Predicting quality of service parameter for Internet of Things network

Abstract

Nowadays, the Internet of Things applications are developed to facilitate users worldwide. Machine learning is very useful to make some decisions for those applications. The main aim of this present work is to provide a model which can predict parameters for IoT network based on various application requirements. The research design mainly includes data collection phase observing a network and learning phase. It is simulated for different parameters of the network which will help to create different network scenarios. Two types of learning models have been applied on the collected data: a linear model and non-linear models. As a part of a linear model, linear regression is applied. The regression tree and random forest are also applied as a part of non-linear models. The results show that there is a non-linear relation between features and the QoS parameter selected. The training and testing score for regression tree are 1.0 and 0.99 and random forest are 0.99 and 0.99 respectively. This shows that regression tree overfits thetraining data. The random forest overcomes that problem and works better. The regression tree and random forest work better for this study.