More in general, fluctuations of backscattered power from sensor-less passive tags placed over the body could provide a coarse information about the subject’s state [21], permitting to distinguish if he’s standing or
moving and eventually to estimate the frequency of periodic movements. This experimental evidence has been engineered in [30] for the kinematics’ recognition of body segments/joints during walking, while gesture classification has demonstrated to be feasible in [31] through the application of state-of-the-art classification techniques borrowed from Machine Learning and Brain Computer Interface background. Fig. 8 shows the experimental
setup for RFID classification by using four tags on the arms and legs, and a typical backscattered pattern to be processed by support vector machine (SVM) algorithm [32] for data mining with possible applications to self-assisted rehabilitation of impaired people in their domestic environment.