Computation efficient sparse DNN nonlinear equalization for IM/DD 112 Gbps PAM4 inter-data center optical interconnects.

In this Letter, we propose and experimentally demonstrate a novel, to the best of our knowledge, sparse deep neural network-based nonlinear equalizer (SDNN-NLE). By identifying only the significant weight coefficients, our approach remarkably reduces the computational complexity, while still upholding the desired transmission accuracy. The insignificant weights are pruned in two phases: identifying the significance of each weight by pre-training the fully connected DNN-NLE with an adaptive L2-regularization and then pruning those insignificant ones away with a pre-defined sparsity. An experimental demonstration is conducted on a 112 Gbps PAM4 link over 40 km standard single-mode fiber with a 25 GHz externally modulated laser in O-band. Our experimental results illustrate that, for the 112 Gbps PAM4 signal at a received optical power of -5dBm over 40 km, the proposed SDNN-NLE exhibits promising solutions to effectively mitigate nonlinear distortions and outperforms a conventional fully connected Volterra equalizer (VE), conventional fully connected DNN-NLE, and sparse VE by providing 71%, 63%, and 41% complexity reduction, respectively, without degrading the system performance.

Reducing computation complexity by using elastic net regularization based pruned Volterra equalization in a 80 Gbps PAM-4 signal for inter-data center interconnects.

Volterra equalization (VE) presents substantial performance enhancement for high-speed optical signals but suffers from high computation complexity which limits its physical implementations. To address these limitations, we propose and experimentally demonstrate an elastic net regularization-based pruned Volterra equalization (ENPVE) to reduce the computation complexity while still maintain system performance. Our proposed scheme prunes redundant weight coefficients with a three-phase configuration. Firstly, we pre-train the VE with an adaptive EN-regularizer to identify significant weights. Next, we prune the insignificant weights away. Finally, we retrain the equalizer by fine-tuning the remaining weight coefficients. Our proposed ENPVE achieves superior performance with reduced computation complexity. Compared with conventional VE and L1 regularization-based Volterra equalizer (L1VE), our approach show a complexity reduction of 97.4% and 20.2%, respectively, for an O-band 80-Gbps PAM4 signal at a received optical power of -4 dBm after 40 km SMF transmission.