High-order modulation signals equalization based on reservoir computing under equivalent-time sampling

For digital signal testing and analysis using a sampling oscilloscope, signal equalization of different frequency components is a difficult task. The discontinuity of the sampling points induces the idea that the equalizers based on high-pass filtering cannot be utilized. In this work, an artificial intelligence equalizer based on reservoir computing (RC) is introduced for processing equivalent-time sampling data of high-order modulation signals. The equalizer is established for the in-phase and quadrature channels of optical quadrature phase shift keying modulated signals, respectively. By training the RC model to aim for equivalent time sampling data and processing, the low-response, high frequency components are compensated. The results show that, compared with the input waveform, the quality of the RC equalized waveform improves significantly, the convergence of the signal constellation diagram is enhanced, and at a signal-to-noise ratio (SNR) of 15 dB, the error vector magnitude is reduced from 17.35% to 1.75%. The equalizer shows robustness; at an SNR of 9 dB, the bit error rate (BER) before equalization is 4.60 × 10-3, and after training with signal-to-noise ratios of 5-15 dB, the BER is reduced to between 1.05 × 10-14 and 1.13 × 10-110.