Reservoir computing-based digital signal equalizer for equivalent-time sampling

Ning Jing; Junpeng Zhao; Minjuan Zhang

doi:10.1063/5.0166523

Reservoir computing-based digital signal equalizer for equivalent-time sampling

Rev Sci Instrum. 2023 Nov 1;94(11):115102. doi: 10.1063/5.0166523.

Authors

Ning Jing¹, Junpeng Zhao¹, Minjuan Zhang¹

Affiliation

¹ School of Information and Communication Engineering, North University of China, Taiyuan 030051, China.

PMID: 37909839
DOI: 10.1063/5.0166523

Abstract

A sampling oscilloscope is an important instrument for evaluating the quality of optical communication signals. Since its working principle is equivalent-time sampling, the data obtained for digital signals with random characteristics do not have continuity, which makes it impossible to use methods such as filtering and averaging to equalize the signal. For this reason, this paper proposes a signal equalization method based on reservoir computing. Through training of the reservoir model, an equivalent-time equalizer is established to solve the problem that the sampling oscilloscope cannot equalize random digital signals. Compared with the continuous-time equalizer, the coincidence degree exceeds 95%. The eye height and eye width are increased by 7 and 1.6, respectively, while the jitter in the eye diagram is reduced by 2.3 times, which solves the problem that the sampling oscilloscope cannot equalize random digital signals.