Thanks for your reply. I almost finish design, and it works well in transient domain and I will do some analysis then.
Since it works at high frequency, I do the s-parameter analysis first.
But after several day's simulation, i am confused now. One question, negative resistance from the active part is more than 3 times than the resistance of LC tank. With Y-parameter analysis, active conductance Ga=1/re{Y11} should be less than 1/3 times of the LC conductance Gp (Gp=1/re{Y11}), right?