Integrated and Real-Time Quantum Randomness Generation for Secure Applications

A key challenge in modern communication is ensuring information security among parties. However, many existing methods are showing their limits and their assumptions are starting to be questioned. Quantum mechanics offers a way to overcome these challenges by enabling protocols that can securely distribute encryption keys over untrusted channels (QKD) and generate truly random numbers (QRNG) that are essential for most cryptographic schemes. In the past years, many experiments have shown the benefits of quantum-based systems over conventional ones. The next step is to make them practical for real-world scenarios. The content of this thesis is divided into three sections and focuses on different practical aspects concerning the realization of QRNG devices and their exploitation by QKD systems. The first part present the implementation in two different integrated photonic platforms of a source-device independent QRNG protocol relying on the POVMs of an optical heterodyne receiver. The former is a custom-made chip engineered for QRNG applications. The second is a commercial device, and the scope was to understand the benefits of exploiting a well-established industry with decades of expertise in integrated photonics for quantum purposes. We also analysed the non-idealities of optical heterodyne receivers, especially the temporal correlations between samples caused by their non-flat frequency response, the fact that they do not make orthogonal measurements, and how they affect QRNG devices. In the second part, we studied the implementation of the binary Toeplitz hashing random extractor. Despite not being the best in terms of computational complexity, it efficiently exploits the resources made available by hardware platforms. In this work, we borrowed some techniques from the high-performance computing field to improve the performance of the binary Toeplitz extractor. The last part of this thesis will report on a field trial experiment of a QKD system that we designed to be compact, easy to deploy, and compatible with the existing fibre network for telecommunication. We will also demonstrate a real-time QRNG system that can produce and stream random numbers through a standard gigabit Ethernet port, making it suitable for integration with our QKD systems. Furthermore, we will describe a flexible hardware architecture that enables us to control various aspects of the QKD and QRNG systems mentioned above and a computer algorithm that we developed to efficiently use the randomness generated by the latter when biasing random bit-strings (which are needed by the efficient three-states BB84 plus decoy QKD protocol implemented by our QKD devices).