For more information on Dr.de Fréin, please click here
Dr Ruairí de Fréin, a Researcher at the Telecommunications Software & Systems Group (TSSG) attended and presented a paper, “Ghostbusters: A Parts-based NMF Algorithm”, at the 24th IET Irish Signals and Systems Conference ISSC in June 2013.
TSSG was also represented at this conference by Mohamed Adel who presented “Online Estimation of Voip Quality-of-Experience via Network Simulation” on behalf of his co-authors Haytham Assem (NUI Maynooth, Ireland), Dr Brendan Jennings (TSSG-WIT), Dr David Malone (NUI Maynooth, Ireland), Jonathan Dunne, and Pat O’Sullivan –both from the Systems and Performance Engineering Lab., IBM Dublin, Ireland.
Over the past 24 years, ISSC has become established as the premier conference in Ireland addressing all aspects of signals and systems. The conference focuses on Digital Signal Processing, Control and Communications, and encompasses algorithm and system modeling, design, and implementation, for a broad range of applications.
Dr de Fréin presented an algorithm –the eponymous “Ghosterbusters”– and supporting results which show how to construct an exact nonnegative matrix decomposition. Traditional methods give a good approximation but the decomposition is not exact. By computing a nonnegative hierarchical closure structure of the matrix (a Galois lattice) Dr de Fréin demonstrated the data had auniqueordered overcomplete dictionary representation. Parts-based constraints on these closures were then introduced to specify and supervise the form of the solution, forming a general framework for guided matrix decomposition. Dr de Fréin presented results on two standard image processing problems. He demonstrated that the results were exact and correct.
Mohamed Adel presented work which was carried out in collaboration with IBM, Ireland which described a testing framework that can provide onlines estimates of audio and video call quality on network paths, without requiring either end-user involvment or prior availability of audio/video sequences or network traces. This framework included a tool that emulates the audio and video traffic of IP calls and employs an extended E-Model to measure the audio quality and VQM to estimate video quality. The tool emulates network impairments to run experiments in different network conditions. Mohamed presented results which demonstrated that the quality measurements acquired using the framework compare well with the most commonly applied industry standard for objective voice and video offline testing –PESQ and PSNR.
Dr de Fréin is currently a research fellow with the Emerging Networks Laboratory at TSSG