Entropy-based optimization to trade-off energy and accuracy for activity mobile sensing

The availability of advanced sensors on smart-phones allows feeding mobile applications with rich contextual information. Continuous sensing mechanisms in smartphones cost high energy consumption to support accurate contextual recognition. Hence, there is a trade-off between the classification accuracy and the energy consumption that needs to be identified and optimized. In this paper, we formulate the energy-accuracy trade-off as an entropy-based optimization problem, in order to propose an efficient algorithm based on user activity and phone sensor parameters. Experiments demonstrate the gains of the proposed algorithm with 43% reduction in energy consumption for a case study based on real data traces.