Quantum Key Distribution
Quantum key distribution (QKD) is the process of establishing a secret shared key between two parties, traditionally named Alice and Bob. The security is based on the laws of quantum mechanics, whereas in classical schemes security relies only on the unproven lack of efficient mathematical algorithms.
We have developed a QKD protocol which is particularly suitable for atmospheric transmission: we employ a local oscillator to perform an optical homodyne detection of weak coherent signal states. Alice utilizes polarization states to combine signal and local oscillator in a single beam. As a consequence, Bob’s detection is very efficient and perfectly shielded against any stray light.
We have experimentally demonstrated the feasibility of this protocol over a distance of 100m on the roof of the MPL building [1,2]. As the next step, we are now establishing a link of length 1.6km between the MPL and the University computer centre.
In contrast to entanglement based QKD schemes, our setup is of the prepare & measure type. Nevertheless Alice and Bob can model their correlations as if they shared an entangled state . Bounds for sophisticated eavesdropping strategies like the manipulation of the local oscillator are derived in ongoing work with the group of Norbert Lütkenhaus. We also experimentally investigated the influence of a simple eavesdropping attack proved to be optimal for a certain class of QKD schemes .
Additionally, we have adapted our successful free-space cryptography scheme to fiber channels. In this setup, we were able to witness non-classical correlations between the sender and the receiver, so called effective entanglement .
ID Quantique to identify loopholes and exploit vulnerabilities by simulating and performing attacks on it. The principal idea is to secure the system in a regenerative sense, so we also propose patches and countermeasures, wherever possible. In this project, we collaborate with the Quantum Hacking group situated at IQC Waterloo in Canada.
So far, we have been able to successfully compromise the security of the system by launching a variety of faked-state attacks. Using tailored bright illumination to blind the detectors , Eve can dictate the measurements performed by Bob and thus, obtain a perfect copy of the raw key while remaining virtually undetected. A more sophisticated version of this employs heating the APDs with bright illumination. Since this thermal blinding  can be done well in advance of the actual key-exchange frames, it is as such harder to catch. We recently also tested another method to control the detection events by sending bright pulses outside the gated region in Bob . A video abstract explaining the basic concepts behind this attack has been made by our team (see right side).
Video abstract: After-gate attack on a commercial quantum cryptosystem
We have also been able to exploit a vulnerability in the implementation of a vital calibration sequence of this commercial QKD system that allows Eve to induce a detector efficiency mismatch (refer figure 2). We demonstrate an optimized faked-state attack on such a hacked system that would cause a QBER below 7% without any reduction in Bob’s expected detection rate for a large range of expected channel transmissions. Most recently, a generalized version of tailored bright illumination that exploits superlinear characteristics of single-photon detectors based on APDs and superconducting nanowires has also been demonstrated .
Due to the linearity of quantum mechanics it is impossible to perfectly clone an arbitrary quantum state. It is however interesting to find the best possible cloning device, i.e. an optimal distributor of quantum information. Our group has already implemented several quantum cloners for coherent states [1–3] that operate at the quantum limit.
In all these implementations, the clones are generated by first amplifying and subsequently splitting the input state. Thereby, the quality of the clones is limited by the inevitable noise penalty imposed by the amplification process. However, this noise penalty can be largely suppressed in a probabilistic setting.
In a recent experiment we could show that the cloning fidelity between a coherent input state and its clones can be pushed beyond the (deterministic) quantum limit in a probabilistic scheme . In the conducted experiment, a schematic of which is shown on the right, coherent states from a phase-covariant alphabet were first subject to a specifically tailored random displacement. Subsequently, a photon number resolving measurement on a small tapped-off part of the state is performed and the remaining part of the state is post-selected based on the outcome of this measurement. This results in a probabilistic noiseless amplification of the state and enables the generation of clones with qualities beyond the deterministic quantum limit.
In the continuous variable regime, the bounds imposed by the laws of quantum mechanics are formulated in terms of ensemble statistics and hence are of a statistical nature. Consequently, these laws need not hold true for every individual observation of a quantum state.