Quantum receiver and method for decoding an optical signal
A quantum receiver for decoding an optical signal includes a beamsplitter for interfering the optical signal with a local-oscillator field to generate a displaced field, and a single-photon detector for detecting the displaced field. The quantum receiver also includes a signal-processing circuit for determining, based on an electrical output of the single-photon detector, a measurement outcome. The signal-processing circuit also determines, based on the measurement outcome and a feed-forward machine-learning model, a next displacement. The quantum receiver also includes at least one modulator for modulating, based on the next displacement, one or both of the optical signal and the local-oscillator field. Like a Dolinar receiver, the quantum receiver implements adaptive measurements to reduce the error probability of the decoded symbol. The use of machine-learning reduces the latency of the signal-processing circuit, thereby increasing the number of measurements that may be performed for each received symbol.
1. A quantum receiver for decoding an optical signal, comprising:
a beamsplitter for interfering the optical signal with a local-oscillator field to generate a displaced field;
a single-photon detector for detecting the displaced field;
a signal-processing circuit for:
determining, based on an electrical output of the single-photon detector, a measurement outcome; and
determining, based on the measurement outcome and a feed-forward machine-learning model, a next displacement; and
at least one modulator for modulating, based on the next displacement, one or both of the optical signal and the local-oscillator field.
2. The quantum receiver of claim 1 , the feed-forward machine-learning model comprising one of an artificial neural network and a probability binary-tree classifier.
3. The quantum receiver of claim 1 , wherein:
the optical signal comprises a received sequence of received segments;
the displaced field comprises a displaced sequence of displaced segments corresponding to the received sequence;
the signal-processing circuit includes a memory storing a historical sequence of measurement outcomes corresponding to the displaced sequence; and
the signal-processing circuit is configured to determine the next displacement based on the historical sequence.
4. The quantum receiver of claim 3 , wherein:
the memory stores a feed-forward look-up table generated from the feed-forward machine-learning model; and
the signal-processing circuit is configured to determine the next displacement by inputting the historical sequence to the feed-forward look-up table.
5. The quantum receiver of claim 3 , wherein:
the memory stores the feed-forward machine-learning model; and
the signal-processing circuit is configured to determine the next displacement by inputting the historical sequence to the feed-forward machine-learning model.
6. The quantum receiver of claim 3 , the signal-processing circuit being configured to determine, based on the historical sequence and a decoding machine-learning model, a decoded state.
7. The quantum receiver of claim 6 , the signal-processing circuit being configured to output the decoded state.
8. The quantum receiver of claim 6 , the decoding machine-learning model comprising a neural network.
9. The quantum receiver of claim 6 , wherein:
the memory stores a decoder look-up table generated from the decoding machine-learning model; and
the signal-processing circuit is configured to determine the decoded state by inputting the historical sequence to the decoder look-up table.
10. The quantum receiver of claim 6 , wherein:
the memory stores the decoding machine-learning model; and
the signal-processing circuit is configured to determine the decoded state by inputting the historical sequence to the decoding machine-learning model.
11. The quantum receiver of claim 1 , the beamsplitter being unbalanced.
12. The quantum receiver of claim 1 , the at least one modulator comprising one amplitude modulator and one phase modulator.
13. A method for decoding an optical signal, comprising:
interfering the optical signal with a local-oscillator field to generate a displaced field;
detecting the displaced field with a single-photon detector;
determining, based on an electrical output of the single-photon detector, a measurement outcome;
determining, based on the measurement outcome and a feed-forward machine-learning model, a next displacement; and
modulating, based on the next displacement, one or both of the optical signal and the local-oscillator field.
14. The method of claim 13 , wherein:
the optical signal comprises a received sequence of received segments;
the displaced field comprises a displaced sequence of displaced segments corresponding to the received sequence; and
said determining the next displacement is further based on a historical sequence of measurement outcomes corresponding to the displaced sequence.
15. The method of claim 14 , wherein said determining the next displacement includes inputting the historical sequence to a feed-forward look-up table.
16. The method of claim 15 , further comprising generating the feed-forward look-up table based on the feed-forward machine-learning model.
17. The method of claim 14 , wherein said determining the next displacement includes inputting the historical sequence into the feed-forward machine-learning model.
18. The method of claim 14 , further comprising determining, based on the historical sequence and a decoding machine-learning model, a decoded state.
19. The method of claim 18 , further comprising outputting the decoded state.
20. The method of claim 18 , the decoded state being one of:
two candidate coherent states of a binary phase-shift keying constellation; and
four candidate coherent states of a quadrature phase-shift keying constellation.