Multimedia Traffic Analysis

Irini Reljin, Branimir Reljin, Marija Zajeganović-Ivančić, Bojana Krstić

This tutorial covers the field of multimedia, in general, with special attention to video compression techniques and standards, and fractal and multifractal nature of multimedia traffic.

Modern multimedia traffic is characterized by a large amount of information, high burstiness and an extraordinary large number of source-destination nodes. The huge amount of information, particularly image and video signals, establishes a hard demand on signal compression.

The main goal in any video compression technique is to achieve as high as possible compression ratio with invisible degradation of video/audio material, implementing as simple as possible algorithm. At the very beginning of the video compression development, the two research directions with opposite goals: low bit rates (for quasi-static video conference sequences, known as 'head and shoulders') and high quality (enabling the adoption of broadcast compression standard), were established. Further development of video compression brought a sharp limit between them, distinguishing the applications targeted to 3G systems (H.261, H.263) and high quality recording and broadcasting, such as the well-established MPEG-2 standard. In the mean time, the multimedia MPEG-4 compression technique, bringing many versions and defocusing the goals of compression (not demanding the lowest bit rate, nor demanding the highest video quality), appeared. A number of individually small improvements, but all producing significant benefits necessary in multimedia, have been introduced. One of the latest compression standards is H.264 AVC that has been adopted by the two standardization bodies: the ITU and the ISO. It brought some kind of bridge between the previous standards - the idea was to enable rather good quality, almost as good as in MPEG-2, at not obviously the smallest bit rates. As it is very convenient for video distribution over Internet and digital video broadcasting for handheld monitors (DVB-H) is the first one in DVB family, which will accept H.264 AVC as a non-MPEG-2 compression. The latest compression standard, called Dirac, brought by BBC, will be discussed here, too.

Analysis of multimedia traffic is based on publicly available video traces, generated by using the bit streams representing the actual bits, and produced as an alternative for video traffic, carrying the number of bits used for the encoding of the individual video frames. Those video traces were analyzed by several methods, including fractal and multifractal analyses of video traces. The simulation results, that will be presented, explore the effects that predictive and bidirectional coding, as well as quantization scales, produce on the multifractal nature of different movies. The importance of this investigation is obvious as video traffic models can be derived from them.

The tutorial will include the three modules:

  1. Overview of video compression techniques - H.265 when and why?

    2.  
  2. Multimedia traffic performances.
     
  3. Fractal and multifractal analyses of compressed video streams.
     

Dr. Irini Reljin

Dr. Irini Reljin teaches courses in Broadcast Technology and Multimedia Technology in undergraduate studies, and Neural Network Applications in Communications, Optical Networks as well as Linear and Non-linear Signal Analysis in the postgraduate studies at the Communications Department at the Faculty of Electrical Engineering, University of Belgrade. She published more than 150 scientific papers, several chapters in international monographs, and several books in Serbian. Papers have more than 20 citations (book, international scientific journals and conference proceedings, PhD thesis, on-line citation by the Ecoregional Fund).

I. Reljin had a number of invited lectures at several universities and institutions in USA, Greece, Romania, Bulgaria, Bosnia and Herzegovina, as well as in Serbia and Montenegro (SCG): Medical Academy of Science, Military Medical Academy, Mathematical Institute of Academy of Sciences. She has been involved in a number of projects, mostly in the field of video and multimedia technology.

I. Reljin is a member of different international professional societies such as IEEE, SMPTE, BSUAE, ProMPEG Forum, as well as national ones: ETRAN, Society of Telecommunications and Gender Team.

Dr. Branimir Reljin

Dr. Branimir Reljin teaches the following courses in graduate and undergraduate studies: Circuit Theory, Digital Image Processing, Medical Informatics, Telemedicine, Artificial Neural Networks, Digital Signal Processing and Network Synthesis.

He has published more than 350 papers in technical journals and conferences, four books and several book chapters, on different aspect of circuit theory, neural networks, digital and medical image processing, fractal- and multifractal analyses. His papers and books were cited more than 60 times in journal and conference papers. He has been awarded twice for the best paper at the largest national (Yugoslav) scientific conference of the Society of Electronics, Telecommunications, Computer Engineering, Automation and Nuclear Engineering (ETRAN). He had a number of invited lectures at several universities and institutions in USA, Greece, Romania, Bulgaria, Bosnia and Herzegovina, as well as in Serbia and Montenegro (SCG): Medical Academy of Science, Military Medical Academy, Mathematical Institute of Academy of Sciences.

Dr. Reljin was a project leader for several projects in the field of Neural Networks, Medical Imaging and Telemedicine, Signal Acquisition and Processing, and currently he is a coordinator of Working Group 5 in European project COST Action 292, "Semantic multimodal analysis of digital media". He is the President of the 3C committee of the Federal Office for Standardization and the member of the Committee of national standards in medical informatics, Group for information technologies, Federal Office for Standardization.

B. Reljin is a member of several scientific and professional societies, an IEEE SCG (former YU) CAS&SP Chairman and the Vice-President of the national professional society ETRAN, as well as a reviewer and a member of the editorial boards of several journals and has been on the program committees of a number of conferences. He is a General Chair of the IEEE co-sponsored conferences Neural Network Applications in Electrical Engineering (NEUREL).

Last updated September 26, 2005