Search

In this paper, a minimum cost scheduling algorithm is proposed for high-quality video streaming in dynamic bandwidth allocation networks. It targets the problem of service cost minimization in optimal smoothing algorithms. Based on the bandwidth usage and the number of bandwidth renegotiations, service cost functions are defined for each transmission segment and a group of segments. The relative cost flags are also defined for each segment, which indicate the direction in which the segment is merged with less service cost. By inspecting the relative cost flags on the boundaries of segments, it is determined whether neighboring segments should be merged, until there is no reduction in the overall service cost. From the simulation results, the proposed algorithm is shown to merge neighboring segments in the direction minimizing the service cost. Compared with the optimal smoothing algorithm, the overall service cost is greatly reduced with a similar level of network resource usage and less transmission rate variability, especially for larger renegotiation cost parameter. Index Terms--minimum cost scheduling, RCBR service, video streaming, segment merging.

Conventional researches on QoS provisioning in HFC networks have focused mainly on the upstream channels. However, due to changing network usage patterns, the provisioned capacity for downstream service flows is no longer sufficient. To provide QoS for downstream VolP service flows in HFC networks, an adaptive committed information rate allocation algorithm is proposed. Given the maximum sustained traffic rate and the minimum reserved rate for each service flow, it allocates service rates to VolP and best effort connections adaptively for possible cases of network congestion. The proposed algorithm is directly tested in a commercial HFC network, and it is shown to reduce packet losses, jitter, and delay in the HFC network, which results in stable VoIP service quality even in network congestion. Index Terms--adaptive committed information rate, VoIP, quality of service, HFC networks.