Human activity recognition through posture identification is increasingly used for medical, surveillance and entertainment applications. This paper proposes a ubiquitous solution to activity recognition through the use of tri-axial accelerometers of smartphones. Use of smartphones for activity recognition poses new challenges such as variation in hardware configuration and usage behavior like where the smartphone is kept. Only a few works address one or more of these challenges. Consequently, in this paper we present an activity recognition framework for identifying both static and dynamic activities addressing above mentioned challenges in order to make the framework ubiquitous. Since accelerometer is widely available in many smartphone configurations, activities are detected based on accelerometer readings only. The framework forms a two-phase classifier to address the variance due to different hardware configuration and usage behavior in terms of where the smartphone is kept (pant pocket, shirt pocket or bag). The framework is implemented and tested on real data set collected from 10 users with 6 different device configurations. It is observed that, with our proposed two phase approach, recognition accuracy increases by 9% on an average than single phase approach in energy efficient manner.