A dynamic bandwidth allocation strategy to support variable bit rate (VBR) video traffic is proposed. This strategy predicts the bandwidth requirements for future frames using adaptive linear prediction that minimizes the mean square error. The adaptive technique does not require any prior knowledge of the traffic statistics nor assume stationarity. Analyses using six one-half-hour video tracts indicate that prediction errors for the bandwidth required for the next frames and group of pictures (GOP) are almost white noise or short memory. The performance of the strategy is studied using renegotiated constant bit rate (RCBR) network service model and methods that control the tradeoff between the number of renegotiations and network utilization are proposed. Simulation results using MPEG-I video traces for predicting GOP rates show that the queue size is reduced by a factor of 15-160 and the network utilization is increased between 190%-300% as compared to a fixed service rate. Results also show that even when renegotiations occur on the average in tens of seconds, the queue size is reduced by a factor between 16-30.

Variable bit rate (VBR) video traffic is expected to be one of the major applications that need to be supported by broadband packet-switched networks. A dynamic bandwidth allocation strategy used to support VBR video traffic is proposed. This strategy predicts the bandwidth requirements for future frames using either adaptive or non-adaptive least mean square (LMS) error linear predictors. The adaptive technique does not require any prior knowledge of the statistics, nor assumes stationarity. Several reservation schemes are also presented. Analysis using six one-half hour video traces indicate that prediction errors for the bandwidth required for the next frame are almost white noise. By reserving a bandwidth equal to the predicted value, only the prediction errors need to be buffered. Because the errors are almost white noise, a small buffer size, high utilization, and a small delay are achieved. Simulation results for 1-step linear predictor show that for the same expected cell loss, the buffer size is reduced by more than a factor of 100 and the network utilization is increased by more than 250% as compared to a fixed service rate.