What happens if you create links that are connected with a gain of 0 and set a constant noise floor at, say, -115. (Don't connect nodes which should not communicate. If a node has a link WORSE than the noise floor, the simulator will choke waiting for a CCA.)
Following the above, do "packet collisions" still play an effect? On Thu, Aug 21, 2008 a t 4:56 PM, <[EMAIL PROTECTED]> wrote: > Hello all, > > I'm simulating a protocl in TOSSIM, which requires me > to modify the TOSSIM source code so that it provides me with a > lossless network. I have tried several things, but I am still > unsuccessful and would appreciate any help on this matter. > > I began by using root permissions to modify the file > UscGainInterferenceModelC.nc under $TOSROOT/tos/lib/tossim/ > > I commented out line 220 to remove the possibility of packetloss: > > // rcv->lost = 1; > > Now, at this point, there is still the possibility of packet loss due > to collisions. The MAC layer proved to be insufficient to eliminate > the problem fully. Therefore I also commented out line 144 to > eliminate collisions. > > // list->lost = 1; > > I put these changes into a copy of the original file called > UscGainInterferenceModelC.nc_no_loss_no_coll, which I keep in my > project directory. The original, unmodified file is copied into my > project directory as UscGainInterferenceModelC.nc_original. Prior to > compiling my project, I copy the modified file into the tinyos > dirctory using the following command in Ubuntu Linux: > > gksudo cp > $PROJECTDIR/UscGainInterferenceModelC.nc_no_loss_no_coll > $TOSROOT/tos/lib/tossim/UscGainInterferenceModelC.nc > > after compilation, the original is copied back in an analogous manner. > > After these changes, I ran my program and found that there was still > packet loss in the network. After some work I began to suspect that my > changes had not taken effect. To make sure, I went back to > UscGainInterferenceModelC.nc and changed it again by removing a > semicolon in an obvious spot. This change breaks the component, thus > if I get a compiler error in the affected spot, then I know the file > is being recompiled. When I attempted to recompile my project, it > succeeded without any such error. Therefore, it is now clear that my > attempt to modify $TOSROOT/tos/lib/tossim/UscGainInterferenceModelC.nc > is not taking effect because the make system will not cause that part > of the tinyos-2.x source tree to recompile. > > So now I attempted to find a way to force that file to recompile. I > first attempted the following command: > > gksudo touch $TOSROOT/tos/lib/tossim/UscGainInterferenceModelC.nc > > This updates the "Modified" date and time on the file to the current > time, in the hope that the make system will recogize that the file has > been changed and then recompile it rather than linking to an existing > object file somewhere. Unfortunately, while the date was changed as > expected, another compilation of my project did not force > UscGainInterferenceModelC.nc to recompile. > > I then searched the various subdirectories under $TOSROOT in hopes of > finding the original object file (presumably with the .o extension). > In theory, if I can delete the file using root permissions, then the > make system will be forced to recompile the source, as it has no objct > file to link to. Yet when I searched, I could not find that file, or > any other obect file whatsoever in the tinyos-2.x tree. I also tried > deleting the tossim.o file that appears in $PROJECTDIR/simbuild after > compilation of the project, but still no success. > > I have also tried manualy modifying the file linkgain.out rather than > running the LinkLayerModel java program. I hoped to get better links > this way. I tried to set the gain at -130.00 if I DON'T want two nodes > to communicate, and I tried setting gains of -1.00 or -5.00 if I DO > want them to communicate. The noise entries were all set with the > value pair -100.00, 3.50 But with these changes, no nodes can > communicate at all. > > It is essential that I get the network to be either perfect or near > perfect if my work is to continue. Any help would be greatly > appreciated. > > Thanks > > _______________________________________________ > Tinyos-help mailing list > [email protected] > https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help > _______________________________________________ Tinyos-help mailing list [email protected] https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
