A Video Frame Dropping Mechanism based on Audio Perception

Video streaming applications are more and more present in our life, but despite the advances of network technologies, several users experience QoS problems. This is mainly due to the high bandwidth requirements of these applications that contrasts with the network bandwidth limitation. To mitigate these QoS problems, video frame dropping mechanisms are often used for adapting the video stream to the network conditions. The selection of the video frames to drop is done considering the perceived quality of the video play out; audio perception is not considered in the selection process. In this paper we show that by taking into account only the video play out quality, audio problems arise very frequently. Hence, we propose a video frame dropping mechanism that takes into consideration the perceived quality of both audio and video play out. A comparison with other video frame dropping techniques is carried out and experimental results show that, although the video play out quality is similar, the audio play out quality is completely different. Our mechanism slightly affects the audio quality, while other techniques strongly affect it. Therefore, by using our mechanism, benefits are remarkable.