Bugzilla – Bug 1629
Make AODV Default to Disable Hello
Last modified: 2013-08-16 01:40:09 EDT
Overview --------- The hello implementation currently has a few behaviors that may not be ideal: - Hellos are always sent by a node, regardless if node is participating in an active route (bug 1188) or not - Other features in AODV RP are interacting with hellos that should not like RecvRequest and SendRequest (bug 1193) although this has been turned off - Hellos are not suppressed if there was a recent broadcast (bug 1190, it was in this bug that the suggestion was made to disable hellos by default if the change could be validated) Hellos have a requirement level of MAY in RFC (see 6.9.). Patches --------- I have a few patches to propose. I ran the aodv test across all of them and they all passed. I monitored the behavior of a modified aodv.cc example program to determine behavior changes and have found some behavior that lead me to develop several patches. 1) Disable hello by default 2) Disable hello by default, and remove the conditional statement on hello bool in the constructor 3) Disable hello by default, and move the action of the conditional statement from the constructor to start method Build --------- This is the dev build I tried the patches on andrew@ubuntu:~/repos/ns-3-allinone/ns-3-dev$ hg summary parent: 9694:eb7335c2423c tip add virtual destructor branch: default commit: 89 unknown (clean) update: (current) Ubuntu 10.04
Created attachment 1560 [details] Disable hello by default
Created attachment 1561 [details] Disable hello by default, and remove conditional statement
Created attachment 1562 [details] Disable hello by default, and move the action of the conditional statement
Created attachment 1563 [details] All patches testing output
Created attachment 1564 [details] Outputs, notes, and comments of disabled hellos behavior in aodv example
Andrew, thanks for these patches; we'll try to get to them soon after the upcoming ns-3.17 release.
When fixing this, remember to declare fixed also bugs: - http://www.nsnam.org/bugzilla/show_bug.cgi?id=1188 - http://www.nsnam.org/bugzilla/show_bug.cgi?id=1190 or at least to assess if they're still valid. From my understanding, the proposed patches should fix these as well.
Created attachment 1637 [details] Diff from bug1190 which might also fix Bug1629 Diff from bug1190 which might also fix Bug1629 Requires Patch for Bug1522
Created attachment 1640 [details] Long run tests (2000s) Stability
author John Abraham <john.abraham.in@gmail.com> Thu, 15 Aug 2013 22:39:03 -0700 (43 seconds ago) changeset 10161 7ea12911298b