Exploiting spatiotemporal and device contexts for energy-efficient mobile embedded systems

Within the past decade, mobile computing has morphed into a principal form of human communication, business, and social interaction. Unfortunately, the energy demands of newer ambient intelligence and collaborative technologies on mobile devices have greatly overwhelmed modern energy storage abilities. This paper proposes several novel techniques that exploit spatiotemporal and device context to predict device interface configurations that can optimize energy consumption in mobile embedded systems. These techniques, which include variants of linear discriminant analysis, linear logistic regression, non-linear logistic regression with neural networks, and k-nearest neighbor are explored and compared on synthetic and user traces from real-world usage studies. The experimental results show that up to 90% successful prediction is possible with neural networks and k-nearest neighbor algorithms, improving upon prediction strategies in prior work by approximately 50%. Further, an average improvement of 24% energy savings is achieved compared to state-of-the-art prior work on energy-efficient location-sensing.