Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

IEEE Transactions on Quantum Engineering

Q-REFINE: A Quantum Datacenter Design Optimization Framework

Chenliang Tian, Raj Kumar Jain, Reza Nejabati, Eneet Kaur, Ramana Rao Kompella, Aiman Erbad, Mohamed M. Abdallah

Published Jan 1, 2026• 0 citations
Publisher

Abstract

Quantum datacenter (QDC) architectures must support communication among quantum processing units (QPUs) under limited communication qubits, optical switching and measurement resources, probabilistic entanglement generation, and physical-layer noise. This paper presents Q-REFINE, a framework for Quantum Resource Exploration via Factorial & Iterative N-factor Experiments. Its measure is the completion rate, the rate at which requests are served with at least a threshold fidelity. For a given QDC topology, Q-REFINE asks which physical-layer factors and their combinations change this rate most, and over what range improving each factor still pays off. Q-REFINE combines a discrete event simulator, controllable inter-QPU request workloads, topology instances with explicit switches and measurement units, and a design of experiments (DOE). We apply it to Fat-tree, Clos, QFly, BCube and DCell. Load selection fixes, with every factor at its best level, the smallest concurrency limit whose completion rate is within 5% of the best while at most 5% of requests go unserved. At that limit, a full-factorial DOE finds which factor combinations serve any request and how much each factor raises the rate, and one-factor-at-a-time sweeps show over what range each improvement pays off. Emitted-pair fidelity, Hong–Ou–Mandel visibility and memory coherence time jointly decide whether any request is served. The emission efficiency raises the rate most, 8.4–9.5 times between its two DOE levels, and a higher attempt rate pays off only with a short communication-qubit cooldown time. Although we use a particular workload, performance metric, and configurations, the framework applies to any workload, metric(s), or configurations.

DOI
10.1109/tqe.2026.3744686

Related papers

Performance Analysis of QAOA Across Distributed Quantum Network Topologies Using SwitchQNet

Samanvay Sharma, Siyuan Niu, Amin Taherkhani +2 more

Quantum data-center (QDC) architectures aim to scale distributed quantum computing (DQC) by interconnecting multiple quantum processing units (QPUs), but their performance depends strongly on how algorithmic communication patterns interact with entanglement generation, switch reconfiguration, and network topology. This paper studies the Quantum Approximate Optimization Algorithm (QAOA) as a graph-structured optimization workload for QDC-based distributed quantum computing. We adapt QAOA to SwitchQNet, a distributed quantum compiler framework that schedules communication and entanglement generation over switch-based QDC networks, by adding a routing generator that converts graph-dependent two-qubit cost interactions into remote-CX communication requests across QPUs. Using this extension, we evaluate QAOA instances across Clos, fat-tree, and spine-leaf topologies, measuring communication latency, EPR-pair overhead, EPR wait time, retry overhead, and sensitivity to buffer size, look-ahead depth, communication-qubit count, EPR latency, and EPR fidelity assumptions. The results show that QAOA obtains modest but consistent latency reductions, highlighting its value as a diagnostic benchmark for studying the interaction between algorithm structure, entanglement management, and quantum-network architecture.

quant-pharXiv:2607.23407•Jul 28, 2026•0 citations

Benchmarking Quantum Data Center Architectures: A Performance and Scalability Perspective

Shahrooz Pouryousef, Eneet Kaur, Hassan Shapourian +3 more

Scalable distributed quantum computing (DQC) has motivated the design of multiple quantum data-center (QDC) architectures that overcome the limitations of single quantum processors through modular interconnection. While these architectures adopt fundamentally different design philosophies, their relative performance under realistic quantum hardware constraints remains poorly understood. In this paper, we present a systematic benchmarking study of four representative QDC architectures-QFly, BCube, Clos, and Fat-Tree-quantifying their impact on distributed quantum circuit execution latency, resource contention, and scalability. Focusing on quantum-specific effects absent from classical data-center evaluations, we analyze how optical-loss-induced Einstein-Podolsky-Rosen (EPR) pair generation delays, coherence-limited entanglement retry windows, and contention from teleportation-based non-local gates shape end-to-end execution performance. Across diverse circuit workloads, we evaluate how architectural properties such as path diversity and path length, and shared BSM (Bell State Measurement) resources interact with optical-switch insertion loss and reconfiguration delay. Our results show that distributed quantum performance is jointly shaped by topology, scheduling policies, and physical-layer parameters, and that these factors interact in nontrivial ways. Together, these insights provide quantitative guidance for the design of scalable and high-performance quantum data-center architectures for DQC.

