Teaching to Fish Rather Than Giving a Fish: The Concentrator Method of Teaching Classic Congestion Control With Learning-Based module

Nowadays, Congestion Control (CC) algorithms are expected to satisfy the diverse demands of applications running over diverse networks. To achieve this goal, the combinations, aiming to inherit both the advantages of classic CC in terms of convergence, overhead, and explainability, and the advantages of learning-based CC on adapting to diverse networks and demands, become a hot topic. In this paper, we reveal the existing combination works are either giving a fish or teaching to fish. Based on the insight of their essential issues, we develop the Concentrator method of teaching to fish. According to this method, we propose Seagull. Specifically, Seagull captures the network characteristics and application demands in a coarse-grained manner via an online learning module. Moreover, this module guides to customize the rate adjustment rules of the classic CC module for fine-grained system evolution. Replacing the assumption on networks by the captured characteristics, the classic CC module of Seagull can fulfill the specified application demands. Real-world experimental results show Seagull respectively outperforms Orca, PCC-Vivace, and CUBIC by 49.3%, 30.4%,and 24.9% interms of throughput ocer the internet, and improves the video quality of experience (QoE) by 12.9 ∼ 33.5% compared to CUBIC over cellular links.