engleski

Increasing throughput in RFID systems with large number of tags

Radio Frequency Identification (RFID) technology, based on the wireless communication between readers and tags became the most popular technology for indoor/outdoor localization and tacking. However, the disadvantages of unreliable tags detection puts significant constraints in its spectra of usage. In scenarios with a large number of tags, it may happen that some of tags passes as undetected due to Medium Access Control (MAC), Physical, or Application level constraints. To solve a part of described issues, this thesis presents solutions which can improve overall RFID quality of service. Thesis emphasize the research on Dynamic Frame Slotted ALOHA (DFSA) MAC and its throughput, i.e. number of read tags in the unit of time. To maximize DFSA throughput, size of its frame should equal the number of competing tags, which is unknown in typical real environment. Therefore, to accomplish fast identification it is necessary to estimate a number of tags which are in the reader interrogation area. The main disadvantage in the implementation of the state-of-art tag estimation algorithms includes the number of required computations, along with temporary storage of large numbers which appear prior estimate. As a consequence such algorithms are energy inefficient and may require specific computer architecture to support calculus. In this thesis we present the new tag estimate method, called Improved Linearized Combinatorial Model (ILCM) which, due to linear property of the estimator allows significant reduction in estimate computation with the acceptable tradeoff in accuracy. Simulation and measurement results shows significant benefit in choosing ILCM over other tag estimate methods. In an addition we present ways on how to improve tags detectability through the energy harvesting, or by employing Software Defined Radio (SDR) technology. Further, once tag data is acquired, it is important to present it users in the real time, with high reliability tag location. Therefore, we apply Markovian approach/Bayesian Belief Networks for tags location modelling, along with add-on to standard architecture that allows tag tracking in the real-time.

Radio Frequency Identification (RFID); Dynamic Frame Slotted ALOHA (DFSA); Tag Estimate Methods; Energy Efficient Computing; Energy Harvesting; Software Defined Radio (SDR); Markov Chain Modelling; Real-Time Protocols

nije evidentirano