Recent Advances in Optimized Sparse Spatial Self-Nested Graph Neural Networks for Secure MU-MIMO-OFDM Systems: Channel Estimation, Attack Detection and Mitigation – A Systematic Review

Abstract

The rapid evolution of sixth-generation (6G) wireless communication
systems has intensified the need for efficient channel estimation, secure
transmission, and intelligent resource allocation in multi-user multiple
input multiple-output orthogonal frequency division multiplexing (MU
MIMO-OFDM) systems. Traditional estimation and security mechanisms
struggle with high-dimensional data, dynamic channel conditions, and
sophisticated cyber-attacks. Recently, graph neural networks (GNNs),
particularly optimized sparse spatial and self-nested architectures, have
emerged as promising solutions due to their ability to model complex
relationships in wireless networks. This paper presents a systematic
review of recent advances (2020–2023) in optimized sparse spatial GNN
frameworks for MU-MIMO-OFDM systems, focusing on channel
estimation, attack detection, and mitigation. The review highlights deep
learning-based approaches including CNNs, RNNs, attention
mechanisms, and deep unfolding models integrated with graph
structures. Additionally, lightweight cryptographic and AI-driven
intrusion detection techniques are examined for securing
communication against adversarial attacks. A comparative analysis of 30
studies is provided, emphasizing performance metrics such as
estimation accuracy, spectral efficiency, computational complexity, and
robustness. Finally, research challenges and future directions are
discussed, including scalability, real-time deployment, and integration
with 6G technologies. The findings demonstrate that optimized GNN
based frameworks significantly enhance system reliability, security, and
efficiency in next-generation wireless networks.