FLIGHT: clock calibration using fluorescent lighting

Zhenjiang Li, Wenwei Chen, Cheng Li, Mo Li, Xiang-Yang Li, Yunhao LiuMobiCom 2012 · DOI

TL;DR

This paper introduces FLIGHT, a novel method for synchronizing clocks in wireless sensor networks by leveraging the periodic fluctuations of fluorescent lighting. FLIGHT offers a low-power, hardware-agnostic solution that achieves tight time synchronization without relying on traditional, energy-intensive radio communication protocols.

What Problem Does This Paper Address?

Maintaining a common notion of time across distributed nodes in a network is crucial for many applications, from real-time gaming to coordinated data processing in healthcare monitoring. However, the crystal oscillators used for generating clocks in most devices are inherently unstable, drifting over time due to environmental factors like temperature and humidity. This drift necessitates frequent clock calibration. Traditional methods for clock synchronization often involve significant communication overhead and energy consumption, especially in large networks, limiting their scalability and practicality for low-power devices. While some approaches use external signals like power lines or radio broadcasts, they often require specialized hardware or still consume considerable energy. This paper addresses the need for an accurate, energy-efficient, and hardware-agnostic clock calibration method suitable for widespread deployment in indoor environments.

Methods

FLIGHT's core innovation lies in its use of ambient fluorescent lighting as a universal time reference. The paper proposes the following key techniques:

  • Period Extraction from Fluorescent Light: Fluorescent lights, powered by alternating current (AC), exhibit intensity fluctuations with a period directly related to the AC frequency (typically 50 or 60 Hz, resulting in a 100 Hz global reference clock). FLIGHT uses a light sensor (or camera) to sample these fluctuations.
  • Lightweight Filtering: To extract the stable periodic signal from noisy light intensity readings, FLIGHT employs a lightweight filtering approach. Specifically, an m-order filter (found to be effective with m=6) is used to smooth the signal, allowing for accurate period detection even in low-light or dynamic conditions. This filtering is designed to be computationally inexpensive, enabling it to run in parallel with sampling.
  • Logic Time Maintenance: The system maintains a "logic clock" that advances based on the native clock and an estimated frequency ratio. FLIGHT continuously estimates this frequency ratio by comparing the native clock's ticks to the detected light periods. This allows for precise compensation of clock drift.
  • Adaptive Calibration Interval: To balance accuracy and energy efficiency, FLIGHT introduces a dynamic calibration interval adjustment mechanism. This scheme uses a drift-controlling factor to adapt the frequency of calibration based on the observed clock drift, reducing calibration frequency when drift is low and increasing it when drift is high.
  • Time Synchronization Upgrade: While FLIGHT primarily focuses on clock calibration (ensuring clocks tick at the same rate), it can be easily upgraded to full time synchronization (ensuring clocks show the same absolute time) by incorporating existing protocols like FTSP with significantly reduced communication overhead due to the pre-calibrated clocks.

Datasets

Not discussed in the paper. The experiments were conducted using a custom test-bed with TelosB motes in laboratory and real-world indoor environments.

Key Insights and Ideas

  • Ubiquitous, Free Time Reference: Fluorescent lighting is a surprisingly robust and widely available "free" time reference, present in most indoor environments, eliminating the need for specialized hardware or external infrastructure.
  • Low-Power Sensing is Sufficient: Accurate clock calibration can be achieved by simply sampling ambient light intensity, which is a low-power operation, making it ideal for resource-constrained sensor nodes.
  • Filtering for Robustness: The proposed lightweight filtering technique is crucial for extracting the periodic signal from noisy and fluctuating light intensity, enabling FLIGHT to work even in challenging mobile or low-light scenarios.
  • Adaptive Calibration for Efficiency: Dynamically adjusting the calibration interval based on observed drift is key to optimizing energy consumption without sacrificing accuracy.
  • Calibration as a Foundation: FLIGHT's focus on clock calibration, rather than full synchronization, provides a fundamental service that can be built upon for more complex time-dependent applications with significantly reduced overhead.

Key Results

  • High Synchronization Accuracy: FLIGHT achieved tightly synchronized time with average logic time errors consistently below 600µs and maximum errors below 950µs in static environments over a week. In dynamic (mobile) environments, errors remained within 1000µs.
  • Low Energy Consumption: The energy consumption of FLIGHT was found to be very low, typically less than 5µW, making it highly suitable for power-constrained devices. This is significantly lower than other communication-based synchronization methods.
  • Robustness to Environmental Changes: Experiments demonstrated that FLIGHT is robust to variations in light intensity, different types of lighting (sunlight, LED, filament), and mobility, including node rotation and movement across different buildings.
  • Minimal Resource Usage: The FLIGHT implementation on TelosB motes occupied only 23% of the available memory, leaving ample space for other applications.

Limitations and Future Work

  • Lighting Availability: FLIGHT relies on the presence of fluorescent lighting. While it can work with very low light intensity, complete darkness would render it inoperable. The paper notes that fluorescent lighting is widely available in most indoor settings, mitigating this concern.
  • Sensor Exposure: Light sensors need to be exposed to the environment. For applications where nodes are permanently enclosed (e.g., inside pockets), a hybrid approach with neighbor-to-neighbor synchronization might be necessary.
  • Robustness to 100Hz Noise: The current filtering focuses on noise outside the 100Hz band. Future work could explore more advanced techniques like Phase-Locked Loops (PLLs) to further enhance robustness against noise near the primary frequency.
  • Interference from Other Light Sources: While FLIGHT showed robustness to mixed light sources, further investigation into more complex interference scenarios could be beneficial.

Summary updated

BibTeX
@inproceedings{li2012flight,
  author = {Zhenjiang Li and Wenwei Chen and Cheng Li and Mo Li and Xiang-Yang Li and Yunhao Liu},
  title = {{FLIGHT: clock calibration using fluorescent lighting}},
  booktitle = {MobiCom},
  year = {2012},
  doi = {10.1145/2348543.2348584},
}
Figure