DDNN Adaptive Framework Deployment

1. Environment Setup

Install DDNN environment by following link:

https://hackmd.io/kN4lb_NCQK-WKO0Rkp8tbg (Install to the Chainer installing step)

Install MQTT Service

• Server :

  • broker (python)

  sudo apt-get install mosquitto mosquitto-clients
  

  • client (python)

  sudo pip install paho-mqtt
  

• End :

  • client (C code)

  sudo apt-get install open-ssl libssl-dev
  git clone https://github.com/eclipse/paho.mqtt.c.git
  cd org.eclipse.paho.mqtt.c.git
  sudo make
  sudo make install
  

Basic requirements:

• Server (cloud):
  • Ubuntu 16.04 or higher version
  • Python 2.7
  • Nvidia vendor GPU or accelerator
• Device (end or edge):
  • Ubuntu 16.04 or higher
  • GCC (linux 5.4.0 or higher)
  • (OpenCL 1.2 or 2.0)

2.Download source code

Download the source code https://drive.google.com/file/d/18q-5_wKKEEDTA4T29CXvDMouxnT-rpPF/view?usp=sharing

3.Deployment

The deployment has two part, Device and Server:

• Server Deployment:

  • Setup DDNN framework into server

  cd Cloud_device
  cd Framework
  sudo ./building.sh
  

• Device Deployment:

  • Compile the C program which is generated.

  cd End_device
  vim mqtt_async_t.h  //Change borker of server IP by '#define ADDRESS "tcp://xxx.xxx.xxx.xxx:1883"'
  sudo ./building.sh
  gcc Source.c -o output -lm -lpaho-mqtt3a
  

4.Execution

First execute Cloud and then execute the End device.

• Server Execution

  Run the broker at first

  cd Cloud_device
  python broker_develop.py
  

• End Execution

  Run the inference request after broker is executed on server

  cd End_device
  sudo chmod +x cifar10*
  ./cifar10_oneinput.sh