Thesis (M.Sc.)--Chulalongkorn University, 2001

This research focuses on Web applications retrieval time caused by a request from a client passed through the Intranet environment. In the Intranet environment, there are some physical factors that are considered as the independent variables such as the number of hubs, the number of switches, and the distance (total cable length) from a client to the server. Furthermore, this delay time also received major effects from various sizes of Web pages. This factor is counted as another independent variable in this research. The experiment is performed in an Intranet for fixed values of the each of independent variables. A delay time is measured and recorded after loading Web applications via an Internet browser. All observation values are applied for creating a model for predicting a proper delay time. The resulting model is in the form of a multiple regression model, which helps us to understand the effects from various physical factors over a network. The accuracy to predict a delay time using a testing data set is approximatly 98%. The regression model, which represents the relationship among mentioned variables, can be used to improve the network's performance or support the adjustment in a network's expansion. In addition, it can lead the way for studying effects from more physical factors in recent networks.