Approaches to human activity recognition via passive radar

TL;DR

Experimental results demonstrate SNN-based neurosymbolic models achieve high accuracy making them a promising alternative for HAR across various domains, and Spiking Neural Networks offer reduced power consumption, ideal for privacy-sensitive applications.

Abstract

The thesis explores novel methods for Human Activity Recognition (HAR) using passive radar with a focus on non-intrusive Wi-Fi Channel State Information (CSI) data. Traditional HAR approaches often use invasive sensors like cameras or wearables, raising privacy issues. This study leverages the non-intrusive nature of CSI, using Spiking Neural Networks (SNN) to interpret signal variations caused by human movements. These networks, integrated with symbolic reasoning frameworks such as DeepProbLog, enhance the adaptability and interpretability of HAR systems. SNNs offer reduced power consumption, ideal for privacy-sensitive applications. Experimental results demonstrate SNN-based neurosymbolic models achieve high accuracy making them a promising alternative for HAR across various domains.

Figures (12)

• Figure 1: Illustrative figure showing a snippet of data captured by one antenna while a person walks.
• Figure 2: Sample of the video dataset for two different activities: a) walking and b) waving both hands. The key points in every video frame help to discern the right side (highlighted with coloured dots) from the left side of the candidate. Image sourced from cominelliFusion2024 with the author’s permission.
• Figure 3: Figure showing the MSE over the frequencies between the mean of the original data and the downsampled data for all the actions
• Figure 4: Time windowed analysis for the Walk action shownig that also considering small windows the mean downsampling performs better.
• Figure 5: Decision tree form of the rules used to classify activities based on the movement characteristics of different limbs.