Diffusion Model of a Non-Integer Order PI Controller with TCP/UDP Streams

Area of Science:

• Computer Science
• Network Engineering
• Control Theory

Background:

• Network congestion impacts performance of Transmission Control Protocol (TCP) and User Datagram Protocol (UDP) flows.
• Active Queue Management (AQM) algorithms are crucial for mitigating congestion in IP routers.
• Existing models often lack the flexibility to represent dynamic flow behaviors.

Purpose of the Study:

• To present a novel diffusion approximation model for analyzing the interplay between TCP and UDP flows.
• To investigate the influence of a fractional PIγ controller on network traffic under AQM.
• To develop a more flexible model capable of handling dynamic flow lifecycles.

Main Methods:

• Employing diffusion approximation to model TCP and UDP flow dynamics.
• Implementing Active Queue Management (AQM) algorithms within IP router simulations.
• Designing and analyzing a fractional PIγ controller based on control theory principles.

Main Results:

• The diffusion model accurately captures the interaction between TCP and UDP flows.
• The fractional PIγ controller demonstrates a significant impact on transport protocol performance.
• The model accommodates the dynamic start and end times of individual TCP and UDP flows.

Conclusions:

• The proposed diffusion model offers a flexible and accurate approach to studying network traffic dynamics.
• AQM with fractional control presents a promising strategy for effective congestion management.
• This research advances the understanding of network protocol behavior under realistic traffic conditions.