Upper bounds on key rates in device-independent quantum key distribution with qudits based on convex-combination attacks

Abstract

Device-independent quantum key distribution (DIQKD) is the pinnacle of secure communication over an untrusted channel. Its security is based solely on the classical data observed by the honest parties attempting to establish a shared secret key. Despite DIQKD’s unrivalled security, real-world implementations are subject to noise, which limits its effective range. Traditionally, DIQKD has been based on measurements performed on an entangled pair of qubits. In this work, we explore the use of higher-dimensional systems as a way to improve its resilience to noise. To do this, we consider convex-combination attacks, which provide easy-to-compute upper bounds on DIQKD key rates. Our results show that using higher-dimensional states only provides a small improvement in resilience to noise, which may not justify the added experimental complexity.