Video traces contain information about encoded video frames, such as frame sizes and qualities, and provide a convenient method to conduct multimedia networking research. Although widely used in networking research, these traces do not allow to determine the video quality in an accurate manner after networking transport that includes losses and delays. In this work, we provide (i) an overview of frame dependencies that have to be taken into consideration when working with video traces, (ii) an algorithmic approach to combine traditional video traces and offset distortion traces to determine the video quality or distortion after lossy network transport, (iii) offset distortion and quality characteristics and (iv) the offset distortion trace format and tools to create offset distortion traces.
Patrick Seeling, Martin Reisslein, Frank H. P. Fit