Quantum-Processing-Assisted Classical Communications

Abstract

We describe a general quantum receiver protocol that maps laser-light-modulated classical communications signals into quantum processors for decoding with quantum logic. The quantum logic enables joint quantum measurements over a codeword to achieve the quantum limit of communications capacity. Our receiver design requires only logarithmically increasing qubit resources with the size of the codeword and accommodates practically relevant coherent-state modulation containing multiple photons per pulse. Focusing on classical-quantum polar codes, we outline the necessary quality of quantum operations and codeword lengths to demonstrate a quantum processing-enhanced communications rate surpassing that of any known classical optical receiver-decoder pair. Specifically, we show that a small quantum receiver of 4 qubits with operational errors of $\sim 0.2\%$ can already provide a $5$ percent gain in the communications rate in the weak signal limit. Additionally, we outline a possible hardware implementation of the receiver where efficient spin-photon interfaces such as cavity-coupled diamond color centers or atomic qubits are used to input the received photonic signal to a small scale quantum processor for decoding. Our results outline a new, promising route for potential quantum advantage in classical communication with near-term, small-scale quantum computers.

Figures (6)

• Figure 1: Schematic representation of the full communication setup including the transmitter, channel and quantum receiver. The receiver, which is the focus of this work, has a modular design to accommodate multiple photon inputs. In each time-bin, the signal is routed to the quantum switches, which will transmit (reflect) the signal if the corresponding memory bank in the quantum processor is empty (full). Single photon receiver: A photon ($\gamma$) arriving in the first time-bin is directed to the transducer qubit. A CNOT gate between the transducer qubit and the quantum switch qubit flips the switch from transmission to reflection. The state of the transducer qubit is mapped to a logical control qubit, and a series of logical CNOT gates flip the processor qubits to encode the photon arrival bin and phase information (the compression step). Next, the control qubit is decoupled from the target qubits with a measurement in the X basis. Finally, gates and measurements are performed on the remaining target qubits to decode the received classical bit string (the decoding step).
• Figure 2: Decoding circuit for $N=4$. The double wires with open circles represent classical control conditioned on a measurement outcome of 0. The $H$ denotes a Hadamard gate.
• Figure 3: A decision tree to decode the message for $N=4$. $M_i$ represents a measurement of qubit $i$ in the $Z$ basis. The binary labels denote measurement outcomes. $CH_{ij}$ is a controlled Hadamard gate with qubit $i$ as control and $j$ as target. $H_i$ is a Hadamard gate on qubit $i$.
• Figure 4: PIE versus mean photon number for CQ polar codes, Dolinar symbol by symbol detection and PPM.
• Figure 5: Receiver performance with (a) transducer errors and (b) gate errors in the compression and decoding steps. The green dotted line denotes the PIE with ideal symbol by symbol detection. In (a), we modeled the errors with depolarizing noise at the physical level on the input photonic state. In (b), we used a total depolarizing channel (unbiased Pauli channel) at the logical level for all interactions.