A thesis submitted to the Faculty of Engineering,
University of the '\Vitwatersrand, Johannesburg,
in fulfilment of the requirements for the degree
of
Doctor of Philosophy / THIS STUDY identifies and evaluates factors that affect call connections in the South
African public data networks, modelling these factors to aid data network planning. The research
shows the relationship between the economic rank of each region served and the data
communication resources required for that region. Moreover, it shows the resources required
between regions.
THE THRUST of this thesis is that the volume of cans from a region can be estimated
from its economic ...k and more than 75% olthe variation in the volume of calls between regions
can be explained using the ranks of the originating and terminating regions. To prove this, records
of more than four million calls are accumulated for all regions of the South African packet
switched data network. An appropriate filtering and aggregation method is developed.
EXISTING growth models including the gravity model are separately examined. Based
on probability and dimensional arguments, the Bell System growth model is selected. It is
revealed that the success of this model depends on one premise being satisfied: this model tacitly
anti implicitly assumes that the originating and terminating calls are statistically independent.
RETURNING to the data network, it is found that the call connections (after filtering
and aggregation) display dependence of destination on origin. Reasons for the dependence are
discovered. Multiple linear regression reveals the nature of this dependence. Surprisingly,
distance is not a factor. The importance of regional ranks and an inter-regional indicator variable
are also discovered.
FINALL Y, call volume from a node is shown to be directly linked with the weighted
Gross Regional and Industrial Product of the region. This quantity, in tum, is inversely related
to the rank of the region. Call connections are then modelled to be equal to the call connections
within the first tanked region divided by the product of the originating region's rank and the
terminating region's rank. This simple and economical model explains 76% of the variations that
occur in call connections. It has proved its use by being included in the data transfer services
product-line report. / Andrew Chakane 2018
Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:wits/oai:wiredspace.wits.ac.za:10539/26144 |
Date | January 1992 |
Creators | Kennedy, Ian Geoffrey |
Source Sets | South African National ETD Portal |
Language | English |
Detected Language | English |
Type | Thesis |
Format | application/pdf |
Page generated in 0.0016 seconds