QUANT-NET Design and Implementation
Key mechanisms
The QUANT-NET design is based on fundamental properties of quantum communications. Here we list a few key mechanisms:
Network-wide time synchronization.
QUANT-NET is a distributed system. A few critical quantum network functions require network-wide synchronization to provide a global notion of time. For example, entanglement provides a means to enables long-distance quantum communications. At present, due to technology immaturity, entanglement has been produced probabilistically over short distances with a short duration. Bell state measurement (BSM) is an essential function for entanglement generation over long distance. However, BSM can only be performed if both photons arrive at a BSMnode simultaneously. In addition, quantum measurement requires accurate timekeeping to correlate physically remote events. QUANT-NET achieves network-wide synchronization by using a master clock. All nodes synchronize with a single master clock. Dedicated high-resolution signal generators (e.g., Marconi signal generator) will be used as clocks.
Pulse-driven operations
QUANT-NET operates on a pulse-driven basis to simplify quantum network operations. Critical pulse-driven quantum network functions include: (1) a trapped ion is driven by a sequence of laser pulses to generate photons, and ion-photon entanglements. Therefore, a Q-node can vary its photon generation time by properly controlling drive pulses. For BSM, because the photons’ time-of-flight from two Q-nodes to a BSM-node are different, the two Q-nodes can vary their photon generation time accordingly so that the generated photons can arrive at the BSM-node simultaneously. (2) Pulse-gated detection of photons helps to reduce dark counts to improve detection accuracy.
Real-time control in a distributed environment
Timing constraints to perform entanglement swapping to generate high-fidelity entanglements between remote ions (5+ km) are stringent. In addition, the accurate control and manipulation of trapped ions requires a time resolution of ns scale. This means that QUANT-NET requires tight timing control, which imposes hard real time constraints at the lowest levels. QUANT-NET applies a systematic approach to meet time constraints:
Time-critical functions running on dedicated ARTIQ-based real-time control systems.
Dedicated ctrl&clk channels established for real-time communication between relevant entities.
Non-real-time functions such as monitoring proper functioning of all components, re-initiating calibration, and data analysis are executed in software in the CPU domain.
Non-real-time communications running through TCP/IP channels.
Active quantum channel monitoring, calibration, and optimization
The single-photon nature of quantum communication signals makes them extremely sensitive to noise on the quantum channels. In particular, QUANT-NET generates and distributed polarization photon qubits that are vulnerable to polarization drifts. Protocols such as teleportation require indistinguishability in spectral, temporal, spatial, and polarization properties of the two photons arriving at a BSMnode. QUANT-NET employs several active and automated quantum-channel calibration and optimization mechanisms to minimize quantum channel loss, reduce background noise, and compensate for polarization and delay drifts.
Active quantum channel monitoring and active quantum signal-to-noise ratio estimation. If necessary, a new quantum channel with less noise will be selected.
Hong-Ou-Mandel (HOM) measurement and calibration is used to ensure quantum indistinguishability. The HOM signal provides feedback to compensate for the photons’ relative time-of-flight, ensuring stable operation.
Active polarization measurement and calibration is used to compensate for polarization drifts in fibers.
Active trapped-ion Q-node monitoring, calibration, and optimization. Realizing practical and useful quantum networks requires long coherence-time qubits and high-fidelity quantum gate operations. To enable such capabilities, each trapped-ion Q-node in QUANT-NET is actively monitored, calibrated, and optimized in a periodical manner. In particular, the cavity length and the cavity drive phases need to be calibrated in sub-ms level. An electronic feedback laser is dedicated to executing such functions. On the other hand, within the cavity of a trapped-ion Q-node, the magnetic field axis/atomic quantization axis defines measurement basis. Because QUANT-NET has multiple trappedion Q-nodes, it is necessary to align the measurement basis of different systems to the same basis. QUANT-NET achieves network-wide basis alignment by using a master Q-node. All nodes synchronize with a single master Q-node.
Trapped-ion Q-node and BSM-node Design and Implementation
A trapped-ion Q-node is a few-qubit quantum computer with a photonic interconnect. It is designed to perform both conventional and quantum functions. Major functions include:
local quantum computations using trapped ions,
single photon generation whose polarization is entangled with one of its trapped-ion qubits,
quantum frequency conversion (QFC) between the native wavelengths of trapped ions (854 nm) and telecom bands (1550 nm), and
conventional computation and communication to control the flow of quantum information.
For example, a Q-node exchanges messages with other quantum network nodes via conventional traffic channels to act on successful entanglement generation and demands for quantum resources, or simply to perform quantum channel calibration and optimization. The former requires real time control while the latter usually is not very time critical.
