Evaluating Heterogeneous and Homogeneous Sampling for Energy-Accuracy Trade-Offs in Multi-Sensor IoT Systems

Abstract

Energy-constrained Internet of Things (IoT) nodes commonly use homogeneous sampling, where all sensors operate at the same fixed intervals. This often leads to inefficient energy utilization and suboptimal sensing accuracy because it does not account for the different temporal characteristics of individual sensors. This limitation highlights the need to adopt adaptive strategies capable of accounting for sensor-specific behaviours. This paper investigates the energy accuracy tradeoff of optimized homogeneous and heterogeneous sampling strategies using Greedy and Random Search methods. Experimental results show that both optimized samplings reduce energy consumption from approximately 70% to below 9% of battery capacity while maintaining comparable sensing accuracy. In addition, heterogeneous sampling improves sensing accuracy by approximately 2 percentage points over optimized homogeneous sampling without increasing energy consumption by adapting sampling intervals to the characteristics of individual sensors. These results highlight that heterogeneous sampling provides a more effective balance between energy efficiency and sensing accuracy, making it better choice for multi-sensor IoT applications.