کنفرانس مهندسی برق ایران

صفحه اصلی / سی و سومین کنفرانس بین المللی مهندسی برق

Soft Decision Adaptive Deep Learning Detection for Enhanced Massive MIMO Performance

نویسندگان :

Farnaz Sedaghati¹ Mojtaba Amiri² Ali Olfat³

1- University of Tehran - Electrical and Computer Engineering 2- University of Tehran - Electrical and Computer Engineering 3- University of Tehran - Electrical and Computer Engineering

کلمات کلیدی :

Massive MIMO،Deep Learning،Signal Detection،Iterative Method،Loss Function،Activation Function

چکیده :

Massive MIMO is a key technology for next-generation wireless systems, offering enhanced spectral efficiency and reliability. However, its high-dimensional nature poses significant signal detection challenges. Conventional detection methods are often ineffective, and optimal schemes such as maximum likelihood (ML) become computationally infeasible. This paper presents a deep learning-based MIMO detection scheme that achieves superior performance while maintaining lower computational complexity. The proposed approach builds on iterative soft-thresholding algorithms, incorporating soft decision mapping to the nearest constellation and an adaptive loss function, and uses a novel training algorithm to enhance detection efficiency and robustness. The results demonstrate a significant improvement in detection performance, as the proposed approach outperforms both conventional techniques and previous state-of-the-art methods. This emphasizes the effectiveness and practicality of the proposed method.