A fuzzy logic approach to congestion control in ATM networks

The principles of fuzzy logic are deployed to tackle the problem of congestion control, an issue that is of prime importance in a multiservice environment like B-ISDN. Using the leaky bucket and moving window mechanisms as building blocks, we develop a fuzzy rule-based system that is transparent to sources that, are compliant to the traffic contract, effective in shutting down noncompliant sources and quick to respond to mean and peak rate violations, as well as to peak rate violations that do not affect the mean rate, compared to the mechanisms previously proposed. The system is intuitive in concept, simple in its implementation and computationally efficient according to the restrictions imposed by the high-speed network environment it operates in.