Real‐Time Monitoring of High‐Speed Signal Quality Using Asynchronous Sampling, Amplitude Sorting, and Artificial Neural Network and Its FPGA Implementation.

In high‐speed data transmission, signal quality monitoring guarantees transmission system reliability and stability. In this study, high‐speed signal waveforms were sampled directly and asynchronously, and then the sampled values were sorted by amplitude. Using these sampled data sets, we propose an artificial neural network (ANN) model to estimate signal quality parameters of high‐speed signals. The sampled values can clearly show the characteristics of the waveform after amplitude sorting, which will significantly reduce the error of parameter calculation, and a relatively simple ANN model can be used. Using 25‐Gb/s non‐return‐to‐zero signals as an example, the ANN inputs 40 signal sampling data through a hidden layer with 7 neurons and can estimate five types of signal quality parameters: Q‐factor, signal‐to‐noise ratio, time jitter, rise time, and fall time. The mean square error of the calculated time jitter was 11.8%, and those of the other parameters were lower than 10%. The simple ANN model will be easier to implement in hardware. A field‐programmable gate array was used to implement the ANN hardware for estimating signal quality parameters. The calculated average errors of the five signal quality parameters between software and hardware methods were less than 1%. The implemented hardware estimation of high‐speed signal quality parameters could be used for real‐time signal quality monitoring in high‐speed data transmission modules and devices. [ABSTRACT FROM AUTHOR]