DISC: A Novel Distributed On-Demand Clustering Protocol for Internet of Multimedia Things

Internet of Multimedia Things (IoMT) are receiving significant attention due to a wide variety of applications, e.g., wildlife habitat monitoring, but they are often highly resource constrained. Compared to Internet of Things, preserving battery power of nodes, and maximizing the lifespan of IoMT are more critical and challenging as sensed data are mostly image/video instead of simple scalar. Recent studies have shown that clustering is an efficient solution to reduce energy consumption. In clusters, the role of each node changes to reduce energy consumption, thereby, prolonging lifespan. In this paper, we address the lifespan maximization problem in IoMT by designing a clustering protocol where clusters are formed dynamically. Specifically, we analyze and solve an optimization problem aiming to maximize the lifespan by reducing the energy consumption among cluster heads. Based on the analysis, we propose a novel DIStributed on-demand Clustering (DISC) protocol. Our cluster head election procedure is not periodic, but adaptive, based on the dynamism of the occurrence of events. This on-demand execution of DISC aims to significantly reduce computation and message overheads. We validate the performance of DISC through extensive experiments. Experimental results show that DISC is 25% more energy balanced and achieves 32% more lifespan as compared to two state-of-the-art solutions.