Algorithm And System Design For Home Activity Recognition Of The Elderly Based On WiFi Signals

With the acceleration of global aging trend,the quality of life and health of elderly people are receiving increasing attention.WiFi technology,as a wireless network technology widely used in various fields,can carry information that characterizes environmental changes during signal propagation.Therefore,research based on WiFi signal recognition behavior has gradually emerged.Using WiFi technology to recognize human behavior can effectively reduce the cost of daily care and provide a more convenient and comfortable living experience for the elderly.This article comprehensively considers the recognition accuracy and robustness of the algorithm,proposes a research on WiFi-based elderly home behavior recognition algorithm,and conducts systematic design to achieve recognition and fall warning for various daily behaviors.The specific research work is as follows:(1)Based on the Fresnel zone theory,a model of signal propagation in the wireless transmission field was established to determine the optimal position of the transmitter and receiver antennas,thus obtaining more reliable CSI data.To address the issue of the lack of a household behavior dataset,we built a data collection platform and collected 10 common daily behaviors of the elderly(such as drinking water and falling)in both laboratory and simulated home environments,totaling 16,800 experimental samples.We then used data preprocessing methods such as phase error correction,filtering,and PCA to remove the influence of random noise in the environment on the raw data,obtaining smooth and stable CSI data,and laying the foundation for subsequent feature extraction for behavior recognition.(2)To address the issues of low recognition rate and unstable performance in elderly household behavior recognition,this paper proposes a new behavior recognition algorithm,At-Bi LSTM.Firstly,a bidirectional structure is added to the traditional LSTM network,which can process continuous CSI signals in two directions and obtain features that can represent richer behavioral information.Then,an attention mechanism is introduced to assign more weights to features and time steps,enabling the model to make more accurate judgments and achieve better recognition results.Through experiments and comparisons with other related algorithms,the results show that the At-Bi LSTM algorithm has stable and efficient recognition performance,as well as robustness and anti-interference ability against external disturbances.(3)A fall detection and alarm system for the elderly was designed based on the At-Bi LSTM algorithm.Firstly,the overall architecture and software development environment of the system were analyzed,and then fall detection software and a user management system were developed based on the Python language.The functionalities of real-time monitoring,viewing historical records,and setting alarm methods for modules were tested.After testing,individual users can log in and use the fall detection software after successful registration,and the administrator can also browse,delete,add,and modify user data through the user management system.The system can accurately detect falling behavior and timely send alarm messages to emergency contacts.