Hi Matthew, On Mon, Feb 1, 2010 at 3:51 AM, Matthew Orlinski <[email protected]> wrote:
> This is my first mail to a mailing list. I hope I'm doing it right. So far so good... :) > To the point. At the university of Manchester in the UK we recently got some > NS2 simulator code running and 802.11s implementation. This code is supposed > to simulate the the 802.11s specification but we are unsure at how accurate > it is as it has not been supported for quite some time. > > Before we release this code out to the general public my supervisor and i > would like to do some validation on its results to make sure its at least a > good start. Would any of you be able to advise me on any benchmarking > figures or techniques which would be appropriate? Apologies if this is > slightly off topic from the current threads but i thought I'd better to ask. You raise two different issues: validation of you model vs. benchmarking. The first would tell you if you model indeed implements 802.11s (whatever the draft version) and therefore if it would be compatible with other implementations. The only way I can think to validate your ns2 model is to look at traffic traces generated by your model. Are the frames in the correct format? Are the different fields updated correctly? Are management frames generated in response to the right events? etc. Benchmarking would tell you how good your implementation is with regards to others (not applicable in your case) or how good is one particular configuration of your mesh is with regards to other possible configurations. For instance, you can measure things like the ratio of path selection frames over the total number of frames or the average path discovery time. And then do the same measurements for different path expiration times, topologies, path selection options (RF/DO), etc. That would provide good guidance on how to configure 802.11s stacks. But of course only if your model is valid :) Anyway, my advice would be not to delay releasing your implementation to the public! Cheers, Javier _______________________________________________ Devel mailing list [email protected] http://open80211s.com/mailman/listinfo/devel
