SK Telecom makes light of random numbers for IoT applications

Quantum random number generators aren't new, but one small enough to provide practical security for Internet of Things applications is interesting. That's what South Korean telco SK Telecom reckons its boffins have created, embedding a full quantum random number generator (QRNG) in a 5x5mm chip. The company's pitch is that …

[John Smith 19]

Cheaper than the cheapest QRNG sounds like a lot of wiggle room.

Less than several $1000US ?

OTOH LEDs are quite complex to fabricate and yet mass production has brought their prices down hugely.

Like the idea.

Hate the application.

[Christoph]

"a quantum noise source ... acting as the input to a deterministic RNG."

That reads as if they are merely using the quantum source to seed the RNG. Which means it still has the same problems with buggy software, and possible predictability if enough samples are gathered.

[Anonymous Coward]

deterministic RNG

Not to mention fake chips that don't use quantum anything, but a collection of seeds kindly provided by a "friendly" government...

[Natalie Gritpants]

Re: deterministic RNG

Doesn't need a fake chip. The real chip will have a debug mode that replaces the Q bits with zero for production testing etc. All the three letter agency has to do is activate the debug mode.

Nobody takes out debug-mode just in case it breaks something and you have no means of debugging.

[Adam 1]

Plenty of room at the table

> At various times, random number failures have hit iOS, Windows XP, Raspberry Pi, and famously, RSA

Don't worry, this is one game that Debian can play too. That meant only 32,768 possible seed values for the RNG and it was unnoticed for nearly 18 months.

[Anonymous Coward]

Is this it?...

I'm guessing that this patent application (published 6 July 2017) might be something to do with it

WO2017116055

[TheElder]

Random numbers?

Quantum Random Number Generators can be very cheap and anyone can build one.

Geiger counter particle detector.

[Paul Uszak]

I'll be queuing to buy one if I can just prove that mine's not simply outputting SHA256[NSAKey || CPUId, time_t].