hi all, just to thank everybody who helped me to solve the problem about packet loss rate. my solution is working. i used some dbg messages to notify each packet loss; then programmed a small script which calculate the number of occurence of this "notification". Hopefully, it works well. Best regards 2008/7/26 Omprakash Gnawali <[EMAIL PROTECTED]>
> On Fri, Jul 25, 2008 at 3:22 PM, fatima zohra <[EMAIL PROTECTED]> > wrote: > > > > 2008/7/25 Omprakash Gnawali <[EMAIL PROTECTED]>: > >> > >> On Fri, Jul 25, 2008 at 7:37 AM, fatima zohra <[EMAIL PROTECTED]> > >> wrote: > >> > > >> > > >> > 2008/7/25 Omprakash Gnawali <[EMAIL PROTECTED]>: > >> >> > >> >> On Wed, Jul 23, 2008 at 11:45 AM, fatima zohra < > [EMAIL PROTECTED]> > >> >> wrote: > >> >> > > >> >> > > >> >> > 2008/7/23 Omprakash Gnawali [EMAIL PROTECTED]: > >> >> > > >> >> >> > >> >> >> If you put sequence numbers in the packet, you can find out which > >> >> >> packets were not received. For example, you sent sequence numbers > 1 > >> >> >> through 10 but and received 1,2,3,4,6,7,8,9,10 then you know > exactly > >> >> >> how many packets were not received. > >> >> >> > >> >> >> - om_p > >> >> > > >> >> > actually this isn't my scenario; > >> >> > here is an example of my protocol mechanism: > >> >> > we assume Node1 neighbor of nodes 2 & 3, and these two nodes are > >> >> > neighbors > >> >> > of source S (S will send 10 msgs for example) > >> >> > if 3 crashes so node 1 won't receive any more packets from 3. which > >> >> > means if > >> >> > >> >> You said there is one source but why is node 3 sending packets? > >> > > >> > because node 3 will forward what the source is sending in order to > reach > >> > the > >> > sink (let's say for example sink = node1 to make it easier) > >> >> > >> >> > S sends messages (from numSeq = 5 to 9) node A will never notice > >> >> > that > >> >> > he > >> >> > >> >> Node A is mentioned here for the first time. Is it node A or 1 or 2 > or > >> >> 3? > >> > > >> > sorry, it's node 1 (which isn't a "direct" neighbor to the source S) > >> >> > >> >> > missed the 5 last nodes (because he doesn't communicate with S > >> >> > directly > >> >> > , > >> >> > >> >> Nodes or packets? > >> > > >> > sorry again, it's "packets" > >> >> > >> >> > and node 3 is turned off for ever and node 2 won't help him to know > >> >> > what > >> >> > node 3 lost as messages (from S or other sources) because seq > number > >> >> > is > >> >> > unique and different for each node (the 5th seqNum for node 2 may > be > >> >> > the > >> >> > 1st > >> >> > message of the source S) > >> >> > now i wonna to calculate packet loss rate in my network since i > have > >> >> > a > >> >> > set > >> >> > of nodes who crash and will lose some sent messages. > >> >> > am i clear now ? if yes, how to handle this situation. > >> >> > thanks in advance. > >> >> > best regards. > >> >> > >> >> Your description has some typos and could use some clarification. > >> >> > >> >> - om_p > >> > > >> > i mention in addition of what was said that: this simulation is done > to > >> > study the behavior of the network after some nodes crash. > >> > so, first i choose some faulty nodes, then , turn them off for the > rest > >> > of > >> > simulation (they won't receive any packet in the future). in the end > of > >> > my > >> > simulation, i calculate my packet loss rate (i.e number of sent > messages > >> > but > >> > not received by faulty nodes once they are stopped). > >> > is it clear now ? > >> > >> Almost. > >> > >> So you have a topology that looks like this: S --> 3 --> 1 but 3 is > >> turned off so 1 is not receiving any packets but you want to find the > >> loss rate on the link 3-->1? > >> > >> - om_p > > > > that's it, > > but S-->3-->1 is just a piece of my topology (i just noticed it to > clarify , > > my topology is larger than that). > > let's work with this small scenario. > > what's the way to find the packet loss rate ? > > Considering your scenario and objective, you should have each node > send packets periodically just for link estimation. Then you will be > able to estimate link quality even in the absence of data packets. > > - om_p >
_______________________________________________ Tinyos-help mailing list [email protected] https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
