Indoor Tracking with Bluetooth Low Energy Devices Using KNN Algorithm

Lie, Koon Kee (2020) Indoor Tracking with Bluetooth Low Energy Devices Using KNN Algorithm. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Lie Koon Kee.pdf Restricted to Registered users only Download (3MB)

Abstract

Global Positioning System (GPS) is a technology that had provided location-based service (LBS) such as mapping, positioning and navigation for a long time. Increased with the number of LBS applications available is the attention towards the possibility of Indoor Positioning System (IPS). However, despite the ability of GPS to work almost perfectly accurate and precise, the technology does not perform the same for indoor environment, showing a very high rate of errors and thus making it an unreliable approach for indoor positioning. Review of literatures shown that there are various kind of technologies suitable for the development of an IPS, including Wi-Fi, Bluetooth Low Energy (BLE), Infrared (IR) and Ultrawideband (UWB), which all the technologies are evaluated based on their complexity of implementation, accuracy, power requirement, cost and area of effectiveness. BLE had been chosen as the technology for this project as it satisfy the aim of the project, which is to create an IPS which required low power and low cost. The localization technique used in this project can be separated into two major parts, the first would be the utilization of KNN algorithm in estimating the region where the BLE beacon is located, and the second would be the trilateration based on the distance that were calculated from RSSI to estimate the exact coordinate of the BLE beacon. For instance, a total of 10 BLE scanners would be deployed in the testing area. The beacon will be transmitting signal at an interval of 100ms. When the scanner detected the signal sent by the beacon, it will send the message to the MQTT broker. The Python scripts will be used to obtain data from the MQTT broker to calculate the average RSSI value received by each scanner to find the 3 scanners with highest average RSSI value. The estimation of the region of the BLE beacon can be performed using the combination of the 3 scanners. This paper also described a number of tests that were performed to validate the accuracy and the reliability of the system. The results are presented in graphical form for easier interpretation. The layout of deployment of the scanners and the number of scanners to be used in the system to yield the highest accuracy will also be discussed. The absence of line of sight is also discussed as a factor that could greatly affects the accuracy of the system.