Skip to content

Project-GrADyS/gradys-gs

 
 

Repository files navigation

Gradys Ground Station

Web application, from Project GrADyS, to monitor, control and display mobile devices networks in field tests

Introduction

This is a repository for the Ground Station framework, developed for the GrADyS project and future IoT projects. It's an extensible and reusable framework to help visualize the location and activity status of interconnected network nodes, monitor and store the flow of data and send commands to a set of devices with different protocols. According to the project's needs, the framework is extensible, introducing ways to insert new buttons, commands, protocols, and functionalities.

Main showcase

Main showcase 2

Installation

Prerequisites

In order to use the components in this repository, you need to have Python 3.0 or higher installed. Also pip, a Python package manager, is recomended to manage and automatically install the required packages of this project. To install Python on Windows, follow these instructions. After installing Python, pip should be installed by default. You can check if it's already installed and it's version:

C:\> python3 -m pip --version

If not installed or need to updgrade, you can get more information here.

Cloning the repository

With Python3 installed, you should be able to clone this repository. More information on how to clone.

Creating a virtual environment

In order to keep this framework in a separate environment, with it's own packages and versions, it's recommended to create a virtual environment. On Windows:

Windows
C:\> python3 -m venv C:\path-to-this-cloned-repository/venv

On Linux, you can check if virtualenv is already installed, install it, if not already installed, and create the venv:

Linux
gradys-gs$ virtualenv --version
  virtualenv xx.x.x
gradys-gs$ sudo pip3 install virtualenv
gradys-gs$ virtualenv venv

This will create a folder called venv, inside the project's folder. Now you have to activate the environment to install/use packages only from this venv.

Windows
C:\> C:\path-to-this-cloned-repository\venv\Scripts\activate
Linux
gradys-gs$ source venv/bin/activate

If you need more information about virtual environments with python, it can be found here.

Installing necessary packages

The list of necessary packages are inside requeriments.txt file, if you are using Windows. It'll be installed automatically, using the Python package manager, pip. You can install, running on Windows console:

Windows
C:\path-to-this-cloned-repository\> pip3 install -r requeriments.txt

On Linux, you should run the compatible script file, requeriments_linux.txt:

Linux
gradys-gs$ pip3 install -r requeriments_linux.txt

Secret variables

This project uses Google Maps services, with paid features IFF used above a threshold. To use these functionalities you need to have or create a Google Maps API Key. Google's guide on how to create an API Key.
This project also use Django Framework that has a secret key variable, for security purposes. You can generate your Django secret key here. The framework will load automatically these as environment variables. With both private keys created,

  • Create a file named /config/.env and insert the secret keys:
    • SECRET_KEY='xxxx' Changing xxxx with your Django secret key
    • GOOGLE_MAPS_API_KEY='xxxx' Changing xxxx with your Google Maps key (Maps javascript API on your Google Cloud API)
    • You shall maintain the ' ' from the 'xxxx'

Usage

Running the server

Django provides lightweight development Web server, that you can use via manage.py file. By default, the server runs on port 8000 on the IP address 127.0.0.1 and should not be used on production. You can run with:

Windows
C:\path-to-this-cloned-repository\> python3 manage.py runserver
Linux
gradys-gs$ python3 manage.py runserver

Or, with diferent IP/PORT, in the example below, Port 8000 on IP address 0.0.0.0. This IP is will listen to all IP adresses the machine supports. So for example, with this server configuration up, you can open the web navigator with localhost:8000 and the inet ip obtainable from ifconfig (linux environment):

C:\path-to-this-cloned-repository\> python3 manage.py runserver 0.0.0.0:8000

Remember to insert, inside config.ini file, the correct IP + Port, on [post] category, if changed to a specific IP, when running the command above.

Connecting to home page

Now you should be able to connect to the home page, acessing, on your browser, the IP/PORT the server is up, on default: localhost:8000.

Project Architecture and customization

Gradys-gs's architecture was designed to make it adaptable and customizable to different needs. The functionalities are modularized, helping to customize and insert new elements, such as command buttons and communication protocols with IoT devices. Gradys Ground Station is structured following the classic concept of web development, with Front-end module, responsible for the interface and visualization, and Back-end module, responsible for server-side information processing.

The project's back-end module was built using the Python programming language, together with the use of Django Framework, a tool written in Python, for the agile development of a web application. This framework provides an easy-to-configure development server, a simple and extensible URL route mapping, and a highly modular architecture.

The front-end module was built using HyperText Markup Language (HTML) or Template language, the Cascade Style styling language Sheets (CSS) and the most popular programming language in use, Javascript.
The technologies mentioned were chosen due to their wide use and popularity, contributing to a framework with more extensible usability. Python and Javascript programming languages ​​also have a vast set of libraries and packages, reflecting an extensive community.
Both modules comunicate with each other via WebSocket channels. A socket connection is a dedicated full-duplex channel based in the Transmission Control Protocol (TCP). This project uses Django Channels library to handle WebSockets communication.

Project Architecture

As shown in Figure above, external devices can send messages to FlexStation via the Connections sub-module. The information processing is done in the Django Channels submodule, also responsible for passing the information to the interface, through an already established WebSocket connection. Messages exchanged between the frontend and backend modules follow the JSON format. Note that the described path, from the external device to the interface, is also possible in reverse, when a command is activated on the interface.
The information gate of the ground station to external devices is through Connections submodule, which constains the routes and logic to receive/send information.

