VBR video frame size prediction using seasonal ARIMA models

Increasing quality of service (QoS) requirements demand continuous development of new and advanced traffic prediction methods. This is especially true for variable bit rate (VBR) video traffic. This paper provides insights into current approaches and solutions in the area of online frame size prediction methods for MPEG4 video, and proposes a prediction method based on seasonal ARIMA models. Proposed method is effective in frame size prediction and promises up to 30% better results than alternative methods.