WiFi-Based Human Activity Recognition and Fall Detection with Taxonomy, Benchmarks, and Future Directions: A Narrative Review

Abstract

WiFi-based human activity recognition (HAR) and fall detection have emerged as promising alternatives to vision- and wearable-based systems for modern assisted living and clinical monitoring for the elderly. These approaches exploit variations in Channel State Information, capturing amplitude and phase fluctuations across subcarriers to characterize how human motion perturbs these patterns through multipath propagation. This narrative review synthesizes 26 peer-reviewed studies published between 2017 and 2025, focusing on systems evaluated in real-world, non-laboratory environments. While recent deep learning models demonstrate near-perfect benchmark accuracy on fine-grained datasets, cross-environment assessments consistently reveal a severe “reality gap.” Specifically, several reviewed studies reported accuracy declines exceeding 20% when deployed in previously unseen buildings, while multi-user interference reduced recognition accuracy by 21–46%. Furthermore, dataset availability and reproducibility are critical considerations; only a limited number of reviewed studies utilized publicly available datasets, and none provided a standardized benchmarking guide-line. To address these gaps, this review presents a comprehensive taxonomy of post-2017 architectures, a comparative synthesis of laboratory versus deployment performance, and a critical analysis of key implementation challenges. We conclude by proposing a set of recommended reporting practices and outlining actionable future research. In doing so, the paper outlines key recommendations for developing adaptive, location-independent models, emphasizing the need for domain-transfer techniques, cross-building validation, and expanding open datasets to enhance generalizability.

Received: 31 August 2025 | Revised: 13 March 2026 | Accepted: 25 June 2026

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.

Author Contribution Statement

Kok Chung Chua: Conceptualization, Methodology, Validation, Formal analysis, Writing – original draft, Writing – review & editing, Visualization. Kai Liang Lew: Resources, Writing – original draft, Writing – review & editing, Project administration. Chean Khim Toa: Conceptualization, Validation, Resources, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition. Matthew Alexander Paudianto: Investigation, Data Curation, Writing– original draft, Writing – review & editing. Ivan Nathanael: Investigation, Data curation, Writing – original draft, Writing – review & editing. Abel Nathanael Hutapea: Investigation, Data curation, Writing – original draft, Writing – review & editing.