The applications of multimedia content analysis (MMCA) operating in real-time environments must run under extremely strict time constraints. To meet these requirements, large-scale multimedia applications are typically executed on Grid systems consisting of large collections of compute clusters. Therefore, it is essential to optimize the utilization of the computing resources by determining the optimal number of compute nodes per cluster, properly balancing the complex tradeoff between the computation versus communication time. Next, once the optimal number of resources are available, one face the ”just-in-time” (JIT) problem of to assigning multi-media video frames at the right times to the server, so as to obtain the highest service utilization possible, while minimizing the buffering time for individual video frames at the server side. Motivated by these observations, we first develop a simple and easy-to-implement method to determine the optimal number of parallel compute nodes. Our method is based on the classical binary search method for non-linear optimization and is independent of the (usually unknown) specifics of the system. Second, we address the JIT problem by introducing a smart adaptive control method that properly reacts to the continuously changing circumstances in Grid systems. 1.
International Journal on Multimedia Tools and Applications
Probability, Networks and Algorithms

Yang, R, van der Mei, R.D, Roubos, D, Seinstra, F.J, & Bal, H. (2012). Modeling Just-in-Time Communication On the Optimal Resource Utilization in Distributed Real-Time Multimedia Applications. International Journal on Multimedia Tools and Applications, 59, 941–971.