Optimal Routing with Machine Learning Classification in Delay Tolerant Networks

Communication networks built to function well in the face of erratic connections, frequent interruptions, or appreciable delays in data transmission are known as delay-tolerant networks or DTNs. Conventional routing protocols have difficulty operating at their best, guaranteeing effective message delivery. Machine learning (ML) techniques have shown great promise in recent years as instruments to enhance DTN performance. To forecast network traffic and identify which source nodes have the best chance of successfully delivering a message to its intended recipient, this makes use of machine learning classifiers. Based on past delivery data, it also locates the nearby nodes engaged in the message delivery process. This research investigates the processing efficiency of Support Vector Machines (SVM), Random Forest (RF), Gradient Boosting (GB), and XGBoost algorithms in combination with ensemble techniques like voting and stacking classifiers. Through a systematic training and assessment process, select features and adjust hyperparameters to improve prediction accuracy. Through this, the suggested machine learning models' adaptability to various network conditions is showcased, along with their robustness in predicting the best routes for message delivery, optimizing routing choices, and cutting down on network overhead, all of which contribute to the DTN environment's overall efficiency.