Efficient Real-time Packet Scheduling for Smoothed Video Streams

Distributed video applications require packet delivery in real time. Compressed video exhibits high variability, so video smoothing techniques have been developed to reduce bit rate variability. This facilitates resource allocation to video streams. Scheduling the packets from various streams, in edge and core routers, requires careful consideration, to support the necessary Quality of Service. We present an extension to previous Round Robin-based scheduling strategies, which guarantees limited delay for video packets in IP networks. The scheme is a version of weighted Round Robin, suited to streams that display constant bit rate (CBR). The smoothing technique considered, the Network Constrained Smoothing, is an algorithm for transmitting pre-recorded video using CBR that can vary from interval to interval. Our scheduler calculates the transmission orders on a per interval basis for smoothed streams. We perform simulation to calculate the efficiency of the scheduler, and the results provide useful information to fine-tune some of the scheduler parameters.