Finite key analysis in quantum cryptography

In view of experimental realization of quantum key distribution schemes, the study of their efficiency becomes as important as the proof of their security. The latter is the subject of most of the theoretical work about quantum key distribution, and many important results such as the proof of unconditional security have been obtained. The efficiency and also the robustness of quantum key distribution protocols against noise can be measured by figures of merit such as the secret key rate (the fraction of input signals that make it into the key) and the threshold quantum bit error rate (the maximal error rate such that one can still create a secret key). It is important to determine these quantities because they tell us whether a certain quantum key distribution scheme can be used at all in a given situation and if so, how many secret key bits it can generate in a given time. However, these figures of merit are usually derived under the ''infinite key limit'' assumption, that is, one assumes that an infinite number of quantum states are send and that all sub-protocols of the scheme (in particular privacy amplification) are carried out on these infinitely large blocks. Such an assumption usually eases the analysis, but also leads to (potentially) too optimistic values for the quantities in question. In this thesis, we are explicitly avoiding the infinite key limit for the analysis of the privacy amplification step, which plays the most important role in a quantum key distribution scheme. We still assume that an optimal error correction code is applied and we do not take into account any statistical errors that might occur in the parameter estimation step. Renner and coworkers derived an explicit formula for the obtainable key rate in terms of Renyi entropies of the quantum states describing Alice's, Bob's, and Eve's systems. This results serves as a starting point for our analysis, and we derive an algorithm that efficiently computes the obtainable key rate for any finite number of input signals, without making any approximations. As an application, we investigate the so-called ''Tomographic Protocol'', which is based on the Six-State Protocol and where Alice and Bob can obtain the additional information which quantum state they share after the distribution step of the protocol. We calculate the obtainable secret key rate under the assumption that the eavesdropper only conducts collective attacks and give a detailed analysis of the dependence of the key rate on various parameters: The number of input signals (the block size), the error rate in the sifted key (the QBER), and the security parameter. Furthermore, we study the influence of multi-photon events which naturally occur in a realistic implementation (orig.)