Neural Network-Based Distributed Adaptive Pre-Assigned Finite-Time Consensus of Multiple TCP/AQM Networks

• Published in: IEEE transactions on circuits and systems. I, Regular papers Regular papers, 2021-01, Vol.68 (1), p.387-395
• Main Authors: Wang, Chunmei, Chen, Xiangyong, Cao, Jinde, Qiu, Jianlong, Liu, Yang, Luo, Yiping
• Format: Article
• Language: English
• Subjects: Active control; Adaptive control; Automation; Backstepping; Closed loop systems; Complex systems; Design parameters; distributed control; Feedback control; Finite-time consensus; Liapunov functions; Multi-agent systems; Multiagent systems; multiple TCP/AQM networks; neural network; Neural networks; TCP (protocol); Time functions; Tracking errors
• ISSN: 1549-8328; 1558-0806
• DOI: 10.1109/TCSI.2020.3031663

In this study, a class of finite-time consensus of multiple transmission control protocol/active queue management (TCP/AQM) networks is investigated on the basis of a design idea of multi-agent systems, and for the first time, to our knowledge, a novel congestion control concept with neural networks is proposed. First, the problem statement and design goal of the consensus of multiple TCP/AQM networks is given. Then, a pre-assigned finite-time function is introduced to ensure that the tracking error approaches a pre-defined area within finite time. Furthermore, by combining a barrier Lyapunov function and backstepping technique, a neural network-based distributed adaptive finite-time control protocol for the output consensus of multiple TCP/AQM networks is presented, which can effectively generate the desired controls and ensure that the convergent time of all errors has nothing to do with the initial condition and design parameters. In addition, all signals in the closed-loop system are bounded. Finally, an example is given to further illustrate the effectiveness of the theoretical finding presented.