Digital Pre-distortion Based on Delta Sigma Modulation Assisted Look-up Table for Optical Transmission

Coherent technologies have been investigated as a promising solution to support beyond Tb/s intra-datacenter interconnections, where the nonlinearity of cost-effective transceiver components is a major impairment. To mitigate the nonlinear distortion, and expand the overall linear range of electronic and opto-electronic devices, the look-up table (LUT) based digital pre-distortion (DPD) is potentially a lightweight solution. However, the conventional LUT methods are designed for symbol sequences, depending on the modulation format of signal. For high-speed transmission, the DPD to samples at the input of digital-to-analog converters (DACs) is essential, especially for the systems with sophisticated DSP at transmitter. For the sample data after DSP, the number of amplitude levels is too large for the LUT operation. In this paper, we propose a DPD method based on delta sigma modulation assisted LUT (DSM-LUT) to pre-distort the sample sequences before DACs. The detailed description of the DSM-LUT scheme as well as extensively experimental demonstrations are provided. For the amplifier-less experiments with 35 Gbaud dual-polarization 64 QAM signals, the proposed DSM-LUT can achieve signal-to-noise ratio (SNR) gains of 0.77 dB and 0.91 dB for 1- and 3-sample memory depths, respectively, which is similar to that of the conventional symbol based LUT (S-LUT) and directly quantized sample based LUT (Qtz-LUT) methods with the same memory depths. Furthermore, the advantages of DSM-LUT against Qtz-LUT and S-LUT are analyzed. The comparisons of LUT table size and the average number of times to look up the table are discussed. Specifically, for the LUT operation with a memory depth of 1, the table size of DSM-LUT is reduced to less than 50% of the size of S-LUT. Compared with Qtz-LUT, the number of quantization bits of DSM-LUT is reduced to around 50%, and the table size of DSM-LUT is reduced to less than 13%.