An approach to recognize and classify arm activities using wearable ultra wide band antennas

Abstract

Continuous observation, recognition and classification of various body movements and activities is essential for the implementation of Wireless Body Area Networks (WBAN) to discern the status of body parts functionalities or abnormalities if any. WBAN can continuously observe various movements of human body parts and human body activities. Recognition and classification of arm activities plays an important role in fitness monitoring, assisted living, and sports tracking, etc. In the presented work, Ultra Wide Band antennas are designed and employed on human body to observe datasets of antenna performance parameters associated with various arm movements and activities. The classification of three arm activities i.e. boxing, rowing, and clapping are implemented using Support Vector Machine, K-Nearest Neighbor, Random Forest and Decision Tree machine learning algorithms. Performance of classification depends on accuracy of implemented algorithm. The highest classification accuracies are found to be 99% in case of Decision Tree algorithm.