Radar-Based Hand Gesture Recognition: Leveraging Raw ADC Data Through Independent Component Analysis and Spiking Neural Networks

In this paper, we introduce a novel low-power radar-based gesture sensing system specifically designed for portable devices with stringent energy efficiency requirements. Traditional approaches often rely on range-Doppler images and deep learning models, which can be computationally intensive and power-hungry. Our system directly processes raw ADC data from radar signals, employing independent component analysis (ICA) to extract essential features for gesture recognition. These features are utilized by a novel spiking neural network (SNN), which is inherently energy-efficient due to its sparse time encoding and event-driven operation. Experimental results demonstrate that our system achieves a remarkable accuracy of 99.98% while maintaining a compact model size, making it highly suitable for deployment in portable devices.