Dynamic Time Warping with moving window

Question

I have two time series, lets say X and Y with 127 and 64 observations in each respectively. I have applied DTW with global constraint to get the distance and then further used that distance to calculate another coefficient. Now I want to calculate DTW using rolling window for the timeseries using one year period for DTW calculations. Lets say, both X and Y starts on January 2012 but X is monthly observation and Y is two-monthly observation. This gives that I will have to calculated DTW for X_1 and Y_1 with 12 and 6 observations respectively and then will shift the window by 2 months for next DTW distance. Now my problem is that I can not find reference from literature that guides about the minimum number of observations needed for DTW algorithm to give meaningful results