A novel Fuzzy Logic Controller (FLC) for shortening the TCP channel roundtrip time by eliminating user buffer overflow adaptively

Abstract

The proposed Fuzzy Logic Controller (FLC) is a novel approach for dynamic buffer tuning at the user/server level. It eliminates buffer overflow by ensuring that the buffer length always cover the queue length adaptively. The FLC and the AQM (active queue management) mechanisms at the router/system level together form a unified solution to stifle TCP (Transmission Control Protocol) channel buffer overflow over the Internet. The FLC contributes to: a) prevent the AQM resources dished out at the system level from being wasted, b) shorten the service roundtrip time (RTT) by reducing retransmission, and c) alleviate network congestion in the process. Combining fuzzy logic and the conventional PIDC (Proportional + Derivative + Integral Controller) model creates the FLC that operates with the {0, Δ}2 objective function. The fuzzy logic maintains the given Δ safety margin about the reference point, symbolically represented by “0” in {0, Δ }2 . The FLC stability and precision is independent of the traffic pattern changes because of its statistical nature. This makes the FLC buffer overflow controller/tuner suitable for applications over the Internet, where the traffic can changes suddenly, for example, from LRD (long-range dependence) to SRD (short-range dependence) or multifractal.