Skip to content

testing_tutorial.md

Latest commit

 

History

History
252 lines (171 loc) · 9.02 KB

testing_tutorial.md

File metadata and controls

252 lines (171 loc) · 9.02 KB

Video streaming testing tutorial combined with EdgeVPN

1. Install and start EdgeVPN.

Install evio and start EdgeVPN all on Raspberry Pi, edge device (arbitrary device can run EdgeVPN closer to Raspberry Pi), and Cloud VM. For transmission delay measure, no need noise removal on cloud. AWS EC2 t2. series is enough.

2. Prepare Python environment and dependencies.

Install SINETStream library:

sudo pip3 install sinetstream-kafka sinetstream-mqtt
sudo pip3 install sinetstream-type-image

Due to sinetstream-type-image depending on opencv-python, you can preinstall lots of its dependencies and opencv-python to prevent error during installing.

sudo apt install libavutil56 libcairo-gobject2 libgtk-3-0 libpango-1.0-0 libqt5core5a libavcodec58 libcairo2 libswscale5 libtiff5 libatk1.0-0 libavformat58 libgdk-pixbuf2.0-0 libilmbase24 libopenexr24 libpangocairo-1.0-0 libwebp6 libgl1-mesa-dev libcairo2-dev

sudo pip3 install opencv-python

Then, download program and configure file.

Prepare reader program and configuration file

Below is the procedure same on all devices.

  1. Create a directory for Reader
  2. Prepare the SINETStream configuration file
  3. Prepare the Reader program

Create a directory and change to that directory.

[user01@reader]$ mkdir -p ~/sinetstream/reader
[user01@reader]$ cd ~/sinetstream/reader

Prepare SINETStream configuration file. Download the configuration file prepared for this tutorial from GitHub.

[user01@reader]$ ss_url=https://raw.githubusercontent.com/Commin/SINETStream-videostreaming/main
[user01@reader]$ curl -O ${ss_url}/reader/.sinetstream_config.yml

Download the sample program of Reader that uses the SINETStream Python3 API from GitHub. Grant execute permission to the program.

[user01@reader]$ curl -O ${ss_url}/reader/consumer.py
[user01@reader]$ curl -O ${ss_url}/reader/video_consumer.py
[user01@reader]$ chmod a+x consumer.py video_consumer.py
  • producer.py can only work with consumer.py to send/receive text message through SINETStream API.
  • rasp_video_producer.py can work with video_consumer.py to send/receive video stream data through SINETStream API.

Verify that the above procedure has been performed correctly. Make sure that the directories and files are the same as in the example below.

[user01@reader]$ pwd
/home/user01/sinetstream/reader
[user01@reader]$ ls -a
.  ..  .sinetstream_config.yml consumer.py video_consumer.py

On edge terminal

Besides, you need to download another program file `` to receive data from Raspberry Pi and send data to cloud broker intermediately.

[user01@reader]$ curl -O ${ss_url}/reader/edge_consumer.py
[user01@reader]$ chmod a+x edge_consumer.py
  • edge_consumer.py can work with rasp_video_producer.py and video_consumer.py to receive data from Raspberry Pi and send video stream data to cloud intermediately.

Prepare writer program and configuration file

Below is the procedure.

  1. Create a directory for Writer
  2. Prepare the SINETStream configuration file
  3. Prepare the Writer program

Create a directory and change to that directory.

[user01@writer]$ mkdir -p ~/sinetstream/writer
[user01@writer]$ cd ~/sinetstream/writer

Prepare SINETStream configuration file. Download the configuration file prepared for this tutorial from GitHub.

[user01@writer]$ ss_url=https://raw.githubusercontent.com/Commin/SINETStream-videostreaming/main
[user01@writer]$ curl -O ${ss_url}/writer/.sinetstream_config.yml

Download the sample program of Reader that uses the SINETStream Python3 API from GitHub. Grant execute permission to the program.

[user01@writer]$ curl -O ${ss_url}/writer/producer.py
[user01@writer]$ curl -O ${ss_url}/writer/rasp_video_producer.py
[user01@writer]$ chmod a+x producer.py rasp_video_producer.py

Verify that the above procedure has been performed correctly. Make sure that the directories and files are the same as in the example below.

[user01@writer]$ pwd
/home/user01/sinetstream/writer
[user01@writer]$ ls -a
.  ..  .sinetstream_config.yml producer.py video_producer.py

Add service to configure file.

Please ignore this part, if configure file includes the service information.