Fig. 7 illustrates a trapped-ion Q-node design. It has one or multiple ions trapped in a cavity. An ARTIQ-based real-time control system controls the amplitude, frequency, and phase of laser light as well as the voltages used to manipulate the internal as well as motional degrees-of-freedom of the trapped ions in real time at the sub-microsecond time scale. Non-real-time operations such as triggering calibration sequences, analysing their results and feedback of the newly determined parameters are executed in software in the CPU domain. A Q-node has three types of communication channels: (1) quantum channel(s) to send quantum signals and messages. (2) ctrl&clk channel(s) to transmit/receive control and clock signals and messages to support time-critical operations. And (3) a TCP/IP connection to transmit conventional messages for non-real-time operations.
Real time control of the Q-nodes is required, allowing the system to make complex decisions based on both external communication as well as from the Q-node itself. Specifically, the Q-node acts on an external trigger when to stop the entanglement attempts, but also when new demands for entanglement arise. Also, the Q-node is designed to react to external demands for teleportation, i.e. to implement a local BSM, communicate its result, and/or act on results on local BSMs of other Q nodes, or BSM nodes.
Figure 7: An ARTIQ-based trapped-ion Q-node.
A BSM-node is designed to perform bell state measurement (BSM) of polarization photon qubits. As illustrated in Fig. 8 below, a photonic BSM is typically implemented using a beam splitter followed by measurement devices. Its major functions include:
Channel calibration and optimization with classical pulses. A BSM node implements an active and automated channel calibration and optimization mechanism to minimize polarization and frequency drifts. This classical calibration will be interleaved with quantum network operation and HOM analyzing unit operation.
HOM analyzing unit. A BSM-node implements a HOM analyzing unit to ensure photon indistinguishability for bell state measurement. The HOM analysis is carried out in temporal, polarization, and spectral degrees of freedom by keeping an updated analysis of interference visibility. (a) In temporal degree of freedom. By real-time monitoring HOM interference visibility at different time delays, the HOM analysis can accurately determine the arrival time difference of two photons at the BSM-node. The time difference is fed back to the related Q-nodes so that they can vary the two photons’ generation time to allow them to arrive at the BSM-node simultaneously. The time difference will also be used to herald successful entanglement generation. Only coincidences in the interference window will be used as successful entanglement heralding events. (b) In polarization and spectral degrees of freedom. After arrival time difference has been actively compensated, the HOM analysis can be similarly performed by varying polarization and frequency, respectively. The analysis results are fed back to the corresponding entities in the quantum network to allow for control and optimization. Depending on the types of visibility reduction, polarization control, or frequency re-optimization can be achieved if interference window is narrowing, to minimize polarization and frequency drifts. Polarization control can be implemented locally in the BSM-node, which is relatively straightforward. However, frequency re-optimization may require a systematic calibration and tuning of laser, trapped-ion cavity, QFC, and filtering, across multiple quantum network nodes.
Bell-state measurement. Most importantly, a BSM-node performs Bell state measurement (BSM) of polarization photon qubits. Two-photon interferometry is used to identify two of the four bell states. The measurement results are transmitted in real time (few 100 ns) to the related quantum nodes in the network. A real-time communication mechanism, either via a dedicated fiber or electrical cable, will be developed to support such a function.
Figure 8: An ARTIQ-based BSM-node .
Each node is assigned an IPv6 address to uniquely identify it. Each channel of a node is numbered and thus uniquely identified. Important guiding principle will be to make each node as autonomous as possible, i.e. it calibrates itself as much as possible and requires only minimal information from the outside. This will ensure that the system is as reliable and scalable as possible. The concept of autonomy will apply both to the real time control level where success of BSMs, requests for teleportation, and readiness will be handled, as well as on the soft time level where resource allocation as well as overall synchronization will be taken care of. However, each node will communicate its status to the central control plane for monitoring and debugging purposes.
Figure 9: QUANT-NET testbed system diagram.
The QUANT-NET testbed
Fig. 9 above illustrates the QUANT-NET testbed system diagram. The testbed runs as a distributed system. A Quantum Network (QN) server coordinates all activities in the network. It controls and manages the underlying Q-nodes and BSM-nodes through a general message bus implementation and broker service. Device models for each class of testbed node have been developed along with well-defined protocol functions to provide validation and verification of expected system behavior. A QN server typically handles global (network-wide) non-real-time functions such as quantum network topology discovery, quantum network monitoring, and periodically scheduling quantum network calibration and optimization etc. It also handles user requests on an event-driven basis. When serving a user request, the server cycles QUANT-NET alternatively between calibration and operational modes to achieve a sustained high-fidelity quantum network. As opposed to the QN Server orchestration layer, ARTIQ-based Q-nodes and BSM-nodes form the underlying data plane (i.e., quantum plane). Each node handles local (node-wide) real-time functions that have tight time constraints.