Fundamental study on improving RFID system operational efficiency
Zheng, Yuan Qing
Date of Issue2014
School of Computer Engineering
Radio Frequency Identification (RFID) technology has recently attracted dramatic attentions from the research and industrial communities. A typical RFID system consists of RFID readers, RFID tags, and the middleware software to support proper working of the system. An RFID tag with its unique ID is a small microchip capable of harvesting energy from reader interrogation, lightweight computation, wireless communication, which transmits data in response to RFID reader queries. RFID tags are usually attached to real objects for explicitly labelling the objects and an RFID reader can thus identify and itemize these objects by verifying the attached RFID tags. Due to the simple structure, small size, and low manufacturing cost of RFID tags, it provides us an economic and competitive method for massive object management. RFID technology is currently becoming ubiquitously available for a variety of applications, including inventory management, transportation and logistics, object identification and tracking. Having been widely adopted across many applications, RFID technology is fast growing as a major element of Internet of Things (IoT) for building future pervasive computing environment. However, opposed to the stringent needs for system efficiency and availability, due to the RFID resource constraints, the working efficiency of some key operations in RFID systems is still severely restricted, concerning RFID identification, estimation, searching, etc. For example, currently an RFID reader has to sequentially interrogate the individual tags and identify these tags one after another. When there are massive tags in the area (e.g., during baggage classification), the operational efficiency could be excessively low. Due to the severe wireless collisions and channel contention backoffs, the operational efficiency could get even lower. This dissertation presents novel techniques and approaches to fundamentally improve the operational efficiency of the most basic and key operations in RFID systems. Such key operations are widely adopted across many RFID systems in practice, and most applications may benefit from the efficiency improvement. 1) We design PET and ZOE protocols to estimate the number of RFID tags in the coverage of RFID readers without identifying the tags. Such approximation protocols trade accuracy for execution time and substantially improve operation efficiency with guaranteed estimation accuracy. PET improves state-of-the-art estimation efficiency from O(log n) to O(loglog n) while ZOE further improves the efficiency to O(1). 2) We further design CATS, the first tag search protocol, to enable efficient tag search in large-scale RFID systems. CATS encodes the tag set using compact approximators to reduce communication overhead, and efficiently aggregates tag responses and extracts useful information without explicit channel arbitration. 3) We develop P-MTI, the first physical layer missing tag identification scheme, to efficiently monitor RFID tags and quickly identify the missing tags. P-MTI leverages the sparsity of missing tag events and recovers physical layer collisions through compressive sensing. To investigate the feasibility of our proposed designs and fundamentally improve our understanding of RFID systems, we build the open source platform, named Open RFID Lab, which allows the full access and control over the RFID systems. The platform unveils several fundamental problems which motivates future researches. For future work, we develop data transfer protocol to efficiently read bulk data using commodity RFID readers from RFID based sensors. Our design allows resource-constrained RFID sensors to meet the stringent response deadlines of commodity communication protocol only with software extensions.
DRNTU::Engineering::Computer science and engineering