Optimized MIMO-OFDM Channel Estimation Using Convolutional Neural Networks

Multiple Input Multiple Output-Orthogonal Frequency Division Multiplexing (MIMO-OFDM) has become a fundamental transmission technique for modern wireless communication systems due to its high spectral efficiency and robustness against multipath fading. Accurate channel estimation is essential for reliable data transmission in MIMO-OFDM systems. Traditional estimation methods such as Least Squares (LS) and Minimum Mean Square Error (MMSE) suffer from performance degradation under low Signal-to-Noise Ratio (SNR) conditions and rapidly varying channels. Recent advancements in deep learning have demonstrated promising capabilities in wireless signal processing tasks. This paper proposes an optimized Convolutional Neural Network (CNN)-based channel estimation framework for MIMO-OFDM systems. The proposed model learns channel characteristics directly from received pilot signals and estimates channel coefficients with improved accuracy. Simulation results demonstrate that the CNN-based estimator significantly outperforms conventional LS and MMSE estimators in terms of Mean Square Error (MSE) and Bit Error Rate (BER). The proposed approach provides enhanced robustness in dynamic wireless environments and is suitable for future 5G and 6G communication systems.

Cited via DOI — this article is permanently archived on Zenodo.