An Integrated and QOS-Aware Framework for Multimedia Service Management

In highly heterogeneous and dynamic distributed environments, the computing and communication resources are not only heterogeneous in their types and capacities, but also dynamic in their instant availability. On the other hand, the delivery of application-level multimedia services typically requires a more substantial and continuous supply of end-to-end resources, compared with non- multimedia services. Meanwhile, multimedia services also exhibit more adaptivity: the same multimedia service may be delivered at different Quality of Service (QoS) levels, depending on the availability of end-to-end resources. Therefore, the provision and management of multimedia services in heterogeneous and dynamic environments poses new challenges. One major challenge is the customized setup of multimedia services: given a multimedia service request, how to determine and configure the end hosts, software modules, and resources involved in the service delivery, so that the subsequent service session will achieve satisfactory QoS under the end-to-end resource availability constraint. To meet the challenge, we in this dissertation present Q-MSM, an integrated and QoS-aware framework for multimedia service management. Q-MSM consists of a polymorphic multimedia service model, a runtime system architecture, and a set of multimedia service setup functions performed by the runtime system. The Q-MSM multimedia service model reflects the polymorphic service delivery, and captures the `QoS-resource'' relation in each multimedia service. The Q-MSM runtime system is deployed as a middleware between the multimedia services and the underlying resource management systems. At runtime, the Q-MSM functions of service configuration, discovery, and reservation will be performed for each multimedia service request, in order to select and set up a customized service delivery. The major contributions of Q-MSM are as follows. First, the multimedia service model allows for service developers to specify different service delivery forms for each service. Second, the Q-MSM runtime system provides generic support for different multimedia services, and service developers do not have to re-implement the Q-MSM service setup functions within each multimedia service. Third, the Q-MSM runtime system fills the gap between existing application-level QoS delivery methods and resource-level QoS management systems. Finally, the feasibility and performance of Q-MSM has been validated via both prototype implementation and simulations.