This article proposes the XGOP-B deterministic traffic model for MPEG video services requiring strict bounds on the quality of service provided by the network. The XGOP-B model takes advantage of two characteristics of MPEG video traffic by performing a two-level analysis of such sources. In our scheme, the traffic parameters are directly obtained by a single-input single-output parameter extracting algorithm. The proposed algorithm is autonomous and leads to a single set of parameters for each video stream avoiding user interaction determine the most appropriate set of parameters. We show through analysis of several video traces that these original features of the XGOP-B model yield accurate parameterizations at a low complexity cost. Our results show that the accuracy of the XGOP-B model leads to significant improvements when compared with the accuracy of the other traffic models. Furthermore, the complexity is evaluated through the number of traffic parameters and we show that it can be substantially decreased.