AWS Media Services for FAST Channel Streaming over SRT

We built an SRT-based contribution and distribution pipeline using AWS Elemental MediaLive and MediaConnect, enabling reliable, encrypted content transport over public internet with broadcast-grade error correction.

Architecture

• Contribution: AWS Elemental MediaConnect for SRT ingest from remote sources
• Transport: SRT protocol with AES encryption and ARQ error correction
• Encoding: AWS Elemental MediaLive for transcoding SRT inputs to multi-bitrate output
• Packaging: AWS Elemental MediaPackage for HLS/DASH packaging for end-viewer delivery
• Distribution: SRT output from MediaConnect for B2B syndication to platform partners
• Monitoring: SRT-specific metrics dashboard (RTT, packet loss, retransmission, jitter)
• CDN: Amazon CloudFront for last-mile HLS delivery to end viewers

SRT Protocol Advantages

vs. RTMP

SRT provides significant advantages over RTMP for contribution feeds: built-in ARQ error correction (tolerating up to 20% packet loss vs. stream breakage at 1-2% with RTMP), native AES encryption, configurable latency control, UDP-based NAT-friendly transport, and minimal bandwidth overhead for error recovery.

vs. Dedicated Circuits

SRT over internet offers dramatically lower cost and faster provisioning compared to dedicated fiber — with the added benefits of multi-path redundancy and geographic flexibility from any internet-connected location.

Pipeline Design

Contribution (Ingest)

1. Remote Sources — Studios, cloud playout, or partners send SRT streams to MediaConnect
2. SRT Listener — MediaConnect endpoint configured as SRT listener
3. Encryption — AES passphrase encryption for content security in transit
4. Error Correction — ARQ recovers lost packets with configurable latency buffer
5. Failover — Dual SRT inputs with automatic failover on primary stream failure

Processing

1. MediaConnect → MediaLive — SRT flow piped to MediaLive for transcoding
2. Transcoding — Multi-bitrate encoding with SCTE-35 passthrough
3. SCTE-35 Injection — Ad break signals inserted at scheduled points
4. Output — Transcoded stream sent to MediaPackage for HLS packaging

Distribution (B2B Syndication via SRT)

For syndication to platform partners who need a broadcast-grade feed:

• SRT output in caller or listener mode based on partner requirements
• Separate encryption passphrases per partner for access control
• Per-partner bandwidth configuration
• Per-output SRT metrics for partner feed health monitoring

SRT Configuration

Latency Tuning

SRT latency is tuned based on network conditions and use case:

• Ultra-Low — Same-region, high-quality networks (studio to cloud)
• Low — Cross-region, good networks
• Standard — International, variable networks
• High Resilience — Poor networks, maximum packet loss tolerance

Settings are optimized per use case with appropriate latency, bandwidth limits, encryption level, and connection mode (caller vs. listener).

Monitoring & Alerting

The platform monitors SRT-specific metrics in real-time:

• Round Trip Time (RTT) — Network latency between sender and receiver
• Retransmission Rate — Percentage of packets requiring ARQ retransmission
• Packet Loss — Pre-ARQ packet loss rate indicating network quality
• Jitter — Variation in packet arrival times
• Bandwidth Utilization — Actual vs. configured maximum bandwidth
• Buffer Level — Receiver buffer fill level (underrun indicates potential stutter)

Automated alerts trigger on degraded metrics for proactive issue resolution.

Key Features

1. SRT Ingest — Receive contribution feeds from any internet-connected source
2. AES Encryption — Built-in content encryption without external VPN or TLS
3. ARQ Recovery — Tolerates up to 20% packet loss with automatic retransmission
4. Configurable Latency — Tunable based on network quality and use case
5. Dual-Input Failover — Automatic switchover on primary SRT feed failure
6. B2B Syndication — SRT output feeds for partner distribution
7. SRT Metrics Dashboard — Real-time RTT, loss, jitter, and retransmission monitoring
8. Hybrid Output — SRT for B2B contribution, HLS via CloudFront for consumer delivery