Akademik Veri Yönetim Sistemi
TELECOMMUNICATION SYSTEMS, cilt.89, sa.2, 2026 (SCI-Expanded, Scopus)
Wireless networks are no longer just sources of data; they have become complex, dynamic ecosystems. As the proliferation of networked devices, such as self-driving cars, has propelled the network into a post-modern era, our traditional methods for building and managing these networks face numerous challenges, including higher bandwidth demands, the need for increased reliability, and optimal resource utilization. Researchers and engineers believe that our networks have to manage themselves using machine learning. The rising complexity of next-generation wireless networks (e.g., 5G-Advanced, 6G, and underwater communication), characterized by ultra-dense deployments, massive connectivity, and diverse service requirements, has made traditional model-based optimization and management approaches less effective. Machine Learning (ML) has appeared as a transformative paradigm to tackle these challenges by leveraging data-driven insights to achieve unprecedented levels of autonomy, efficiency, and performance. This paper presents a comprehensive comparative evaluation of machine learning classifiers for symbol detection in Orthogonal Frequency Division Multiplexing OFDM systems under varying SNR and fading conditions. Additionally, the impact of targeted SNR enhancement is investigated as a supplementary performance analysis. The classifiers were trained using datasets containing up to 105 modulated symbols under controlled channel conditions, with offline training and SNR-dependent testing to ensure reproducibility.