Ultra-Low latency video transport

We are planning to build new streaming system for a cloud-based video surveillance service. In the new system, Secure Reliable Transport (SRT) will be used as a base protocol to archive the critical requirements; ultra-low latency, packet loss recovery. SRT is considered to meet these requirements, but it is not enough to integrate to the existing service, or newly developing business model. Hence, this project aims to develop SRT based ecosystem to satisfy various requirements; high performance, high scalability, and 3rd-party friendly SDK. The final goal is to implement whole software stacks from embedded system to user application.

Overview

The ecosystem consists of a distributed system that allows clients to access Edge devices streaming using SRT protocol to provide ultra low latency video.

Libraries

• Gaeguli: SRT streaming library

• Hwangsae: Stream relay and access control library

Services

• Source - provides access to capture devices, encodes and streams video

• Relay - provides stream forwarding and accesss control

• MJPEG - transcodes H.264 MPEG-TS into MJPEG stream

All services are implemented in Gaeul component on top of libgaeguli and libhwangsae.

Description

To allow N different clients to access the streaming genearted by an Edge device, this project implements a distributed system.

Edge devices running Gaeul Agent or Servers running Hwangsaeul Relay Agent provide D-Bus APIs that third party applications can use to interact with Hwangsaeul components.

Visualized Effort

This video is compact visualization of our efforts from the beginning. We will try to keep updating every month.

$ gource \
        -1920x1080 \
        -s .2 --auto-skip-seconds .2 \
        --multi-sampling \
        --stop-at-end \
        --output-framerate 30 \
        --hide mouse,progress \
        --highlight-users \
        --output-ppm-stream - \
        combined.log  | \
        avconv -y -r 30 -f image2pipe -vcodec ppm -i - \
        -b 32768k hwangsaeul.mp4