Device-independent randomness expansion experiments

Short answer: Device-independent (DI) randomness generation produces random bits whose security does not depend on trusting the devices, but it needs a Bell inequality violation. A 2018 Nature paper reports DI random-number generation with a loophole-free Bell test, and a 2021 Physical Review Letters paper and a Nature Physics paper each report DI randomness expansion, meaning more certified random bits out than entropy in. These are laboratory experiments, not products.

What device-independent means

The Nature Physics paper says DI randomness generation does not require trusted devices and so provides strong guarantees of the security of the output, but it comes at the price of requiring the violation of a Bell inequality. The 2018 Nature paper describes DI random-number generation based on the loophole-free violation of a Bell inequality as producing genuine, unpredictable randomness without assumptions about the inner workings of the devices.

The 2018 loophole-free experiment

The Nature authors say earlier DI demonstrations were not provably secure against the most general adversaries or did not close the locality loophole of the Bell test. They report DI random-number generation secure against quantum and classical adversaries. They use quantum optical technology to create, modulate and detect entangled photon pairs, with an efficiency of more than 78 per cent from creation to detection at a distance of about 200 metres, which they say greatly exceeds the threshold for closing the detection loophole. They choose measurement settings independently and randomly, and ensure space-like separation between the measurement events, to close the locality loophole. After applying a large Toeplitz-matrix hashing technique they report 62.469 million quantum-certified random bits in 96 hours.

Expansion experiments

The Nature Physics authors say that although randomness had been generated in earlier experiments, the amount consumed was too high to certify expansion under existing theory. They report a Bell test setup with a single-photon detection efficiency of around 84% and a spot-checking protocol. They report a net gain of 2.57 times 10 to the 8 certified bits, over a run of 19.2 hours, an average rate of 13,527 bits per second, and use the entropy accumulation theorem to establish security against quantum adversaries.

The 2021 Physical Review Letters paper reports what it calls the first experimental realization of DI quantum randomness expansion secure against quantum side information, established through quantum probability estimation. The authors report generating 5.47 times 10 to the 8 quantum-proof random bits while consuming 4.39 times 10 to the 8 bits of entropy, a gain of 1.08 times 10 to the 8 bits, at a latency of about 13.1 hours.

Limits

Each figure comes from a specific lab setup and run time, and the three papers use different security proofs and assumptions. The Nature Physics authors anticipate further improvements to push device-independence towards commercial viability, which implies it is not there yet. The papers here do not describe a commercial product.

FAQ

Does DI randomness need trusted devices?

No. That is its defining feature, per the Nature Physics paper, but it requires a Bell inequality violation.

What is randomness expansion?

Generating a longer random sequence from a short one. The Physical Review Letters paper says this is viable in quantum mechanics but not allowed classically.

How many bits did the experiments report?

2.57 times 10 to the 8 net certified bits in one, and 1.08 times 10 to the 8 bits of expansion in another.

How long did the runs take?

19.2 hours in the Nature Physics experiment, and a latency of about 13.1 hours in the Physical Review Letters one.

Is it commercial?

The authors of one paper anticipate further improvements towards commercial viability, so it is a research result.