| Real-time video applications, such as videoconferencing, videophony, and on-demand video streaming, have gained increased popularity. However, a key problem of video transmission over the existing Internet and wireless networks is the incompatibility between the nature of the network conditions and the QoS (Qualify of Service) requirements (such as those in bandwidth, delay, and packet loss) of real-time video applications. Cross-layer design is a natural approach to deal with the incompatibility problem. This approach aims to efficiently perform cross-layer resource allocation such as bandwidth, transmission energy, and buffers) by increasing the communication efficiency of multiple network layers.; Our focus is on the end-system design. We assume that the lower layers provide a set of given adaptation components; from the encoder's point of view, these components can be regarded as network resource allocation "knobs". Assuming that our encoder can access and specify those adaptation components, we propose a resource-istortion optimization framework, which optimally assigns cross-layer resources to each video packet according to its level of importance.; The proposed framework is general and flexible. Within the framework, we can jointly consider the available error control components in different network infrastructures. In particular, we study the following problems: (1) joint source-channel coding (JSCC) for real-time Internet video transmission, (2) joint source-channel coding and power adaptation (JSCCPA) for real-time wireless video transmission, and (3) joint source coding and packet classification (JSCCPC) for real-time DiffServ (differentiated services) network video transmission. Besides single layer video source coding, we also consider using scalable video source coding for further error resilience. In addressing each of the above problems, we propose efficient algorithms for obtaining the optimal solutions. The simulation results, as expected, demonstrate the benefits of joint design of source coding and cross-layer resource allocation. In addition; the proposed framework serves as an optimization benchmark against which the performances of other sub-optimal systems can be evaluated, and also provides a useful tool in assessing the effectiveness of different error control components in practical system design. |
