Comparison of Denoising Methods in Improving V2V/V2X Communication

Abstract

Vehicle-to-vehicle (V2V/V2X) communication is essential to our current transportation systems; it enables vehicles to exchange crucial data for better efficiency and safety. However, communication channels in these networks are susceptible to different forms of interference and noise, which causes a deterioration in signal quality and communication reliability. This paper compares different signal denoising techniques for V2V communication channels, focusing on four prominent methods: Fast Fourier Transform (FFT), Discrete Wavelet Transform (DWT), machine learning, and deep residual networks. We evaluate the denoising performance of each method using simulated signals corrupted by different noises and interference. Our experimental results demonstrate the effectiveness of each approach in mitigating noise and possibly improving communication reliability. Specifically, we observe that FFT and DWT offer efficient frequency and time-frequency domain representations for denoising signals. Traditional machine learning methods and residual networks (ResNets) demonstrate superior denoising performance. Our analysis provides insights into the strengths of each denoising technique, and the advantages one can have over the other. Overall, this study contributes to the advancement of signal processing techniques for improving the reliability of V2V communication systems in real-world scenarios.