Dynamic content-based adaptation of streamed multimedia

Most adaptive delivery mechanisms for streaming multimedia content do not explicitly consider user-perceived quality when making adaptation decisions. We show that an optimal adaptation trajectory (OAT) through the set of possible encodings exists, and that it indicates how to adapt encoding quality in response to changes in network conditions in order to maximise user-perceived quality. The OAT is related to the characteristics of the content, in terms of spatial and temporal complexity. We describe a method to automatically determine the OAT in response to the time-varying characteristics of the content. In this way, as the characteristics of the content change over time, the system can dynamically and intelligently adjust the adaptation process in order to maximise the user-perceived quality. The OAT can be used with any sender-based transmission adaptation policy. We demonstrate content-based adaptation using the OAT in a practical system using two different adaptation algorithms. Furthermore, we show how this form of adaptation can result in differing adaptation behaviour not only as a result of the dynamics of the content but also as a result of the adaptation algorithm being used. Finally, we show how increased feedback frequency does not necessarily improve the behaviour of the adaptation algorithm being used.