The Mobile Network Congestion Prediction Using Machine Learning: The Case of Abuja Telecom

The Mobile Network Congestion Prediction Using Machine Learning: The Case of Abuja Telecom

A mobile network, also known as a cellular network, is a radio network that is distributed over land areas known as cells, each of which is supplied by at least one fixed-position transceiver, also known as a cell site or base station. Congestion, fraud, and delay on international calls are among the issues that these networks confront. For practically all telecom service providers across the world, these issues are severe threats and customer churn issues. In the context of Abuja telecom network data, this paper seeks to handle mobile network congestion problems using machine learning techniques termed multilayer perceptron neural networks. The network data used in this article was obtained from Abuja Telecom’s key performance indicator database over a six-month period. For the aim of constructing the machine learning models, a total of 3080 data sets with 15 attributes are employed after removing unnecessary data, formatting the data organization, and clustering the data into three independent data sets for each site. By conducting a performance analysis of Multilayer Perception Neural Network models with different combinations of training algorithms, activation functions, learning rate, and momentum, it was found that the Multiple Layer Perception Neural Network model has 15 hidden layers each having 200 neurons with Adam optimizer training algorithm and Relu activation function delivered the lowest mean absolute error of 0.272 while another Multilayer Perception Neural Network model having 10 hidden layers having 200 neurons in each layer, the same activation function and training algorithm had the mean absolute error of 0.345. The results of this research showed that performance analysis of Multilayer Perception Neural Network models is a crucial process in model implementation of Multilayer Perception Neural Networks for mobile network congestion prediction and a multilayer perceptron having 15 layers can give a comparable prediction of the real mobile network congestion situation. The lack of sufficient data and enough expert knowledge of the performance parameters of the network were the major challenges faced during the craft of this research paper. Finally, through the results found in this paper, we recommend Abuja Telecom to implement this mobile network congestion prediction techniques and avoid such types of irregularities throughout the network which will improve user experience and reduce customer churn.

The Mobile Network Congestion Prediction Using Machine Learning: The Case of Abuja Telecom, GET MORE COMPUTER SCIENCE PROJECT TOPICS AND MATERIALS

Sharing is caring!