Denoising Method for Wireless Communication Signals Based on Convolutional AutoEncoder

Denoising Method for Wireless Communication Signals Based on Convolutional AutoEncoder

연구 분야: Infrastructure

학회: 2025 International Conference on Artificial Intelligence in Information and Communication (ICAIIC)

In this paper, we propose a method to effectively remove noise from signals by utilizing convolutional autoen-coder. In particular, the proposed method focuses on converting sequence signals into images and removing noise through image-based signal processing methods to improve signal quality while preserving key signal characteristics. Experiments were conducted using 5G demodulation reference signal data simulating real-world wireless communication environments, and a test bed was built based on two rdio universal software peripherals to obtain reliable results. Experimental results show that the proposed method achieves an average classification accuracy improvement of 32.6% and a maximum of 47.9%, and also performs well in low signal-to-noise ratio environments, especially in the -2.81dB signal-to-noise ratio environment, where it achieves a significant accuracy improvement over the model without denoising. This work demonstrates the potential to effectively improve the quality of wireless signals in various noisy environments and suggests its applicability to real-world communication systems.