A nonblocking multistage switching network for distributed quantum computing

Abstract

Quantum computing, utilizing the unique properties of quantum mechanics, has the potential to revolutionize various fields. However, current quantum processors face challenges in scaling the number of qubits, limiting their practical applications. In response, Distributed Quantum Computing (DQC) has emerged as a promising paradigm where multiple interconnected Quantum Processing Units (QPUs) collaborate to execute quantum circuits. In this paper, we focus on designing networks to interconnect QPUs for the implementation of DQC. We find that in real-world experiments and systems, the photon collection and coupling efficiency is low, leading to significant performance degradation in direct connection networks. To address this limitation, we propose a novel multistage switching network tailored for DQC, which has low system complexity and high entanglement generation rates. The proposed switching network comprises log2(N ) stages and N/2 binary switches at each stage, where N represents the number of QPUs. We prove that the proposed network is nonblocking and develop an efficient routing algorithm with a time complexity of O(N log(N )). Additionally, we show the success probability of entanglement generation in the proposed switching network. Extensive simulations demonstrate that our network significantly outperforms the highly efficient circuit-switching Beneš network and three direct connection networks.