AI-Powered Route Selection in Ad Hoc Wireless Networks

Ad hoc wireless networks (AHWNs) are decentralized and change without warning; therefore, selecting a good route presents a key problem. Established routing protocols battle to change to many network topology changes. They also have a hard time with diverse link conditions. In this study, we propose an AI-supported method for picking routes that uses a Random Forest Classifier (RFC) to make the best choice of paths in AHWNs. We use network graph modeling, feature extraction, and machine learning to forecast the optimal routes according to node attributes. We use NetworkX to create a network graph; every node is a wireless device and includes location coordinates, number of neighbors, and average shortest path length as features. A Random Forest Classifier is used to train a supervised learning model, and this classifier learns certain routing patterns from a labeled dataset that is produced through shortest path computations. The normalized network data trains a quantity of models, and each model is assessed for its predictive performance regarding routes.