Presentation Information

[B-16-23]Per-CCA Flow Ratio-Based Transmission Control for Inter-Flow Fairness in Programmable Data Plane Technologies

◎Yuya Kotoku1, Daiki Nobayashi1, Takeshi Ikenaga1 (1. Kyushu Institute of Technology)

Keywords:

BBR,CUBIC,Congestion control,Machine learning,Programmable data plane,P4

End-to-end communication performance on the Internet depends on the congestion control algorithm (CCA) employed by the transport protocol. Previous studies have reported that when heterogeneous CCAs compete over the same bottleneck link, throughput fairness among flows deteriorates depending on the buffer size of the bottleneck router. Therefore, this study investigates a method for maintaining throughput fairness among competing flows even in environments where heterogeneous CCAs coexist.