PREDICTION OF NETWORK CONGESTION USING ARTIFICAL NEURAL NETWORK

BUHARI ODUNMOLORUN ANUOLUWA
Computer Science, Federal University of Agriculture, Abeokuta
March, 2015
Full text (external site)
 

Abstract

In telecommunication service provision, the problem of network congestion has remained a
major challenge both to the service providers as well as the subscribers. In this project work,
a research design methodology using Time Series Analysis and an Artificial Neural Network
(ANN) model for predicting the possible outcomes of data being considered as factors that
determine network congestion was gotten and tested. These factors include number of
channels available, number of calls served and the number of calls dropped, where number of
channels available and number of calls served was used as input variables for the ANN
model. The Levenberg – Marquardt backpropagation algorithm present in Matlab neural
network toolbox was used to compute the model and the least square method was also use.
The two results were compared to each other and the result shows that the predicted ability of
the neural network tool was good.