A QoS guarantee approach for multimedia software system

Abstract

Quality of service (QoS) is the overall performance of a computer network, which not support or guaranteed well with the current Internet technology. It is hard to track, locate, and recognize the reason of a QoS violation in volatile situations. Even though this issue is solved, other issues may rise, such as the Internet service providers not willing to open their management interfaces. Besides, not all Internet hardware can support QoS controlling. End-to-end QoS controlling is an effective way to overcome these issues. The existing end-to-end QoS solutions have a problem that they analyze and exploit the relationships between the QoS metrics and contexts in an isolated way. This paper proposes a novel comprehensively context-aware approach to exploit these relationships based on the Bayesian networks and fuzzy set. This approach divides into three processes: context discretization, context and QoS metric mapping, and QoS quantitative guarantee. We validate the effectiveness of the proposed approach in our real multimedia software system and compare the performance of our approach with other existing bandwidth-aware and playback-buffer-aware schemes. The experimental results demonstrate that the proposed approach performs better than existing solutions in improving the video Peak Signal-to-Noise Ratio.