Tracking Wild Bats with SDRs – Featured in Science Magazine
Recently research from Tel-Aviv University by Sivan Toledo et al. involving the use of USRP SDRs to track wild bats was published in Science. The Journal Science (aka Science Magazine) is one of the world's top peer reviewed academic journals.
Sivan and his collaborators developed inexpensive 434 MHz band tracking tags for bats that emit radio pings every few seconds. These pings do not contain any location data, however the location is accurately tracked by several USRP SDRs with high accuracy GPSDO oscillators set up around the target tracking area. A radio direction finding technique known as "time difference of arrival" or TDoA is used to pinpoint the location of each tag. Sivan writes:
A wildlife tracking system called ATLAS, developed by Sivan Toledo from Tel-Aviv University in collaboration with Ran Nathan from the Hebrew university, enabled a science breakthrough reported in an article in Science that was published yesterday.
The system uses miniature tracking tags that transmit radio pings in the 434 MHz bands and SDR receivers (Ettus USRP N200 or B200). Software processes the samples from receivers to detect the pings and to estimate their time of arrival. The overall system is a "reverse-GPS" system, in the sense that the principles and math are similar to GPS, but the role of transmitters and receivers is reversed. A youtube video explains how the system works. SDR-RTL dongles can certainly detect the pings, but their oscillators are not stable enough to accurately localize the tags.
The system has been used to track 172 wild bats (in batches, some consisting of 60 simultaneously-tagged bats). The results showed that bats can make novel shortcuts, which indicates that they navigate using a cognitive map, like humans. The system, and other ATLAS systems in the Netherlands, England, Germany, and Israel are also tracking many different animals, mostly small birds and bats.
The video below shows the bats being tracked on a map accelerated to 100x.
The Science article itself is mostly about the discoveries on bat behaviour that were made by the system. However the YouTube video embedded below explains a bit more about how the technical radio side works.
I would never handle a bat with my bare hands. Bats are famous for hosting many diseases, some of which can directly infect or eventually cross-over to Humans. Handling a bat without protection exposes you to direct contact with bat urine and feces which shed disease particles, worse yet the bat can bite you. In a bat colony’s resting environment you should always wear at-least a properly fitting N99 respirator and sealed goggles. For obvious reasons, when entering a closed space infested with bats, you should seriously consider wearing a disposable Tyvek coverall with hood and shoe protection. A wide brim hat and/or face shield will prevent droppings from reaching your exposed facial skin. Bats are nasty critters, you can never be sure they are safe to handle unprotected.
Would be interesting to see their results with the RTL-SDR regarding the clock drift.
There is another project, that successfully applied TDOA with the RTL-SDR v3 with impressive results:
http://www.panoradio-sdr.de/tdoa-transmitter-localization-with-rtl-sdrs/
Also, some interesting details on the correlation operation under frequency offsets are provided.
Thanks a lot jo, very interesting website. We do the clock synchronziation differently, in a stand-alone way, but the principle is I think similar. The details are here: https://www.tau.ac.il/~stoledo/Bib/Pubs/accuracy-ipsn16-ieee-copyright.pdf
Also, we do correct for frequency errors in the transmitters, but only for PSK tags. This is explained here:
https://www.tau.ac.il/~stoledo/Bib/Pubs/modulation-ieee-copyright.pdf
“SDR-RTL dongles can certainly detect the pings, but their oscillators are not stable enough to accurately localize the tags.” This is not true, the SDR-RTL V3 can also do the job if you compensate for the clock drifts, as was shown in the study below:
https://repository.nwu.ac.za/bitstream/handle/10394/25449/Krüger_SW_2017.pdf
Hi Paulus,
You are write that the way I wrote the post is not technically completely correct. I should have written that the current version of ATLAS requires more accurate and stable clocks. I am well aware of Kruger’s thesis, which is very impressive work, certainly for an MS thesis. I cited it in a recent paper:
Andrey Leshchenko and Sivan Toledo.
Modulation and signal-processing tradeoffs for reverse-GPS wildlife localization systems.
In Proceedings of the European Navigation Conference (ENC), pages 154-165, June 2018.
(https://www.tau.ac.il/~stoledo/Bib/Pubs/modulation-ieee-copyright.pdf).
We also have a paper showing the same thing, in ATLAS, using post-processing:
Saeed Shojaee, Johannes Schmitz, Rudolf Mathar, and Sivan Toledo.
On the accuracy of passive hyperbolic localization in the presence of clock drift.
In Proceedings of the IEEE International Symposium on Personal, Indoor, adn Mobile Radio Communication (PIMRC), pages 1-6, October 2017.
(https://www.tau.ac.il/~stoledo/Bib/Pubs/pimrc_2017_paper.pdf).
However, putting that technique into a system that works 24/7, localizes both in real time and from database records, and is tolerant of receiving disappearing and coming back up proved very challenging, and this work is not complete yet. But you are certainly right that it can be done.
ICARUS Global Monitoring with Animals and international space station (ISS): https://giammaiot.blogspot.com/2019/10/icarus-global-monitoring-with-animals.html?m=1