Wearable Ultra Wideband Technology for Daily Activity Recognition

This paper presents daily activity recognition using wearable ultra-wideband technology. Channel parameters are analyzed for various postures occurring during the daily activity which act as key features to estimate the activity trend. Compact wearable antennas are placed on suitable locations on the human subject for each activity (walking, standing and sitting) studied in order to have maximum direct path propagation between the two wearable on-body nodes. It is observed that the three activities analyzed show distinct variation in the channel features making it possible to classify the activities through statistical analysis and inter-distance measurements between the wearable nodes. Low correlation is observed between the activity patterns with 0.01-0.3 correlation coefficient values. This indicates that the activities can be easily distinguished from each other using channel information. The work will be suitable for tracking, rehabilitation and activity monitoring applications in the healthcare domain.