Predicting energy consumption of MPEG video playback on handhelds

One of the principal concerns of wireless computing is the limited battery life of mobile devices. To address the concern, applicationaware adaptation has emerged as one of the effective strategies for controlling energy usage. But for adaptation to be effective, we must be able to predict the energy requirement of running applications, in an application-specific manner. This paper reports a set of experiments performed to understand the energy usage pattern of handheld devices while decoding and displaying MPEG compressed video in software. Experiments are designed to bring forth parameters that can be used to predict the energy requirement for MPEG playback. Based on our experimental results, we find that it is possible to construct simple polynomial models of energy usage with good least square fit (R value of 0.92 and higher). These models can be used to predict application energy requirement under various combinations of controllable video parameters.