U.S.-Turkey Collaborative Research: Cross-Layer Approaches for Interactive Media Streaming

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

0352824 Altunbasak Description: This project supports collaborative research between Dr. Yucel Altunbasak, Department of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, Georgia and Dr. Mehmet Civanlar, Department of Computer Engineering, Koc University, Istanbul, Turkey. They propose to study the development of a new framework for multimedia streaming that integrates the application and network layer functionality. Timely delivery of the media being a critical task, new application-aware streaming methods have to be developed to support these services better. Real-time media distribution imposes stringent bounds on the packet loss and delay to achieve a minimally acceptable quality. Under no quality-of-service guarantees, satisfying these demands requires intelligent and network adaptive source coding plus innovative use of the existing network resources. The PIs will investigate methods based on vertical integration in the conventional protocol stack. The PIs propose models for multi-path streaming and plan to develop a multi-path selection method that chooses a set of paths maximizing the overall quality at the client under various constraints. Dr. Altunbasak and his students have worked extensively on video processing and video communications over error-prone networks and developed several new algorithms to improve the existing standards. Their strength in this area is complemented by the extensive experience of Dr. Civanlar, who is an expert in the areas of computer network architectures and protocols for multimedia. This combination of expertise is appropriate for the project. Scope: The main contributions of this research will include: development of multi-path streaming models, particularly for multiple description encoded video streaming; development and evaluation of techniques for the use of network support to aid the path selection function; characterization of multi-path streaming methods and their evaluation for path selection problem; use of existing overlay infrastructure to provide proof of concept for the proposed path selection method; design of a reusable experimental platform for experimenting with proposed path selection method over intercontinental links. The work will go beyond the path selection methods in today's simpler environments into the near-future environments that will involve much richer and more demanding multimedia communication services. The research results will enhance performance of important applications over the Internet. The project will encourage student exchange, and specific classes and projects in both institutions that will involve undergraduate students. The project will promote developing solutions for distance learning classes. The society will indirectly benefit from any improvements on the 'information highways' and the related services.