SVCEval-RA: an evaluation framework for adaptive scalable video streaming

Wilder E. Castellanos, Juan C. Guerri, Pau Arce
2015 Multimedia tools and applications  
Castellanos Hernández, WE.; Guerri Cebollada, JC.; Arce Vila, P. (2017). SVCEval-RA: an evaluation framework for adaptive scalable video streaming. Multimedia Tools and Applications. 76(1):437-461. Abstract. Multimedia content adaption strategies are becoming increasingly important for effective video streaming over the actual heterogeneous networks. Thus, evaluation frameworks for adaptive video play an important role in the designing and deploying process of adaptive multimedia streaming
more » ... ms. This paper describes a novel simulation framework for rate-adaptive video transmission using the Scalable Video Coding standard (H.264/SVC). Our approach uses feedback information about the available bandwidth to allow the video source to select the most suitable combination of SVC layers for the transmission of a video sequence. The proposed solution has been integrated into the network simulator NS-2 in order to support realistic network simulations. To demonstrate the usefulness of the proposed solution we perform a simulation study where a video sequence was transmitted over a three network scenarios. The experimental results show that the Adaptive SVC scheme implemented in our framework provides an efficient alternative that helps to avoid an increase in the network congestion in resource-constrained networks. Improvements in video quality, in terms of PSNR (Peak Signal to Noise Ratio) and SSIM (Structural Similarity Index) are also obtained. (Dynamic Adaptive Streaming over HTTP) [5]. All these solutions have different objectives, but agree on the target goal of assisting the network to provide adaptive media streaming. Regarding the content adaptation, Scalable Video Coding (H.264/SVC) [6] is a coding technique that offers an efficient alternative for adaptive content distribution. It was standardized in 2007 as an extension of the H.264/AVC (Advanced Video Coding) standard, and several SVC-based solutions for video transmission have been proposed so far, for example the works presented in [7-12]. A video stream encoded with H.264/SVC consist of several layers (one base layer and multiple enhancement layers) each with different characteristics of quality. Moreover, the adaptability feature of the bit rate, inherent of the scalable coding scheme, provides a natural and efficient way to adapt the video source rate to the available network throughput. This bit-rate adaptation has a low computational cost in SVC since it is performed without the need to transcode: layers can be added and removed with low complexity operations. In addition, improvements in the subjective quality of videos encoded with SVC, compared with H.264/AVC, has been demonstrated in some studies such as [13, 14]. An open issue in the development of adaptive video delivery systems is the lack of freely available video evaluation toolset where researchers and network developers can test the performance of their network-adaptive techniques. One possibility for support of the latter is to use the Evalvid-RA (Evalvid Rate Adaptive) framework developed by Arne Lie et. al. [15]. Evalvid-RA is a simulation tool that supports rate adaptive video transmission using MPEG-4 VBR (Variable Bit Rate) videos. The adaptive mechanism of this platform involves switching between different pre-encoded versions of the video (each with different levels of quality) according to bandwidth variations. However, a more efficient solution is provided by H.264/SVC, which supports encoding of a video in different qualities within the same bit-stream. This includes different resolutions, different frame rates (fps) and different quality levels. Although some evaluation platforms have been proposed to assess the transmission of SVC videos, such as SVEF (Scalable Video-streaming Evaluation Framework) [16], EvalSVC [17] and myEvalSVC [18]. However, none of these evaluation platforms enables networkadaptive streaming operation at the video source, which always injects video streams with the highest video quality (i.e. at the highest bit rate) to the network. We propose in this paper a novel simulation framework called SVCEval-RA (SVC Evaluation Platform for Rate-Adaptive Video), which can be used in the assessment of rate-adaptive video streaming using Scalable Video Coding. Our approach is a modified and enhanced version of the myEvalSVC tool-set. More precisely, we have implemented in SVCEVal-RA a procedure to adapt the bit rate of the traffic source adding or removing SVC layers from the video stream based on the estimation of the available bandwidth. This procedure (referred to hereafter as the Adaptive SVC scheme of SVCEval-RA)
doi:10.1007/s11042-015-3046-y fatcat:cofzma7yxjdqfnja5gwkjie3sa