Fuzzy Explicit Marking: A Unified Congestion Controller For Best-effort And Diff-serv Networks

摘要

This paper, proposes a generic active queue management (AQM) control methodology in TCP/IP networks, based on Fuzzy logic control principles. A simple, effective and efficient non-linear control law is built, using a linguistic model of the system, rather than a traditional mathematical model, which is easily adapted in different network environments (e.g. Best-Effort and Differentiated-Services architectures). We demonstrate, via extensive simulative evaluation, that the proposed Fuzzy control methodology offers inherent robustness with effective control of the system under widely differing operating conditions, without the need to (re)tune any parameters. As demonstrated, this is in contrast with its well-known conventional counterparts A-RED, REM, PI, AVQ for Best-Effort, and two-level PI, RIO for Differentiated-Services based networks, where the proposed approach outperforms all of them both in Best-Effort and Differentiated-Services architecture application scenarios. Furthermore, the proposed approach is applied without any retuning for the two architectures using the same Fuzzy logic controller for both, with the only difference being in the setting of the target queue lengths for different precedence levels in Differentiated-Services.