1208.2106 (Osamu Hirota)

Incompleteness and Limit of Quantum Key Distribution Theory    [PDF]

Osamu Hirota

It is claimed in the many papers that a trace distance ($d$) guarantees the universal composition security in quantum key distribution (QKD). In this introduction paper, at first, it is explicitly explained what is the main misconception in the claim of the unconditional security for QKD theory. In general terms, the cause of the misunderstanding on the security claim is the Lemma in the paper of Renner. It suggests that the generation of the perfect random key is assured by the probability $ (1-d)$, and its failure probability is $d$. Thus, it concludes that the generated key provides the perfect random key sequence when the protocol suceeds. So the QKD provides perfect secrecy to the one time pad. This is the reason for the composition claim. However, the quantity of the trace distance (or variational distance) is not the probability for such an event. If $d $ is not small enough, always the generated key sequence is not uniform. Now one needs the reconstruction of the evaluation of the trace distance if one wants to use it. One should first go back to the indistinguishability theory in the computational complexity based, and to clarify the meaning of the value of the variational distance. In addition, the same analysis for the information theoretic case is necessary. The recent serial papers by H.P.Yuen have given the answer on such questions. In this paper, we show more concise description of Yuen's theory, and clarify that the recent upper bound theories for the trace distance by Tomamichel et al and Hayashi et al are constructed based on the reasoning of Renner and it is unsuitable as the analysis for information theoretic security. Finally, we introduce a macroscopic quantum communication with different principle to replace Q-bit QKD.

View original: http://arxiv.org/abs/1208.2106