Scheduling the retrievals of continuous media objects

In multi-user multimedia information systems (e.g., movie-on-demand, digital-editing), scheduling the retrievals of continuous media objects becomes a challenging task. This is due to both intra and inter object time dependencies. Intra-object time dependency refers to the real-time display requirement of a continuous media object. Inter-object time dependency is the temporal relationships defined among multiple continuous media objects. In order to compose tailored multimedia presentations, a user might define complex time dependencies among multiple continuous media objects having various length and display bandwidth requirement. Scheduling the retrieval tasks corresponding to the components of such a presentation in order to respect both inter and intra task time dependencies is the focus of this dissertation. To tackle this problem, we introduce a new class of task scheduling problems and distinguish it from the conventional scheduling problems. The uniqueness of these problems is mostly due to our assumed framework. The framework enables tasks to share resources (disk bandwidth) in a regular manner once activated. Hence, tasks compete with each other over the resources resulting in bandwidth contention. We identify two types of contention depending on the load imposed on the system. Both types of contention might be observed by any member of our class of task scheduling problems. Each scheduling problem is formally defined, and shown to be an intractable problem (NP-hard in strong sense). Subsequently, an on-line scheduling algorithm is proposed corresponding to each problem. The algorithms consist of some mechanisms and techniques that are independent of the system load. However, policies knowledgeable of the load are required to employ the mechanisms effectively. Hence, for reasonable combinations of scheduling problems and contention types, motivated by real-world applications, a number of alternative policies and heuristics are proposed. Finally, the policies are evaluated and compared by means of simulation studies. Briefly, we present and evaluate a set of mechanisms and policies to solve a new class of task scheduling problems, where a task corresponds to the retrieval of a continuous media object in a multi-disk/multi-user environment.