In this week, I mainly work on two parts:
1. Decoupling paper, which has been submitted.

2. Decaps. 
For this part, I am still working on obtaining proper target impedance bound. However, this task seems not so easy as I thought before. 
1. Loss bound can be obtained by assuming all the frequency components overlap at the maximum voltage. However this is too conservative.
2. Considering different components, different impedance patterns can lead to same maximum voltage variation. Thus obtain a tight bound is hard if not impossible. 
3. Coupling between current source complicates the problem even more. 

Based on the observation above, I think directly considering voltage variation in the optimization proccess will be more realistic if we have to consider noise directly. One way is simulation, but is time consuming. I am still thinking about more efficient approach.