Omprakash, Thanks for your detailed comments.
Interesting point about the LPL. I believe the nodes are sleeping looking into the low power consumption (which is still more than what we calculated/estimated and this may be the reason). I'll take the multimeter and measure the amount of current drawn by nodes which are a hop away from the root. So, what is your suggestion here? Increase the root beacon interval? We tried that (increase to 10 and 20 sec) and the amount of losses seemed to increase (no hard numbers though). We keep the root plugged into the electricity so no concerns for its power consumption. Other than the power issue you pointed out, have you ever observed reliability problems with CTP when LPL was enabled? What were the numbers (i.e. sleep interval etc.) you used for testing? As for the snooping, what should I expect to see if I snoop next to the lost node? Could you refer me to a document describing the LEEP frame? We are using a platform with msp430 and cc2420. Minor modification of TelosB. Finally, about the packet loss when the nodes are multi-hops away: we definitely observe packet loss much more than just 1%. It is more like 30-40%. Still no hard numbers though. Any idea how we can debug this problem? Thanks, Bulut On Mon, Apr 7, 2008 at 12:16 PM, Omprakash Gnawali <[EMAIL PROTECTED]> wrote: > On Sun, Apr 6, 2008 at 2:00 PM, Bulut ERSAVAS <[EMAIL PROTECTED]> wrote: > > Hi Phil, > > > > Thank you for your comments as well. Our application is based on > > MultihopOscilloscope with no significant changes. We added LPL > > functionality, would this cause any issues (sleep interval is 1000ms)? > In > > addition, we inserted Dissemination capability. > > We have done some testing with CTP+LPL. The problem is the root is > generating beacons every second and if you are using LPL with a sleep > interval of 1000ms, the nodes are probably not sleeping. Are the nodes > sleeping much at all? My guess is, they are not because the root is > keeping the medium busy all the time. > > > > > > If, let's say, the RSSI worsens by about 10dB and a node looses its > > neighbor; it will be able to re-discover the neighbor if RSSI get backs > to > > normal, right? > > > > Yes, that is what should happen. When you observe this problem next > time, you can snoop the packets. The CTP beacon is encapsulated in a > LEEP frame so you should be able to see if the node has any neighbor > or not... > > If you reboot and the problem goes away, CTP has a bug... > > What platform/radio are you using? > > To answer your earlier question, yes, we do observe fewer packets from > nodes farther away, but we are talking about a difference of less than > 1% between nodes that are 1 hop and 4 or 5 hops away, in our > experimental scenarios. Yours is different, of course.What kind of > loss are you seeing? > > - om_p >
_______________________________________________ Tinyos-help mailing list [email protected] https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
