Machine Learning Framework for Evaluating and Predicting Routing Protocol Performance in Mobile Ad Hoc Networks

Self-organizing, infrastructure-much less, wireless networks known as Mobile Ad Hoc Networks (MANETs) enable communication in fairly dynamic settings. On the other hand, common topology adjustments and node mobility frequently impair community dependability and routing performance. To deal with those concerning occasions, this test presents a hybrid simulation tool, a framework that combines a Random Forest classifier for adaptive routing protocol prediction with NS-2 simulations. Within the Random Waypoint Mobility Model (RWM), four routing protocols – AODV, DSDV, AOMDV, and DSR – are assessed in terms of the use of the subsequent essential Quality of Service (QoS) common overall performance metrics: throughput, packet delivery ratio (PDR), provide up to stop remove, and routing overhead. The Random Forest model may also then learn the use of the simulation results, which show the excellent behavior patterns of each method across a range of node densities. With a general accuracy of 96.84%, the advised model suggests strong predictive ability in identifying the most reliable routing protocol in various mobility conditions. The architecture decreases routing cost, reduces stop-to-save delay, maintains greater PDR stability, and increases throughput by up to 12% when compared to conventional heuristic strategies. By providing scalable and reliable routing optimization in dynamic network situations, this hybrid method advances self-analyzing and adaptive MANET designs.