quant-pharXiv:2601.01353•Jan 6, 2026•8 citations

ReFINE: Scheduling of Distillation and Coding for Rate-Fidelity Tradeoff in Quantum Networks

Narges Alavisamani, Matthieu Bloch, Moinuddin Qureshi

In quantum networks, nodes are connected via sharing of Einstein-Podolsky-Rosen (EPR) pairs, ideally with high fidelity and high rate. However, the fidelity of EPR pairs degrades due to imperfect generation and decoherence errors. Entanglement Distillation is a method that increases the fidelity but operates probabilistically and may destroy all involved EPR pairs upon failure. This failure reduces available EPR pairs for application use, thereby decreasing the service rate. Quantum Error Correction (QEC) is another mechanism to protect EPR pairs against error by forming what we term as Coding-Enhanced Memory (CEM). While effective, CEM requires extra time and resources to form the code, which also reduces the service rate. Existing methods often use static combinations of distillation and CEM, ignoring demand variations. This results in a low service rate without significant fidelity gain. Limited resources together with this rate-fidelity tradeoff make it essential to schedule when to run distillation, form CEM, or serve requests. We propose ReFINE, a demand-aware preemptive scheduler that based on application requirements either serves an available EPR pair immediately or preserves it in CEM. This selective use of CEM, only when needed, enables a better balance for rate-fidelity tradeoff than always using CEM. Between request arrivals, ReFINE either schedules distilling EPR pairs or forming CEM to protect distilled pairs, following one of the three priority policies: ReFINE-D (Distillation-First) first generates EPR pairs for distillation and then forms the CEM, prioritizing service rate. ReFINE-M (Memory-First) first forms the CEM, then generates the EPR pairs for distillation, prioritizing fidelity. ReFINE-C (Concurrent) performs both distillation and CEM formation concurrently, balancing between fidelity and service rate.

quant-pharXiv:2609.21152•Sep 21, 2026•0 citations

HI-QDC: An Isometric Modular Scalable Architecture for Quantum Data Centers

Mohadeseh Azari, Anoosha Fayyaz, Amy Babay +3 more

Server-centric quantum data-center architectures offer scalability by distributing communication tasks across QPUs rather than concentrating complexity in a centralized switching core. However, scaling such architectures increases the path length, the number of Bell-state measurements, and the loss of end-to-end fidelity. We ask whether the path diversity of a server-centric topology can be converted into a mechanism for preserving not only rate, but also fidelity. We study this through end-to-end purification as a fidelity-restoration mechanism. First, in a black-box model, we determine the minimum number of raw end-to-end Werner-state copies, each carrying the degraded Werner parameter of a distance-ell path, that purification must consume to recover a single copy matching an elementary link, comparing recursive 2-to-1 and optimized nested r-to-1 purification. Second, we instantiate these requirements in a probabilistic BCube architecture, whose edge-disjoint path diversity supplies the raw copies. Because purification imposes a lower bound on input fidelity, no path redundancy can raise the output below this threshold, which limits scaling. To address it we present the Hop-Independent Quantum Data Center (HI-QDC), an isometric, modular, scalable architecture in which purification transforms end-to-end entanglement across a module into an effective link-level resource for the inter-module topology. Our results identify the regimes in which a BCube module, under end-to-end purification, preserves both the fidelity and the yield of an elementary link. The module then hides its internal hop count and acts as an effective elementary link for a higher-level network. Thus topology supplies the path multiplicity purification requires, while purification converts it into fidelity recovery, enabling recursively scalable quantum data-center networks.

quant-pharXiv:2607.20633•Jul 24, 2026•0 citations

Efficient entanglement scheduling in quantum data centers

Shahrooz Pouryousef, Eneet Kaur, Don Towsley +1 more

Modular quantum computing scales quantum systems by connecting Quantum Processing Units (QPUs) through a quantum network. A key challenge in such architectures is efficiently scheduling entanglement generation to enable distributed circuit execution. This work investigates how architectural factors—such as network topology, switch placement, and communication qubit allocation—impact entanglement scheduling and resource utilization. Through simulation, we evaluate static and dynamic scheduling strategies under realistic constraints including probabilistic entanglement generation, limited communication qubits, reconfiguration delays, and qubit coherence time. Our results show that dynamic scheduling performs better in highly connected architectures, while sparse topologies lead to contention and entanglement loss. We also profile Bell State Measurement (BSM) demand to identify architectural bottlenecks and inform provisioning. These insights highlight the critical role of architecture in enabling scalable, efficient distributed quantum computing.

