We present an in-depth analysis regarding the error resistance and optimization of our all-optical Bell measurement and ultrafast long-distance quantum communication scheme proposed by Ewert, Bergmann, and van Loock [Phys. Rev. Lett. 117, 210501 (2016)]. In order to promote our previous proposal from loss to fault tolerance, we introduce a general and compact formalism that can also be applied to other related schemes {including non-all-optical ones such as that of Muralidharan et al. [Phys. Rev. Lett. 112, 250501 (2014)]}. With the help of this new representation we show that our communication protocol does not only counteract the inevitable photon loss during channel transmission, but is also able to resist common experimental errors such as Pauli-type errors (bit and phase flips) and detector inefficiencies (losses and dark counts). Furthermore, we demonstrate that on the physical level of photonic qubits the choice of the standard linear optical Bell measurement with its limited efficiency is optimal for our setting in the sense that, apart from their potential use in state preparation, more advanced Bell measurements yield only a small decrease in resource consumption. We devise two state generation schemes that provide the required ancillary encoded Bell states (quasi-)on-demand at every station. The schemes are either based on nonlinear optics or on linear optics with multiplexing and exhibit resource costs that scale linearly or less than quadratic with the number of photons per encoded qubit, respectively. Finally, we show that it is possible to operate our communication scheme with on-off detectors instead of employing photon-number-resolving detectors.

• Received 4 November 2016

DOI:https://doi.org/10.1103/PhysRevA.95.012327