Thesis Topic Details

Topic ID:
New Measurement Techniques for a Science of Networks
Guillaume Jourjon
Research Area:
Networks, Embedded Systems, Configurable system
Associated Staff
Mahbub Hassan
Topic Details
R & D
Group Suitable:
The last five years have seen the production of nearly 85% of the information humankind ever produced. In order to use this data new methods in the broad field of "big data" have been generalised from facebook to the banking industry. Nevertheless, performing data analysis in an ah-hoc manner is neither efficient in terms of computation nor remains accurate as new data may alter the produced result.

In the field of networking, and in particular over the Internet, an accurate estimation of the system is required to perform a proper evaluation as well as a reconfiguration in case of malfunction. As a first attempt to solve this problem, we have developed at NICTA novel measurement architecture based on the idea of a modular and dynamic filtering of measurement streams [1, 2]. Recently this measurement architecture has been chosen to be the cornerstone of a large international project including the University of Massachusetts and NYU among others. We believe that the forthcoming iteration on this measurement architecture, on which your honours project will be based on and you will contribute greatly, will allow to perform a more
scientific sounded research in networking as it should allow the experimenter as well as the network administrators to retrieve only the significant measurement that a large distributed system produces.

During this honours project, the student will be included in small but very dynamic team of researchers in which they will have a large freedom to propose new measurement schemes (e.g. filtering) as well as new measurement applications that could range from basic network impairments discovery to more sophisticated metrics such as security levels.

A more detailed roadmap of this honours project is as follows:
1. Literature review of network measurement architectures
2. Design/modification of our existing measurement architecture for a NetFPGA card. This could include the dynamic filtering of the measurement stream.
3. Implementation in the NetFPGA card of the basic feature of the measurement architecture.
4. Performance evaluation in the context of a Deep Packet Inspection scenario.
5. Going further with the filtering at the lower level.

One of the impacts of this work resides in the development of an open source solution for the monitoring of networking systems. This solution will be cheaper than the existing industry stand (DAG card in particular). Furthermore, it will offer more modularity for the definition of the significant measurements.
Past Student Reports
No Reports Available. Contact the supervisor for more information.

Check out all available reports in the CSE Thesis Report Library.

NOTE: only current CSE students can login to view and select reports to download.