Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Hercules: Heterogeneous Requirements Congestion Control Protocol

Neta Rozen-Schiff, Itzcak Pechtalt, Amit Navon, Leon Bruckman

TL;DR

The paper tackles the problem of highly heterogeneous QoS requirements under congestion by introducing Heterogeneous Requirements Fairness (HRF) and an online-learning congestion control protocol called Hercules. HRF defines fairness as the lexicographical max-min of normalized per-connection rates $\overline{x_i} = (x_i-a_i)/(b_i-a_i)$, guiding a rate-control strategy that combines an arctan-based penalty $H(\overline{x_i})$ with per-flow metrics (loss and RTT). Hercules implements this strategy as a QUIC module with a rate-control loop (slow-start, probing, moving) that converges to HRF, and it is evaluated against CUBIC, BBR, Reno, and Vivace, showing up to $3.5\times$ QoS gains and robust convergence. The results suggest a scalable path to fine-grained, heterogeneous congestion control in modern networks, with potential extensions to infer requirements from traffic content.

Abstract

Future network services present a significant challenge for network providers due to high number and high variety of co-existing requirements. Despite many advancements in network architectures and management schemes, congested network links continue to constrain the Quality of Service (QoS) for critical applications like tele-surgery and autonomous driving. A prominent, complimentary approach consists of congestion control (CC) protocols which regulate bandwidth at the endpoints before network congestion occurs. However, existing CC protocols, including recent ones, are primarily designed to handle small numbers of requirement classes, highlighting the need for a more granular and flexible congestion control solution. In this paper we introduce Hercules, a novel CC protocol designed to handle heterogeneous requirements. Hercules is based on an online learning approach and has the capability to support any combination of requirements within an unbounded and continuous requirements space. We have implemented Hercules as a QUIC module and demonstrate, through extensive analysis and real-world experiments, that Hercules can achieve up to 3.5-fold improvement in QoS compared to state-of-the-art CC protocols.

Hercules: Heterogeneous Requirements Congestion Control Protocol

TL;DR

The paper tackles the problem of highly heterogeneous QoS requirements under congestion by introducing Heterogeneous Requirements Fairness (HRF) and an online-learning congestion control protocol called Hercules. HRF defines fairness as the lexicographical max-min of normalized per-connection rates , guiding a rate-control strategy that combines an arctan-based penalty with per-flow metrics (loss and RTT). Hercules implements this strategy as a QUIC module with a rate-control loop (slow-start, probing, moving) that converges to HRF, and it is evaluated against CUBIC, BBR, Reno, and Vivace, showing up to QoS gains and robust convergence. The results suggest a scalable path to fine-grained, heterogeneous congestion control in modern networks, with potential extensions to infer requirements from traffic content.

Abstract

Future network services present a significant challenge for network providers due to high number and high variety of co-existing requirements. Despite many advancements in network architectures and management schemes, congested network links continue to constrain the Quality of Service (QoS) for critical applications like tele-surgery and autonomous driving. A prominent, complimentary approach consists of congestion control (CC) protocols which regulate bandwidth at the endpoints before network congestion occurs. However, existing CC protocols, including recent ones, are primarily designed to handle small numbers of requirement classes, highlighting the need for a more granular and flexible congestion control solution. In this paper we introduce Hercules, a novel CC protocol designed to handle heterogeneous requirements. Hercules is based on an online learning approach and has the capability to support any combination of requirements within an unbounded and continuous requirements space. We have implemented Hercules as a QUIC module and demonstrate, through extensive analysis and real-world experiments, that Hercules can achieve up to 3.5-fold improvement in QoS compared to state-of-the-art CC protocols.
Paper Structure (15 sections, 2 theorems, 2 equations, 12 figures)

This paper contains 15 sections, 2 theorems, 2 equations, 12 figures.

Table of Contents

  1. Introduction
  2. Preliminaries and Motivation
  3. Existing Fairness Definitions
  4. Heterogeneous Requirements and Fairness
  5. Hercules Design Overview
  6. Heterogeneous Requirements Fairness
  7. Hercules Utility Function
  8. Hercules Rate-Control Module
  9. Implementation
  10. Evaluation
  11. Satisfying Heterogeneous Requirements
  12. Convergence Time
  13. Fairness with Other Protocols
  14. Related work
  15. Conclusion and Further Research

Key Result

Theorem 3.1

In a network with $n$ connections competing over a bottleneck, where the connections are regulated by Hercules per connection utility function (eq:Hercules_utility) with per connection requirements $\{(a_i,b_i)\}_{i \in [n]}$, any equilibrium is HRF.

Figures (12)

  • Figure 1: Delay, bandwidth and reliability requirements for different applications
  • Figure 2: Different fairness objectives
  • Figure 3: Different requirement penalties as functions of normalized rates, for different values of D. The normalized rate values $0$ and $1$ correspond to the minimum and maximum requirements respectively, and are denoted in red dashed lines.
  • Figure 4: Hercules' extension over QUIC objects
  • Figure 5: Satisfaction ratio for the two scenarios at different network congestion conditions
  • ...and 7 more figures

Theorems & Definitions (2)

  • Theorem 3.1
  • Theorem 3.2