Adaptive Video Streaming over an Event Notfication Service

Abstract

The goal of this thesis is to find a strategy for adaptive video streaming suitable for the Distributed Media Journaling project (DMJ). DMJ focuses on capturing and indexing distributed multimedia content in real time. An important aspect of the project is scalable multi-receiver video streaming with fine granularity selectivity in multiple video quality dimensions.

The scalability comes from routing video data as notifications in a content-based network. In content-based networking data is routed based on content rather than addresses. This means that the sender needs no information about individual users.

The selectivity is realized by hierarchical video coding and by mapping video data into notifications. Video quality is specified in subscriptions, which act as filters for notifications. Receivers have different interests and resource capabilities and subscriptions are the mechanism they use to specify the quality they desire. However, resource availability vary over time and it is inconvenient for a receiver to change the subscription in response to these variations manually.

This thesis focuses on the development of automatic adaptation of video quality in DMJ. Users should be able to specify their preferences in adaptation polices. When resources are scarce, the appropriate video quality is determined based on the adaptation policy, as well as the availability of resources. The goal is maximization of user utility.

It is necessary that both resource requirements and the resource availability are known. While monitoring resource availability is outside the scope of this thesis, a mathematical function for approximation of resource requirements has been suggested as an alternative to performing exhaustive measurements.