Quantum Computing, Communication, and Simulation•Mar 5, 2026•0 citations

Towards a High-Performance Quantum Data Center Network Architecture

Yufeng Xin, Liang Zhang

Quantum Data Centers (QDCs) are needed to support large-scale quantum processing for both academic and commercial applications. While large-scale quantum computers are constrained by technological and financial barriers, a modular approach that clusters small quantum computers offers an alternative. This approach, however, introduces new challenges in network scalability, entanglement generation, and quantum memory management. In this paper, we propose a three-layer fat-tree network architecture for QDCs, designed to address these challenges. Our architecture features a unique leaf switch and an advanced swapping spine switch design, optimized to handle high volumes of entanglement requests as well as a queue scheduling mechanism that efficiently manages quantum memory to prevent decoherence. Through queuing-theoretical models and simulations in NetSquid, we demonstrate the proposed architecture's scalability and effectiveness in maintaining high entanglement fidelity, offering a practical path forward for modular QDC networks.

International Conference on Communications•Jun 8, 2025•2 citations
Quantum datacenter (QDC) architectures must support communication among quantum processing units (QPUs) under limited communication qubits, optical switching and measurement resources, probabilistic entanglement generation, and physical-layer noise. This paper presents Q-REFINE, a framework for Quantum Resource Exploration via Factorial & Iterative N-factor Experiments. Its measure is the completion rate, the rate at which requests are served with at least a threshold fidelity. For a given QDC topology, Q-REFINE asks which physical-layer factors and their combinations change this rate most, and over what range improving each factor still pays off. Q-REFINE combines a discrete event simulator, controllable inter-QPU request workloads, topology instances with explicit switches and measurement units, and a design of experiments (DOE). We apply it to Fat-tree, Clos, QFly, BCube and DCell. Load selection fixes, with every factor at its best level, the smallest concurrency limit whose completion rate is within 5% of the best while at most 5% of requests go unserved. At that limit, a full-factorial DOE finds which factor combinations serve any request and how much each factor raises the rate, and one-factor-at-a-time sweeps show over what range each improvement pays off. Emitted-pair fidelity, Hong–Ou–Mandel visibility and memory coherence time jointly decide whether any request is served. The emission efficiency raises the rate most, 8.4–9.5 times between its two DOE levels, and a higher attempt rate pays off only with a short communication-qubit cooldown time. Although we use a particular workload, performance metric, and configurations, the framework applies to any workload, metric(s), or configurations.
Quantum data-center (QDC) architectures aim to scale distributed quantum computing (DQC) by interconnecting multiple quantum processing units (QPUs), but their performance depends strongly on how algorithmic communication patterns interact with entanglement generation, switch reconfiguration, and network topology. This paper studies the Quantum Approximate Optimization Algorithm (QAOA) as a graph-structured optimization workload for QDC-based distributed quantum computing. We adapt QAOA to SwitchQNet, a distributed quantum compiler framework that schedules communication and entanglement generation over switch-based QDC networks, by adding a routing generator that converts graph-dependent two-qubit cost interactions into remote-CX communication requests across QPUs. Using this extension, we evaluate QAOA instances across Clos, fat-tree, and spine-leaf topologies, measuring communication latency, EPR-pair overhead, EPR wait time, retry overhead, and sensitivity to buffer size, look-ahead depth, communication-qubit count, EPR latency, and EPR fidelity assumptions. The results show that QAOA obtains modest but consistent latency reductions, highlighting its value as a diagnostic benchmark for studying the interaction between algorithm structure, entanglement management, and quantum-network architecture.
Scalable distributed quantum computing (DQC) has motivated the design of multiple quantum data-center (QDC) architectures that overcome the limitations of single quantum processors through modular interconnection. While these architectures adopt fundamentally different design philosophies, their relative performance under realistic quantum hardware constraints remains poorly understood. In this paper, we present a systematic benchmarking study of four representative QDC architectures-QFly, BCube, Clos, and Fat-Tree-quantifying their impact on distributed quantum circuit execution latency, resource contention, and scalability. Focusing on quantum-specific effects absent from classical data-center evaluations, we analyze how optical-loss-induced Einstein-Podolsky-Rosen (EPR) pair generation delays, coherence-limited entanglement retry windows, and contention from teleportation-based non-local gates shape end-to-end execution performance. Across diverse circuit workloads, we evaluate how architectural properties such as path diversity and path length, and shared BSM (Bell State Measurement) resources interact with optical-switch insertion loss and reconfiguration delay. Our results show that distributed quantum performance is jointly shaped by topology, scheduling policies, and physical-layer parameters, and that these factors interact in nontrivial ways. Together, these insights provide quantitative guidance for the design of scalable and high-performance quantum data-center architectures for DQC.
