基於擴展卡爾曼濾波器之RFID系統標籤數量估測

Abstract

無線射頻技術(Radio Frequency Identification, RFID)如今已被廣泛運用於許多場合，具體應用包含了鈔票防偽、電子收費系統、悠遊卡、動物識別追蹤、倉庫的貨品盤點等。然而在RFID系統的標籤辨識工作進行時，仍有一項常被拿來討論的問題---如何減少標籤之間的碰撞。因此有許多防碰撞演算法被提出，可概分為Tree-based演算法與ALOHA-based演算法，但前者會造成辨識效率不佳，因此現今主要以ALOHA-based演算法之中的動態訊框時隙ALOHA(Dynamic Framed Slotted ALOHA, DFSA)演算法為最普遍被使用的防碰撞演算法。 為了讓防碰撞演算法能夠有效發揮其功用，必須還要搭配精準的標籤數量估測法，才能夠真正有效減少碰撞的發生並提升系統的效能。但現今用來作標籤數量估測的方法都仍有改善的空間。因此本篇論文提出了一種基於擴展卡爾曼濾波器(Extended Kalman Filter, EKF)來進行標籤數量估測的演算法，希望藉由其可以達到估測結果之均方差最小值的特性，來提升估測的精準度。並且我們也結合了下界法來對某些偏差的估測值做檢查及修正，以達到更進一步改善估測結果的目的。最後我們也透過實驗證明了本篇論文所提出的演算法在標籤數量估測上確實擁有相當高的精準度。

Radio Frequency Identification (RFID) systems are now widely used in many occasions, such as banknote security, Electronic Toll Collection (ETC) systems, Easy Card, animal tracking, tracking of goods, among many others. Generally, the so-called tag collision problem is still a significant issue in RFID tag identification systems. In order to solve this problem, many anti-collision algorithms have been proposed that can generally be divided into two categories, namely, Tree-based algorithms and ALOHA-based algorithms. Since the former has low performance in identification, advanced ALOHA-based algorithms with a Dynamic Framed Slotted design have been widely used in RFID anti-collision nowadays. Although algorithms based on Dynamic Framed Slotted ALOHA (DFSA) work effectively as far as anti-collision is concerned, the number of tags must be accurately estimated in advance to reduce the incidence of collisions and to promote system performance. Even in state-of-the-art tag estimation methods proposed in the literature, one can find that there is still room for improvement. In this thesis, an efficient tag estimation algorithm is proposed using an extended Kalman filter to compute the Minimum Mean Square Error (MMSE) as an accurate estimation. For further refinement, we also employ a lower bound method to check the produced estimation. Experimental results show that the proposed tag estimation scheme has a very high accuracy in the application to RFID systems.