Computing and Information Systems - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemProximity-based location sensing using RFID technology( 2014)This thesis investigates the use of RFID (Radio Frequency IDentification) technology for localization and tracking. RFID is an automatic identification technology that plays a key role in ubiquitous computing applications. It is a promising technology for the purpose of object localization and tracking due to its advantages such as reasonable price, contactless communications, high data rate and non line-of-sight readability. In their simplest form, RFID readers provide only presence information of the tags in their vicinity. Space partitioning utilizes this simple capability and provides a fundamental technique for a variety of localization and tracking applications. The underlying idea is to discretize a physical space into uniquely identifiable partitions. There are two ways of achieving space partitions. (I) An RIFD-enabled object can be sensed via different RFID readers, and each partition is uniquely determined by those readers that can detect the object. (II) Alternatively, the environment can be supplied with RFID tags, for example a carpet may be fitted with hundreds or thousands of RFID tags, and a position can be computed via a mobile RFID reader using the tags in reading range. Obtaining a deeper fundamental understanding of the space partitioning technique using RFID technology is the overall objective of this thesis. The results of this thesis are presented in three chapters. Reader-based space partitioning in RFID systems: This chapter includes a theoretical and computational investigation of reader arrangements to provide an optimal partitioning of a deployment area. The number of generated distinguishable partitions can be used as an approximate measure to determine the localization accuracy. We provide an upperbound on the number of distinguishable partitions a deployment area can be divided into, given a specific number of readers equipped with either omni-directional or directional ntennas. Omni-directional antennas can detect the presence of a tag from any direction within a specific distance; whereas, directional antennas have a limited range and can only determine the presence of a tag within a sector. Further, we propose arrangements of both directional and omni-directional antennas which create the number of distinguishable partitions asymptotically equivalent to the calculated upperbounds. Tag-based space partitioning in RFID systems: This chapter includes a theoretical and computational investigation of the k-coverage problem in tag-based partitioned spaces utilizing regular tag arrangements. K-coverage is a fundamental problem in tag-based partitioned spaces and provides many advantages including increased robustness with regard of tag failures and higher degree of resolution. It ensures that at least k tags can be read from any point within the deployment area. We investigate and compare the three regular tag arrangements – triangular, square and hexagonal tag arrangements – to provide k-coverage of an area for every k greater than one. Further, we show that the regular tag arrangements are preferable to uniform random tag arrangements to k-cover a deployment area in terms of tag density. Further, we compare different regular tag arrangements in terms of their robustness to tag placement errors and inclination of the deployment area. A real-time gesture recognition system employing RFID reader-based space partitioning technique: Because of the various error sources in RFID systems, reliable operation as the tags move in the environment is inherently difficult and presents a significant challenge. To verify and evaluate RFID localization in real application scenarios, we design and implement a real-time gesture recognition technique. Our proposed technique uses reader-based space partitioning to track passive RFID tag motions in practical situations. We use multiple hypothesis tracking and combined tags to overcome uncertainties inherent in RFID systems. The experiments show that the proposed real-time gesture recognition technique has an accuracy between 88% and 96%.