Congestion Control of TCP/AQM Networks using Predictive Control and On-Line Linearized Models

T. Alvarez, A. Salim, and M. Bolajraf (Spain)


Predictive control, TCP/AQM, congestion control, PID.


Internet users rely on the good capabilities of TCP/IP networks for dealing with congestion. Classic algorithms, such as drop tail or even RED, do not give good results under many situations. The network delay and the number of users fluctuate from one moment to the next. Predictive control can deal with these conditions so performance can be improved. This paper presents how predictive control, with on-line linearized models, performs better than PID or classic predictive control when dealing with control congestion, because changes in the network delay and the current number of users can be updated in the model. Non-linear simulations illustrate the advantages and disadvantages of the various methods under different working scenarios.

Important Links:

Go Back