BBRv3 + FEC (Phase 1 & 3) Testing Results — +10% Goodput @ 5% Loss

[Maksim Lanies]

Dear BBR Development Group,

I’d like to share our recent Phase 1 & Phase 3 testing report on the integration of BBRv3, Forward Error Correction (FEC), and Post-Quantum Cryptography (PQC) simulation in the CloudBridge QUIC stack.

This report summarizes real measurements from November 2025 validating BBRv3 under mobile and high-loss network conditions.
Highlights:

• +10.0 % goodput improvement with XOR-FEC (10 %) at 5 % channel loss
• No latency or jitter penalty (51.25 ms RTT, 0.72 ms jitter)
• Stable fairness (0.82–0.94) and low bufferbloat (0.025)
• Full-stack validation: BBRv3 + FEC + 0-RTT + Key Update + Datagrams
• Resource use: ≈65 % CPU, 3–4 MB per connection
• Limitations: fec_recovered = 0 (current loss injector not aligned with FEC groups); next step — group-aligned loss validation and RS-FEC (k-loss recovery).

The full technical report includes formulas, metrics, and reproduction details.

We would greatly appreciate peer review and feedback on:

• BBRv3 behavior with XOR-FEC under burst loss patterns
• Recommended approaches for RS-FEC integration (GF(2⁸) or Reed-Solomon libraries)
• Any insights from prior FEC + BBR experiments within QUIC or MASQUE contexts.

Many thanks to the community for maintaining such a rigorous technical forum.
Your feedback will help us finalize the RS-FEC + BBRv3 roadmap and contribute comparative data back to the group.