Evoluční fuzzy pravidla pro adaptivní řízení spotřeby energie v IoT senzorech

Abstract

Wireless sensor networks (WSNs) represent a key technology for autonomous data collection within Internet of Things (IoT) systems; however, their operation is constrained by the available energy and the variability of environmental conditions. This thesis focuses on controlling the energy consumption of wireless sensor nodes powered through energy harvesting. Based on the conducted analysis, research objectives were formulated to design an innovative method inspired by genetic and evolutionary principles, and to validate its performance. The proposed method employs evolutionary fuzzy rules (EFR), which enable the automatic synthesis of both the structure and the parameters of controllers using genetic programming. The method was evaluated in an extensive software-based simulation experiment using historical meteorological data and compared against reference energy-management strategies. The results demonstrate improvements in the evaluation metric, and cross-testing confirmed the transferability of the trained controllers across different locations. The computational requirements were further verified on real embedded hardware. The main contribution of this dissertation is the design of a novel method for energy-consumption control in IoT nodes that combines evolutionary algorithms with fuzzy logic, eliminates the need for manual controller design, and enables automatic synthesis of the control strategy. A practical contribution is the verified implementability of the method on cost-effective embedded hardware.