Optimisation of Acknowledgements Handling for CMT-SCTP
The Stream Control Transmission Protocol (SCTP) is a general-purpose, reliable transport protocol like TCP. However, unlike TCP, it provides support for so-called multi-homing. That is, endpoints (e.g. PCs, mobile devices, etc.) may be connected to multiple networks simultaneously. In conjunction with the Concurrent Multipath Transfer for SCTP (CMT-SCTP) extension, these multiple paths can be used simultaneously, in order to improve the application payload throughput. CMT-SCTP is currently the subject of research on multi-homed transport protocols, with the goal to standardise it in the Internet Engineering Task Force (IETF) as RFC.
As for TCP, SCTP realises reliable transmission by using acknowledgements. That is, when data has been successfully received, the receiving instance confirms the reception to the sender. The sender may then go on with transmitting new data, or retry sending data that has been lost. Current practical tests with CMT-SCTP shows that in certain cases the acknowledgements processing becomes expensive in terms of CPU processing overhead needed. Therefore, for CMT-SCTP, improved data structures and algorithms are needed in comparison to plain SCTP.
There are already some ideas by the CMT-SCTP developers to make the acknowledgements processing more efficient. However, these ideas have not been realised and evaluated, yet. Based on these ideas, the subject of this masteroppgave will therefore be to design, develop and evaluate an improved strategy to handle the acknowledgements for CMT-SCTP, in order to allow for efficient high-speed multi-path transmissions. The new approach will be implemented into an existing OMNeT++ simulation of state-of-the-art CMT-SCTP. Support by the main developers of this implementation is available. Particularly, this implementation is intended to be used in the context of the ongoing IETF standardisation process as well as for research on multi-homed transport protocol performance in the NorNet project.
This project is *your* chance to get involved into SCTP development and worldwide IETF protocol standardisation as well as into international, top-level research at the Simula Research Laboratory in Fornebu! Are you interested in this challenge? Do not hesitate to contact us!
For more information, please contact Thomas Dreibholz (email@example.com)
What you should know:
- Some knowledge about data structures, algorithms and performance is necessary.
- You should have some programming experience.
What you will do:
- Design and develop a strategy, based on some ideas of the CMT-SCTP developers, for more efficient CMT-SCTP acknowledgements handling.
- Implementation of the new strategy into the existing OMNeT++ framework.
- Evaluation of the developed strategy.
What you will learn:
- You will obtain an understanding of SCTP, particularly of its acknowledgements handling.
- You will get to know the OMNeT++ simulation environment.
- You will learn about runtime performance and efficiency.
- Dreibholz, Thomas: ``Evaluation and Optimisation of Multi-Path Transport using the Stream Control Transmission Protocol´´
- Dreibholz, Thomas et al: ``Implementation and Evaluation of Concurrent Multipath Transfer for SCTP in the INET Framework´´
- Iyengar, Janardhan et al: ``Concurrent Multipath Transfer using SCTP Multihoming over Independent End-to-End Paths´´
- NorNet -- A Programmable Testbed for Measurements and Experimental Networking Research