Multi-time-scale, multi-service-class fuzzy control design for ATM networks

Abstract EN

This paper proposes a fuzzy traffic and congestion controller (FTCC) for asynchronous transfer mode (ATM) networks that includes multiple flow and congestion control algorithms that operate at different time-scales and for different classes of services.

The FTCC allocates capacity to each call in the call time-scale and then admits the call if there are enough resources (buffer and capacity), otherwise it rejects the call.

In the cell time-scale, the FTCC classifies the cells of the accepted calls into three classes and then indicates the dynamic thresholds that manage these classes in the buffer.

The input linguistic variables of the FTCC are appropriately chosen so that the controller is a closed-loop system.

As a result, the system not only works as a preventive controller but also reacts to the performance degradation.

Simulation results that are based on the Poisson arrivals, the ON/OFF sources, and the Continuous State Autoregressive sources show that the proposed FTCC achieves high utilization with satisfactory quality of service (QoS) of all traffic classes.

Also, at moderate load the cell loss for ATM services is quite compliant with the QoS requirements of the real applications.