Quantum Strategies to Overcome Classical Multiplexing Limits

TL;DR

Near-term quantum networks face bottlenecks from low rates and decoherence; the paper derives semiclassical multiplexing limits and introduces quantum multiplexing and multi-server multiplexing to surpass them. It validates the concepts through entanglement-generation and remote-state-preparation use-cases, including scenarios with asymmetric memories and many clients, and a multi-server hub model. The findings show potential superlinear improvements in multi-qubit protocol execution rates, highlighting how distributing tasks across multiple quantum resources can significantly boost network throughput. The work provides a roadmap for combining semiclassical, quantum, and multi-server multiplexing to approach broadband-like quantum networks with existing technologies.

Abstract

Near-term quantum networks face a bottleneck due to low quantum communication rates. This degrades performance both by lowering operating speeds and increasing qubit storage time in noisy memories, making some quantum internet applications infeasible. One way to circumvent this bottleneck is multiplexing: combining multiple signals into a single signal to improve the overall rate. Standard multiplexing techniques are classical in that they do not make use of coherence between quantum channels nor account for decoherence rates that vary during a protocol's execution. In this paper, we first derive semiclassical limits to multiplexing for many-qubit protocols, and then introduce new techniques: quantum multiplexing and multi-server multiplexing. These can enable beyond-classical multiplexing advantages. We illustrate these techniques through three example applications: 1) entanglement generation between two asymetric quantum network nodes (i.e., repeaters or quantum servers with inequal memories), 2) remote state preparation between many end user devices and a single quantum node, and 3) remote state preparation between one end user device and many internetworked quantum nodes. By utilizing many noisy internetworked quantum devices instead of fewer low-noise devices, our multiplexing strategies enable new paths towards achieving high-speed many-qubit quantum network applications.

Figures (6)

• Figure 1: Four schematic illustrations of multiplexing between two parties. a) Remote state preparation between a client equipped with a laser (left) and a node equipped with a quantum memory (right) is performed using an intermediate measurement station as in vanDam2025. Traditional time-bin multiplexing allows many attempts to be performed provided one can fit more pulses in the beamline. b) Quantum multiplexing between two nodes (left and right) with asymmetric available memories enable higher entanglement generation rates. c) Quantum multiplexing between several clients (left) and one server (right) allows the server to provide higher rates of remote state preparation in the face of high demand. d) Multi-server multiplexing improves the rate at which a single client (left) can perform $s$-qubit remote state preparation on a collection of quantum server nodes (right) due to fast inter-server connections, especially for color centers where entanglement generation with a node degrades the quality of qubits already stored in that node.
• Figure 2: Quantum multiplexing between two nodes in a quantum network. (a) Fidelity-rate curves parameterized by $\xi_A$ for symmetric channel efficiencies $\eta_A=\eta_B=\eta$. Solid lines show $M=1$, dashed lines show $M=5$. The dimensionless rate given here is in units of the duration of an EG attempt. (b) Multiplexing gain $m$ as function of $M$ for generating link with minimum fidelity $F_\mathrm{min}=0.95$ between node $A$ and $B$ using $M$-to-$1$ quantum multiplexed heralded entanglement generation. The theoretical maximum in the small $\eta$ limit is $m_\mathrm{max}=\frac{2M}{M+1}$. Any $m\geq 1$ exceeds the classical bound.
• Figure 3: Quantum multiplexing between several client devices held by a single client and a single node in a quantum network with a continuous demand for single-qubit RSP. (a) Fidelity-rate curves parameterized by $\gamma=\eta_c |\alpha|^2$. Solid lines show $M=1$, dashed lines show $M=5$. The dimensionless rate given here is in units of the duration of an RSP attempt. (b) Multiplexing gain $m$ as function of $M$ for RSP with fidelity $F_\mathrm{min}=1-10^{-6}$ using $M$ clients. The theoretical maximum with a perfect server and the classical bound is plotted using the magenta and black dashed lines, respectively. Similar results are observed for modified network protocols where the client devices are held by different users with continuous demand (Fig. \ref{['fig: fidelity rate curves for clients 2']}) as well as one-time demand (Fig. \ref{['fig: single use']}) are calculated and discussed in Appendix B.
• Figure 4: Numerical results using the multi-server multiplexing strategy $\sigma = m$ for distant links for the client-server connection ($\eta_c = 10^{-3}$) and short links for the server-server connections $\eta_s = 10^{-1}$. Coherence times with and without EG/RSP are $20$ ms and $2.8$ s, respectively, with the full parameter values given in Table \ref{['tab:appC_tab1']} of Appendix D. (a) Rate-fidelity tradeoff for a scenario with $s=2$ and various numbers of servers, with rate measured in units of the duration of an EG/RSP attempt $\tau_e = 300$ ns. Note that for lower fidelities the tradeoff saturates; finite memory lifetimes effect the tradeoff by linking reduced single-qubit rates (which effect the multi-qubit rates) to a reduced the achievable final state fidelity. (b) Comparison between the $m_s$ obtained for the cases $s=2$ through $s=4$ with their corresponding classical bounds $M^s$. We included the point $m_s(M=1)=1$ for visual reference, and otherwise only consider $M\geq s$ where the multiplexing strategy we define is unambiguously specified. Note the logarithmic scaling; even for modest values of $s$ and $M$, multi-server multiplexing results in several orders of magnitude improvement in protocol execution rate.
• Figure 5: Quantum multiplexing between several client devices held by different clients and a single node in a quantum network with continuous RSP demand. (a) Fidelity-rate curves parameterized by $\gamma=\eta_c |\alpha|^2$. Here, $M=5$ clients each with their own photonic client device continuously compete for access to a single processing node, with both time sharing (solid) and quantum multiplexing (dashed) strategies. (b) Multiplexing gain in this case $m$ as function of $M$ for RSP with fidelity $F_\mathrm{min}=1-10^{-3}$ using $M$ competing clients. The theoretical maximum with a perfect server and the classical bound is plotted using the magenta and red black lines, respectively.