growth.

An original approach to the solution of such systems would be to partition the overall problem into smaller units and solve these in isolation(and as a further improvement use parallel processing in solving the derived subsystems). Once the solution of the smaller units are known, they are combined in a coordinating routine to yield the overall solution.

The main objective of this research was therefore, to develop a suitable distributed simulation algorithm and to implement it in a distributed computing system. The research resulted in the development and implementation of a nonlinear diakoptics algorithm. This research has established a framework for the development of a distributed computing system based on the generalization of ADA rendezvous mechanism.

The overall distributed computing scheme was implemented on a tightly-coupled transputer network and on a network of loosely-coupled workstations connected using Ethernet communication link. The computational efficiency of the algorithm is evaluated using two realistic networks and results are extrapolated to large scale systems.

The overall computational efficiency and the amount of storage required in the network tearing method is strongly influenced by the way the system is partitioned. Two graph partitioning techniques were evaluated and their performances in partitioning the given example networks are compared and presented in the form of graphs in which derived partitions are clearly depicted.

Item Type:	Thesis
Creators:	Hosseinzaman, A.
Date:	1995
ISBN:	9781369312928
Identifiers:	Number Type PQ10182994 Other
Rights:	This copy of the thesis has been supplied on condition that anyone who consults it is understood to recognise that its copyright rests with the author and that no quotation from the thesis and no information derived from it may be published without the author’s prior consent.
Divisions:	Schools > School of Science and Technology
Record created by:	Linda Sullivan
Date Added:	28 Aug 2020 10:15
Last Modified:	15 Jun 2023 09:41
URI:	https://irep.ntu.ac.uk/id/eprint/40565

Actions (login required)

Edit View

Views

Downloads