To date all realized MDI QKD systems have relied on attenuated laser pulses assisted by the decoy state analysis to exchange qubits and generate a secure key [XM05]. This has the advantage that inexpensive and reliable off the shelf components can be used, however, the decoy state analysis results in a rather large post-processing overhead. Moreover, for typical values of the security parameter the bound that the decoy state analysis provides is only statistically significant for very large raw key data sets. If the acquisition time is limited, e.g. duration of a satellite fly-over, then the decoy state analysis will not yield secure keys with enough statistical weight. In this case pure single-photon sources are more practical. It is in principle possible to generate pure single-photons at room-temperature from a heralded source with multiplexing [MGP17]. This approach is, however, only practical if it can be developed in a compact integrated platform. Otherwise single emitter based sources are the most obvious choices for generating pure-single photons. In this category, quantum dot based sources are particularly interesting because they are solid-state based and thus (in principle) easily packaged and integrated [KDJ15, AWE17]. In addition the emission wavelengths can be tuned (either during growth or actively during operation) [PS17, JJB18] and the few GHz bandwidth is well suited for fast QKD systems including the detectors. One caveat for quantum dots is that photons emitted from different sources are not completely indistinguishable due to spectral jitter. In the worst case this can be overcome with filtering to the lifetime limited spectral linewidth. As a note, colour centres in diamond are interesting alternatives for single-emitter sources that could offer room T operation [SW18].

In the QCloudLab we are developing a quantum dot based MDI-QKD system in collaboration with the National Research Council of Canada for the fabrication of quantum dots. The NRC quantum dots are based on a InAsP layer embedded in InP nanowires. These quantum dots have shown very large collection efficiencies [GB14] and are currently being integrated with waveguide structures to produce single-photon emitters that could easily be transported and installed in a cryostat. Currently the optimal emission wavelength is at around 900 nm, i.e. not well suited for neither free space or fibre long-distance transmission. Thus, a proof-of-principle demonstration over shorter fibres is a first step. In parallel, quantum dot emission around 1300 nm (telecom O-band) [SH18] will be further studied with the aim of improving the collection efficiency. Finally, we will collaborate on efforts to perform frequency conversion in commercial waveguide-based non-linear crystals [AD18].