UC Riverside

Multi-sensor IoT devices are used to monitor different environmental phenomena however these devices depend on batteries and renewable energy sources for power. Therefore, efficient energy management solutions are needed to maximize device lifetime while ensuring appropriate and sufficient data is captured to be processed by the IoT application. To address this problem at the device we use two strategies. Our first strategy adapts the IoT device load to the available energy ensuring the device lifetime is extended while meeting the application information requirements. Our on-device solution called Predictive EneRgy Management for IoT (PERMIT) is a low-complexity Model Predictive Control (MPC) based approach that optimizes sensing tasks while ensuring battery charging is prioritized. Evaluations using our prototype IoT device and simulations demonstrate the effectiveness of PERMIT. Our second strategy leverages cooperative sensing to further increase energy efficiency and minimize temporal overlap. Cooperative sensing allows multiple IoT devices to collaborate and coordinate their sensing operations to obtain the required data while minimizing energy use. Our Distributed Task Adaptation (DTA) algorithm empowers devices to modify their sensing task operations using information from neighboring devices. DTA works with our distributed Block Scheduler that minimizes overlap across all executions of a sensing task among neighboring devices by reframing the scheduling problem as a block placement problem and minimizing total block overlap every epoch. Experimental results demonstrate the effectiveness of DTA and the Block scheduler in saving energy and reducing temporal overlap among IoT devices while using tokens to share minimal information between IoT devices.