11 ms·
We need to talk about the security implications of low-power devices broadcasting data. I cringed when the article talked about broadcasting "medically relevant
by andrewflnr 9y ago
We need to talk about the security implications of low-power devices broadcasting data. I cringed when the article talked about broadcasting "medically relevant data" for kilometers, in the clear. That's not OK. Everyone needs to be clear on the fact that that's not OK.
I'm ecstatic at the possibilities this has for reducing the power cost of communication[0]. Please use the savings on some (at least half-decent) crypto.
[0] Seriously, can this work for a consumer handset communication network?
- andrepd 9y agoEncryption is a easy. I doubt this will be a problem.
- gwern 9y agoThe Internet of Shitty Things is always a problem, and you know people will be disabling encryption to save 10% energy.
- pavel_lishin 9y agoOr two days' of developer time. Or a nickel on the manufacturing costs.
- vvanders 9y agoOh the stories I've seen/been involved in over half a cent on a BoM.
- LeifCarrotson 9y agoIt's one thing to have a problem that can be solved mathematically. It's a much more difficult thing to implement the solution correctly and uniformly.
- andrewflnr 9y agoEncryption is easy on platforms with plenty of power and transistors and well-established libraries, which are the exact opposite of what we're talking about here. Encryption on very low-powered systems is a well-known problem.
- boznz 9y agoEncryption for low data rate systems is actually a very easy problem. I have sensors in the field which transmit a 64 byte package once a day only 18 bytes of which is data and is encrypted. the encryption an one time pad in ROM which lasts the life of the device (8 years) the device does not receive. I agree encryption is totally necessary for medical/sensitive data or two way devices but for other cases (such as irrigation data) no encryption or an encrypted ID is just fine and a problem that does not need to be solved.
- forapurpose 9y agoSeemingly trivial data can be combined to form a detailed picture of someone. Regarding irrigation: How is the farmer's business doing? What kind of seed are they using and what is their schedule? Is the homeowner home? Are they following local water use laws or guidelines?
- skummetmaelk 9y agoNo wonder the internet of shit exists when people have this attitude.
- dfox 9y agoDoing reasonable encryption and authentication on interfaces with MTUs on the order of tens of bytes (which is common for various low-power IoT radio interfaces) is actually pretty complex problem. Essentially all widely used and safe cryptographic constructions have overhead (IV and MAC tag for example) that would consume significant portion of your packet (or in some cases is actually larger that the packet).
- foepys 9y agoIf you have low power devices with basically no external input but a build in clock and a network interface that receives one packet every few minutes (often it even only sends data!), how do you create cryptographically secure random numbers for your encryption?
- HankB99 9y agoNoise. Most - if not all - of these chips have analog to digital convertors because most of the applications involve monitoring some sort of signal. There's going to at least a bit or two of noise in the signal (or if not otherwise used, a floating input.) That's where I'd start if I was looking for entropy.
- boznz 9y agoSame "noise" has to be at receiving crypto otherwise no key to decode it.
- Ar-Curunir 9y agoThat's... not how modern crypto works.
- wahern 9y agoFor the sake of argument, I'll grant that encryption is easy. Key management is not easy. Most people don't even understand, let alone appreciate, the problems, which is why there's substantially less bike-shedding in this domain. In the context of sell-and-forget (intentionally or unintentionally) IoT devices, there often may be more real security in simply giving up, to mitigate a false sense of security.