PowerForecaster: Predicting Power Impact of Mobile Sensing Applications at Pre-Installation Time

Abstract

This article presents a hierarchical context monitoring and composition framework that effectively supports next-generation context-aware services. The upcoming ubiquitous space will be covered with innumerable sensors and tiny devices, which ceaselessly pump out a huge volume of data. This data gives us an opportunity for numerous proactive and intelligent services. The services require extensive understanding of rich and comprehensive contexts in real time. The framework provides three hierarchical abstractions: PocketMon (personal), HiperMon (regional), and EGI (global). The framework provides effective approaches to combining context from each level, thereby allowing us to create a rich set of applications, not possible otherwise. It deals with an extensively broad spectrum of contexts, from personal to worldwide in terms of scale, and from crude to highly processed in terms of complexity. It also facilitates efficient context monitoring and addresses the performance issues, achieving a high level of scalability. We have prototyped the proposed framework and several applications running on top of it in order to demonstrate its effectiveness.