Generalising Location-Centric Variations to Enhance Contactless Human Activity Recognition

FAWAD KHAN^1*Syed Yaseen Shah²Jawad Ahmad³Alanoud Al Mazroa⁴Adnan Zahid⁵Muhammed Ilyas⁶Qammer H. Abbasi⁷Syed Aziz Shah¹

• ¹Coventry University, Coventry, United Kingdom
• ²Glasgow Caledonian University, Glasgow, Scotland, United Kingdom
• ³Prince Mohammad bin Fahd University, Khobar, Saudi Arabia
• ⁴Princess Nourah bint Abdulrahman University, Riyadh, Saudi Arabia
• ⁵Heriot-Watt University, Edinburgh, Scotland, United Kingdom
• ⁶Al Ain University, Al Ain, Abu Dhabi, United Arab Emirates
• ⁷University of Glasgow, Glasgow, Scotland, United Kingdom

Contactless Human Activity Recognition (HAR) has played a critical role in smart healthcare and elderly care homes to monitor patient behaviour, detect falls or abnormal activities in real time.The effectiveness of non-invasive HAR is often hindered by location-centric variations in Channel State Information (CSI). These variations limit the ability of HAR models to generalise across new unseen cross-domain environments, for instance a model trained in one location might not perform well in another physical location. To address this challenge, in this paper, we present a novel Federated Learning (FL) algorithm designed to train a robust global model from local datasets in different localisations. The proposed Federated Weighted Averaging for HAR (Fed-WAHAR) algorithm mitigates location-induced disparities, including heterogeneity and non-Independent and Identically Distributed (non-IID) data distributions. Fed-WAHAR employs a dynamic weighting approach based on local models' accuracy to improve global model classification accuracy and reduce convergence time effectively. We evaluated the performance of Fed-WAHAR using various metrics, including accuracy, precision, recall, F1 score, confusion matrix, and convergence analysis. Experimental results demonstrate that Fed-WAHAR achieves an accuracy of 85% in recognising human activities across different locations, enhancing the ability of model to infer across new unseen locations.