Design of an Effective 3-Dimension RFID Location Identification Methodology and Its Applications
| Autor: | Sumalee Chaisit, 蘇瑪琍 |
|---|---|
| Rok vydání: | 2014 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 102 Radio Frequency Identification (RFID) technology has a high potential to improve performance in many industrial practices. LANDMARC has been well known as indoor positioning technology using different power levels of readers and fixed reference tags placed. This approach uses Radio Signal Strength Indicator (RSSI) applied to estimate the location of the objects. The challenges of this scheme extensively encourage many researchers in exploring the positional estimation of indoor positioning systems. As a result, many methods have been improved through an application of LANDMARC based. However, the indoor positioning system reveals that several factors affect location identification accuracy. Since the accuracy is influenced by tag distance, tag orientation, tag position of the objects, the power level of transmission, dense of reference tag, and its related environment. Furthermore, the localization system design is another one factor that affect on the performance of its estimation algorithm and applying for the real implementation system. Therefore, this study aims to enhance the localization system accuracy based on the integration of Landmarc algorithm, statistical filtering techniques, and artificial neural network. Including, developing the three dimensional RFID localization system to be deployed in the agricultural application in a real practical site. The technique of probability density function is applied to filter fluctuate RSSI data, followed by the enhancing of the system with the artificial neural networks (ANN), back propagation network (BPN) is used to optimize the estimated location in the post process. This proposed technique is investigated to challenge the effectiveness of 3D RFID location identification methodology and applications, that face with localization identification technology and a heterogeneous technology relying on dynamic environments. Dog house of Working Dog Training school, NPUST is a selected agricultural site to conduct the experiment on the RFID application deployment. This application was used to locate dogs when they are coming out from the safe zones as their cages. It is found that the system was able to capture and locate some dogs at the time around 15.00PM to 17.30PM which is dog’s scheduled activities, and during night time at 0.01-3.00AM. Particularly, there are two cats living in the same house at different zone. Thus, this 3D RFID application is suitable for supporting a kennel monitoring system of the school. Consequently, the results reveal that the statistical filtering techniques are minimized error estimation through location estimation algorithm. However, the BPN results indicate the technical computing algorithm in the key training data set that is insufficient from its required configuration caused by the environmental conditions. This BPN method is supposed to optimize the localization system in the post-process after Landmarc scheme even if the configuration requirements are sufficiently provided. In summary, this study implies that the proposed methods are the optional systems to be a selection location method in an agricultural deployment, based on the conditions of its environments, and how its deployments as theirs acceptable on suitability requirement. In addition, the application model in agriculture requires flexibility on adjusted configuration to be practically run on the optional function as possible relied on its environmental areas. As well as, this 3D RFID passive application model also meets the economic cost on a system implementation and maintenance. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|