InterQ: Communication-Aware Scheduling Across Modular QPUs with Classical and Quantum Links

Abstract: As quantum computing scales toward practical workloads, future systems are expected to move beyond single monolithic processors toward modular architectures that connect multiple QPUs. Different platforms realize this modularity through different communication models: superconducting systems rely on real-time classical links and dynamic-circuit coordination, trapped-ion systems use photonic interconnects for remote entanglement, and neutral-atom systems provide strong intra-core connectivity with proposed optical links for inter-core communication. This heterogeneity makes communication-aware scheduling essential for shared modular quantum cloud environments. We present InterQ, a communication-aware scheduler for modular QPU architectures with heterogeneous communication models. InterQ jointly considers qubit capacity, placement, parallel execution, and communication-driven dependencies across distributed subcircuits, while enabling adaptive circuit cutting to reduce makespan while balancing fidelity and communication overhead. The framework distinguishes classical-link execution, where measurement and feedforward impose synchronization constraints, from quantum-link execution, where entanglement distribution and state transfer determine coordination cost. Using a unified simulation framework to compare superconducting, trapped-ion, and neutral-atom modular systems, InterQ shows how communication models and scheduler-driven cutting decisions affect throughput, latency, and fidelity. Across evaluated workloads, InterQ exposes an architecture-dependent tradeoff: neutral-atom modular QPUs achieve the highest fidelity, superconducting systems minimize runtime, and trapped-ion systems provide a balanced intermediate profile across fidelity and makespan.

• The paper presents InterQ, a scheduler that incorporates communication constraints as a core factor in modular QPU scheduling.
• It employs architecture-aware partitioning across LO, LOCC, and QComm models to optimize runtime, fidelity, and throughput.
• Experimental results across superconducting, trapped-ion, and neutral-atom systems show significant performance gains over traditional scheduling.

Communication-Aware Scheduling of Modular QPUs: The InterQ Framework

Motivation and Problem Formulation

The paper introduces InterQ, an integrated scheduling framework for modular quantum processing units (QPUs) that incorporates communication as a primary constraint in resource management and scheduling. As quantum computing scales from monolithic devices to modular systems with distributed QPUs, efficient scheduling requires accounting for both qubit placement and inter-module communication (classical and quantum). Key architectural platforms exhibit heterogeneous communication models: superconducting devices utilize real-time classical links and dynamic circuits, trapped-ion systems rely on photonic quantum interconnects for remote entanglement, and neutral-atom modules support strong intra-core connectivity with proposed optical links.

InterQ addresses the scheduling problem by modeling qubit capacity, circuit partitioning, placement, parallel execution, and communication-driven dependencies. It differentiates between local operations (LO), local operations with classical communication (LOCC), and quantum communication (QComm). The scheduler uniquely adapts to the hardware-specific communication modality, recognizing that the scheduling strategy for superconducting, trapped-ion, and neutral-atom modular QPUs requires fundamentally different coordination, fidelity, and throughput tradeoffs.

Execution Model Abstractions

InterQ formalizes three distributed execution models:

• Local Operations (LO): Fragments execute independently, with nonlocal correlations reconstructed offline. Sampling overhead scales exponentially with the number of circuit cuts.
• Local Operations with Classical Communication (LOCC): Real-time measurement and feed-forward establish causal dependencies between fragments. Sampling overhead is reduced to $O(4^k)$, but synchronization constraints limit parallelism and impose additional latency.
• Quantum Communication (QComm): Remote operations are mediated by entanglement distribution. Scheduling must account for stochastic Bell-pair generation, link contention, and fidelity reduction, further constraining throughput.

For LO, fragments are maximally parallel but can incur substantial offline reconstruction costs. For LOCC, synchronization delays and causality restrict parallel scheduling. In QComm, shared quantum links are explicit resources, and contention (e.g., for Bell-pair distribution) leads to nontrivial coordination cost and fidelity degradation.

