Predictive functional control for active queue management in congested tcp/ip systems. - pubmed - ncbi

Predictive functional control (PFC) like a new active queue management (AQM) method in dynamic TCP systems supporting explicit congestion notification (ECN) is suggested. Ale the controller in handling system delay and its simplicity and occasional computational load makes PFC a fortunate AQM method within the high-speed systems. Besides, thinking about the disturbance term (addressing model/process mismatches, exterior disturbances, and existing noise) within the control formulation adds some degree of sturdiness in to the PFC-AQM controller. It is really an important and preferred property within the charge of dynamically-different computer systems. Within this paper, the controller was created with different small signal linearized fluid-flow type of the TCP/AQM systems. Then, closed-loop transfer function representation from the product is derived to evaluate the sturdiness with regards to the network and controller parameters. The analytical along with the packet-level ns-2 simulation results show the out-performance from the developed controller for queue regulation and resource utilization. Fast response, low queue fluctuations (and therefore low delay jitter), high link utilization, good disturbance rejection, scalability, and occasional packet marking probability are also options that come with the developed method regarding other well-known AQM methods for example RED, PI, and REM that are also simulated to compare.