Photon Science Center, Graduate School of Engineering, The University of Tokyo, Bunkyo-ku, Tokyo 113-8656, Japan.
1] E. L. Ginzton Laboratory, Stanford University, Stanford, California 94305, USA [2] National Institute of Informatics, Hitotsubashi, Chiyoda-ku, Tokyo 101-8430, Japan.
Nature. 2014 May 22;509(7501):475-8. doi: 10.1038/nature13303.
Quantum cryptography exploits the fundamental laws of quantum mechanics to provide a secure way to exchange private information. Such an exchange requires a common random bit sequence, called a key, to be shared secretly between the sender and the receiver. The basic idea behind quantum key distribution (QKD) has widely been understood as the property that any attempt to distinguish encoded quantum states causes a disturbance in the signal. As a result, implementation of a QKD protocol involves an estimation of the experimental parameters influenced by the eavesdropper's intervention, which is achieved by randomly sampling the signal. If the estimation of many parameters with high precision is required, the portion of the signal that is sacrificed increases, thus decreasing the efficiency of the protocol. Here we propose a QKD protocol based on an entirely different principle. The sender encodes a bit sequence onto non-orthogonal quantum states and the receiver randomly dictates how a single bit should be calculated from the sequence. The eavesdropper, who is unable to learn the whole of the sequence, cannot guess the bit value correctly. An achievable rate of secure key distribution is calculated by considering complementary choices between quantum measurements of two conjugate observables. We found that a practical implementation using a laser pulse train achieves a key rate comparable to a decoy-state QKD protocol, an often-used technique for lasers. It also has a better tolerance of bit errors and of finite-sized-key effects. We anticipate that this finding will give new insight into how the probabilistic nature of quantum mechanics can be related to secure communication, and will facilitate the simple and efficient use of conventional lasers for QKD.
量子密码学利用量子力学的基本定律提供了一种安全交换私人信息的方法。这种交换需要一个共同的随机比特序列，称为密钥，由发送方和接收方秘密共享。量子密钥分发（QKD）的基本思想是，任何试图区分编码量子态的尝试都会导致信号的干扰。因此，实现 QKD 协议涉及对实验参数的估计，这些参数受到窃听者干预的影响，这是通过随机采样信号来实现的。如果需要高精度地估计许多参数，那么被牺牲的信号部分就会增加，从而降低协议的效率。在这里，我们提出了一种基于完全不同原理的 QKD 协议。发送方将一个比特序列编码到非正交量子态上，接收方随机指定如何从序列中计算单个比特。窃听者无法了解整个序列，因此无法正确猜测比特值。通过考虑两个共轭可观测量的量子测量之间的互补选择，计算了安全密钥分发的可实现速率。我们发现，使用激光脉冲序列的实际实现可以达到与诱骗态 QKD 协议相当的密钥速率，诱骗态 QKD 协议是一种常用于激光的常用技术。它还具有更好的抗比特错误和有限大小密钥效应的能力。我们预计，这一发现将为量子力学的概率性质如何与安全通信相关提供新的见解，并促进传统激光在 QKD 中的简单高效使用。
