Energy Saving Data Abstraction and Reformation Algorithms for Human Movement Monitoring

Detecting human movement is an important issue in monitoring and studying human activities especially for elderly and patients. The availability of wireless sensor network eases the monitoring work. Generally, a wireless sensor node for movement detection is embedded with an accelerometer and powered by batteries. The sensor node needs to transmit the sensed data from accelerometer wirelessly to other nodes or directly to a base station. If there are more data to be transmitted, it consumes more energy and thus the batteries drain out more quickly. Thus, an energy saving scheme, called Data Abstraction and Reformation (DAR), is proposed in this paper to reduce data transmission. Through data abstraction, sensor nodes filter out insignificant sensed data but only report those significant to a base station, and with data reformation, the complete data will be reconstructed at the base station with only the received data. We show that with a good selection of data abstraction and data reformation criteria, a movement detecting sensor node will only need to report 10%_~30% of the sensed data in order to provide full human movement monitoring at a base station.