AI-driven autonomous task scheduling for energy harvesting IoT systems using piezoelectric sensors

A smart and sustainable energy-harvesting Internet of Things (IoT) task scheduling framework is proposed to address limited energy availability, dynamic workloads, and system stability. The framework develops an AI-enabled autonomous scheduler that adapts to real-time variations in harvested energy, task queues, and system conditions. The approach combines the principles of piezoelectric energy harvesting, Priority-based Real-Time scheduling, Reinforcement Learning and Multi-objective optimization in a distributed architecture for battery-less scalable operation. The system constantly monitors the available energy, workload and system state to make optimum scheduling decisions. The results indicate that utility improved by 0.91, task success was 94.2%, and energy utilization was 88.7%. Under heavy workloads, priority-based scheduling met the performance requirement of 85.2% mission success with missed deadlines at < 9.5%, and stable operation held up to an energy level of 0.4 J. The piezoelectric module was able to produce a power of up to 2.30 J, which facilitated the continuous use of the module. Reinforcement learning enabled a rewards increase from −95 to 85 and the Q values from 0.15 to 0.91, reducing latency to 130 ms and failure rate to 3.2%. Overall, the framework achieved 94.6% task completion and 98.6% uptime.