Date Added: Nov 2010
The technology of Radio Frequency IDentification (RFID) is being deployed in many applications, such as logistics and inventorying. However, a generic problem in all RFID systems is to ensure that the RFID readers can reliably read a set of RFID tags, such that the probability of missing tags stays below some acceptable value. This paper introduces statistical methods to deal with the problem of missing RFID tags. These methods are applied at the reliability layer, which initiates multiple reader sessions by invoking certain MAC-layer arbitration (anti-collision) protocols to collect tag responses. The reliability layer obtains a running estimate of the probability of having at least one tag missing. This estimate is used to detect if an additional reader session is required.