Ergodicity-Informed Adaptive Sensing for Energy-Constrained Urban IoT Networks

Urban environments pose invisible yet pervasive health hazards, including exposure to air pollutants, noise, and temperature extremes. Monitoring these hazards requires high-resolution data to identify hotspots and support timely interventions. However, conventional stationary monitoring networks lack spatial resolution, while mobile drive-by sensing systems suffer from uneven data distribution and energy constraints. Here we show that an ergodicity-informed adaptive sensing framework can significantly improve both the coverage uniformity and energy efficiency of mobile environmental monitoring in cities. Using solar-powered sensors mounted on taxis in Stockholm, Sweden, our system regulates sensor activation based on recent sampling history and air quality guidelines, reducing redundant measurements in over-sampled street segments, resulting in a natural redistribution of sensing counts towards under-sampled regions. Drawing from the ergodic properties of urban mobility, we formally demonstrate—through the Random Waypoint Model—that temporally aggregated sensor data converges to a stable spatial distribution governed by long-term mobility patterns. Our approach achieves a 10.74% increase in energy efficiency and 72.14% reduction in spatial variability of sensing coverage. Validation through statistical tests confirm significantly improved uniformity, and a clustering optimization revealed coherent regions of coverage enhancement. These findings position adaptive sensing as a scalable and theoretically sound solution for equitable, resource-efficient environmental monitoring in complex urban systems.