Optimal wakeups clustering for highly-efficient operation of WSNs periodic applications

In this paper we present an efficient algorithm for clustering the system wakeups in Wireless Sensor Networks (WSNs), running periodic tasks from multiple heterogeneous sources. The clustering algorithm breaks the overall problem space of a system running periodic tasks in self-consistent sub problems, referred to as clusters, on which the optimal sequence of system wakeups can be found without loss of global optimality. Once a cluster is found through the presented algorithm, it is fed to the subsequent optimization phases for determining the optimal sequence of system wakeups. Experimental results show that the algorithm effectively reduces the problem sizes, strongly reducing the complexity of the subsequent task merging optimization phases.