Dynamic-scale transformation for UHF RFID tag collision separation on physical layer

In a passive ultra-high frequency (UHF) radio-frequency identification (RFID) system, when multiple tags select a single time slot to backscatter their signals simultaneously to a reader, the tags will collide with each other. In this case, the slot is a useless slot and the collided tags will re-select other slots to backscatter their signals until no collision happens. However, an MAC-physical (PHY) cross-layer approach does not view the collision slot as a useless one2, where collided tags could be separated and identified. Therefore, the cross-layer approach can enhance the identification efficiency. This paper proposes a separated algorithm on a PHY layer, which firstly utilizes a dynamic-scale transformation to transform the collided signals to time-scale domains. And then, a transforming matrix would be obtained from the transformation. Finally, LS criterion and MMSE criterion are applied to the collided tag signal separation. Simulation results show that, in contrast to existing tag signal separation algorithms, the proposed algorithm can separate more than two collided tag signals under single-antenna condition. Also, the algorithm has better performance of the separation even without capture effect.