Abstract
Present-day Internet Protocol (IP) networks employ Quality of Service (QoS) architectures that can mainly provide network-level support for multimedia communications. Although these approaches are effective in delivering both soft and hard service guarantees, they are usually inefficient in satisfying specific user and application-level QoS requirements. This is due to the fact that such networks are designed with the principle of implementing the main intelligence and media processing at the end-points. Moreover, introduction of new services depend on cumbersome standardisation process which hinders their timely deployment. Furthermore, increasing heterogeneity of networks and end systems makes it difficult to provide end-to-end QoS guarantees. Therefore, providing advanced support for user applications require the networks to evolve into a more flexible architecture which can shift some of the intelligence and processing load away from the end-user terminals. It is envisaged that this evolution will take place in line with the active networking approaches where distributed processing of multimedia applications will be performed inside the networks through the use of dynamically deployable content adaptation services. Looking from a user-centric point-of-view, this thesis proposes to enhance the perceived quality of video-based applications by utilising a series of visual content adaptation services that can be deployed in the active network nodes. These services are based on a real-time video transcoding system that is capable of applying certain rate-control and error-resilience operations on the input video streams. This enables video transmission over heterogeneous networks according to their respective bandwidth and error-resilience requirements. In addition, rate-control functionality of the transcoder is presented as an alternative congestion control mechanism to the traditional traffic conditioning approaches. Furthermore, an adaptive error resilience scheme is presented, which is responsive to the detected video scene activity and the reported channel error conditions of the wireless network. Extensive computer simulations demonstrate the effectiveness of the visual content adaptation services in terms of improving the perceived QoS of multimedia applications over both wired and wireless networks.