Optical decoder learning for fiber communication at the quantum limit

Optical communication technology can be enhanced by using quantum signals to transfer classical bits. This requires the message-carrying signals to interact coherently at the decoder via a joint-detection receiver (JDR). To date, the realization of a JDR using optical technologies has remained elusive: the only explicit design, called a Green–Hadamard receiver (GHR), increases distinguishability at the cost of reducing the code size. We introduce a supervised-learning framework for the systematic discovery of optical JDR designs based on parametrized photonic integrated circuits. We find JDR designs with higher decoding success probability than any single-symbol receiver, including homodyne, Kennedy, and Dolinar. Furthermore, our new receiver families surpass the GHR receiver for mean photon number >0.1, both in terms of code size and decoding probability, paving the way for practical applications of JDR in optical fiber networks and free-space.