A Dynamic Adaptive Streaming System ForProviding Video Contents to Handheld Device

This paper proposes a dynamic adaptive streaming system for providing video contents to handheld devices. The system has two major features to achieve high quality service. One is the session mobility and the other is adaptability. Session mobility makes a user keep own session even though user changes his/her location and display terminal. Adaptability means that the streaming system fits a streaming content according to the client terminal's capability such as network condition, display resolution, and user's preference. In the system, video contents are adapted to lower or higher quality service according to network conditions. It is done by predicting network bandwidth using a nonlinear autoregressive predictor modeled by neural network. In the implementation section, the paper demonstrated the proposed system on Linux and WinCE environment.