Abstract
The high-speed networks, especially the ATM-based Broadband ISDN, are expected to integrate through statistical multiplexing large numbers of traffic sources, which have a broad range of burstiness characteristics. In order for the estimation of cell loss and decide bandwidth allocation for congestion control and bandwidth management in an ATM-based BISDN, in this paper, we propose a new approach to model the network traffic. We use the hidden Markov model to approximate the actual traffic process coming from different LAN and then model the ATM multiplexer as an MMDP/D/1/K queuing system. Using queuing analysis, we derive a formula associated with cell loss probability and bandwidth allocation in term of parameters λ(A,B,π) of HMM. Bandwidth allocation is specified, after MMDP/D/1/K queuing system, in order to improve the performance cell loss rate and decide bandwidth allocation, an adaptive feedback control algorithm is proposed in time to throttle the congestion of traffic data. The feedback control signal is proportional to the predicted congestion level of the multiplexing buffer. The congestion control scheme reduces the peak rate of traffic sources in order that excess cells are not allowed to enter the network. Therefore, the congestion downstream can be prevented and the cell loss rate can be reduced to guarantee a desired QoS. Simulation results from practical multimedia traffic data from video sources or mixed video/audio sources confirm that the proposed scheme provides a simple and efficient traffic management method for broadband networks.