Virtual GPS: a middleware for power efficient localization of smartphones using cross layer approach

Abstract

Location based smartphone services such as maps and context-aware advertisements are increasingly becoming popular. However, a major limitation of these services is that they tend to consume significant power due to the usage of power-hungry GPS receivers. This paper proposes Virtual GPS, a middleware layer that provides current location to the applications in a power efficient manner using our novel cross layer approach. A number of mechanisms such as GPS, WiFi signatures[14], crowd sourcing, mobility tracking, etc. can be used to determine the current smartphone location. All these mechanisms differ from one another in terms of their power consumption and location accuracy. Virtual GPS requires higher level applications to indicate the desired location accuracy, e.g. low, medium or high, and then chooses the mechanism that satisfies the desired location accuracy with minimum power consumption. A prototype of Virtual GPS has been implemented. The paper describes the design and implementation of this prototype, and provides an extensive evaluation that includes lab-controlled settings as well as real world settings. Primary contributions of this work is in energy-efficient location estimation by considering application's accuracy requirements and dynamically switching sensors based on estimation algorithm in Virtual GPS layer. Another contribution of our V-GPS layer is to provide location to multiple applications simultaneously from stored location data instead of sampling sensors when the location errors accumulated satisfies application's accuracy requirements. We evaluate our approach using multiple experiments in moderate mobile and highly mobile environments. Results and evaluations show that Virtual-GPS layer can save over 28% compared to direct sensor usage with least overhead.