Database Abstraction Layer for Smart Phone Sensors

Abstract

Today's Smart phones have variety of sensors enable them to collect and use them in various monitoring applications. With the explosion of low priced smart phones and their widespread usage can be tapped to build phones sensor networks. Already there are sensor networks that are used in monitoring and tracking in health, entertainment and in other various application areas. All of these networked systems have core functions of reading, aggregating and communicating data; in which each of application system use different approaches to implement them, and give the application programmer abstractions or API’s to programmer to access these core functions. We propose SQL Query like Database abstraction layer architecture on top the phone sensors providing a higher level abstraction and user friendly query language to transparently access and query sensor data from one device to another device using existing infrastructure and technologies. The proposed project demonstrates a prototype where a SQL-like query can be entered into interface provided to an Android phone to access and query data from another Android phone. Also we propose messaging mechanism based on XMPP protocol which provides a medium for data transfer. This messaging component builds on top of XMPP provides basic primitives for communication in near real-time. The project draws inspiration from work done in the area of sensor networks specifically projects like Tiny DB[1] and Tikiri DB[2]. We borrowed ideas like acquisitional query processing and applied in the context of smart phones. We created SQL query like language, processed queries using acquisitional methods and communicated them over XMPP using our basic high-level primitives. We implemented our design consisting about three android phones all of which were 3g enabled. We used the publicly accessible GTalk from Google, as our XMPP provider to communicate. We ran several query types in different usage scenarios, these included basic select, relational, logical, and action type queries. Thus we provided a prototype and proved its viability as sensor network. During the evaluation of system we compared query usage scenarios and we concluded that our smart phone network is viable concept in social related applications, but unreliable in mission critical applications. Finally we talk about future of this area, and how this research provides basis for future smart phone sensor networks for variety of social aware smart phone sensor network applications.