[Cryptography] doorbells for Ukraine

William Allen Simpson william.allen.simpson at gmail.com
Tue Dec 20 06:33:42 EST 2022


On 12/19/22 10:47 PM, Peter Gutmann wrote:
> William Allen Simpson <william.allen.simpson at gmail.com> writes:
> 
>> I'm daydreaming more about very cheap Ring (and/or Nest) devices distributed
>> to Ukrainian households, mounted on buildings, scattered widely across an
>> area.
> 
> If they're using the near-universal Ambarella devices for this

Ring and Nest both use Ambarella, according to Ambarella's web site.


> then AFAIK those are optimised for (relatively)
> close-up work and recognition of images like humans, vehicles, and similar (I
> have no idea how good a CNN is at picking that five blurry pixels are a drone,
> vs. identifying closeup human faces).  I would imagine that picking tiny
> moving objects out of the sky and dealing with background clutter wouldn't be
> a good fit.  And then beyond that the Ring devices would have optics and other
> hardware designed to function as doorbells, not distant small moving object
> recognisers and trackers.
> 

Exactly what I was wondering.  Beyond my expertise.

My initial communication security interest was because the Ring in particular can
communicate via wired ethernet, with WiFi backup, and cellular backup backup.

Also supposedly can mesh communicate directly with other Ring devices without a
WiFi base station.  Pretty much ideal for field deployment.

Reports are that current Ring is much more secure than previously.


> Having said that, with a greenfields design and a lot of time you could
> probably adapt Armbarella hardware for this use... 

If the processing is mostly in the cloud, this design development should be
both quicker and more scalable than a cell phone app.

> and then run into problems
> with export restrictions (it's a US company) because the only really
> significant application for this is probably military.
> 

Since Ring had/has engineering in Ukraine, and the hardware is manufactured in
China, this shouldn't be a problem.



More information about the cryptography mailing list