Recent Advances in Prediction of scenarios for routing in IoT based MANETs on expanding ring search and random early detection parameters using global pooling dilated convolutional neural network: A Systematic Review

Abstract

Mobile Ad Hoc Networks (MANETs) integrated with Internet of Things (IoT) devices create highly dynamic and decentralized communication environments where efficient routing is essential for maintaining reliable network performance. Traditional routing protocols often struggle with challenges such as node mobility, congestion, routing overhead, and frequent topology changes. Techniques like Expanding Ring Search (ERS) and Random Early Detection (RED) have been widely adopted to improve routing efficiency and congestion control. ERS reduces unnecessary network flooding by incrementally expanding the search radius during route discovery, while RED proactively manages congestion by controlling packet queue behavior, thereby enhancing overall network stability and energy efficiency.Recent advancements have introduced machine learning and deep learning techniques for predicting routing behavior based on network parameters. These models analyze ERS and RED metrics to estimate performance indicators such as packet delivery ratio, throughput, and end-to-end delay, enabling adaptive routing decisions under varying conditions. Deep learning architectures, particularly Convolutional Neural Networks (CNNs) and dilated CNNs, effectively capture spatial and temporal dependencies in network data. Dilated convolutions expand the receptive field without increasing computational cost, improving prediction accuracy. This review provides a comprehensive analysis of routing optimization approaches and highlights future directions for intelligent, scalable, and adaptive routing in IoT-based MANET systems.