Comprehensively context-aware approach to guaranteeing multimedia conferencing services

Abstract

Service-Oriented Communication (SOC) is a key research issue to enable media communications using the Service-Oriented Architecture (SOA). Motivated by the necessity to guarantee the service quality of our web-based multimedia conferencing system, we present a Comprehensively Context-Aware (CoCA) approach in this paper. One major problem in the existing end-to-end Quality of Service (QoS) management solutions is that they analyse and exploit the relationships between the QoS metrics and corresponding contexts in an isolated manner. In this paper, we propose a novel approach to leveraging such relationships in a comprehensive manner based on Bayesian networks and the fuzzy set theory. This approach includes three phases: 1) information feedback and training, 2) QoS-to-context mapping, and 3) optimal context adaption. We implement the proposed CoCA in the real multimedia conferencing system and compare its performance with the existing bandwidth aware and playback buffer aware schemes. Experimental results show that the proposed CoCA outperforms the competing approaches in improving the average video Peak Signal-to-Noise Ratio (PSNR). It also exhibits good performance in preventing the playback buffer starvation.