Implemented various machine learning algorithms to calculate path loss of signals and concluded random forest as the best algorithm to predict path loss. Documentation for this research work has been submitted to the reviewers of VTC conference which is scheduled at Belgium in May 2020.
Path loss models are important to predict signal accessibility, use limited network resources effectively and optimize performance of wireless communications. Many researches have been carried out taking deployment of machine learning for modelling of path loss propagation in different scenarios such as railway, indoor and outdoor environment but contribution towards building path loss models for smart campus environment has not been processed yet. Our efforts in modelling the pass loss for campus environment through machine learning algorithms resulted in surprisingly highly accurate . In the presented work, measured data from Covenant University, Ota, Ogun State, Nigeria are used to train and evaluate the performance of different machine learning methods such as random forest, k-nearest neighbour, artificial neural networks (ANN) to prepare a suitable path loss model for smart campus environment. The statistical analysis shows that machine learning models outperform traditional propagation models such as log distance model but in this paper the research has been further extended to compare machine learning models with standard empirical models such as COST-231 Hata with regard to accuracy, complexity and prediction time. The prediction performance of trained models is assessed on both the train set and test set according to the metric such as mean squared error.