Wireless Fall Detection: An Integrative Investigation of Deep Learning Innovations, Challenges and Future Directions

The fall-detection equipment is critical for addressing accidental falls among the aging population, which has a huge health effect. The traditional wearable devices and monitors were not widely developed due to limitations in convenience and privacy. Therefore, non-invasive detection equipment leveraging Wi-Fi Channel State Information has become increasingly popular as a promising alternative to protect older adults efficiently. However, the practical deployment proved to be more challenging, including the generalization problem across different environments, users, and actions. This review provides an integrative analysis of wireless falling detection methods leveraging deep-learning models, information modelling, and cross-modal learning to study their core mechanisms in signal processing and data augmentation. Moreover, for challenges such as studying similar behaviors, data scarcity, and computational processes, this work also provides future study directions like lightweight architecture and cross-modal systems. This review aims to offer critical insights to support the development of more robust fall detection systems.