Users are changing their traditional communication paradigm based on voice calls or text messages to real-time or on demand video services consumed on mobile devices. In this sense, the transmission of video content considering an adequate Quality of Experience (QoE) in mobile wireless networking infrastructures is a critical issue in both academic and industrial communities. Furthermore, video on demand have a growing consumption over Internet requiring higher bandwidth and lower latency. In this context, a fog computing paradigm can enhance the user experience in wireless networks. Fog computing for video on demand streaming can improve QoE by both video caching and adaptation schemes. However, it is important to evaluate the performance of downloading the videos with diferent codec con_guration and cached closer to the user to measure the gain from the user perspective. We designed a multi-tier fog computing architecture with three levels located in the cloud, nearer the edge and in mobile devices. We evaluated the performance of downloading the video from multiple tier located in distinct geographical with a multimedia application. We assessed in an experimental environment with idle and congested network of streamed videos coded into H.264 and H.265 with three bitrates in a scenario deployed in the FIBRE testbed. We collected QoE metrics, playback start time, freeze times, QoS metric, round-time trip, and energy consumption to analyze the gains for each video configuration. These results showed an important understanding about cache, codec and bitrate schemes in multimedia networking scenarios.
|
