Improving semi-device-independent randomness certification by entropy accumulation

Certified randomness guaranteed to be unpredictable by adversaries is central to information security. The fundamental randomness inherent in quantum physics makes certification possible from devices that are only weakly characterised, i.e. requiring little trust in their implementation. It was recently shown that the amount of certifiable randomness can be greatly improved using the so-called Entropy Accumulation Theorem generalised to prepare-and-measure settings. Furthermore, this approach allows a finite-size analysis which avoids assuming that all rounds are independent and identically distributed. Here, we demonstrate this improvement in semi-device-independent randomness certification from untrusted measurements.