Philipp Koller

Quantum communication over global distances relies on the faithful distribution of entanglement, which can be achieved either via satellites or through fibre-based quantum networks with quantum memories. While satellite links are fundamentally constrained in coverage and by cost, memory-based networks offer a scalable alternative—provided their core components reach sufficient technological maturity. In this work, we investigate the loading process of spin-based quantum memories, focusing on the noise introduced when attempts to store incoming photonic qubits fail. While decoherence and readout errors have been the subject of substantial prior research, loading-induced noise—in particular the effect of lost heralding photons after spin interaction—remains largely unexplored in repeater protocol design. We close this gap by providing a detailed analysis of this specific source of noise, which has so far received limited attention. To mitigate the loading-induced noise, we propose a reinitialisation strategy that resets the memory after a maximal cutoff time if loading fails. We model the associated noise channel, quantify the trade-offs between rate and fidelity, and optimise the strategy across a range of link distances for memory-assisted MDI-QKD. The results demonstrate that protocol-level control of memory loading substantially improves entanglement distribution performance, paving the way for more reliable near-term quantum networks.