Contracts: A unified lens on congestion control robustness, fairness, congestion, and generality

2025-04-25

AI Summary

This paper addresses the fundamental design challenge of implicit bandwidth negotiation—termed “contracts”—in congestion control algorithms (CCAs). We propose the first formal contract theory framework that characterizes intrinsic trade-offs and lower bounds among four key dimensions: signal error robustness, inter-flow fairness, congestion metrics (delay/loss), and link-rate adaptability. Leveraging control-theoretic modeling, steady-state analysis, game-inspired signal encoding abstraction, and large-scale simulation, we identify critical design pitfalls—such as flow starvation—that arise from contract misalignment. We formally prove that contract structure dictates fundamental performance limits. Empirical validation across mainstream CCAs—including Reno, CUBIC, and BBR—quantifies their implicit contract properties and reveals their practical performance ceilings. Our framework establishes verifiable, theoretically grounded criteria for CCA design and network measurement. (149 words)

Technology Category

• Multiagent Systems: Mechanism Design
• Game Theory and Economic Paradigms: Mechanism Design
• Machine Learning: Online Learning & Bandits

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Security and Privacy: Large-scale security measurements
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Congestion control algorithms (CCAs) operate in partially observable environments, lacking direct visibility into link capacities, or competing flows. To ensure fair sharing of network resources, CCAs communicate their fair share through observable signals. For instance, Reno's fair share is encoded as $propto 1/sqrt{ exttt{loss rate}}$. We call such communication mechanisms emph{contracts}. We show that the design choice of contracts fixes key steady-state performance metrics, including robustness to errors in congestion signals, fairness, amount of congestion (e.g., delay, loss), and generality (e.g., range of supported link rates). This results in fundamental tradeoffs between these metrics. We also discover some properties of contracts that describe CCA design pitfalls that can lead to starvation (extreme unfairness). We empirically validate our findings and discuss their implications on CCA design and network measurement.

Problem

• Analyzing congestion control contracts' impact on fairness and robustness
• Exploring tradeoffs between congestion signal accuracy and network performance
• Identifying contract design pitfalls causing starvation in resource sharing

Innovation

• Contracts ensure fair network resource sharing
• Contracts fix key steady-state performance metrics
• Contracts reveal CCA design pitfalls and tradeoffs