Most human actions are a direct response to stimuli from their five senses. In the past few decades there has been a growing interest in capturing and storing the information that is obtained from the senses using analog and digital sensors. By storing this data it is possible to further analyze and better understand human perception. While many devices have been created for capturing and storing data, existing software and hardware architectures are aimed towards specialized devices and require expensive high-performance systems. This thesis aims to create a framework that supports capture and monitoring of a variety of sensors and can be scaled to run on low and high-performance systems such as netbooks, laptops and desktop systems. The proposed architecture was tested using aural and visual sensors due to their availability and higher bandwidth requirements compared to other sensors. Four different portable computing devices were used for testing with a varied set of hardware capabilities. On each of the systems the same suite of tests were run to benchmark and analyze CPU, memory, network, and storage usage statistics. From the results it was shown that on all of these platforms capturing data from multiple video, audio and other sensor sources was possible in real-time. Performance was shown to scale based on several factors, but the most important were CPU architecture, network topology and data interfaces used.

Library of Congress Subject Headings

Multisensor data fusion; Perception--Data processing; Data transmission systems

Publication Date


Document Type


Department, Program, or Center

Computer Engineering (KGCOE)


Cockburn, Juan


Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TK7872.D48 A43 2011


RIT – Main Campus