Date of Award
Master of Science (MS)
Computer and Information Technology
Committee Member 1
Thomas J. Hacker
User location can be leveraged to provide a wide range of services in a variety of indoor locations including retails stores, hospitals, airports, museums and libraries etc. The widescale proliferation of user devices such as smart phones and the interconnectivity among different entities, powered by Internet of Things (IoT), makes user device-based localization a viable approach to provide Location Based Services (LBS). Location based services can be broadly classified into 1) Proximity based services that provides services based on a rough estimate of users distance to any entity, and 2) Indoor localization that locates a user's exact location in the indoor environment rather than a rough estimate of the distance. The primary requirements of these services are higher energy efficiency, localization accuracy, wide reception range, low cost and availability. Technologies such as WiFi, Radio Frequency Identification (RFID) and Ultra Wideband (UWB) have been used to provide both indoor localization and proximity based services. Since these technologies are not primarily intended for LBS, they do not fulfill the aforementioned requirements. Bluetooth Low Energy (BLE) enabled beacons that use Apple's proprietary iBeacon protocol are mainly intended to provide proximity based services. iBeacons satisfy the energy efficiency, wide reception range and availability requirements of LBS. However, iBeacons are prone to noise due to their reliance on Received Signal Strength Indicator (RSSI), which drastically fluctuates in indoor environments due to interference from different obstructions. This limits its proximity detection accuracy.
In this thesis, we present an iBeacon based proximity and indoor localization system. We present our two server-based algorithms to improve the proximity detection accuracy by reducing the variation in the RSSI and using the RSSI-estimated distance, rather than the RSSI itself, for proximity classification. Our algorithms Server-side Running Average and Server-side Kalman Filter improves the proximity detection accuracy by 29% and 32% respectively in contrast to Apple's current approach of using moving average of RSSI values for proximity classification. We utilize a server-based approach because of the greater computing power of servers. Furthermore, server-based approach helps reduce the energy consumption of user device. We describe our cloud based architecture for iBeacon based proximity detection.
We also use iBeacons for indoor localization. iBeacons are not primarily intended for indoor localization as their reliance on RSSI makes them unsuitable for accurate indoor localization. To improve the localization accuracy, we use Bayesian filtering algorithms such as Particle Filter (PF), Kalman Filter (KF), and Extended Kalman Filter (EKF). We show that by cascading Kalman Filter and Extended Kalman Filter with Particle Filter, the indoor localization accuracy can be improved by 28% and 33.94% respectively when compared with only using PF. The PF, KFPF and PFEKF algorithm on the server side have average localization error of 1.441 meters, 1.0351 meters and 0.9519 meters respectively.
Faheem, Faheem, "Ibeacon based proximity and indoor localization system" (2016). Open Access Theses. 767.