Demand prepaging is the fetching of multiple pages at each page fault in a virtual memory system. That is, when a process references a page that is not resident, the virtual memory system fetches not only the demanded page (that is, the one that the process referenced), but also some number of other pages that it speculates will be used soon. We have concentrated on understanding when demand prepaging will reduce the number of page faults, and when it will increase it.
We have written a paper on demand prepaging that will be presented at International Symposium on Memory Management 2002. You can get a copy of the paper here in PDF or in Postscript form. If you are interested in the code or traces for our experiments, email me.