MQTT-Driven Node Discovery for Integrated IoT-Fog Settings Revisited: The Impact of Advertiser Dynamicity

The recent advances in telecommunications, wide availability of powerful and always-connected smart devices, and the wide adoption of cloud computing services are paving the way towards the concept of fog computing. An outstanding problem in this area is the availability of effective application-layer protocols between service seekers and providers at the Internet of Things (IoT) level because most IoT nodes run on batteries. Hence, IoT-Fog discovery service must be energy efficient to prolong the lifetime of smart objects. This paper presents a thorough study of our previously proposed MQTT-driven IoT-fog integration, namely the Power Efficient Node Discovery (PEND), and investigates the impact of dynamic arrival patterns on its performance. The MQTT broker serves as a fog node to trigger turning on/off the Bluetooth Low Energy (BLE) interfaces of the surrounding objects by monitoring their trajectories. Furthermore it leverages this additional location awareness to significantly reduce the power consumption of the device discovery process for mobile devices. With this motivation, we present a detailed performance study of PEND under various settings and provide in-depth discussions of some lessons about the obtainable energy saving and the effectiveness of the our enhanced BLE device discovery solution. The results we present are valuable for the fog community to design new optimizations and to refine the whole IoT device discovery process to the purpose of better efficiency and scalability.