Enchancing RFID data quality and reliability using approximate filtering techniques

• Main Author: Kamaludin, Hazalila
• Format: Thesis
• Language: English; English; English
• Published: 2018
• Subjects: TK5101-6720 Telecommunication. Including telegraphy, telephone, radio, radar, television
• Online Access: http://eprints.uthm.edu.my/190/1/24p%20HAZALILA%20KAMALUDIN.pdf; http://eprints.uthm.edu.my/190/2/HAZALILA%20KAMALUDIN%20COPYRIGHT%20DECLARATION.pdf; http://eprints.uthm.edu.my/190/3/HAZALILA%20KAMALUDIN%20WATERMARK.pdf; http://eprints.uthm.edu.my/190/

Radio Frequency Identification (RFID) is an emerging auto-identification technology that uses radio waves to identify and track physical objects without the line of sight. While delivering significant improvements in various aspects, such as, stock management and inventory accuracy, there are serious data management issues that affect RFID data quality in preparing reliable solutions. The raw read rate in real world RFID deployments is often in the 60-70% range and naturally unreliable because of redundant and false readings. The redundant readings result in unnecessary storage and affect the efficiency of data processing. Furthermore, false readings that focused on false positive readings generated by cloned tag could be mistakenly considered as valid and affects the final results and decisions. Therefore, two approaches to enhance the RFID data quality and reliability were proposed. A redundant reading filtering approach based on modified Bloom Filter is presented as the existing Bloom Filter based approaches are quite intricate. Meanwhile, even though tag cloning has been identified as one of the serious RFID security issue, it only received little attention in the literature. Therefore we developed a lightweight anti-cloning approach based on modified Count- Min sketch vector and tag reading frequency from e-pedigree in observing identical Electronic Product Code (EPC) of the low cost tag in local site and distributed region in supply chain. Experimental results showed, that the first proposed approach, Duplicate Filtering Hash (DFH) achieved the lowest false positive rate of 0.06% and the highest true positive rate of 89.94% as compared to other baseline approaches. DFH is 71.1% faster than d-Left Time Bloom Filter (DLTBF) while reducing amount of hashing and achieved 100% true negative rate. The second proposed approach, Managing Counterfeit Hash (MCH) performs fastest and 25.7% faster than baseline protocol (BASE) and achieved 99% detection accuracy while DeClone 64% and BASE 77%. Thus, this study successfully proposed approaches that can enhance the RFID data quality and reliability.