|Supervisor:||Dr.-Ing. Christian Herglotz (Room 06.026)|
|Faculty:||Prof. Dr.-Ing. André Kaup|
Video decoders on portable devices like smartphones or tablet PCs consume a large amount of energy such that, depending on the bit stream, the battery charge can be drained quickly. Recent research showed that so-called bit stream features can be used to estimate the required decoding energy. With the help of this information, it is possible to generate energy saving bit streams.
To give a vivid explanation for a bit stream feature, one can imagine that such a feature corresponds to a certain routine in the decoding process that can be executed multiple times during the decoding of one video bit stream. This routine consumes a rather constant amount of energy upon each execution. Hence, summing up over all features, the energy to decode a complete bit stream can be estimated.
In this work it shall be investigated which of the features are relevant for estimation and what their value means. Furthermore, depending on the method the values were determined, they can be checked for plausibility. As a goal of this research we want to obtain deeper insights into the decoding process and the energy estimation model.