Low-rank adaptation of deep learning-based receiver for cross-domain channel estimation

• Main Author: Wang, Yilong
• Other Authors: Teh Kah Chan
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2025
• Subjects: Engineering; Low-rank backbone; Deep learning; Channel estimation; OFDM
• Online Access: https://hdl.handle.net/10356/183554
• Institution: Nanyang Technological University

This project presents a low-rank adaptive method for deep learning based receivers used in OFDM systems for channel estimation and signal detection. While traditional DL-based receivers perform well under complex channel conditions, they often require a substantial amount of labeled training data. To mitigate this reliance on labeled data, the proposed method leverages the low-rank structure of the channel matrix, aiming to effectively bridge the source domain and the target domain. Specifically, this method enhances the receiver's performance by retraining the low-rank backbone of the DL based receiver to estimate missing entries in the target domain. By addressing the issue of domain transformation, this approach significantly reduces the dependence on labeled training data, making deep learning-based receivers more practical for deployment in real-world scenarios.