.png){kind=icon}
.png){kind=icon}
Return
Radar Complex Intermediate Frequency Signal Denosing Based on Convolutional Auto-Encoder Network
.png){kind=icon}
OA
AI
DOI：10.1109/ACCESS.2023.3309643.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In radar systems, target state features are commonly extracted from intermediate frequency signals. However, these signals often have a low signal-to-noise ratio due to noisy environments and limitations of the radar hardware. This can lead to a significant loss in performance during target state feature extraction. Therefore, improving the signal-to-noise ratio of intermediate frequency signals is crucial for the effective operation of radar systems. To solve this problem, we developed a deep learning-based method for denoising intermediate frequency signals in this paper. Our approach involves using an auto-encoder network to remove unstructured noise and recover the original signal. During the signal preprocessing stage, it is important to ensure that the phase of the complex signal remains undistorted and that differences in signal amplitudes do not negatively affect the denoising performance. To achieve this, the real and imaginary parts of the complex signal are separated and subjected to 0-1 normalization. The loss function of the denoising network is then established based on signal correlation. The numerical results demonstrate that the proposed method outperforms other denoising techniques in terms of mean square error and denoising performance.
Keywords:
Noise reduction
Signal to noise ratio
Radar
Convolutional neural networks
Decoding
Frequency-domain analysis
Training
Complex signal denosing
deep learning
intermediate frequency signals
convolutional denoising auto-encoder