I'm not sure that "anti-aliasing of time series data" is the correct terminology to use, so let me explain:
I have some sources of data that are aligned quarterly, mostly to do with quarterly running costs. I have some other sources of data, some of which are aligned weekly, and some monthly, and these are mostly related to transactions that took place. By the term 'anti-aliasing', I am referring to the problem of how to represent this data on a common time granularity, so that it can all be integrated together into a data warehouse.
How do I get this data which does not align through time, into a data warehouse? The alignment problem mostly stems from the fact that weeks do not exactly fit into months or quarters. The months against quarters issue is less of a problem, since 3 months make a quarter.
For example, if I know the quarterly running cost, should I divide that up, and attribute an equal share of it to each day in the quarter. Then if I know the number of transactions completed on a weekly basis, should I divide them up and attribute them equally to each day of the week. That would artificially break down the grain of the data into days, and allow me to roll it up by week, month or quarter.
(Is there a more correct terminology in data-warehousing for this problem?)