Machine Learning-Based Congestion Detection and Traffic Management in MANET

Abstract

Mobile Ad Hoc Networks (MANETs) have become an important communication technology for infrastructure-less wireless networking due to their ability to support dynamic topology, decentralized communication, and rapid network deployment in military operations, disaster recovery, emergency response, intelligent transportation, and mobile computing environments. However, the continuously changing network topology, limited bandwidth, node mobility, and shared wireless communication medium frequently lead to network congestion, packet loss, increased routing overhead, communication delays, and degraded Quality of Service (QoS). Traditional congestion control mechanisms primarily rely on predefined threshold values and reactive routing strategies, which often fail to adapt efficiently to highly dynamic MANET environments. Machine Learning has emerged as an intelligent computational approach capable of predicting network congestion, identifying traffic patterns, optimizing routing decisions, and improving communication performance through data-driven learning. This experimental study proposes a Machine Learning-Based Congestion Detection and Traffic Management Framework for Mobile Ad Hoc Networks that integrates network monitoring, traffic feature extraction, congestion prediction, machine learning-based classification, adaptive traffic management, intelligent routing optimization, and performance evaluation into a unified computational architecture. The proposed framework employs supervised machine learning algorithms to classify congestion states and dynamically optimize traffic flow while minimizing packet loss and communication delays. A mathematical framework and algorithmic strategy are developed to evaluate congestion prediction accuracy, packet delivery ratio, throughput, end-to-end delay, routing overhead, network utilization, and Quality of Service. Experimental evaluation demonstrates that the proposed framework significantly improves congestion detection accuracy, communication reliability, traffic distribution, routing efficiency, and overall MANET performance while reducing packet loss, communication delay, and network congestion. The proposed framework provides valuable guidance for researchers, communication engineers, and network designers seeking to develop intelligent, scalable, adaptive, and computationally efficient congestion management systems for Mobile Ad Hoc Networks.