Abstract
Mobile Ad hoc Networks (MANETs) play a vital role in tactical and military communication scenarios where fixed infrastructure is unavailable or unreliable. However, the highly dynamic nature of such environments—characterized by frequent topology changes, node mobility, limited energy resources and fluctuating link quality—poses significant challenges to conventional MANET routing protocols. Traditional routing approaches such as AODV, DSR and OLSR rely on static or rule-based mechanisms and often fail to achieve optimal performance under high-mobility and resource-constrained conditions. Motivated by these limitations, this paper proposes a machine-learning-assisted optimization framework for tactical MANET routing to improve key network performance indicators, including throughput, end-to-end latency, packet delivery ratio and energy efficiency. The proposed framework leverages realtime network data on node mobility, traffic patterns and link quality to build predictive machine learning models that anticipate network state variations. These predictions are then used to enable adaptive, proactive routing decisions that respond effectively to both current and anticipated future network conditions. A comprehensive simulationbased evaluation is conducted under diverse traffic loads and mobility scenarios and the proposed approach is benchmarked against established MANET routing protocols. The results demonstrate that the machine-learningassisted framework consistently outperforms conventional protocols, achieving higher throughput, reduced latency, improved packet delivery ratio and lower energy consumption, particularly in high-mobility environments. Overall, the proposed framework provides a scalable, energy-aware and adaptive routing solution suitable for mission-critical tactical MANET deployments.
Keywords: Latency, Machine Learning, Packet Delivery Ratio, Routing Algorithms, Tactical Networks, Throughput.