Dandelion inspired sensors

Environmental sensors are becoming ubiquitous, for better or worse.  As sensors become smaller and cheaper, one key challenge is how to deploy them.  People voluntarily carry them in their devices, and so-called “smart” cities and buildings have sensors built in.  But many parts of the planet are not conveniently engineeered, so sensors have to be sent there in one way or another.  This is one of the killer apps for small UAVs, for example.

This spring researchers at U. Washington report a bioinspired method for dispersing tiny sensors:  wind dispersal inspired by dandelion seeds [1]. These small devices are designed to be strewn from a UAV, and float randomly on the wind.  The solar powered environmental sensor collect and relay data during the day, and rest at night, so they can persist for quite a while with no need for external power or replacement.

The wind dispersal is inspirited by wind blown seeds, and in keeping with the biological models, each package is slightly different [2].  Between the chaos of wind currents and the variability of the sensors, the group will disperse rather randomly.  Like seed dispersal, this strategy achieves coverage and works around ‘hard places’.

This project caught my eye for a slightly different reason.  I’ve been watching the development of tiny battery free devices for many years now (OK—probably decades).  But I’m more interested in signaling than sensing.

In particular, I have a vision of a cloud of decorative artificial fireflies.  Not that natural fireflies aren’t awesome, but I’d like to be able to deploy them at parties, indoors, or on a cityscape.  Zillions of tiny lights, emitting charming patterns (and sounds).

I’ve also imagined jewelry that sings softly, just to you.  Image earrings that chirp like a bird.  Or whisper sweet nothings from a lover. 

The dandelion sensors show that power and electronics are getting small enough to begin to realize my visions. Substitute tiny LEDs and speakers, and you’ve got my firefly / earrings.

Now, there are challenges with the dandelion sensors.  Networking and data retrieval are certainly a problem.  These little guys don’t have much bandwidth, so you need to get a relay station pretty close—another job for a UAV.  But still, there are distinct limits to how much data you can actually get out.

Another big problem with the dandelion sensors is that they need to be biodegradable.  Spewing these over, say, a farm field, is fine for sensing.  But it leaves behind all those sensors which may remain for a long time.  They are probably at least a little toxic.  And since they are tiny and randomly dispersed, it’s going to be pretty impossible to pick them up one by one.  So we really want them to be biodegradable.

I’ll note that my jewelry concept doesn’t suffer from these challenges.  The jewelry is intended to last a long time, and doesn’t need to transmit.  (In fact, we really don’t want our personal effects spying on us.)  And should the jewelry need to be recharged or reprogrammed, it can be done on a table in controlled conditions. 

(Why would it need to be reprogrammed?  Well, you might need to change the sound recording to erase the voice of a former amour, no?)

Anyway, this is very inspiring, if not quite for the applications highlighted by the researchers.

---

1. Vikram Iyer, Hans Gaensbauer, Thomas L. Daniel, and Shyamnath Gollakota, Wind dispersal of battery-free wireless devices. Nature, 603 (7901):427-433, 2022/03/01 2022. https://doi.org/10.1038/s41586-021-04363-9
2. Sarah McQuate, Tiny battery-free devices float in the wind like dandelion seeds, in UW News, March 16, 2022. https://www.washington.edu/news/2022/03/16/battery-free-devices-float-in-wind-like-dandelion-seeds/

DNA Snooping Online

I’m not a gigantic fan of popular DNA testing, that enable people to submit DNA and get reports on supposed health risks and alleged geohistorical links.  To me, this is using weak science to help people self-diagnose and, worse, to reinforce dubious notions of “heritage”.  (Look, a few arbitrary gene markers don’t make you “Irish” or “German” or “West African”.  What matters is where you live now and what culture you live in, and who you really are.)

This fall researchers at University of Washington report on an even more problematic variation on this theme [2].  Not content to peddle cargo cult anthropology to the masses, some sites are letting people do their own DNA searching, looking for relatives in a database of uploaded DNA.  This is basically the technology that police forces use to try to identify unknown DNA samples.  Indeed, these commercial databases are used by police and private investigators for just this purpose.

You upload your DNA, and let the world search it.  To satisfy your vanity, you voluntarily put your DNA in the police database.  What could possibly go wrong?

Well, the researcher shows that there are some pretty serious privacy issues, at least with the specific service they worked on.

The big problem is that the API in question (they studied GEDmatch) is excellent, providing a lot of features and options, and presenting the results in very usable forms.  Unfortunately, this means that they are also abusable.  Another key issue is that the system will accept syntactically valid data, but does not (and probably cannot) verify the source.  This means that adversaries can upload synthetic samples which are entered in the system the same as data from real individuals.

I would say that, at base, publishing “anonymized” data of any kind is difficult in any case.

The research showed that by creating a relative handful comparison samples, they are able to play twenty questions with the data and identify over 90% of the specific DNA sequences.  This means that they can determine the presence of specific genetic markers in an individual’s DNA.  This might indicate medical conditions or some other genetically linked characteristic.  Ooops.

It is also possible to spoof relationships in the data by creating synthetic DNA profiles that match a target at whatever degree desired.  Such a false match could be used for several kinds of mischief, including fraud (your long lost cousin!  Luke, I am your father!) and disrupting identiy searches by injecting chaff into the system.

Yoiks!

I’m not really sure why anyone would upload data to such a system in the first place, but clearly you might want to rethink.

The researchers outline a number of responses to reduce these hazards.

One obvious issue is authenticating the provenance of the data.  This will need to be done because sooner or later these services will face denial of service attacks, with bot nets flooding the dataset with bogus data.  That would be the end of the business.

A second issue it the storage format (lossless compression) which is intended to mask the original data but in fact does not.  (I don’t know a technical answer to this one, but I’m not a cryptographer.)

The researchers also suggest that the search features should be reconsidered.  Obviously, it would be wise to limit the number and rate of queries.  Again, sooner or later, there will be DDOS attacks by bot nets flooding the system with spam queries.

The rich data returned is cool, but probably should be reviewed carefully.  This is a more difficult challenge, because these are useful and valuable features, indeed, they are the crux of the usability of the systems.  Still, anything that reveals detailed values of the original data need to be scrutinized.

The researchers speculate that a similar data service to the one in the study is probably harder to suborn precisely because it has a much less featureful interface.

In short, its back to the drawing board, and until its fixed, just say no.

---

1. Sarah McQuate, Popular third-party genetic genealogy site is vulnerable to compromised data, impersonations in UW News. 2019. https://www.washington.edu/news/2019/10/29/genetic-genealogy-site-vulnerable-compromised-data-impersonations/
2. Peter Ney, Luis Ceze, and Tadayoshi Kohno, Genotype Extraction and False Relative Attacks: Security Risks to Third-Party Genetic Genealogy Services Beyond Identity Inference (accepted), in Network and Distributed System Security Symposium (NDSS). 2020: San Diego. (preprint) https://dnasec.cs.washington.edu/genetic-genealogy/ney_ndss.pdf