Welcome to the web site of QOLSR


Updated draft submission
2006-03-15
An update to the individual draft version (02 to 03) has been submitted. The draft focuses on QoS in ad hoc networks using OLSR (QOLSR).
Major changes from version 02 to version 03:
  • Standard MPRs are used to reduce the flooding of broadcast packets in the network instead of QMPRs.
  • TC messages are generated by QMPRs and relayed by MPRs.
  • A new algorithm for QMPRs selection.
  • QOLSR routes contain only QMPRs as intermediate nodes between a source destination pair.
  • Using the QOLSR protocol, a node keeps information about the next hop alone. So, it loses the ability to distinguish the part of used capacity that is induced by the given flow from other ones. There are cases in which a source node continuously changes a flow's next hop in response to the change of available bandwidth on its path and cannot tell apart the traffic induced by itself from traffic generated by other nodes. This situation is called the oscillation problem. Two solutions are proposed to resolve this problem: at the source or distributedly by each hop in the path.
Updated draft submission
2005-04-15
An update to the individual draft version (00 to 01) has been submitted. The draft focuses on QoS in ad hoc networks using OLSR (QOLSR).
Major changes from version 00 to version 01:
  • Generalization of all possible QoS metrics. Available bandwidth and delay metrics are the default parameters for QOLSR. All other metrics are optional and used according to QoS flow requirements or network scenarios.
  • Introduction of an efficient QoS routing algorithm based on Lagrange relaxation, which provides a polynomial heuristic solution to multiple-metric constraints.
Admission control and reservation mechanisms under multiple interfaces will be integrated in next versions.
Updated website
2005-02-04
The website has now up-to-date information. The project description is made more readable but goes less into details. The research team is updated (we miss Anelise already and welcome Thomas). Take a look inside the publications section to get more detailed information about the project. Simulation code for the OPNET simulator is now available for download. The "Working" code section changed into "Implementation". As usual, please send any remarks/questions to .
Usability fixes
2005-01-31
Now Qolyester should be more easily usable. All the /proc/sys/net/... stuff is managed automatically at startup, to make it easier to use. Still some things related to route management should be done automatically but a proper way to do it is yet to be found (in doubt, read the README file).
Bughunting and fixes season
2004-05-19
Lots of nasty bugs have been fixed lately, so you may give the snapshots a try, each time one goes out. =)
Even more fixes
2004-05-06
If you've been testing Qolyester before, and dumped it your project wastebasket because it did not route packets correctly after all, well you can give it a second chance: a major bug regarding flushing routing tables to the kernel has been fixed.
Fixes again and again
2004-05-04
Now routing should be performed properly, the topology graph being correctly managed.
New snapshot, lots of changes (again)
2004-05-03
This time the network interface management has been changed. Structures are provided to allow easy in-flight interface removal and addition. Along this, tools for interface address addition and removal have been added. These two features should enable easy change of interface address, for node autoconfiguration schemes to work. 2-hop neighbors are now better taken care of during routing table calculation and that makes Qolyester more strictly compliant with the RFC (thanks to the people from FTR&D for letting me come across this bug). This snapshot has almost not been tested, so please report any problems. =)
New snapshot, lots of changes
2004-04-07
A lot of changes in this snapshot, that may have broken many things. These are mostly improvements of the precision with which events get triggered in time. This has not been tested extensively, so if you experience problems, feel free to email us. If you need to use a more "stable" version, you're probably better off using some older snapshot.
Link Hysteresis is in the place
2004-04-03
The Link Hysteresis feature has been implemented in the latest snapshot. IIANM, this is the last lacking feature from the RFC. Now all the mandatory and optional OLSR features are supported.
A new snapshot with Doxygen documentation included
2004-03-16
It seems that Doxygen 1.3.6 works. Dunno if any older version works too, though. So this snapshot has documentation generated with Doxygen included in the tarball. Some cleaning up of the comments and some more comments overall...
More cleaning up...
2004-03-02
The project description has been cleaned up. References and figures remain to be dealt with, though (any volunteers?).
Cleaning up the project description
2004-03-01
The project description is being cleaned up. Mostly mathematical formulae are being rewritted using HTML constructs that make them more readable. There is some cleaning work left to be done and several figures left to be drawn and put into place. This last idea might not be realized that soon, because we have more important work to do right now.
New QOLSR web site under construction
2004-02-03
A new web site is being worked on by yours truly. For the moment it is based on HTML 4.01 Strict and CSS 2.1, so if you're using some lousy browser (read: IE), the page might not render as expected. IE compatibility will be worked on as soon as the global structure is finished.
New Qolyester snapshot: 20040130
2004-01-30
This new release is merely a small "feature"-fix to disable Doxygen by default and allow users to specify the --with-doxygen flag to force doxygen use.
Bandwidth calculation using HELLOs
2003-12-04
There would be a way indeed, if we suppose that any radio interference is produced by one of the neighbors participating in the MANET. While this is utterly unrealistic, it proves quite accurate in theory, at least.
Delay measurement using HELLOs
2003-11-28
We have tested a simple way to measure RTT (round-trip-time) delay using only HELLOs. In fact, the drift of the clocks between two HELLO messages is neglected.
Webmaster: Ignacy Gawędzki