Here is edge-video-kafka information:

edge-video-kafka:
    type: kafka
    brokers: "edge-broker:9092"
    topic: video-kafka
    value_type: image

Here is cloud-video-kafka information:

cloud-video-kafka:
    type: kafka
    brokers: "broker:9092"
    topic: video-kafka
    value_type: image

Add service edge-video-kafka and cloud-video-kafka information to sinetstream/reader/.sinetstream_config.yml and sinetstream/writer/.sinetstream_config.yml on Raspberry Pi.

Add service edge-video-kafka and cloud-video-kafka information to sinetstream/reader/.sinetstream_config.yml on edge device.

Add service cloud-video-kafka information to sinetstream/reader/.sinetstream_config.yml on cloud instance.

3. Start broker container.

Prepare Broker container

Run the backend messaging systems (Kafka and MQTT) used by SINETStream in a Docker container.

Execute the following command in cloud instance.

[user00@host-broker]$ docker run -d --name broker --hostname broker \
                      -p 1883:1883 -p 9092:9092 sinetstream/tutorial:latest

Show the status to confirm that the container has started successfully.

[user00@host-broker]$ docker ps -l
CONTAINER ID        IMAGE                        COMMAND                  CREATED              STATUS              PORTS                                            NAMES
xxxxxxxxxxxx        sinetstream/tutorial:latest  "/usr/local/bin/supe…"   About a minute ago   Up About a minute   0.0.0.0:1883->1883/tcp, 0.0.0.0:9092->9092/tcp   broker

Change hostname to edge-broker and run edge broker container on edge device (same device as edge-consumer.py is OK).

[user00@host-broker]$ docker run -d --name broker --hostname edge-broker \
                      -p 1883:1883 -p 9092:9092 sinetstream/tutorial:latest

4. Add IPs to hosts file

On Raspberry Pi:

Add IP of broker, edge-broker to /etc/hosts.

sudo vim /etc/hosts

[broker-ip] broker
[edge-broker-ip] edge-broker

On Edge:

Add IP of broker to /etc/hosts.

sudo vim /etc/hosts

[broker-ip] broker

5. Evaluate transmission latency

Evaluate transmission latency (not including weather classification & image processing) with / without EdgeVPN (Need editing IPs in /etc/hosts file).

Variables: Three video solutions: 240p, 360p, 480p Three pre-set fps: 60p, 30p, 20p Three transmission routes:

  1. Rasp - Edge - Rasp
  2. Rasp - Cloud - Rasp
  3. Rasp - Edge - Cloud - Rasp

First, run video_consumer.py on Raspberry Pi

For instance, for Rasp - Edge - Cloud - Rasp route, you can run video_consumer.py as follows:

sudo python3 video_consumer.py -s edge-video-kafka --csv ./csv/edge_cloud_reader.csv

Second, run edge_consumer.py on edge terminal (only need for the third routes)

you can run edge_consumer.py as follows:

sudo python3 edge_consumer.py --edgeservice edge-video-kafka --cloudservice cloud-video-kafka

Third, run exp_rasp_video_producer.py on Raspberry Pi (for Round-trip time)

you can run exp_rasp_video_producer.py as follows:

sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 426 --height 240 --fps 60 --csv ./csv/rasp_edge_cloud_rasp_240p_60fps.csv --frame 100
sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 426 --height 240 --fps 30 --csv ./csv/rasp_edge_cloud_rasp_240p_30fps.csv --frame 100
sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 426 --height 240 --fps 20 --csv ./csv/rasp_edge_cloud_rasp_240p_20fps.csv --frame 100

sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 640 --height 360 --fps 60 --csv ./csv/rasp_edge_cloud_rasp_360p_60fps.csv --frame 100
sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 640 --height 360 --fps 30 --csv ./csv/rasp_edge_cloud_rasp_360p_30fps.csv --frame 100
sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 640 --height 360 --fps 20 --csv ./csv/rasp_edge_cloud_rasp_360p_20fps.csv --frame 100

sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 854 --height 480 --fps 60 --csv ./csv/rasp_edge_cloud_rasp_480p_60fps.csv --frame 100
sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 854 --height 480 --fps 30 --csv ./csv/rasp_edge_cloud_rasp_480p_30fps.csv --frame 100
sudo python3 exp_rasp_video_producer.py -s edge-video-kafka --width 854 --height 480 --fps 20 --csv ./csv/rasp_edge_cloud_rasp_480p_20fps.csv --frame 100

For the first & second routes, please edit --csv parameters correspondingly. Particularly, for the second route: Rasp - Cloud - Rasp, please edit -s parameter to cloud-video-kafka.

Finally, list all data from .csv file and calculate average latency (RTT) for all conditions.