Frequency Matching in Spiking Neural Networks for mmWave Sensing

TL;DR

This paper demonstrates that spiking neural networks, leveraging their inherent low-pass filtering, can effectively process mmWave sensing data by matching their frequency response to the data's spectral content, leading to improved accuracy and energy efficiency.

Contribution

It introduces a frequency-matching criterion for configuring LIF neuron parameters in SNNs, validated across multiple datasets, enhancing robustness and efficiency in mmWave sensing.

Findings

SNNs outperform ANNs when data's discriminative info is in low-to-mid frequencies.

Proper frequency matching improves test accuracy by an average of 6.22%.

Energy consumption is reduced by 3.64 times compared to baseline ANNs.

Abstract

Millimeter-wave (mmWave) sensing enables privacy-preserving, always-on edge perception, but its measurements are often sparse, temporally irregular, and corrupted by high-frequency noise. Existing mmWave pipelines predominantly rely on artificial neural networks (ANNs), which achieve robustness through extensive preprocessing or deep architectures, thereby limiting their efficiency on edge devices. In this work, we study spiking neural networks (SNNs) for mmWave sensing from a mechanism-data alignment perspective. By leveraging the low-pass filtering behavior of leaky integrate-and-fire (LIF) dynamics, we analyze how their implicit temporal filtering interacts with the frequency structure of mmWave signals. Our analysis shows that when discriminative information resides in low-to-mid frequencies, LIF dynamics can inherently suppress high-frequency noise, clarifying when and why SNNs outperform ANNs. Based on this insight, we derive a principled criterion for configuring the membrane decay factor by matching the effective bandwidth of LIF dynamics to the data's discriminative spectral content. Experimental results across four widely used mmWave datasets validate the proposed frequency-matching hypothesis, yielding an average test-accuracy improvement of 6.22% and a 3.64$\times$ reduction in theoretical energy consumption relative to ANN baselines, under a unified evaluation protocol.