The Big Night Out: Experiences from Tracking Flying Foxes with Delay-Tolerant Wireless Networking

Philipp Sommer, Branislav Kusy, Adam McKeown, Raja JurdakREALWSN 2013 · DOI

TL;DR

This paper presents a novel delay-tolerant wireless networking system designed to track the movements of free-living flying foxes. By developing a low-power, energy-efficient collar and a network of base stations, the system enables long-term, perpetual monitoring of these animals, addressing the challenges of limited energy budgets and intermittent connectivity.

What Problem Does This Paper Address?

Tracking the movements of small, wild animals presents significant challenges, primarily due to the stringent weight and energy constraints for on-animal devices. For flying foxes, which are highly mobile and nocturnal, this is further complicated by their ability to travel long distances and the need for data to be collected and transmitted reliably over extended periods, often without direct access to the devices. Existing solutions are often too power-hungry, bulky, or lack the necessary delay-tolerant capabilities to handle the intermittent communication windows that arise when animals are far from data collection points. This paper tackles the problem of designing and deploying a sustainable, low-power wireless sensor network for continuous monitoring of flying foxes, overcoming these limitations.

Methods

The authors developed a system comprising three main layers:

  • Mobile Sensing Layer: Custom-designed, low-power printed circuit boards (PCBs) were integrated into collars for the flying foxes. These nodes run a modified Contiki operating system with extensions for logging and remote procedure calls (RPCs). The core challenge here was minimizing size, weight, and power consumption while incorporating sensors like GPS, accelerometers, and pressure sensors.
  • Base Station Layer: Spatially distributed base stations are deployed in bat roosting camps. These stations use short-range wireless communication (TI CC1101) to download data from the mobile nodes when they are in range. They are powered by solar panels and batteries for autonomous operation and connect to a central server via 2G/3G modems for data upload.
  • Backend Storage and Control Layer: A central database stores all collected sensor data and maintains a synchronized view of downloaded data across all base stations, preventing duplicate downloads.

A key novel contribution is the BatMac protocol, a time-synchronized medium access protocol designed for intermittent connectivity. It features:

  • Slotted Communication: Mobile nodes transmit in scheduled sub-slots within communication rounds, determined by their node ID and synchronized to GPS time. This allows nodes to aggressively duty-cycle their radios while base stations remain more active.
  • Announcement Beacon: Mobile nodes periodically broadcast an announcement beacon containing their ID and current flash page number.
  • Node Selection: Base stations listen for beacons and initiate communication if new data is available or configuration updates are needed, keeping the node's radio awake.

For data storage, they adapted the Tagged Data Format (TDF) to efficiently pack heterogeneous sensor readings into a byte stream, minimizing overhead. Data retrieval uses an RPC-based approach, where base stations request data page by page.

Datasets

The paper evaluates the system using empirical data collected from three free-living flying foxes over a 7-day period. This data includes GPS positions, battery voltage, solar panel voltage, and charge current. The authors also refer to previously published work for data formats and protocols.

Key Insights and Ideas

  • Delay-tolerant networking is crucial for mobile wildlife tracking: The intermittent connectivity between animals and base stations necessitates protocols that can handle delayed data delivery and asynchronous communication.
  • Energy efficiency is paramount: Every component and protocol decision is driven by the need to conserve power, enabling long-term deployments on animals with strict weight limits.
  • BatMac protocol effectively manages intermittent connectivity: The time-synchronized, slotted approach allows mobile nodes to sleep for extended periods while ensuring data can be offloaded when animals return to camp areas.
  • TDF provides a flexible and compact data storage solution: The ability to store diverse sensor data with minimal overhead is essential for maximizing data retention on limited flash memory.
  • Remote configuration and debugging are vital: Given the difficulty of physically accessing deployed nodes, over-the-air configuration and debugging tools (RPCs, logging) are indispensable for system maintenance and troubleshooting.

Key Results

  • The system successfully collected data from three flying foxes, demonstrating the feasibility of perpetual monitoring.
  • The BatMac protocol allows mobile nodes to aggressively duty-cycle their radios, saving significant energy.
  • The TDF format was found to be efficient, with actual sensor data accounting for 65% of flash page size, headers for 31%, and fragmentation for only 4%.
  • Packet Reception Rate (PRR) for announcement beacons ranged from 0.35 to 0.78, indicating reasonable connectivity.
  • GPS Time to First Fix (TTFF) averaged 4.17 seconds, with longer times observed after extended GPS inactivity.
  • The system demonstrated the ability to track animal movements, with GPS sampling rates adjusted based on time of day and battery levels.

Limitations and Future Work

  • Limited number of nodes tested: The experiments were conducted on only three flying foxes. Scaling to hundreds or thousands of nodes will require further validation.
  • Co-existence of software versions: The authors acknowledge that staggered software releases mean collars with older versions will remain active, necessitating a mechanism to identify capabilities via software version in announcement beacons.
  • Future work includes:
    • Fine-tuning system operations based on gathered empirical data.
  • Deploying a larger number of collars (hundreds) in future.
  • Further development of diagnostic tools and remote procedure calls for flexible node inspection.
  • Improving the accuracy of remaining energy estimation.

Summary updated

BibTeX
@inproceedings{sommer2013big,
  author = {Philipp Sommer and Branislav Kusy and Adam McKeown and Raja Jurdak},
  title = {{The Big Night Out: Experiences from Tracking Flying Foxes with Delay-Tolerant Wireless Networking}},
  booktitle = {REALWSN},
  year = {2013},
  doi = {10.1007/978-3-319-03071-5\_2},
}
Figure