RSSIFilter: Selective RFID Tag Reading via RSSI Thresholding

RFID systems face fundamental efficiency limitations in dense deployments as traditional protocols treat all tags equally, regardless of signal strength or proximity. We present RSSIFilter, the first system enabling selective RFID tag reading through physical-layer RSSI thresholding without protocol or hardware modifications. Our key insight exploits an under-investigated phenomenon: when tags collide, those with a suitable RSSI advantage remain decodable, allowing intentional collision scheduling to suppress non-target tags as noise. We derive theoretical and practical bounds for collision recovery and develop a C1G2-compliant mechanism that dynamically engineers productive collisions. We implement RSSIFilter on commercial RFID readers and tags. Experimental results show that RSSIFilter reduces inventory time by 4.31× compared to slotted ALOHA while maintaining 98.2% target tag recall. This work establishes a new paradigm for physical-layer aware RFID arbitration, bridging a gap between tag selection requirements and protocol capabilities.