Indoor Next Location Prediction with Wi-Fi Boon-Khai Ang Institute of Systems Science National University of Singapore 25 Heng Mui Keng Terrace, 119615 Singapore A0092597@nus.edu.sg Daniel Dahlmeier SAP Research and Innovation CREATE Tower 1 Create Way, 138602 Singapore d.dahlmeier@sap.com Ziheng Lin Singapore Press Holdings 1000 Toa Payoh North, 318994 Singapore linziheng@gmail.com Jian Huang Institute of Systems Science National University of Singapore A0092599@nus.edu.sg Mun-Lie Seeto Institute of Systems Science National University of Singapore A0092671@nus.edu.sg Hendy Shi Institute of Systems Science National University of Singapore A0092709@nus.edu.sg ABSTRACT Indoor Location Intelligence is a novel application that relates indoor localization technology to business data to allow for better decision making for retail businesses. In this context, Wi-Fi technology has a big potential for localization of customers who move through the store. With this information, retailers are able to analyze shopper movement behavior when formalizing their business strategies. This paper evaluates the accuracy of next location prediction based on a Markov-chain model for forecasting the next location of a customer in a shop based on the last n locations he has visited. We report experiments on a real data set and achieve prediction accuracies of up to 37% for n=1 and 49% for n=2. KEYWORDS Indoor Location Intelligence, Location Prediction Model, Wi-Fi, MAC Address, Markov Chain, Mobility Behaviors, Transition Traces 1 INTRODUCTION People are increasingly accessing information via mobile devices, such as tablets and smartphones. As part of this trend, customers’ shopping experience is no longer limited to physical in-store shopping − they wish to search for product information and reviews as well as compare prices anytime and anywhere. Therefore, many retailers have facilitated in-store free Wi-Fi services to attract and retain customers. The giant supermarket chain Tesco, for example, rolled out free Wi-Fi services in hundreds of its stores in the UK [1]. In addition, many U.S. retailers such as American Apparel, Nordstrom and Family Dollar have started indoor location intelligence programs using various technologies, such as Bluetooth radio signal, radio-frequency identification (RFID), video cameras or sound waves, which allow them to track shoppers’ foot-traffic and physical movements [2]. However, the most extensive technique used is to capture Wi-Fi signals emitted by shoppers’ smartphones within a few meters, and collect the MAC address while customer moves through the store. Chan and Baciu [3] described a number of Wi-Fi indoor positioning systems in detail. In addition to the benefit of free Wi-Fi access, Wi-Fi based client-side positioning system also provides guides to shoppers to navigate in shopping malls, airports, museums, etc. For example, Google provides an indoor Google Map service to partner with property owners to build indoor 3D maps to benefit Android-phone users [4]. With Wi-Fi indoor positioning, property owners, for example owners of a large shopping mall, are able to analyze the footfall of visitors and to ISBN: 978-0-9891305-4-7 ©2014 SDIWC 107