In order to achieve high-sensitivity time-domain diffuse correlation spectroscopy (TD-DCS) measurement of functional changes in cerebral blood flow, this study applied simulation methods to optimize the TD-DCS system under real experimental conditions (including the consideration of the effects of finite coherence length and non-ideal instrument response function IRF). Under a real experimental condition where the incident power is 75 mW, the source-detector distance is 1.0 cm, and the full width at half maxima of the IRF is 160 ps, we used simulation experiments to investigate the relationship between the contrast of the intensity autocorrelation function ( 2 ) in two brain functional states (i.e., baseline and activation) and TD-DCS system parameters (including , IRF, source-detector distance, gate opening time and gate width).Our simulation results show that both longer and longer integration time are beneficial to a more sensitive detection. With a fixed and integration time, the optimal parameters of gate opening time is 800 ps (relative to the peak time of IRF), and gate width is equal to or larger than 800 ps. This study may be useful for guiding the sensitive measurement of human brain functions (e.g., changes in cerebral blood flow) using the TD-DCS technology.