Analysis of fuzzy logic and autoregressive video source predictors using T-tests

Abstract

This paper presents performance comparison of a fuzzy logic predictor and an autoregressive predictor. Multi-step ahead predictions were made on the traffic intensity of digital video sources coded with an MPEG coder (hybrid motion compensation/differential pulse code modulation/discrete cosine transform method). Although current coding standards recommend constant bit rate (CBR) output by means of a smoothing buffer, the method inherently produces variable bit rate (VBR) output, and VBR transmission is necessary for high quality delivery. Prediction results described in this paper have been obtained using a fuzzy logic prediction scheme, and the conventional linear autoregressive (AR) algorithm. Prediction errors are analysed with low order statistical parameters and hypothesis tests. The proposed fuzzy prediction methods offer higher accuracy and can be applied to the development of usage parameter control (UPC), connection admission control (CAC) and congestion control algorithms in ATM networks.