Single-step nonlinearity compensation using artificial intelligence for digital coherent transmission systems
Aspects of the present disclosure describe a method for digital coherent transmission systems that advantageously provides low-complexity, single-step nonlinearity compensation based on artificial intelligence (AI) implemented in a deep neuron network (DNN).
1. A nonlinearity compensation method for optical transmission systems for optical networks employing digital coherent receivers performing analog-to-digital conversion (A/D), synchronization and resampling, chromatic dispersion compensation (CD), polarization demultiplexing (PD), carrier phase recovery (CPR), and forward error correction decoding (FEC) operations, said method comprising:
performing, after the carrier phase recovery and before the forward error correction operation, an estimation of nonlinearity of received signals, wherein outputs of the carrier phase recovery operation are input to a deep neural network (DNN) to estimate the nonlinearity H NL ,
determining recovered symbol {tilde over (H)} according to the following relationship:
{tilde over (H)}=H−H NL
wherein {tilde over (H)} is the recovered symbol, H is the received symbol, and H NL is the estimated nonlinearity of the received symbol as determined by the DNN;
wherein intra-channel cross-phase modulation (IXPM) and intra-channel four-wave-mixing (IFWM) triplets are computed from recovered symbols spanning a symbol window length L around a symbol of interest H 0 and the recovered symbols in the symbol window length L, together with the computed triplets are input to the DNN to estimate fiber nonlinearity of received signal H 0 .
2. The method of claim 1 wherein there are in total (3L 2 +1)/4 triplets computed.