Statistical Control for Real-Time Video Multiplexing

We are concerned about the transmission of Variable Bit Rate (VBR) video traffic over Asynchronous Transfer Mode (ATM) networks. A good characterization of VBR sources allows the implementation of efficient Connection Admission Control (CAC) and Police functions. We show that the Gamma distribution presents a reliable tool for modeling the aggregate video traffic entering the multiplexer. It therefore allows an easy and efficient implementation of the CAC function., based only on the sources Mean Rates and Variances as traffic descriptors. In order to make these traffic descriptors policeable, we propose a Rate Control scheme to enforce the Mean Rate and Variance at the encoder output to comply with LeakyBucket-Like parameters at the network interface.