Machine Learning Based Routing for the Internet of Things

Larouci, Nour Elhouda (2025) Machine Learning Based Routing for the Internet of Things. Doctoral thesis, Université Mohamed Khider (Biskra - Algérie).

Text Larouci_Thesis_Last_Version.pdf Download (31MB)

Abstract

As the Internet of Things (IoT) continues to evolve and expand globally, it presents growing challenges for achieving efficient, scalable, and reliable data communication, particularly in mobile and resource-constrained environments. Mobile Ad hoc Network (MANET) routing protocols, commonly used in such settings, often rely on local information exchange and are not optimised for the dynamic and context-aware nature of smart city networks. In this thesis, we are interested in enhancing routing efficiency in mobile IoT networks by incorporating learning-based decision models tailored to smart city environments, where mobility patterns can be leveraged to improve routing outcomes. To this end, two model-based routing contributions have been proposed. The first introduces a Machine Learning-Based Routing Protocol (MLBRP) that uses historical routing information to train predictive models. Decision Trees (DT), Support Vector Machines (SVM), and Artificial Neural Networks (ANN) are employed to guide packet forwarding decisions. Simulation results show that MLBRP can reduce control message overhead by up to 73%, while enhancing energy efficiency, load balancing, and packet delivery performance. The second contribution refines this approach by improving routing accuracy through deep learning. A Convolutional Neural Network (CNN)-based framework is proposed, enabling forwarding decisions based solely on local contextual and geographic information. By exploiting the regular mobility patterns observed in smart city scenarios, this model enhances path optimality, reduces end-to-end delay, and extends network lifetime. Overall, the results demonstrate that machine learning, and particularly deep learning, offers strong potential for improving the intelligence, adaptability, and efficiency of routing in next-generation mobile IoT networks.

Actions (login required)

View Item