Scheduler Design and Architecture Awareness

InterQ operates as a constraint-aware optimizer that simultaneously considers qubit capacity, communication modality, group placement, and workload parallelization. It partitions circuits adaptively—only when partitioning reduces scheduling cost and preserves feasibility—using architecture-specific expansions (LO, LOCC, QComm). The scheduler employs a unified cost function that penalizes runtime imbalance, synchronization delay, communication overhead, and sampling cost, weighted by user-tunable parameters.

Placement decisions reflect hardware distinctions:

• Superconducting modular systems: Favor LOCC execution with synchronization-sensitive scheduling, exploiting classical coordination.
• Trapped-ion modular systems: Prefer QComm execution, shaped by remote entanglement generation rates, fidelity, and resource contention.
• Neutral-atom modular systems: Exhibit hierarchical scheduling due to disparate intra-core and inter-core operation costs, necessitating nuanced placement logic.

Grouping rules are execution-model specific, ensuring that classical synchronization chains and quantum-link budgets are never oversubscribed.

Experimental Evaluation and Numerical Results

Evaluation leverages SimPy-based discrete-event simulation, incorporating backend profiles for IBM-style (LOCC), IonQ-style (QComm), and Atomic QComm systems. Benchmarks include MQT Bench, QUEKO, RevLib, and random circuits. Metrics include queue length, queue time, runtime, response time, partition count, remote operation demand, throughput ratio (TRF), fidelity factor (TiIF), and log-probability of success (LPST).

Key findings:

• Superconducting (IBM LOCC): Achieves minimal makespan due to large qubit per module and low classical reconstruction overhead. For circuits that fit backend capacity, LOCC avoids partitioning, minimizing synchronization overhead.
• Trapped-ion (IonQ QComm): Exhibits higher runtime owing to smaller QPU capacity and increased quantum-link contention, but achieves superior LPST (fidelity metric) when Bell-pair generation is robust.
• Neutral-atom QComm: Provides the strongest fidelity profile for partitioned workloads due to high-fidelity interconnects, but incurs moderate runtime.
• InterQ consistently outperforms classic round-robin scheduling in throughput and response time when adaptive partitioning is applied, particularly for large circuits that exceed device capacity.
• For quantum-link execution, the scheduler tracks remote-operation count and teleportation-induced qubit overhead, adjusting partitioning numerically so as not to exceed communication-resource budgets.

Contradictory claims are demonstrated: maximum parallelism in LO mode may be offset by exponential offline reconstruction cost, and reduced sampling overhead in LOCC may be negated by increased synchronization delay and reduced throughput under heavy causal dependencies.

Implications and Future Directions

InterQ establishes that scalable modular quantum computing is fundamentally constrained by communication modality, not just computational capacity. The framework provides a unified scheduling optimization across heterogeneous platforms, allowing direct comparison of classical vs. quantum link costs and exposing critical bottlenecks—such as link contention and fidelity degradation—that shape practical throughput and latency.

Practically, InterQ enables quantum cloud systems to balance makespan, fidelity, and communication-resource utilization as workloads scale and hardware modularity increases. The scheduler’s architecture awareness is crucial for exploiting parallelism, minimizing overhead, and maintaining system-wide performance.

Theoretically, this work demonstrates that communication-driven scheduling is a necessary foundation for distributed quantum computing, especially as quantum error correction and physical qubit requirements push system modularity. In future, as Bell-pair generation rates and quantum interconnect fidelity improve, the scheduling frontier will shift toward even more granular, adaptive partitioning—potentially employing reinforcement learning or look-ahead search for optimal groupings. Integration with advanced circuit cutting, stitching, and hybrid reconstruction techniques will further improve throughput and scalability.

Conclusion

InterQ provides a communication-aware scheduling solution for modular quantum computing, jointly optimizing circuit partitioning, QPU placement, parallel execution, and communication resource allocation across LO, LOCC, and QComm regimes. Strong numerical results reveal platform-dependent tradeoffs: superconducting systems minimize runtime, neutral-atom architectures yield highest fidelity, and trapped-ion systems balance both. The framework highlights the centrality of communication constraints in distributed quantum execution and lays a foundation for future scalable, heterogeneous quantum cloud infrastructures.

For further technical details and numerical results, see "InterQ: Communication-Aware Scheduling Across Modular QPUs with Classical and Quantum Links" (2605.17769).