End-to-end Deep Learning of Optical Fiber Communications

We implement the optical fiber communication system as an end-to-end deep neural network, which includes the complete chain of transmitter, receiver and channel. This approach allows transceiver functions to be optimized in a single end-to-end process. Our work shows that such a deep learning fiber-optic system can achieve reliable communication below forward error correction threshold. In particular we show that for intensity modulation/direct detection systems information rates of 76 Gbps can be achieved at distances beyond 80 km.