Network Performance Lab

Optimization of QUIC/MASQUE, BBRv3, FEC to achieve target performance metrics: RTT ≤ 50 ms, jitter < 1 ms, goodput +15%

Active Started: Q4 2025

Project Overview

Network Performance Lab is a comprehensive platform for testing and analyzing network protocols with focus on optimizing QUIC, MASQUE, BBRv3, and FEC. The project provides real-time metric visualization and professional analysis of network protocol performance.

The quic-test platform provides a complete set of tools for testing QUIC protocol, including support for various congestion control algorithms (BBRv2, BBRv3, CUBIC), network condition emulation, and metrics export to Prometheus and JSON formats.

Key Objectives

RTT ≤ 50 ms

Achieve target round-trip time of no more than 50 milliseconds

Jitter < 1 ms

Maintain latency variability of less than 1 millisecond

Goodput +15%

Increase useful throughput by 15% compared to baseline

Current Status

Completed

  • BBRv3 integrated according to draft-ietf-ccwg-bbr-04 with support for all metrics
  • Implemented FEC (Forward Error Correction) for improved loss tolerance
  • Integrated PQC (Post-Quantum Cryptography) with Kyber/Dilithium support
  • Created TUI (Terminal UI) for real-time QUIC metrics monitoring
  • Implemented network emulator with support for various profiles (WiFi, LTE, 5G, Satellite)
  • Metrics export to Prometheus and JSON formats

Achieved Results

  • Throughput: +10-12% improvement compared to BBRv2 (RTT > 80 ms)
  • Jitter: -40-50% reduction thanks to pacing gain optimization
  • Fairness: ≥ 0.9 (Jain Index) achieved

BBRv2 vs BBRv3 Comparison

* Throughput: +10-12% improvement (RTT > 80 ms), Jitter: -40-50% reduction, Fairness: ≥ 0.9 (Jain Index)

In Progress

  • Validation of packet loss tolerance (-60% target)
  • Measuring recovery time (-60%, ≈ 200 ms target)
  • Checking bufferbloat factor (≤ 0.2 target)

Technical Details

Protocols

Congestion Control

  • • BBRv2 (baseline)
  • • BBRv3 (optimized)
  • • CUBIC (comparison)
  • • Experimental algorithms

Technologies

Go Rust QUIC BBRv3 FEC PQC TUI Prometheus

Metrics

  • • RTT (p50, p95, p99, p999)
  • • Throughput (Mbps)
  • • Jitter (ms)
  • • Packet Loss (%)
  • • Fairness Index (Jain)
  • • Recovery Time (ms)

Metrics Over Time

* Target metrics: RTT ≤ 50 ms, jitter < 1 ms, goodput +15%

QUIC Performance Visualization

* Data based on real metrics from quic-test platform

Key Features

QUIC/MASQUE Optimization

Measurement and optimization of stack performance with BBRv2 and BBRv3 . MASQUE tunneling support for CONNECT-UDP and CONNECT-IP.

Real-time Monitoring

Professional TUI visualization with live QUIC metrics, heatmaps, correlation analysis. Integration with Prometheus and Grafana.

Network Simulation

Linux tc integration with preset profiles for testing various conditions: WiFi, LTE, 5G, Satellite, Datacenter.

FEC & PQC

Forward Error Correction for improved loss tolerance and Post-Quantum Cryptography (Kyber/Dilithium) for future security.

Publications & Reports

Lab Report 2025-11

CloudBridge Network Technology Laboratory Research 2025

Summary of 2025 laboratory research results on network technologies. Includes experiments with QUIC, MASQUE, FEC, BBRv3.

Read report
Article 2025-10

BBRv3, FEC and QUIC: How We Maintained jitter <1ms and Stabilized RU<->EU

Article on optimization of BBRv3, MASQUE, FEC, QUIC network protocols on real RU↔EU routes. Achieved jitter <1ms PoP↔PoP, P50 ~20–21ms end-to-end RU↔EU.

Read on Habr
All Publications

Related Research

AI Routing Lab

Predictive route selection by latency/jitter using machine learning. Integrates with quic-test for model validation on real QUIC traffic.

Learn more

Project Resources

Repositories

Technologies

GitHub Repository