Open Source · GPL v3.0

🔄 Redfire Switch

Complete Class 4 SIP switch platform with libraries for building telecom software

Enterprise-oriented telecommunications platform with an optional GPU-accelerated transcoding path, ML-powered threat detection, and enterprise-grade clustering. Comprehensive RFC compliance, completely free and open source.

Get Started Learn More

Project Overview

🚀 Enterprise-Oriented Platform

Enterprise telecommunications platform with an optional GPU transcoding path, ML-powered security, and comprehensive RFC compliance, built in memory-safe Rust.

🦀 Optional GPU Acceleration

GPU transcoding engine covering 56 codec pairs, available behind the cuda/rocm build features (not required - CPU fallbacks are always available). Memory-safe Rust implementation for carrier-grade reliability.

🔓 GPL v3.0 Licensed

Completely free and open source software. You have the freedom to use, modify, and distribute the software for any purpose, including commercial use.

Production Features

🚀 Optional GPU-Accelerated Transcoding

G.711 & PCM16

G.711 μ-law/A-law and PCM16 transcoding are complete and production-ready, CPU or GPU.

56 Codec Pairs (GPU, Feature-Gated)

GPU-accelerated transcoding across 56 codec pairs is available behind the cuda/rocm build features. Per the project's own README, GPU transcoding is feature-gated and currently unverified in CI - CPU fallbacks are always available.

G.729 & G.722.2/AMR-WB

Simplified reference implementations, not bit-exact ITU-T codecs - intended for signaling/plumbing tests rather than production audio quality, per the project README.

Design-Target Performance

15-20x GPU speedup and 3.2KB-per-stream memory figures are the project's documented design targets for the GPU path, not independently verified production benchmarks.

🛡️ Enterprise Security & Compliance

ML Threat Detection

Advanced machine learning-powered anomaly detection and behavioral analysis for real-time threat identification.

5 Vulnerabilities Fixed

Comprehensive security hardening addressing log injection, JWT confusion, memory exhaustion, and panic-based attacks (5 internally-tracked critical issues, not externally-assigned CVE IDs).

J-STD-025 & CALEA Compliance

U.S. lawful intercept standard implementation with CALEA compliance for law enforcement assistance.

ETSI LI Implementation

European lawful intercept standards (TS 102 232, TS 133 108) with warrant validation and HI2/HI3 delivery.

STIR/SHAKEN

Complete RFC 8224/8225 implementation with PASSporT tokens and call authentication.

📚 Telecommunications Libraries

Complete RFC Implementations

Full compliance with RFC 3261 (SIP), RFC 8224/8225 (STIR/SHAKEN), RFC 3398 (SIP-I), and additional standards for building telecom applications.

Modular Architecture

Reusable libraries and components for building custom telecommunications software and applications.

Real-time Monitoring

Continuous security surveillance with automated blocking and predictive threat response.

🏢 Enterprise Platform

High Availability

Multi-node clustering with automatic failover, call state synchronization, and split-brain detection.

Operational Dashboard

Real-time performance analytics, security monitoring, and comprehensive alerting system.

SIP-I/PSTN Integration

RFC 3398 compliant SIP-I with ISUP interworking for carrier-grade PSTN connectivity.

CDR & Analytics

Comprehensive Call Detail Records with ClickHouse integration and business intelligence.

Architecture

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│   SIP Clients   │    │ Redfire Switch  │    │  Termination    │
│                 │◄──►│   Enterprise    │◄──►│   Providers     │
│ • Softphones    │    │ • GPU Transcoding│    │ • Carriers      │
│ • PBX Systems   │    │ • ML Security   │    │ • Gateways      │
│ • Applications  │    │ • B2BUA/SIP-I   │    │ • PSTN/SS7      │
└─────────────────┘    └─────────────────┘    └─────────────────┘
                               │
                       ┌─────────────────┐
                       │ Enterprise Ops  │
                       │                 │
                       │ • HA Clustering │
                       │ • ML Monitoring │
                       │ • Live Dashboard│
                       │ • Auto-scaling  │
                       └─────────────────┘

Platform Components

GPU Transcoder (optional) - 56 codec pairs behind feature-gated GPU builds, CPU fallback always available
ML Security Engine - Real-time threat detection and response
Class 4 Switch Core - Production-ready SIP routing and processing
HA Clustering - Multi-node with automatic failover
SIP-I Gateway - PSTN/SS7 integration with ISUP
CALEA/J-STD-025/ETSI LI Compliance - U.S. and European lawful intercept
Telecommunications Libraries - Modular components for building telecom software
Operational Intelligence - Real-time dashboard and analytics

An Open Source Alternative to Commercial Class 4 Softswitches

Teams evaluating Redfire Switch are often comparing it against commercial Class 4 softswitch and session border controller platforms such as PortaOne PortaSwitch, Cataleya, Sansay VSXi, Ribbon Communications SBC Core (formerly Sonus), Metaswitch Perimeta, Oracle Communications Session Border Controller (Acme Packet), and VOS3000/VOS5000. Unlike those licensed products, Redfire Switch's source is fully open under GPL v3.0 - free to inspect, self-host, and modify, with commercial support available - reach out if your team wants it.

