Artificial Intelligence Techniques for Similarity-Navigated Graph Neural Networks and Lightweight Cryptography for Preventing Black Hole Attacks in MANET: Trends and Challenges

Abstract

Mobile Ad Hoc Networks (MANETs) are decentralized wireless systems that operate without fixed infrastructure, making them highly flexible but vulnerable to security threats such as black hole attacks. In such attacks, malicious nodes falsely advertise optimal routes and drop packets, severely degrading network performance. Traditional security mechanisms, including trust-based routing and conventional cryptography, often fail to provide efficient and scalable protection due to high computational overhead and limited adaptability. Recent advancements in Artificial Intelligence (AI), particularly Graph Neural Networks (GNNs), have introduced topology-aware security solutions capable of detecting anomalous node behavior through similarity-based learning. Additionally, lightweight cryptographic techniques have emerged as effective methods to ensure secure communication with minimal resource consumption. This paper presents a comprehensive analysis of AI-driven approaches, focusing on similarity-navigated GNN models combined with lightweight cryptographic mechanisms for preventing black hole attacks in MANETs. The study reviews recent literature, highlighting key methodologies, performance improvements, and research gaps. Comparative analysis demonstrates that hybrid frameworks integrating GNN-based detection and lightweight encryption achieve superior accuracy, improved packet delivery ratio, and reduced latency. The paper also discusses challenges such as scalability, energy efficiency, and adversarial robustness, and outlines future research directions for secure MANET architectures.