Configuring Transmission Thresholds in IIoT Alarm Scenarios for Energy-Efficient Event Reporting

Abstract

Industrial Internet of Things (IIoT) applications involve real-time monitoring, detection, and data analysis. However, the intermittent activity of IIoT devices and limited battery capacity pose critical challenges. This paper addresses these interconnected issues, focusing on extending the battery life of IIoT devices sensing events/alarms by minimizing the number of unnecessary transmissions. We propose a threshold-based transmission-decision policy based on the sensing quality and the network spatial deployment. We optimize the transmission thresholds using several approaches such as successive convex approximation, block coordinate descent methods, Voronoi diagrams, explainable machine learning, algorithms based on natural selection and social behavior, and Q-learning. Through numerical evaluation, we demonstrate significant performance enhancements in low-power IIoT environments, with Q-learning performing the best, while the block coordinate descending method performs the worst. We compare the proposed methods to a benchmark that assigns the same transmission threshold to all devices. In low-density scenarios, all proposed methods outperform the benchmark, while in high-density scenarios, only Voronoi-(i), K-nearest neighbors, and Q-learning show better performance. Power consumption is reduced by up to 95% in low-density scenarios compared to the benchmark and by 63% in high-density scenarios.