PostureTag: RFID-Enabled Human Sitting Posture Recognition Using a Tag-Embedded Chair

This paper presents a cost-effective wireless approach for identifying human postures using RFID technology. The system uses UHF RFID tags, some embedded in a chair and others integrated inside footwear, to collect information about sitting postures. The novelty lies in the detection procedure which is a simple and straightforward algorithm that calculates the root mean square error (RMSE) between any test data set and a reference one, encoded by the states of the tags. These states are represented by the normalized counting of tag electronic product code (EPC) by a reader within a specific time duration. The algorithm is made so that of all postures under test, the one having the closest data pattern (i.e. minimum error) to the reference will be selected as the correct one. The required quantity and placement of the tags are optimized by looking at a trade-off between the accuracy and complexity of the system. The proposed method in this stage can reliably differentiate between six primary sitting postures and an empty chair with up to 94-100% accuracy. It also ensures correct identification even if the user shifts position or rotates relative to the reader.