Video decoding for reduced jitter and improved resource usage

Abstract. In this paper we propose a novel video stream adaption strategy based on the gravitational task model [Guerra and Fohler 2008] for reduced jitter and improved resource usage. In a competitive market, electronics producers follow the trend of minimizing production cost by constraining the amount of resources in devices, but QoS should be kept. Multimedia applications are intrinsically target sensitive because frames must be displayed in specific points is time for maximized perceived quality of video. If system resources are insufficient to allow all frames to be displayed at these points, reducing the overall deviation will minimize the impact on the perceived quality of video. In some situations resources are not even enough to decode all frames, and hence, some frames should be skipped to avoid wasting resources. Skipping or delaying frames, however, should be performed only when accounting for a better resource usage. Our video adaptation strategy addresses the trade-off between delaying and dropping frames depending on the resource availability of the system based on the gravitational task model. The gravitational task model is exploited in this work for its capability of expressing target points for the execution of jobs and an elliptical utility degradation as a function of this point. Simulation results using real trace files show how this approach accounts for the jitter and allows a better usage of system resources. 1