AFR: Accurate and fast RFID estimation

Abstract

A hot and primary problem in RFID system is the estimation of RFID tag cardinality and many algorithms have been proposed for estimating RFID cardinality. However, we find that there is a serious problem that has been widely neglected by some of previous works of RFID cardinality estimation. They incorrectly assume that states of different time slots in the same frame are independent of each other. The consequence of such incorrect assumption would lead to low estimation accuracy. To demonstrate how we correct this mistake, we propose an accurate and fast estimating algorithm called AFR based on the pattern of the number of consecutive empty slots before the first non-empty slot. We theoretically derive the expected number and the variance of consecutive empty slots before the first non-empty slot as a function of the number of tags. Based on such theoretical derivation, we determine the best estimation of the number of tags by only looking at the numbers of consecutive empty slots before the first non-empty slot. Simulation results demonstrate that our algorithm makes more accurate estimation of RFID tag cardinality than an existing state-of-the-art algorithm while using shorter estimation time.