Bluetooth Low Energy (BLE) beacons are one of the most common choices to realize IoT and smart city applications because of their scalability and affordability. Moreover, the proliferation of Bluetooth technology in our modern lifestyle has made the deployment of BLE beacon-based solutions easier than its competitors. However, BLE beacon devices suffer from a short battery lifetime, which induces additional maintenance costs and complex operations. Such limitations will ultimately hinder the growing momentum of beacon technology adoption. In this thesis, we investigate various methods to extend the lifetime of a beacon device from multiple angles to realize a self-sustainable beacon infrastructure that is less costly to maintain.

Firstly, we address the lifetime issue from a hardware pe...[ Read more ]

Bluetooth Low Energy (BLE) beacons are one of the most common choices to realize IoT and smart city applications because of their scalability and affordability. Moreover, the proliferation of Bluetooth technology in our modern lifestyle has made the deployment of BLE beacon-based solutions easier than its competitors. However, BLE beacon devices suffer from a short battery lifetime, which induces additional maintenance costs and complex operations. Such limitations will ultimately hinder the growing momentum of beacon technology adoption. In this thesis, we investigate various methods to extend the lifetime of a beacon device from multiple angles to realize a self-sustainable beacon infrastructure that is less costly to maintain.

Firstly, we address the lifetime issue from a hardware perspective by proposing an energy harvesting design for BLE beacons. To this end, we investigate and model the behaviors of solar panels, supercapacitors, and BLE beacons to propose a design methodology for selecting hardware component parameters. We also approach the problem from the firmware design perspective by proposing a novel user existence-aware BLE beacon firmware, User-B, that extends BLE beacon lifetime by changing its operating configuration. Furthermore, we present a novel machine learning architecture to predict user existence probability and leverage this information to extend the lifetime of the beacon device. Finally, noting the importance of sensing tasks in various IoT applications, we propose a novel framework that extends the lifetime of IoT devices by adaptively adjusting the sensing task interval based on the predicted chances of sensor data changes. To facilitate energy-efficient prediction, we also propose a novel neural network architecture that reduces the energy consumption of the neural network significantly. Our proposed framework is validated and proven effective through experiments with a real-life dataset.