Human activity classification via Doppler shift estimation from 60 GHz OFDM transmission

Human presence detection along with classifi-cation of activities and gestures are important features in emergent security systems, smart home appliances, and home entertainment gadgets. This paper focuses on person activity classification through estimation of Doppler shift in the 60 GHz millimeter wave band radio communication channel. The OFDM symbol transmission with known pilot sub carriers is considered, the received signal is post-processed by a large fast Fourier transform spanning for thousands of OFDM symbols to increase the frequency resolution. Typical Doppler frequency shift patterns for human activities such as gait, run and in-air hand-waving along and towards the receiver are presented. Simple features are chosen and their ability to distinguish among the activities is demonstrated via k-means clustering and KNN classifier.