Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Transformer-Based Multipath Congestion Control: A Decoupled Approach for Wireless Uplinks

Zongyuan Zhang, Tianyang Duan, Liang Wang, Zihan Fang, Zheng Lin, Yijun Lu, Jiening Wu, Xia Du, Miao Yang, Zhe Chen, Heming Cui, Jun Luo

TL;DR

Extensive evaluation on both simulated and real dual-band Wi-Fi testbeds demonstrates that TCCO achieves superior adaptability and performance than state-of-the-art baselines, validating the feasibility and effectiveness of TCCO for wireless networks.

Abstract

The proliferation of artificial intelligence applications on edge devices necessitates efficient transport protocols that leverage multi-homed connectivity across heterogeneous networks. While Multipath TCP enables bandwidth aggregation, its in-kernel congestion control mechanisms lack the programmability and flexibility needed for achieving efficient transmission. Additionally, inherent measurement noise renders network state partially observable, challenging data-driven approaches like deep reinforcement learning (DRL). To address these challenges, we propose a Transformer-based Congestion Control Optimization (TCCO) framework for multipath transport. TCCO employs a decoupled architecture that offloads control decisions to an external decision engine via a lightweight in-kernel client and user-space proxy, enabling edge devices to leverage external computational resources while maintaining TCP/IP compatibility. The Transformer-based DRL agent in the external decision engine uses self-attention to capture temporal dependencies, filter noise, and coordinate control across subflows through a unified policy. Extensive evaluation on both simulated and real dual-band Wi-Fi testbeds demonstrates that TCCO achieves superior adaptability and performance than state-of-the-art baselines, validating the feasibility and effectiveness of TCCO for wireless networks.

Transformer-Based Multipath Congestion Control: A Decoupled Approach for Wireless Uplinks

TL;DR

Extensive evaluation on both simulated and real dual-band Wi-Fi testbeds demonstrates that TCCO achieves superior adaptability and performance than state-of-the-art baselines, validating the feasibility and effectiveness of TCCO for wireless networks.

Abstract

The proliferation of artificial intelligence applications on edge devices necessitates efficient transport protocols that leverage multi-homed connectivity across heterogeneous networks. While Multipath TCP enables bandwidth aggregation, its in-kernel congestion control mechanisms lack the programmability and flexibility needed for achieving efficient transmission. Additionally, inherent measurement noise renders network state partially observable, challenging data-driven approaches like deep reinforcement learning (DRL). To address these challenges, we propose a Transformer-based Congestion Control Optimization (TCCO) framework for multipath transport. TCCO employs a decoupled architecture that offloads control decisions to an external decision engine via a lightweight in-kernel client and user-space proxy, enabling edge devices to leverage external computational resources while maintaining TCP/IP compatibility. The Transformer-based DRL agent in the external decision engine uses self-attention to capture temporal dependencies, filter noise, and coordinate control across subflows through a unified policy. Extensive evaluation on both simulated and real dual-band Wi-Fi testbeds demonstrates that TCCO achieves superior adaptability and performance than state-of-the-art baselines, validating the feasibility and effectiveness of TCCO for wireless networks.
Paper Structure (17 sections, 15 equations, 14 figures, 2 tables)

This paper contains 17 sections, 15 equations, 14 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Challenge and Motivation
  3. From Traditional Algorithms to DRL-based Methods
  4. Limitations of DRL-based Methods under Partial Observability
  5. TCCO Framework
  6. System Overview
  7. System Model
  8. Cross-Subflow Coupling Dynamics
  9. POMDP Formulation for Multipath CC
  10. System Design
  11. Observation Pipeline
  12. Transformer-based DRL Agent in External Decision Engine
  13. Evaluation and Validation
  14. Evaluation in emulated environment
  15. Validation in Physical Testbed
  16. ...and 2 more sections

Figures (14)

  • Figure 1: Performance comparison between CUBIC and BBR. The queue buffer was simulated using the Linux Traffic Control (TC). During the test in (b), the buffer size was set to 1000 packets (greater than the actual requirement).
  • Figure 2: Statistics of ACK packets during a real MPTCP upload. The test was conducted on a dual-band, dual-link Wi-Fi network supporting 5GHz and 6GHz.
  • Figure 3: Responsiveness with different control intervals. TCCO (1000ms) uses artificially introduced delay. The dashed line shows the total bandwidth (altered via Linux TC) of paths.
  • Figure 4: The concept of network function decoupling in TCCO. The external decision engine can operate in edge devices. Communication methods between modules are flexible.
  • Figure 5: An overview of TCCO framework.
  • ...and 9 more figures