Cerebral autoregulation is a mechanism that blood flow keeps constantly steady in spite of blood pressure variability in the brain. This mechanism has been modeled as a control system with ABP as input and CBFV as output. Linear methods of assessing CA suffer from non-linearity and non-stationarity, while newly developed methods, wavelet and MMPF, have their inherent drawbacks. Wavelet is limited by uncertainty principle, whilst Hilbert-Huang transform (HHT) in MMPF is restricted to mono-component signal. We therefore used the time-frequency distribution which can track instantaneous dynamics of CA. In this paper, we will be focus on the analysis of ABP, as the dynamics of the input of a control system is important in terms of system identification. Three different TFD methods, smoothed pseudo Wigner-Ville distribution (SMWVD), Zhao-Atlas-Marks distribution (ZAMD), and Choi-Williams distribution (CWD), are compared to show the embedded dynamics of ABP signal signals properly. ABP signals are collected with Nexfin monitor from eight health volunteers and multi-component is produced by deep breath with supervision. Experiment results shows that Choi-Williams distribution has better performance over other methods and is qualified optimal time-frequency distribution.