Hullo mate, When the error on the link is small, you will not notice a difference.
Snoop just like any other link layer solution was designed to enhance TCP performance in wireless or random error environments. In a lossless environment, the setup with or without snoop will yield very similar, if not the same results. So, to notice a difference you need to keep varying the packet loss rate on the link where snoop is in charge. Then you do the same for an end to end connection (i.e. without snoop) and compare the results. The example "MySnoop.tcl" has an uniformly distributed random variable error model that you can integrate into the example "Snoopns2.tcl" that worked for you. You'll notice that as the variable "set loss_prob 10" grows, there'll be a difference in performance when snoop is used, and when snoop is absent. That should get you there. Kind regards, Nicholas Mbonimpa On Fri, Nov 16, 2012 at 6:15 AM, shiny <shiny....@gmail.com> wrote: > hi frnd, > yes.am struggling to differentiate the performance with and without > snoop... > > > > On Thu, Nov 15, 2012 at 9:19 AM, nicholas mbonimpa <nmboni...@gmail.com>wrote: > >> Unfortunately I don't. >> >> But, just for the beginning, you can use xgraph to visualise the outputs >> then have a rough idea on the throughput / bandwidth behaviour. >> >> I guess you can now move on from here, and try out a couple of other >> interesting scenarios. >> >> I'm glad I've been of help. >> >> Cheers mate. >> >> >> On Thu, Nov 15, 2012 at 8:15 PM, shiny <shiny....@gmail.com> wrote: >> >>> hi frnd, >>> thanks for ur help.snoopns2.tcl script is running sucessfully. while >>> executing awk script for throughput calculation am getting 0 as output. do >>> u have awk script for throughput calculation.... >>> >>> >>> Thu, Nov 15, 2012 at 8:49 AM, shiny <shiny....@gmail.com> wrote: >>> >>>> thanks a lot for ur help..... >>>> >>>> >>>> On Wed, Nov 14, 2012 at 11:12 AM, nicholas mbonimpa < >>>> nmboni...@gmail.com> wrote: >>>> >>>>> Hullo Shiny, >>>>> >>>>> I also forgot to mention that how you can tell there's a difference is >>>>> by running one protocol end to end (for example Reno), and then running >>>>> the >>>>> same setup but with snoop at an intermediate node. Then monitor some >>>>> statistics like throughput at the sink. One of the examples I sent you has >>>>> an error model for the channel, so you can vary that and see what happens. >>>>> >>>>> The examples also have scripts to compute throughput and some other >>>>> parameters you may want to look at. >>>>> >>>>> It's from such comparisons that you can ascertain whether there's a >>>>> difference or not. >>>>> >>>>> I hope this will help. >>>>> >>>>> Kind regards, >>>>> >>>>> Nicholas Mbonimpa >>>>> >>>>> >>>>> >>>>> On Wed, Nov 14, 2012 at 4:31 PM, nicholas mbonimpa < >>>>> nmboni...@gmail.com> wrote: >>>>> >>>>>> Hullo Shiny, >>>>>> >>>>>> Apologies for failing to get back to you in time, I've been so caught >>>>>> up. >>>>>> >>>>>> Please find attached two examples running snoop. >>>>>> >>>>>> In example snoopns2.tcl, the slight difference in configuration for >>>>>> snoop is explained in a comment. >>>>>> >>>>>> Please pay attention to the source path at the beginning of the code. >>>>>> >>>>>> I hope that will help. >>>>>> >>>>>> Kind regards, >>>>>> >>>>>> Nicholas Mbonimpa >>>>>> >>>>>> P.S: I've run this code on my side and it works just fine. >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> On Wed, Nov 14, 2012 at 3:25 PM, shiny <shiny....@gmail.com> wrote: >>>>>> >>>>>>> >>>>>>> hi frndz, >>>>>>> i have added LL/LLSnoop to my tcl script for snoop agent deployment >>>>>>> at >>>>>>> base station.there is no difference with or without LL/LLSnoop. Is >>>>>>> there >>>>>>> any more code to add snoop agent in tcl script.plz help me frndz... >>>>>>> >>>>>>> -- >>>>>>> With Regards, >>>>>>> John Shiny.J >>>>>>> >>>>>> >>>>>> >>>>> >>>> >>>> >>>> -- >>>> With Regards, >>>> John Shiny.J >>>> >>>> >>> >>> >>> -- >>> With Regards, >>> John Shiny.J >>> >>> >> > > > -- > With Regards, > John Shiny.J > >
