Fusion Channels: A Multi-sensor Data Fusion Architecture

Abstract

Due to the falling price and availability of sensors, information capture
\nand processing at a realtime or soft realtime rate is emerging as a
\ndominating application space. This class includes interactive multimedia,
\nrobotics, security and surveillance applications and many more. A common 
\ndenominator of these applications is fusion of data gathered by various sensors 
\nand data aggregators. In this paper we propose a Data Fusion architecture, 
\nspecifically geared toward such multi-sensor data fusion applications and 
\nreport on the prototype we have built. Our infrastructure provides a 
\nprogramming abstraction that offers programming ease, at the same time provides 
\nbuilt-in optimizations that are quite complicated to implement from scratch. We 
\nshow the ease of programming through two sample applications and also 
\ndemonstrate through various experiments that our system has low overhead and 
\noffers better performance compared to otherwise naively written fusion routines. 
\nWe also demonstrate improved scalability.