Adaptive algorithm for microsensor in sustainable environmental monitoring

TL;DR

The paper tackles the problem of excessive data and power use in continuous environmental sensing by introducing an FFT-based harmonic-analysis algorithm that identifies dominant frequency components via $X[k]$ and $A[k]$, enabling event-driven sensor activation. It then reconstructs the signal using a subset of harmonics with $x[n] = (1/N) \sum_{k=0}^{Kopt} X[k] exp(j 2*pi*k*n/N)$ and applies an energy-based criterion $\sum_{k=0}^{Kopt} |X[k]|^2 \ge 0.5 \sum_{k=0}^{N-1} |X[k]|^2$ (with a 90% energy variant) to minimize data while preserving essential information. The approach focuses on CO$_2$, humidity, and temperature measurements, demonstrating how derivative-based activation moments can indicate events and enable accurate anomaly detection for building-surface monitoring. The findings suggest substantial reductions in data transmission and storage requirements while maintaining predictive accuracy, making microsensor networks more sustainable for environmental monitoring.

Abstract

Traditional data collection from sensors produce a lot of data, which lead to constant power consumption and require more storage space. This study proposes an algorithm for a data acquisition and processing method based on Fourier transform (DFT), which extracts dominant frequency components using harmonic analysis (HA) to identify frequency peaks. This algorithm allows sensors to activate only when an event occurs, while preserving critical information for detecting defects, such as those in the surface structures of buildings and ensuring accuracy for further predictions.

Figures (5)

• Figure 1: Multivariate Microsensor
• Figure 2: Microsensor Measurements
• Figure 3: $CO_2$ Modeling
• Figure 4: Reconstructed $CO_2$ Signal
• Figure 5: Microsensor activation time for $CO_2$