CLEANING AND ANALYSIS OF RFID DATA

Abstract

RFID (Radio Frequency Identification) is an Automatic Identification and Data Capture (AIDC) technology that uses radio frequency waves to read RFID tags by RFID readers. These tags provide unique identification to the items to be tracked. If we consider the RFID deployments in real world scenarios, read-rate has been observed to be around 60% to 70% only. Thus data captured by the RFID readers is * noisy and erroneous. It needs to be cleaned because noise in data leads to misleading results in data analytics. Efficient cleaning algorithms are available which handle the noises like false negatives (missed readings arising when tag is not detected),false positives (arising when captured tag moves out of range of reader and assumed to be still present). These noises can be significantly minimized by adaptive window based algorithm SMURF (Statistical Smoothing of RFID Data). in the present work, an improvement over the basic SMURF algorithm has been proposed. It improves the performance in cleaning of raw RFID Data which has been verified through experimental results. Then, patterns detected in RFID Data have been discussed which provide significant information about tag movements.