Abstract:
Video streaming over error-prone wireless networks can result in unsatisfactory user experience, which includes two factors: video distortion due to video compression and...Show MoreMetadata
Abstract:
Video streaming over error-prone wireless networks can result in unsatisfactory user experience, which includes two factors: video distortion due to video compression and packet loss and video smoothness due to the mismatch of video bitrate and channel bandwidth, packet retransmission delay. Existing algorithms have addressed the problem of adaptive video bitrate and optimization of transport layer parameters separately, which can lead to non-optimal solution in terms of user experience. In this paper, we propose a cross-layer design, which integrates an optimization framework for selecting optimal video bitrate and redundancy models under varying channel conditions. In this design, to enhance video quality and satisfy delay constraint of streaming applications, different redundancy levels are configured for different types of video frames based on their importance and video bitrate also is configured adaptively according to selected redundancy and channel bandwidth. Our cross-layer design features a neural network model-based distortion estimator to facilitate the decision of redundancy and video bitrate. This distortion estimator takes into account video characteristics, redundancy models, encoder settings, and channel packet error rate to accurately predict the video distortion for each option of video bitrate and redundancy model under varying channel conditions. Simulation shows that our approach can improve the distortion of video streaming with different types of videos in the different wireless network enviroments.
Date of Conference: 12-15 January 2022
Date Added to IEEE Xplore: 26 January 2022
ISBN Information:
Print on Demand(PoD) ISSN: 1976-7684