Adaptive Multicast of Multi-Layered Video: Rate-Based and Credit-Based Approaches

is the highest rate that both paths can support. However, in a multicast connection with hundreds or even thousands of destinations, there is likely to be at least one very congested path. Limiting the video rate according to the most congested path penalizes the quality of video offered across all the other paths, regardless of how much bandwidth is available on them. A more scalable solution to the problem of available band-width variation is to use multi-layered video. A multi-layered video encoder encodes raw video data into one or more streams, or layers, of differing priority. The layer with the highest priority, called the base layer, contains the most important portions of the video stream. One or more enhancement layers with progressively lower priorities may then be encoded to further refine the quality of the base layer stream. For instance, in the example of Figure 1, the ideal deployment of multi-layered video results in a base layer stream transmitted at 4 Mbps and a single enhancement layer stream transmitted at 6 Mbps. There are two primary advantages to using multi-layered video encoding in multicast-capable networks. First is the ability to perform graceful degradation of video quality when loss occurs. Because each video layer is prioritized, a network experiencing congestion may discard packets from low priority layers, thereby protecting the important base layer and higher priority enhancement layers from corruption. The second advantage, which is related to the first, is the ability to support multiple destinations with different bandwidth constraints or end-system capabilities. For each source-to-destination path with a unique bandwidth constraint, an enhancement layer of video may be generated. Multi-layered video is not by itself sufficient to provide ideal network bandwidth utilization or video quality, however. To improve the bandwidth utilization of the network and optimize the quality of video received by each of the destinations, the source must respond to constantly changing network conditions by dynamically adjusting the number of video layers it generates as well as the rate at which each layer is transmitted. For the source to do this, it must have congestion feedback from the destinations and the network. In this paper, we study two novel and promising feedback mechanisms, both of which rely on adaptive, multi-layered video encoding. The first is a rate-based mechanism that uses a closed Figure 1: Example multicast video session. Abstract Network architectures that can efficiently transport high quality, multicast video …