Dynamic reconfiguration for complex multimedia applications

Current state-of-the-art distributed multimedia applications require an environment that provides stable quality of service (QoS). However, these applications typically run on best-effort heterogeneous platforms, and compete with other applications or connections for end system and network resources, hence suffer from instability and dynamic variations with regards to QoS. We present a dynamically reconfigurable middleware control architecture to enhance adaptation awareness of these applications. Our middleware architecture allows for detecting QoS fluctuations in the surrounding environment and signaling optimal control actions to the application. Furthermore, we discuss the design of a core component, the Configurator, that adopts a fuzzy control approach to compute optimal control actions. Preliminary experimental results with a distributed visual tracking application show that our approach is viable and effective in controlling adaptive applications