On Fri, Sep 12, 2008 at 11:12 AM, Gaurav Chandwani <[EMAIL PROTECTED]> wrote: > Hi All, > > Thanks Om and Razvan for the LPL papers, would go through them ... > > Chetan, i have tried multiple permutations and combinations in my network > with maximum number of nodes being 30. > > Single-Hop - RadioCountToLeds(original and also with modifications specified > in the previous mail) + BaseStation > > Multi-Hop - MultiHopOscilloscope(CTP - 4bitle,le and LQI) > > SingleHop + MultiHop - RadioCountToLeds + MultiHopOscilloscope(modified to > also receive and forward broadcast messages from RadioCountToLeds to the > Base Station). LQI worked better in this scenario than CTP while it should > be otherwise according to the mailing lists. > > As I described in the previous mail, little changes in the topology/codes > have significant impact on packet-loss behaviour. > For example I added another snoop interface in MultiHopOscilloscope to > receive and forward > RadioCountToLeds messages and this leads to a big increase in packet-loss, > probably the forwarding/snooping engine at the lower level(queues) etc > interferes with the snooping(receive and forward of RadioCountToLeds > messages) at the application level. I am trying to use the debugging > information to pinpoint the issues.
I don't think there is any software/buffer interference here. Any problem might be due to potential channel saturation you are causing. > I think if other people have done similar tests, we should share the results > to understand whether its our network/topology-specific problems or its a > general problem faced by everyone. So, if you can tell me abt your testbed > and results, wud b helpful. If you want more information about my testbeds, > can send you the details. People probably did lots of tests similar to yours before developing LPL and other dutycycing protocols. So, I would look at them and see if they do not satisfy your needs. If you want to build another dutycycling protocols, it might be good to use the existing dutycycling protocols as a base unless you are trying to do something radically different. If you come up with ways to improve/extend the existing dutycycling approaches, that will be a great contribution to the community. - om_p _______________________________________________ Tinyos-help mailing list Tinyos-help@millennium.berkeley.edu https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
