Recognition of Human Activity Based on Compressed Sensing in Body Sensor Networks

Energy efficiency is a primary challenge in wireless body sensor networks for the long-term physical movement monitoring. In order to reduce the energy consumption while maintaining the sufficient classification accuracy of the human activity, a compressed classification approach is explored combining classification with data compressing based on sparse representation and compressed sensing. The proposed approach firstly compresses the sensing data by random projection on the sensor nodes, and then recognizes activities on compressed samples after transmitting to the central node by sparse representation, which can reduce the energy transmission of original data. The performance of the method is evaluated on the opened Wearable Action Recognition Database (WARD). Experimental results are validated that the compressed classifier achieves comparable recognition accuracy on the compressed sensing data.