Robust Congestion Control for TCP/AQM Networks via a Predictive GPIO-Based Approach

Wei Xiong,Xuechao Qiu,Zuo Wang,Shihua Li
DOI: https://doi.org/10.1109/IECON51785.2023.10311975
2023-01-01
Abstract:In this paper, a robust predictive generalized proportional-integral observer (PGPIO)-based control is investigated for transmission control protocol (TCP)/active queue management (AQM) networks. In the design of the congestion control scheme, the round-trip time (RTT) is considered as the input time-delay, while the non-response flows (e.g., user datagram protocol flows) are regarded as external disturbances. A novel PGPIO is proposed to provide future information about the disturbance as well as the system dynamics in the presence of time-delay. Based on the predictions, a robust composite controller is constructed as the AQM algorithm to avoid congestion. The performance is demonstrated through rigorous mathematical analysis. Finally, some simulation results are presented to verify the feasibility and effectiveness of the proposed method.
What problem does this paper attempt to address?