Sundial: Using Sunlight to Reconstruct Global Timestamps

TL;DR

Sundial addresses the challenge of reconstructing global timestamps in environmental sensor networks without reliable online clock synchronization. By jointly estimating clock drift and offset through a robust combination of RGTR and solar-pattern anchoring, Sundial leverages light-derived length-of-day and solar noon to align local measurements to a global timeline, even across segments separated by reboots. The method yields accurate reconstructions across long deployments, achieving RMSEs on the order of a minute and demonstrating resilience to missing anchor points and desynchronized basestations. This approach enables reliable postmortem time reconstruction in harsh, disconnected deployments and offers a practical, hardware-light alternative or complement to traditional synchronization techniques.

Abstract

This paper investigates postmortem timestamp reconstruction in environmental monitoring networks. In the absence of a time-synchronization protocol, these networks use multiple pairs of (local, global) timestamps to retroactively estimate the motes' clock drift and offset and thus reconstruct the measurement time series. We present Sundial, a novel offline algorithm for reconstructing global timestamps that is robust to unreliable global clock sources. Sundial reconstructs timestamps by correlating annual solar patterns with measurements provided by the motes' inexpensive light sensors. The surprising ability to accurately estimate the length of day using light intensity measurements enables Sundial to be robust to arbitrary mote clock restarts. Experimental results, based on multiple environmental network deployments spanning a period of over 2.5 years, show that Sundial achieves accuracy as high as 10 parts per million (ppm), using solar radiation readings recorded at 20 minute intervals.

Figures (16)

• Figure 1: An illustration of mote reboots, indicated by clock resets. Arrows indicate the segments for which anchor points are collected.
• Figure 2: Time reconstruction error due to $\alpha$ estimation errors as a function of the deployment lifetime.
• Figure 3: Ambient temperature data from two motes from the L deployment. The correlation of temperature readings in the left panel indicates consistent timestamps at the segment's start. After two months, the mote's reading become inconsistent due to inaccurate $\alpha$ estimates.
• Figure 4: The solar (model) length of day (LOD) and noon pattern for a period of two years for the latitude of our deployments.
• Figure 5: The light time series (raw and smoothed) and its first derivative. The inflection points represent sunrise and sunset.