Quick Start

GPU-Accelerated Build

# Clone repository
git clone https://github.com/carrierone/redfire-switch.git
cd redfire-switch

# Setup GPU development environment
./setup-gpu-dev.sh

# Build with CUDA support
cargo build --features cuda --release

# Generate production config
./target/release/redfire-switch gen-config -o config.json

Enterprise Deployment

# Start enterprise platform
docker-compose -f docker-compose.gpu.yml up -d

# Enable clustering
./target/release/redfire-switch --config config.json \
  --enable-clustering --cluster-nodes 3 start

# Monitor performance
redfire-cli performance --real-time

Performance Benchmarks

The figures below are the project's documented design targets for the optional GPU transcoding path, not independently verified production benchmarks - GPU transcoding is feature-gated and currently unverified in CI per the project's own README.

🚀 GPU Transcoding (Design Target)

15-20x target speedup vs. CPU-only, on the feature-gated GPU path

G.729 → G.711: 19x target speedup (45μs vs 850μs)
G.722.2 → G.711: 17x target speedup (55μs vs 920μs)
G.729/G.722.2 are simplified reference codecs, not production audio quality

🛡️ Security Processing

Real-time ML threat detection

Behavioral analysis with auto-blocking
5 internally-tracked critical vulnerabilities fixed
Continuous security monitoring

🏢 Enterprise Architecture

Carrier-grade reliability design

Multi-node clustering with automatic failover
Split-brain detection
BGP anycast clustering support

🏛️ Regulatory Compliance & Lawful Intercept

Enterprise-grade telecommunications compliance with comprehensive lawful intercept capabilities meeting international standards.

J-STD-025 & CALEA Compliance

U.S. Lawful Intercept Standard

CALEA Implementation - Communications Assistance for Law Enforcement Act compliance
Warrant Processing - Automated court-ordered call record disclosure
Emergency Call Tracking - E911 records with priority intercept support

ETSI LI (Lawful Intercept)

European Standards Implementation

ETSI TS 102 232 - LI handover specification for IP delivery
ETSI TS 133 108 - 3GPP UMTS 3G security handover interface
HI2/HI3 Delivery - Intercept-related information and content delivery

Security & Call Authentication

Enterprise-Grade Security

STIR/SHAKEN Implementation - RFC 8224/8225 call authentication with PASSporT tokens
Warrant Validation - Automated legal authorization verification
AES-256-GCM Encryption - ETSI-compliant end-to-end encryption
Mutual TLS Authentication - X.509 certificate-based security
Comprehensive Audit Trails - Complete compliance logging and reporting

⚠️ Legal & Compliance Notice

SECURITY WARNING: CALEA and lawful intercept functionality requires proper legal authorization and court orders. Warrant validation, audit trails, and regulatory compliance are mandatory under U.S. and international law. Unauthorized access or misuse may violate federal privacy statutes and telecommunications regulations.

Codec Support Matrix

Codec Matrix - 56 Pairs, Optional GPU Path

Codec	Description	Sample Rate	GPU Accelerated	Performance
G.711 μ-law/A-law	Standard telephony - complete, production-ready	8 kHz	✅	15x design target (GPU)
G.729/G.729A/G.729B	CELP compression - simplified reference implementation, not bit-exact	8 kHz	⚠️ gated	19x design target (GPU)
G.722.2 (AMR-WB)	Wideband ACELP - simplified reference implementation, not bit-exact	16 kHz	⚠️ gated	17x design target (GPU)
G.722	ADPCM wideband - partial implementation	16 kHz	⚠️ gated	12x design target (GPU)
PCM16	Linear PCM - complete, hub codec	8/16 kHz	✅	Hub codec
Opus	Modern codec - placeholder, not yet implemented	48 kHz	❌	Not available

GPU acceleration is behind the cuda/rocm build features and is currently unverified in CI; CPU fallback is always available. Patent-Free: G.722.2/AMR-WB patents expired in 2023, enabling free commercial use of the reference implementation.

🏆 Achievement Status

✅ Core Platform (COMPLETE)

✅ G.711/PCM16 transcoding, CPU or GPU
✅ ML-powered security (5 vulnerabilities fixed)
✅ Class 4 switching platform
✅ High availability clustering with split-brain detection

🚧 GPU Transcoding & Extended Codecs (FEATURE-GATED, UNVERIFIED IN CI)

GPU-accelerated transcoding across 56 codec pairs (behind cuda/rocm build features)
G.729, G.722.2/AMR-WB - simplified reference implementations, not bit-exact
G.722 - partial implementation
Opus - placeholder, not yet implemented

✅ Standards Compliance (COMPLETE)

✅ RFC 3261 (Core SIP) - Full implementation
✅ RFC 8224/8225 (STIR/SHAKEN) - Complete
✅ RFC 3398 (SIP-I/ISUP) - PSTN integration
✅ Multi-RFC integration validated

🔮 Future Enhancements

CI-verified GPU benchmarks
Bit-exact G.729/G.722.2 codec implementations
ROCm/AMD GPU support
Advanced AI/ML integration
Opus codec support

Deploy Enterprise Platform

Experience the Class 4 SIP switch platform with optional GPU acceleration, ML security, carrier-grade clustering, and comprehensive libraries for building telecom software.

Contribute on GitHub Report Issues