A methodological approach to optimizing RFID deployment Benny C. F. Cheung & S. L. Ting & Albert H. C. Tsang & W. B. Lee Published online: 31 October 2012 # Springer Science+Business Media New York 2012 Abstract The interest of adopting RFID continues to grow in many industries, ranging from supply chain automation to healthcare management. However, dynamics of the operat- ing environment is one of the major challenges that impede RFID deployment. Even though numerous researchers focus on controlled laboratory experiments to enhance the success of deployment, it is found that system performance in the actual production environment may differ significantly from that conducted in a controlled laboratory, resulting in poor deployment result. To cope with this situation, this paper proposes an RFID Deployment Optimizer (RFIDDO), which is a generic methodology for optimizing the RFID configura- tion to provide objective, quantifiable data about the data capture performance of RFID readers for comparing and optimizing RFID applications in a scientific manner. A case study has also been conducted in a logistics company to demonstrate the implementation of RFIDDO and provide contextual details to help other firms in coping with the environmental dynamics in the journey of RFID deployment. Keywords Radio-frequency identification (RFID) . Supply chain automation . Performance measurement . Deployment optimization 1 Introduction Radio Frequency Identification (RFID) technology provides quality data and enhance enterprise collaboration by auto- matically acquiring product information (Angeles 2007; Lekakos 2007). In supporting various types of business automation which build upon RFID technology, system performance in providing accurate and reliable data is fun- damental in gaining user acceptance (Wu et al. 2006). For instance, the RFID data enables users to identify genuine products intelligently (Kwok et al. 2008) and even monitor animal behaviors ubiquitously (Ting et al. 2007). With the increasing importance of real time information provided by the RFID technology, reliability requirements for RFID systems deployment have become more critical. In particu- lar, high readability performance is the primary factor that guarantees the success of system deployment. However, numerous studies reported that RFID pilot tests were plagued by the technology’ s inability to get 100 % read rates (Gilbert 2004; Heese 2007; Çakıcı et al. 2011). For example, Wal-Mart’ s pilot test achieved 95 % accuracy on conveyors in distribution centers, while only 66 % accuracy in reading RFID tags on individual cases on fully loaded pallets (IDTechEx 2005). To improve the situation, Bardaki et al. (2012) argued that re-designing the tag antenna solu- tion and adjusting it to the environment is one of the sol- utions to ensure 100 % readability. Metro’ s pilot RFID read rate on its distribution center in Germany only considered pallet tagging, and excluded case or carton tagging from their scope, to ensure success (Everett 2005). Furthermore, Bai et al. (2006) propose means to filter and clean data streams from RFID applications that contain false (e.g., false positive, false negative) readings and duplicates in order to improve the RFID tag identification accuracy. Darcy et al. (2011) discover that the Non-Monotonic Reasoning classi- fier obtains the highest cleaning rate when handling false- positive RFID readings. One of the major factors causing the low reading rate is the invisible radio frequent waves; thus RFID practitioners find it difficult to adjust and fine tune the RFID settings (especially the reading range of tag and reader). In particular, users of B. C. F. Cheung : S. L. Ting (*) : A. H. C. Tsang : W. B. Lee Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong e-mail: jacky.ting@connect.polyu.hk Inf Syst Front (2014) 16:923–937 DOI 10.1007/s10796-012-9391-8