LeoCC: Making Internet Congestion Control Robust to LEO Satellite Dynamics

The recent renaissance of LEO satellite networks expands the boundaries of global Internet access, but also introduces substantial new challenges for existing end-to-end CCAs. The rapid and continuous movement of LEO satellites leads to infrastructure-level dynamics, resulting in frequent, LEO-dynamics-induced changes in link capacity, delay, and packet loss rate, which can further mislead the rate control in existing CCAs and cause self-limited performance.

Therefore, we presents LeoCC, a novel CCA that addresses the above challenges and is robust to LEO satellite dynamics. The core idea behind LeoCC lies in a critical characteristic of emerging LEO networks called “connection reconfiguration”, which implicitly reflects satellite path changes and is strongly correlated to network variations. Specifically, LeoCC employs a suite of new techniques to: (i) efficiently detect reconfiguration on the endpoint; (ii) apply a reconfiguration-aware model to characterize and estimate network conditions accurately; and (iii) precisely regulate the sending rate.

Overview

The directories list our contributions, comprising LeoCC, LeoReplayer and traces we collect in real world.

LeoCC

Inspired by the design of BBR, LeoCC integrates new mechanisms specifically designed for LEO satellite environments. There are several major modifications:

• Detecting connection reconfigurations efficiently at the endpoint.
• Applying a reconfiguration-aware model for accurate bandwidth and delay estimation.
• Precisely regulating the sending rate.

There are two variants of LeoCC:

• Live-Network version: designed for deployment in live networks, using Netlink to exchange RTT and response interval information between kernel space and user space.
• Simulation version: adapted to the LeoReplayer framework, requiring additional parameters (min_rtt_fluctuation, offset) to model realistic reconfiguration behaviors observed in satellite networks.

Note: This implementation is based on a BBRv3-derived kernel. As a result, .tso_segs is used instead of the standard .min_tso_segs, though this difference does not affect LeoCC's functionality.

LeoReplayer

LeoReplayer is a record-and-replay tool designed to reproduce the highly dynamic conditions of LEO satellite networks.

• It records real network dynamics using two concurrent flows:
  • a heavy UDP flow to capture time-varying maximum capacity;
  • a light ICMP flow to measure base RTT and packet loss.
• The recorded traces are then extracted and replayed to precisely emulate bandwidth, delay, and loss variations observed in real Starlink networks.

The system consists of two major parts:

1. Recorder scripts for collecting real-world measurements.
2. Extended Replayer Environment. Our replayer extends Mahimahi by introducing the ability to reproduce time-varying RTTs and bandwidth conditions, while still reusing Mahimahi’s original features. This enables accurate replay of dynamic link behaviors observed in real-world satellite networks.

Trace

We collect a large-scale dataset from real-world Starlink experiments. They are organized into 8 directories, each containing 600 individual traces plus a per-directory statistics file. These traces serve as the foundation for reproducing realistic network dynamics in LeoReplayer.

Citation

If you find our contributions helpful in your project or research, please cite with the following BibTeX entry:

@inproceedings{lai2025leocc,
  title={LeoCC: Making Internet Congestion Control Robust to LEO Satellite Dynamics},
  author={Lai, Zeqi and Li, Zonglun and Wu, Qian and Li, Hewu and Li, Jihao and Xie, Xin and Li, Yuanjie and Liu, Jun and Wu, Jianping},
  booktitle={Proceedings of the ACM SIGCOMM 2025 Conference},
  pages={129--146},
  year={2025}
}

Contact Us

For any questions or feedback related to this project, please feel free to get in touch with us. (Email:zeqilai@tsinghua.edu.cn, lzl24@mails.tsinghua.edu.cn).