MI tracking


Now in 3-D with a resolution of 30cm over 15m x 15 m – click here for more.


My presentation of MI tracking at SenSys 2010 won the best presentation award! We also presented a demo, showcasing the low power operation of MI, in hostile (to RF) environments, such as underwater.

Core Idea

Track animals in their underground burrows using low frequency magnetic fields


Conventional methods of animal tracking are tailored to work above ground, for example VHF tracking and GPS. However, many animals spend large portions of their lives in underground burrows or dens. As radio-waves cannot adequately penetrate layers of soil (e.g. consider GPS performance in a basement of a building) their behaviour is hidden, preventing monitoring and tracking by zoologists and other scientists. To this end, we have developed a novel method of localizing an animal underground by using low-frequency magnetic fields. This technology has been trialled on badgers (Meles meles) in Wytham Woods, Oxfordshire. This is part of the WildSensing project.

System Description

A number of antennas are placed above the sett. These generate time-multiplexed digitally coded signals which identify the antenna currently active:

Badgers are fitted with tracking collars which have three perpendicular detection coils. This makes the measurements rotationally invariant i.e. the same field strength will be detected regardless of the orientation of the badger or the collar:

The problem now is – how to retrieve the field strength readings that have been detected by the collar? One option would be to retrap the badger, but there is always the possibility of collar loss, failure or destruction, which would result in a total loss of data. Instead, the tracking collar is also equipped with a Zigbee (2.4GHz) transceiver which allows it to transfer the stored information when the animal emerges at night to forage:

Once the field strength information is delivered to the basestation, the locations of the animal can be calculated:


In January 2010, four collars were placed on adult badgers in Wytham Woods, Oxfordshire.

The tracking collars are small and lightweight, and operate for approximately three months, obtaining location estimates once per second:


Overall, we obtained more than 5 million individual field strength measurements from our four badgers. These revealed some very interesting badger behaviour.

At 07:00, you can see two badgers are in two separate locations of the sett:

At 08:00, Badger D moves through the sett to share a chamber with Badger C. Note that we know for sure that this movement took place underground, as the badger tags emit radio beacons, which we would have picked up with our surface detection nodes.

From then on, Badgers C and D are co-located within the same area:

Another output from the system are aggregated animal locations. Not only does this show space/resource usage, it actually outlines the underground structure of the tunnel!


Although the early results from this system demonstrates that MI tracking has a great deal of potential in this challenging environment, there were a number of issues that were faced in the deployment.

  • Premature collar failure: None of the collars obtained their 3 month target lifetime. We now know that this was due to the type of battery selected (lithium thionyl chloride) which suffers from a phenomenon called passivation, where crystals of insulating material build up around the anode, leading to a voltage delay
  • 2-D localization: Although the system is theoretically capable of full 3-D localization, in this deployment, we only could obtain good accuracy in the horizontal plane and had quite poor accuracy in the vertical plane. We are investigating ways of placing our antennas to provide good 3-D coverage.
  • Limited range of operation: With the four antennas that we used, we covered an area of about 8 x 8 m. The entire sett covers an area of about 30 x 20 m, so we only provided partial coverage. We are now looking at ways of extending the coverage, the simplest being just to increase the number of antennas.
These issues have been addressed in the latest deployment which operated for over 9 months, with an accuracy of 30cm in 3-D over a 15m x 15m range – click here for more.


Our proof of concept deployment demonstrated the feasibility of MI tracking for localizing burrowing animals. There are still a lot of avenues that need to be explored to make this a robust system, but the early information gathered has revealed some interesting badger behaviour. Have a look at the full papers to get more information about how the system works.


[wp-publications bib=”citations.bib” search=”sensys”]



Andrew Markham (Oxford University Computing Laboratory)
Niki Trigoni (Oxford University Computing Laboratory)
Stephen Ellwood (Wildlife Conservation Research Unit, Department of Zoology, University of Oxford)
David Macdonald (Wildlife Conservation Research Unit, Department of Zoology, University of Oxford)