A Comprehensive Review of Methods and Architectures for Similarity-Navigated Graph Neural Networks and Lightweight Cryptography for Preventing Black Hole Attacks in MANET

Abstract

Mobile Ad Hoc Networks (MANETs) are decentralized wireless systems characterized by dynamic topology and absence of centralized infrastructure, making them highly vulnerable to routing-based attacks. Among these, black hole attacks pose a severe threat by allowing malicious nodes to advertise false routes and drop packets, significantly degrading network performance such as packet delivery ratio and throughput. This paper presents a comprehensive review of advanced methods for detecting and preventing black hole attacks, focusing on Similarity-Navigated Graph Neural Networks (SNGNN) and lightweight cryptographic mechanisms. Recent research between 2020 and 2023 shows a paradigm shift from traditional trust-based and rule-based approaches to intelligent models such as machine learning, deep learning, and graph-based learning. Graph Neural Networks (GNNs) effectively model MANET topology and detect anomalous nodes, while SNGNN enhances performance by incorporating similarity-based embeddings. Lightweight cryptography ensures secure communication with minimal computational overhead, addressing the constraints of resource-limited nodes. Furthermore, hybrid frameworks combining GNN, optimization, and cryptographic techniques demonstrate detection accuracy above 95% with improved efficiency. This study analyzes these approaches, provides comparative insights, and identifies research gaps for future development of scalable, energy-efficient, and secure MANET systems.