Adaptive Real-Time VBR Video Traffic Predictor for IEEE 802.15.3 Wireless Ad Hoc Networks.

This paper proposes a new real-time video traffic predictor to meet increasing consumer demand for a high speed high performance wireless broadband network. It analyzes the behaviors and the problems of previous adaptive LMS-type predictors using fixed step size in detail and then proposes an adaptive predictor using variable step size for predicting bandwidth requirement of real-time VBR videos. The proposed adaptive predictor has better ability for handling scene changes and needs not change its parameters for different VBR videos. The simulation shows that the performance of the proposed adaptive predictor is better or near the optimal performance of previous adaptive LMS-type predictors using fixed step size. © 2006 International Federation for Information Processing.