Towards More Efficient Cardinality Estimation for Large-Scale RFID Systems

Abstract

Radio frequency identification (RFID) cardinality estimation with an accuracy guarantee is of practical importance in various large-scale RFID applications. This paper proposes a fast RFID cardinality estimation protocol, named Zero-One Estimator (ZOE). ZOE only requires 1-bit response from the RFID tags per estimation round. More importantly, ZOE rapidly converges to optimal parameter configurations and achieves higher estimation efficiency compared to existing protocols. ZOE guarantees arbitrary accuracy requirement without imposing heavy computation and memory overhead at RFID tags except the routine operations of C1G2 standard. ZOE also provides reliable cardinality estimation with unreliable channels due to the robust protocol design. We prototype ZOE using the USRP software defined radio and the Intel WISP tags. We extensively evaluate the performance of ZOE compared to existing protocols, which demonstrates encouraging results in terms of estimation accuracy, time efficiency, as well as robustness over a large range of tag population.