Performance Analysis of the ARIA Adaptive Media Processing Workflows using Colored Petri Nets

Multimedia systems are one of the most complex and interesting applications that are nowadays proposed to the users. Their complexity derives mainly from the fact that multimedia systems have to process huge amounts of data, while respecting real-time deadlines. For this reason performance evaluation of the underlaying workflow is a key issue in the design process of a new Multimedia system. In this paper we consider the ARchitecture for Interactive Arts (ARIA), an adaptive media processing workflow, developed at the Arizona State University, and outline a semi-automatic procedure to translate its specification into Colored Petri Nets. We then provide guidelines on how to compute the parameters for the performance models, and apply the proposed methodology to a realistic example of a face recognition application.