Classification of a video stream is an essential preliminary step to estimate the bit loss assess the Quality of Service when the video stream is transmitted over a communication network. In this paper, we classify the video by the average frame size and estimate the bit loss for each class when the bitrate exceeds the capacity of the bottleneck link. The video stream under study is encoded using the explicit slice-based H.264/AVC encoding scheme.which is recently developed. This scheme reduces the burstiness of regular H.264/AVC encoded video by removing the traditional GoP structure. Instead, a repetitive combination of intracoded and predicted slices is employed, thereby introducing a specific dependence structure in the video data. We consider a bufferless model of the communication system and evaluate the channel capacity required to give a maximum allowed loss rate for each class.Due to the high variability, non-stationarity and non-homogeneity in the underlying video data, the classes obtained are checked regarding the dependence and the distribution structure of the data.We evaluate the channel capacity required to fulfill a maximum allowed loss rate for each class.The high quantiles of the losses for each class are estimated.
