View source for Slicer/Features/Middleware

<< [[Slicer/Features]]


  For updates see [[OpenIGTLink]]


=Title=
Middleware

=Names=
*session leader

*participants

=Technology/Function description=
== Goal ==
* Slicer + IGSTK

== Technologies ==
* Look at Publish subscribe ORB/ACE - [http://portals.omg.org/dds DDS] [Peter]
* Socket IGSTK [Junichi Patrick] 
* Navitrack - phase out?
** Hard to debug
** Not modular (some modules can't be disabled)
** No error handling - just halt
** Not as simple and clean as just using the standard TCP/IP socket
** Still problems with image transfer under Windows and possibly other OS (packet size is not based on the size allowed by OS)
** Security: can be flooded with packets, no authentication

==Requirements==
* Reliable for command transfer (all data transferred, in order)
* Fast for tracking data transfer (only the latest coordinate is required, can lose data)
* Image transfer (500K or more)
* Data types (struct: string, integer, float)
* Data type checking (number is correctly formatted etc)
* Time out (device is down)

==Would be nice to have==
* Priority
* Point - multi-point data transfer

==Questions==
* UDP - fast, no delay due to ACK packages
* TCP/IP - reliable, easy
* Security
** malicious packets or sent by mistake
** CRC and auth.?
** pop up a dialog box before accepting data? (just one, to avoid flood)
** same issues with [[Slicer3:Slicer_Daemon]]
* Firewall?

==See also==
* [http://www.cisst.org/wiki/MRI_robot/MIT_2007_06_27_Meeting discussion about data transfer]

=Potential application in IGT=
* Tracker data transfer
* Image transfer from MRI
* Robot coordinate transfer
* Robot control
* Periferials: OpenTracker/Navitrack has 3D mouse driver etc.

=Which part is already in Slicer=
* IGSTK intra process communication

=What is not=
* IGSTK inter-process communication

=Who among us will take leading role?=
* Junichi
* Georgetown
* JHU
* Queen's

=Links=
* [[2007_Slier_IGTWS_homework_GT]]
* [[2007_Slier_IGTWS_homework_NCIGT]]