Interaction-Aware Dynamic Power Optimization Scheme for Wireless Network Interface Cards

Power optimization of wireless network cards plays a significant role in the development of mobile and pervasive computing. However current wireless power management often substantially degrades performance when used with interactive applications, which account for a large fraction of software usage on modern mobile devices. We observed that user initiated interface event drives the execution of interactive application and wireless device, in turn is one critical factor to determine the energy consumption. Therefore, we propose an interaction-aware power optimization scheme, which exploits user interface events, both mouse clicks and key strokes, to guide power management of wireless network cards. We develop a user interface events analysis technology to provide better understanding of user intents and accurate hints for later prediction. Then, we predict network demand by using the correspondence of analyzed user events with underlying network traffic patterns, and adapt the power mode of wireless cards based on those predictions. Real-world evaluation results show that, compared with state-of-the-art power management methods, our scheme achieves a better balance for the competing goals of interactive performance and energy conservation.