Performance Analysis of Routing Protocols in Manets under Higher Density and Realistic Mobility

Mobile Ad Hoc Networks (MANETs) are self-organizing wireless systems without fixed infrastructure, widely applied in military operations, disaster response, and mobile collaboration. Their dynamic topology, high mobility, and varying node densities make routing highly challenging. This study evaluates the performance of the traditional Ad Hoc On- Demand Distance Vector (AODV) protocol and its machine learning–enhanced variant (ML-AODV) under realistic mobility patterns and high-density conditions. Simulations were conducted in NS-2.35 within a 2000 × 2000 m area using node densities of 100, 500, and 1000, three mobility models (Random Waypoint, Random Walk, and Levy Walk), and varying node speeds. Constant Bit Rate (CBR) traffic with 512-byte packets was used, and performance was assessed through Packet Delivery Ratio (PDR), throughput, end-to-end delay, routing overhead, and jitter. Results show that ML-AODV outperforms AODV in Random Waypoint and Random Walk scenarios, achieving up to 23% higher PDR, lower jitter, and more than 50% reduction in routing overhead. However, AODV performs better under the Levy Walk model at medium and high speeds, especially in dense networks, due to its lightweight route discovery mechanism. Overall, ML-AODV is more effective in unpredictable or human-like mobility environments, while AODV remains advantageous in dense and structured conditions.