Mr Gwen Maudet from the SRCD department and IRISA department, will present his research about :
"Exploiting Sensor Similarity to Enhance Data Collection in Massive IoT Networks"
The Internet of Things (IoT) are commonly employed for monitoring various physical quantities. In the innovative approach of Massive IoT (MIoT), a massive deployment of highly constrained sensors is considered to reduce deployment and maintenance costs. Aligned with this scenario, this thesis focuses on the development of mechanisms to reduce sensor energy consumption. The method relies on the principle of similarity: sensors can be considered similar if they provide similar observations. This approach enables the transmission of a subset of sensors to fulfill the monitoring requirements. First, we identified and synthesized existing methods from the literature based on the principle of similarity. We established that this approach can be decomposed into three components, which we studied in the context of MIoT. Next, we examined methods for managing sensor observations to maintain a constant stream of messages over time. Our first method involves transmitting a specified number of sensors in a round-robin fashion. The second method achieves precision results comparable to the first while reducing the number of sensor updates when the sensor fleet changes. Finally, we propose a solution to form groups of sensors identified as similar by analyzing their observations. To this end, we introduce a new similarity measure based on interpolation, coupled with a hierarchical clustering method.
Thesis acreditation from IMT Atlantique with the Doctoral School SPIN
Keywords: Massive Internet of Thing, IoT Efficiency, Observation Collection Management