Deep Learning-Based Phase Noise Mitigation in FMCW LiDAR Sensors for Precise Distance Measurement during Sweeping

Phase noise in Frequency Modulated Continuous Wave (FMCW) Light Detection and Ranging (LiDAR) systems degrades measurement accuracy by distorting the received signal during frequency sweeping. This leads to errors in distance estimation and reduces system performance, particularly in complex and dynamic environments. To address this challenge, this manuscript proposes a Binarized Simplicial Convolutional Neural Network (Bi-SCNN)-based framework for effective phase noise mitigation and accurate distance estimation in FMCW LiDAR systems. The objective is to enhance Frequency Modulated Continuous Wave (FMCW) Light Detection and Ranging (LiDAR) performance by compensating phase noise and improving distance accuracy, where signals are preprocessed using an Adaptive Two-Stage Unscented Kalman Filter (ATSUKF), followed by Bi-SCNN for phase distortion prediction, and optimized using Electric Eel Foraging Optimization (EEFO) for accurate compensation and distance estimation. The proposed method is implemented in Python, and its performance is evaluated utilizing Accuracy, Root Mean Squared Error (RMSE), Mean Squared Error (MSE) and Signal-to-Noise Ratio (SNR). The Bi-SCNN achieves 97.5% accuracy, 0.02 RMSE, 0.023 MSE, 44 dB SNR, and 86 seconds computational time, outperforming existing methods such as Phase Noise Compensation (PNC), Iterative Learning Control (ILC), and Coherently Coupled Orbital Angular Momentum (CCOAM).