In quantum networks, nodes are connected via sharing of Einstein-Podolsky-Rosen (EPR) pairs, ideally with high fidelity and high rate. However, the fidelity of EPR pairs degrades due to imperfect generation and decoherence errors. Entanglement Distillation is a method that increases the fidelity but operates probabilistically and may destroy all involved EPR pairs upon failure. This failure reduces available EPR pairs for application use, thereby decreasing the service rate. Quantum Error Correction (QEC) is another mechanism to protect EPR pairs against error by forming what we term as Coding-Enhanced Memory (CEM). While effective, CEM requires extra time and resources to form the code, which also reduces the service rate. Existing methods often use static combinations of distillation and CEM, ignoring demand variations. This results in a low service rate without significant fidelity gain. Limited resources together with this rate-fidelity tradeoff make it essential to schedule when to run distillation, form CEM, or serve requests. We propose ReFINE, a demand-aware preemptive scheduler that based on application requirements either serves an available EPR pair immediately or preserves it in CEM. This selective use of CEM, only when needed, enables a better balance for rate-fidelity tradeoff than always using CEM. Between request arrivals, ReFINE either schedules distilling EPR pairs or forming CEM to protect distilled pairs, following one of the three priority policies: ReFINE-D (Distillation-First) first generates EPR pairs for distillation and then forms the CEM, prioritizing service rate. ReFINE-M (Memory-First) first forms the CEM, then generates the EPR pairs for distillation, prioritizing fidelity. ReFINE-C (Concurrent) performs both distillation and CEM formation concurrently, balancing between fidelity and service rate.
Server-centric quantum data-center architectures offer scalability by distributing communication tasks across QPUs rather than concentrating complexity in a centralized switching core. However, scaling such architectures increases the path length, the number of Bell-state measurements, and the loss of end-to-end fidelity. We ask whether the path diversity of a server-centric topology can be converted into a mechanism for preserving not only rate, but also fidelity. We study this through end-to-end purification as a fidelity-restoration mechanism. First, in a black-box model, we determine the minimum number of raw end-to-end Werner-state copies, each carrying the degraded Werner parameter of a distance-ell path, that purification must consume to recover a single copy matching an elementary link, comparing recursive 2-to-1 and optimized nested r-to-1 purification. Second, we instantiate these requirements in a probabilistic BCube architecture, whose edge-disjoint path diversity supplies the raw copies. Because purification imposes a lower bound on input fidelity, no path redundancy can raise the output below this threshold, which limits scaling. To address it we present the Hop-Independent Quantum Data Center (HI-QDC), an isometric, modular, scalable architecture in which purification transforms end-to-end entanglement across a module into an effective link-level resource for the inter-module topology. Our results identify the regimes in which a BCube module, under end-to-end purification, preserves both the fidelity and the yield of an elementary link. The module then hides its internal hop count and acts as an effective elementary link for a higher-level network. Thus topology supplies the path multiplicity purification requires, while purification converts it into fidelity recovery, enabling recursively scalable quantum data-center networks.
Modular quantum computing scales quantum systems by connecting Quantum Processing Units (QPUs) through a quantum network. A key challenge in such architectures is efficiently scheduling entanglement generation to enable distributed circuit execution. This work investigates how architectural factors—such as network topology, switch placement, and communication qubit allocation—impact entanglement scheduling and resource utilization. Through simulation, we evaluate static and dynamic scheduling strategies under realistic constraints including probabilistic entanglement generation, limited communication qubits, reconfiguration delays, and qubit coherence time. Our results show that dynamic scheduling performs better in highly connected architectures, while sparse topologies lead to contention and entanglement loss. We also profile Bell State Measurement (BSM) demand to identify architectural bottlenecks and inform provisioning. These insights highlight the critical role of architecture in enabling scalable, efficient distributed quantum computing.
Quantum Data Centers (QDCs) are needed to support large-scale quantum processing for both academic and commercial applications. While large-scale quantum computers are constrained by technological and financial barriers, a modular approach that clusters small quantum computers offers an alternative. This approach, however, introduces new challenges in network scalability, entanglement generation, and quantum memory management. In this paper, we propose a three-layer fat-tree network architecture for QDCs, designed to address these challenges. Our architecture features a unique leaf switch and an advanced swapping spine switch design, optimized to handle high volumes of entanglement requests as well as a queue scheduling mechanism that efficiently manages quantum memory to prevent decoherence. Through queuing-theoretical models and simulations in NetSquid, we demonstrate the proposed architecture's scalability and effectiveness in maintaining high entanglement fidelity, offering a practical path forward for modular QDC networks.