From shankar_setty04 at yahoo.com Tue Apr 18 00:52:22 2006 From: shankar_setty04 at yahoo.com (shankar setty) Date: Tue, 18 Apr 2006 00:52:22 -0700 (PDT) Subject: [Diffusion-users] Diffusion API doubt Message-ID: <20060418075222.74486.qmail@web52205.mail.yahoo.com> Hello, I have a small problem, i.e., how will Diffusion API decide which node will be the next to send the interest. Lets consider that the initially interest is disseminated by node1 and say within the radio range of node1 are node2,3,4 which get the interest from node1. Now say node2 is the first to disseminated the interest to its neighbors, now my question is how is that node2 is selected for disseminating interest but why not node 3 or 4 although say node 3 and 4 are close to node1 but node 2 is far from node 1. Well how will this Diffusion API select which node will send the next set of interest when all the 3 received at the same time from node1. Waiting for your reply. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From johnh at ISI.EDU Tue Apr 18 10:17:24 2006 From: johnh at ISI.EDU (John Heidemann) Date: Tue, 18 Apr 2006 10:17:24 -0700 Subject: [Diffusion-users] Diffusion API doubt In-Reply-To: <20060418075222.74486.qmail@web52205.mail.yahoo.com> Message-ID: <200604181717.k3IHHOri005798@dash.isi.edu> On Tue, 18 Apr 2006 00:52:22 PDT, shankar setty wrote: >Hello, > >I have a small problem, i.e., how will Diffusion API >decide which node will be the next to send the >interest. > >Lets consider that the initially interest is >disseminated by node1 and say within the radio range >of node1 are node2,3,4 which get the interest from >node1. Now say node2 is the first to disseminated the >interest to its neighbors, now my question is how is >that node2 is selected for disseminating interest but >why not node 3 or 4 although say node 3 and 4 are >close to node1 but node 2 is far from node 1. Well how >will this Diffusion API select which node will send >the next set of interest when all the 3 received at >the same time from node1. > >Waiting for your reply. > I believe this question is answered specifically in several of the diffusion papers. See for example the section on reinforcement in the 2000 MOBICOM paper by Intanagonwiwat et al, or the section summarizing diffusion (3.1) in the 2001 SOSP paper by Heidemann et al. -John Heidemann P.S. Please do not cross-post questions across many mailing lists. That defeats the purpose of having multiple, separate lists. From shankar_setty04 at yahoo.com Tue Apr 18 20:53:34 2006 From: shankar_setty04 at yahoo.com (shankar setty) Date: Tue, 18 Apr 2006 20:53:34 -0700 (PDT) Subject: [Diffusion-users] Diffusion API doubt In-Reply-To: <200604181717.k3IHHOri005798@dash.isi.edu> Message-ID: <20060419035334.41134.qmail@web52212.mail.yahoo.com> Hello Sir, My doubt is at the initial setup state of the interest flood. 3 4 1 2 Consider the above example where node1 is the sink and it has disseminated the interest by flooding and nodes 2,3,4 are its neighbors which listen to them first. Now from 2,3 and 4 the interest needs to be flooded to next neighbor that come in there radio range. The Question is here, once node2,3,4 gets the interest from 1, which will be the node outoff 2,3 and 4 to flood the interest ? In my case say 2 is flooding the interest first and then 3 or 4, now my doubt is how is 2 flooding the interest first where exactly in the code is this happening that 2 is selected first but why not 3 or 4 ( is this at the 2pp_ping_application or is it at the two_phase_pull algorithm) the decission is being taken place or is it decided by the MAC layer or are there File descriptors set for the ouput devices. When i see the trace file I see that when the interest is flood from 1 initially the time in the trace file shows that 4, 3 and 2 (4 received first, 3 received next, and 2 received last) and after this the trace file shows 2 sends the next interest message to its neighbors, I mean how come this is happening ? I hope u got my point.... Please do reply I checked out ur specified papers but those deal with reinforcement but my doubt is at Initial phase of interest setup. Waiting for your reply. --- John Heidemann wrote: > On Tue, 18 Apr 2006 00:52:22 PDT, shankar setty > wrote: > >Hello, > > > >I have a small problem, i.e., how will Diffusion > API > >decide which node will be the next to send the > >interest. > > > >Lets consider that the initially interest is > >disseminated by node1 and say within the radio > range > >of node1 are node2,3,4 which get the interest from > >node1. Now say node2 is the first to disseminated > the > >interest to its neighbors, now my question is how > is > >that node2 is selected for disseminating interest > but > >why not node 3 or 4 although say node 3 and 4 are > >close to node1 but node 2 is far from node 1. Well > how > >will this Diffusion API select which node will send > >the next set of interest when all the 3 received at > >the same time from node1. > > > >Waiting for your reply. > > > > I believe this question is answered specifically in > several of the > diffusion papers. See for example the section on > reinforcement in the > 2000 MOBICOM paper by Intanagonwiwat et al, or the > section summarizing > diffusion (3.1) in the 2001 SOSP paper by Heidemann > et al. > > -John Heidemann > > P.S. Please do not cross-post questions across many > mailing lists. > That defeats the purpose of having multiple, > separate lists. > > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com From ldbanks at ucdavis.edu Mon Apr 24 14:29:12 2006 From: ldbanks at ucdavis.edu (Lerone Banks) Date: Mon, 24 Apr 2006 14:29:12 -0700 (PDT) Subject: [Diffusion-users] tracefile questions Message-ID: <200604242129.k3OLTCsi003384@cucujus.ucdavis.edu> Hi, I have seen several questions in the archives about tracefiles but no definite answers. My particular question is about packets that are received but based on the tracefile appear to never have been sent. For most packet ids, I see a packet being sent and eventually received (although not always received). Packets not being received is troubling but I just assume this is due to the simulation of a wireless network. More troubling is the reciept of packets with ids that, based on the trace, have never been sent. What causes this to occur? Perhaps I am confused about how ids are assigned to packets. I am running simple-diffusion.tcl as-is with MAC tracing turned off to reduce clutter. Enclosed is a portion of the tracefile (*) mark lines of interest. + -t 0.000001000 -s 0 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 30504276 -k AGT - -t 0.000001000 -s 0 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 30504276 -k AGT *** h -t 0.000001000 -s 0 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 30504276 -k AGT [NEVER RECEIVED] + -t 0.000001000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803425 -k AGT - -t 0.000001000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803425 -k AGT h -t 0.000001000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803425 -k AGT + -t 1.010001000 -s 0 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 30504277 -k AGT - -t 1.010001000 -s 0 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 30504277 -k AGT h -t 1.010001000 -s 0 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 30504277 -k AGT *** r -t 1.010002000 -s 0 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 729236 -k AGT [NEVER TRANSMITTED] + -t 2.510001000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803426 -k AGT - -t 2.510001000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803426 -k AGT h -t 2.510001000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803426 -k AGT r -t 2.510002000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 672080287 -k AGT + -t 2.591003000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803427 -k AGT - -t 2.591003000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803427 -k AGT h -t 2.591003000 -s 1 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 503803427 -k AGT r -t 2.591760827 -s 0 -d -1 -p diffusion -e 0 -c 2 -a 1 -i 729237 -k AGT thanks, L. Banks