Fast and reliable identification of Radio Frequency Indentification (RFID) tags by means of anticollision (MAC) protocols has been a problem of substantial interest for more than a decade. However, improvements in identification rate have been slow, as most solutions rely on sequential approaches that try to avoid collisions, which have limited margin for performance improvement. Recently, there has been growing interest in concurrent techniques that exploit the structure of collisions to recover tag IDs. While these techniques promise substantial improvements in speed, a key question that remains unaddressed is how to deal with noise or interference that might introduce errors in the recovery process at the reader. Our goal in this paper is to consider a noisy wireless channel and add robustness to concurrent RFID identification techniques. We propose a new protocol, called CIRF (Concurrent Identification of RFids), which uses multiple antennas to add robustness to noise and leverages block sparsity-based optimization to recover EPC IDs of transmitting tags. We include fail-safe methods to handle errors that persist after the optimization stage. Extensive simulations show that CIRF achieves substantial resilience improvement in a range of very low to medium Signal-to-Noise (SNR) situations, being able to always correctly recover 99% of tags.
Keywords: MAC protocols; RFID identification; noise resiliency; sparse recovery.