Hi Michael, greenON means that the buffer where i store the Messages for transmission is full. This could happen because the sending task is much slower than the speed at which new packets are generated. This led never lights up so the buffer is never overflowed. (This buffering algorythm is the same one as used on TOSBase to receive and forward packets)
redOn is a send failed. The motes are also sending extra data fields: each failed sendDone() increases an errorcount variable, and each packet has a packetnumber variable. So on the host side i can review these numbers. But only about 13 sendDone() events return with a fail. Calculated over the total amount of packets this is 0.00000001 %. In the period where packets are missing not all packets are missing but too much. When i check the log file on the host, i can clearly see missing packetnumbers in those bad periods. I can sit next to the motes and watch them for 2 hours observing the leds as you suggest. But i don't think i will see any stange pattern in the flashing leds. I think i'm going to start a new test with the radio tuned to another channel and dB. Thank you -----Original Message----- From: Michael Schippling [mailto:[EMAIL PROTECTED] Sent: vrijdag 11 mei 2007 19:00 To: Oussama Chougna Cc: tinyos-help@Millennium.Berkeley.EDU Subject: Re: [Tinyos-help] Radio communication problem issue Just a few questions and suggestions... The greenON means that you need the message buffer before transmission is done? And redON is a send() failed, or exactly what? When you say "The missing packets period" are all packets lost or just some? If some can you see a pattern? There must be something 'beating' between the two transmitters. You could try running them at slightly different sample rates to see if that changes the behavior. Maybe flash an LED in sendDone() and watch it with a scope to see jitter in normal times and any drop outs when things are bad. Do the same thing with send() calls and compare against the done()s. I would also use a least two message buffers so you have some breathing room between messages, and it's a bad idea to count on getting all the messages anyway because some get dropped even in the best of times. MS Oussama Chougna wrote: > Hello, > > > > Using: Tmote Sky, TinyOS 1.x, Cygwin WinXP > > > > I have two motes sampling 4 channels of the AD converter module at 200HZ. > > On the ADChannels a function generator is putting a sine with an offset > applied. > > When 10 samples are collected from each channel (10 * 2 bytes * 4 > channels = 80 bytes), they are packaged in a TOS_Msg and buffered in a > TOS_Msg buffer. > > Another process with tasks is sending the TOS_Msg's from the buffer over > the radio. > > > > *_THIS IS NOT A CIRCULAR BUFFER!!_* > > *_ _* > > To understand my story you should understand that when using a high AD > sampling frequency, the frequency with which you send packets will > increase. More packets are generated! > > > > The TOS_Msg's are received by a TOSBase mote and the data is stored on > the computer, for later visualization. > > > > Leds debug on the mote: > > greenON : TOS_Msg Buffer is overflowed. > > redOn: A packet could not be sent. > > > > > > > > > > *_Scenario 1:_* > > 2 motes sampling the ADChannels @ 20HZ works fine. When I visualize the > sine, it looks good (Like the sine of the function generator). This is > 20HZ sampling so the frequency of packets sending is low. > > No debug leds are lighting up. > > > > *_Scenario 2:_* > > 2 motes sampling the ADChannels @ 50HZ. > > This works fine *for about 2.5 hours*, after this period packets are > missing from *both motes*. The missing packets period is about *30 > minutes*. After this *30 minutes* it works fine again for *2.5 hours*. > > No debug leds are lighting up. > > > > > > *_Scenario 3:_* > > 2 motes sampling the ADChannels @ 200HZ. > > This works fine for *1 hour*, after this period packets are missing from > *both motes*. The missing packets period is *about 45 minutes*. After > this 45 minutes it works fine again for 1 hour. > > No debug leds are lighting up. > > > > *_ _* > > *_Scenario 4: (WorkLoad is too much for TOSBase maybe?????????)_* > > 1 mote sampling ADChannels @ *500HZ*. This works *fine* for the whole > period. So in this scenario packet sending is done in a very high > frequency, *but TOSBase can handle them!!!* > > No debug leds are lighting up. > > > > > > > > *So TOSBase seems not to be the problem because it can handle one mote > sampling @ 500HZ. The packet receive frequency is very high here!!!!* > > *When I increase the sampling frequency, the period where no packets are > missing decreases.* > > * * > > *So the problem here is that when using two motes there is a period with > packets not arriving at TOSBase (missing packets), maybe the motes are > fighting then for the radio channel or something?* > > *Why is it working fine when using low frequency sampling or when using > just 1 mote??? 1 mote has no opponent to fight with. * > > * * > > *These are my conclusions but I I still don't know the exact explanation > for this problem. Why is there a period FOR BOTH MOTES at the same time > missing packets? It is not random missing packets!* > > * * > > * * > > *I hope someone could help me out with this.* > > * * > > * * > > *Thank you* > > * * > > * * > > *Chougna* > > > > > > > > > ------------------------------------------------------------------------ > > _______________________________________________ > Tinyos-help mailing list > Tinyos-help@Millennium.Berkeley.EDU > https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help _______________________________________________ Tinyos-help mailing list Tinyos-help@Millennium.Berkeley.EDU https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
