A Novel Statistical Analysis Method For Nonlinear High-Speed Links

With the commercialization of 5G,and the rapid development of artificial intelligence,the number of interfaces increases quickly,and the High-Speed I/O trends to high density to meet the demand for high memory bandwidth.At high data rates,the signal quality is limited to signal integrity and power integrity problems,due to impedance mismatch,channel losses,cross-talks,Inter Symbol Interference(ISI),and so on.The signal integrity analysis of the high-speed links becomes more important.At the same time,circuit nonlinearity ha to be taken into consideration,because the nonlinear drivers are imbalanced,and the saturation characteristics of component.In above case,it may be hard to be great accuracy with actual behavior.In order to solve above problems,integrated simulation with a full transistor model is computationally complicated and time-consuming,and it's difficult to have a precise analysis result in low Bit Error Rate(BER).Various fast and accurate eye diagram prediction methods have been proposed to replace the conventional time-consuming pseudo-random bit sequence(PRBS)simulation.Statistical link analysis and link budget calculation is the essential part for current high-speed system design.A fast and accurate statistical eye diagram estimation method for high-speed nonlinear links is proposed in this thesis.Probability density functions of output responses is derived based on multiple edge responses.According to the property that the influence of nonlinearity in high-speed links will not last for long time,a new scheme for calculating the probability density functions of responses is presented.The convolution process is divided into nonlinear section,transition section,and linear section.High order of multiple edge responses convolution is only used in nonlinear section,low order of multiple edge responses convolution is used in transition section,and double edge responses convolution is used in linear section.The new scheme can drastically reduce the amount of computation,shorten analysis time.In this thesis,the proposed method is verified by comparing the probability density distributions of statistical eye diagram,the bathtub curves,and simulation time with that of traditional MER-based statistical eye diagram.Results show that the simulation speed of the proposed method has been improved by about fourteen times,and the accuracy is almost the same as traditional statistical eye diagram.There are double great advantages of being both accuracy and effective compared with traditional statistical analysis method.