5GMETA Sensors and Devices

Introduction

In this page, we introduce examples, including scenario, pre-conditions and scripts/Dockers, to produce video streams.

The examples presented are located in the examples/video-stream-broker folder.

Scenario

A video sensor wants to push a UDP video stream including a RTP video with H264

Pre-Conditions

The following systems must be configured and running:

• The Message Broker (AMQP) is up and running
• The Kakfa Broker (KAFKA) is up and running and ready at EKS. It polls data through Kafka Connector and KSQLDB from AMQP
• The GPS positions in examples/video-stream-broker/video_sensor.py to get the tiles are hardcoded in the example removing the integration with a specific GPS device.
• The component for the Video Sensor is built but not running

Usage

• To Push a UDP Source to 5GMETA MEC

1. First, we need a UDP SOURCE or RTSP Stream with an H.264 video

   • $ gst-launch-1.0 videotestsrc is-live=true do-timestamp=true ! videoconvert ! videorate ! video/x-raw, width=640, height=480, framerate=10/1 ! textoverlay font-desc="Arial 40px" text="container TX" valignment=2 ! timeoverlay font-desc="Arial 60px" valignment=2 ! videoconvert ! tee name=t ! queue max-size-buffers=1 ! x264enc bitrate=2000 speed-preset=ultrafast tune=zerolatency key-int-max=5 ! video/x-h264,profile=constrained-baseline,stream-format=byte-stream ! h264parse ! rtph264pay pt=96 config-interval=-1 name=payloader ! application/x-rtp,media=video,encoding-name=H264,payload=96 ! udpsink host=127.0.0.1 port=7000 sync=false enable-last-sample=false send-duplicates=false

2. Second, we need to run the Video Sensor Docker sending the UDP video stream

   • $ docker run --net=host --env AMQP_USER="<user>" --env AMQP_PASS="<password>" --env VIDEO_SOURCE="udp" --env VIDEO_PARAM="7000" --env VIDEO_TTL="100" 5gmeta/video_sensor
   • Remember to configure the <user>, <password> to your local environment
   • NB:
     • Note that 7000 is the UDP source Port
     • Note that 100 is the timeout until the Video Sensor will stop sending a video stream

3. Third, we can play the video from the 5GMETA MEC infrastructure, produced by the WebRTC Proxy, allowing data pipelines at the 5GMETA MEC infrastructure to process

   • $ gst-launch-1.0 udpsrc port=5045 ! application/x-rtp,encoding-name=H264,payload=96 ! rtpjitterbuffer latency=50 ! rtph264depay ! decodebin ! videoconvert ! textoverlay font-desc="Arial 40px" text="local RX" valignment=1 ! videoconvert ! autovideosink
   • NB:
     • Note that 5045 is the UDP Port produced by the WebRTC proxy based on the provided ID (from the registration process)

Local Consume a AMQP Source

1. First, change amqp2video.py code to set the target ID

   • if (event.message.properties['sourceId'] == 45):
   • NB: Note that 45 should be changed to the dataflow ID identified in the logs coming from registration and WebRTC proxy logs

2. Second, launch the player

   • $ AMQP_USER="<user>" AMQP_PASS="<password>" AMQP_IP="<AAA.BBB.CCC.DDD>" AMQP_PORT="<port>" GST_DEBUG=3 python3 ./amqp2video.py
   • Remember to configure the <user>, <password>, <AAA.BBB.CCC.DDD> and <port> to your local environment

Deployment of the S&D Modules

In this section, a Sensors&Devices instance is created to send data to the 5GMETA MEC platform though the message and video brokers hosted in the MEC.

Prerequisities

• VirtualBox
• Vagrant

Deploy an Instances of a S&D

Installation requirements can be found in the repository. Clone the repository using

git clone ssh://git@gitlab.akka.eu:22522/5gmeta-new/sensors-and-devices.git
cd sensors-and-devices/vagrant
vagrant up

Vagrant will provision the VM with the following Senders:

• cits_sender_python: a python sender and receiver that run an AMQP sender simulating a number of vehicles that send messages with some properties attached.

• image_sender_python: a python sender and receiver that run an AMQP sender simulating a number of vehicle that sends images (different size, but all the same)

• cits_receiver_python: a python example to receive messages from AMQP events