UAN AUV models review

82 views
Skip to first unread message

Tom Henderson

unread,
May 11, 2011, 1:03:25 AM5/11/11
to ns-3-reviews, andrea....@gmail.com
copied from ns-developers list:

On 04/19/2011 11:37 PM, Andrea Sacco wrote:
> Hi all,
> I just want to alert the community I have (finally) released the UAN AUV
> modules, including mobility models, energy models and battery models, for
> Autonomous Underwater Vehicles; in particular, there are two models (but
> their number can be extended), one for Seaglider AUV and the other for Remus
> AUV.
> I've pushed the source files into my personal repository at
> http://code.nsnam.org/andrea/ns-3-dev-uan-framework/ but I'm waiting for the
> green light to merge them into ns-3-dev.
> The new files are located in src/uan/auv
>
> For any question or problem, please email me here in the list or directly.
>
> Best Regards,
> Andrea Sacco

Andrea, I reviewed this AUV extension again (it has already been
reviewed in the past during GSoC) but have some issues that need to be
resolved before merging:

1) I would recommend to merge the auv/ directory files into the
existing uan/ or else make auv/ a separate module, but do not embed
auv/ as a module within the module. It seems to me that just merging
it into uan/ would be appropriate.
2) you are introducing some new API to class WaypointMobilityModel;
namely:

* The user can set an underlying mobility model to use for position
validation,
* the only constraint is that the model must be a
* ConstantVelocityMobilityModel or a child of it.

/**
* \param model mobility model to be used as waypoint validator
*/
void SetMobilityModel (Ptr<ConstantVelocityMobilityModel> model);

/**
* \return the underlying mobility model
*/
Ptr<ConstantVelocityMobilityModel> GetMobilityModel (void) const;

however, in reading the new code, it wasn't clear to me why you want
to do this vs. handling this coupling between the two mobility models
outside of the class instead of embedding it within the
WaypointMobilityModel class, or creating a new hybrid class with some
Waypoint and some ConstantVelocity properties. Could you live without
modifying WaypointMobilityModel (what are the implications of doing
so)?
Reply all
Reply to author
Forward
0 new messages