Optimization of bluetooth low energy-based indoor localization using smart systems

Thesis Type: Postgraduate

Institution Of The Thesis: Yildiz Technical University, Graduate School Of Natural And Applied Sciences, Turkey

Approval Date: 2022

Thesis Language: English


Supervisor: Tülay Yıldırım


Bluetooth is one of the several technologies to cater to indoor localization. It has the lowest power consumption and good accuracy performance. In the world of IoT, data from sensors and software help in giving meaning to physical objects connected to the internet.
In this thesis, data gathered using Bluetooth Low-Energy sensors is used in predicting an agent's location in an indoor environment. A Bluetooth-based model that is divided into two parts is proposed: a Convolutional Neural Network that trains on data transformed into images and ideas from Game Theory that uses the Markov Decision Process to determine the exact location of the agent. The data to image transformation uses the Image Generator for Tabular Data algorithm, which considers the Euclidean distances between the access points in creating the images.
The results show that the Convolutional Neural Network trains well on  transformed images and offers a solid approach to determining every beacon used for Bluetooth-based indoor localization. After a beacon is found, Markov Decision Process finds the optimal policy to locate the access point under which the agent lies.