Affiliations: [a] Group of Computer Networks, Software Engineering, and Systems (GREat), Federal University of Ceará, Fortaleza-CE, Brazil | [b] Virtual University Institute, Federal University of Ceará, Fortaleza-CE, Brazil | [c] Mathematics Institute, Federal University of Rio de Janeiro, RJ, Brazil
Abstract: Recently, semantic clustering has been proposed to save energy in wireless sensor networks. Semantic clustering organizes the topology in clusters composed of semantically correlated nodes whose leader (collector) is periodically elected. Since collectors’ energy depletion is faster than the inner cluster nodes, suitable election mechanisms are required to avoid the energy hole problem. A potential drawback in adopting traditional election mechanisms is their reactive features since they usually wait for the leader nodes death to then elect a new leader. This behavior may cause holes in the network. The semantic clustering has presented better energy efficiency than other classical clustering methods. We propose PALES to reduce power consumption of Wireless Sensor and Actuator Network (WSAN) through a predictive election of semantic collectors using the ARIMA method. PALES extends an existing decentralized semantic clustering mechanism, inheriting its properties of self-adaptation, self-reconfiguration, and self-organization through a collaborative process. Results show that the PALES election increased up to 73% the collector battery saving in comparison to reactive election methods.