Django

URL/View

To understand the server-side structure of this project, first it's required a basic understanding of how Django is structured and how it operates. Building a URL scheme with Django is a simple task, thanks to the URL/View mapping that the python web framework provides. When a user requests a page from the URL schema, Django does a mapping to the corresponding Python function, that's called View.
So, for example, the URL scheme below has a mapping between the home page path and index view, also between /command/ path (note that 'command' is a simple integer) and receive_command_test view.

urlpatterns = [
    path('', index),
    path('<int:command>/', receive_command_test),
]

Inside the main app's folder, connections, there is urls.py and views.py files. The urls.py file is responsible for making the association between a URL address and a view. Note that there is another urls.py file, inside the config folder, that is responsible for the whole project's pathing. So, for example, if there was another app in our project, we could create a prefix path to that specific app. Our main app has the default path, so there's no prefix attached. If you want to add a new URL path, it should be added a new path() item inside the urlpatterns list, in /connections/urls.py. For example:

urlpatterns = [
    path('', index),
    path('<int:command>/', receive_command_test),
    path('new-path/', new_view)
]

Now we want a view to handle the new url path request. A view is a Python function that takes a Web request and returns a Web response. This response can be the HTML contents of a Web page, or a JSON or a redirect, or a 404 error, anything, really. The view itself contains whatever arbitrary logic is necessary to return that response.

def index(request):
  context = {
    'google_maps_key': settings.GOOGLE_MAPS_API_KEY
  }
  return render(request, 'index.html', context=context)

The example above is the index view, accessed when home page is loaded. It receives a request, creates a context variable, with the google maps key from .env, and load the index.html template, attached with the context. We store our views inside /connections/views.py. If you want to create the new view, it should receive a request and return something (could be anything). To send additional parameters, you can send via the url, for example, the url localhost:8000/new-path/10/, needs to be declared inside the connections/urls.py as integer as:

path('new-path/<int:id>/', new_view)

And our view can receive an id parameter, as follows:

def new_view(request, id):
  # Function Logic
  return 

Now, accessing the default server 127.0.0.1:8000/new-path/5/ is going to call our new_view method, sending the parameter 5.

Routing/Consumers

Our main app form of communication with templates, or HTMLs, is using websocket connections. Django Channels package mediates these connections. The logic to stablish a websocket connection is similar with the URL/View logic presented on the topic above. The connections/routing.py file contains the websocket url patterns, or schema:

ws_urlpatterns = [
  path('ws/connection/', ConnectionConsumer.as_asgi()),
  path('ws/receive/', ReceiveCommandConsumer.as_asgi()),
]

As said, Django Channels makes a mapping, associating an url with a Consumer. A Consumer is a Python Class that handles a websocket connection. So, when our Javascript is loaded, it tries to connect with a specific Consumer, accessing a specific URL, inside our ws_urlpatters.

// Javascript stablishing new connection
var socket = new WebSocket('ws://localhost:8000/ws/connection/');

When this command is read, the ConnectionConsumer class is called and a connection is initated. Our Consumers are inside connections/consumers_wrappers/ and a new one can be created, inheriting WebsocketConsumer or AsyncWebsocketConsumer, depending on it's functionality. You can substitute three main methods:

  • connect: called when the specific url is accessed and start a dedicated connection with self.accept. This is the only method you NEED to override.
  • receive: called when a message is sent via socket connection.
  • disconnect: called when the connection is closed.

Creating a new Consumer, is simple as creating a new file inside connections/consumers_wrapper/ with a Class like:

class NewConsumer(AsyncWebsocketConsumer):
  async def connect(self):
    await self.accept()

  async def receive(self, message):
    # Handle the message

  async def disconnect(self, close_code):
    # Handle disconnection

  async def additional_method(self, *args):
    # Additional method's logic

To send a message to the other side of connection (Django -> Javascript) it can be done using the send method, inherted from WebSocket class:

await self.send(data)

Creating the new path can be done adding a new path to ws_urlpatters list:

ws_urlpatterns = [
  path('ws/connection/', ConnectionConsumer.as_asgi()),
  path('ws/receive/', ReceiveCommandConsumer.as_asgi()),
  path('ws/new-socket/', NewConsumer.as_asgi()),
]

Finnaly, accessing ws://localhost:8000/ws/new-socket/, a dedicated full-duplex connection should be stablished and our two ends can communicate with each other.

Front-end

Our front-end consists of templates files, CSS styling files and Javascript files. The home page template file is rendered when the default ip+port is accessed, as showed above. New templates files can be added inside the /templates/ folder. They work very similarly to HTML files, with some add-ons.

{% load static %}
<link rel="stylesheet"  href="{% static 'connections/css/connection.css' %}">

The code above introduces the '{% %}' tag (that's not HTML native), in this case, to load a css file to the page. For more information about templates, you can access here. To load a Javascript file in a template, the logic is the same, as long this Javascript file is inside the folder that STATIC_URL variable is pointing to. This variable is inside config/settings.py. In our case STATIC_URL variable is pointing to /static/ folder.

STATIC_URL = '/static/'

Our index.html home page template loads gmap.js, responsible for Google Map's virtual map and main.js, responsible for starting websockets connections with the back-end and the main button's logic. To start a websocket connection, you have to create a new object, sending an available URL in the routing schema (see Routing/Consumers topic).

var socket = new WebSocket('ws://localhost:8000/ws/connection/');

This object has methods to interact with the socket connection. Here are the main:

  • send: Transmits data to the server via the WebSocket connection.
socket.send(jsonToSend);
if (socket.readyState == WebSocket.OPEN) {
    // Handle connection OPEN
}
  • onclose: An event listener to be called when the readyState of the WebSocket connection changes to CLOSED.
socket.onclose = function(e) {
  // Handle connection closed
};
  • onmessage: An event listener to be called when a message is received from the server. Receives a message parameter.
socket.onmessage = function(msg) {
  // Handle message received
}

External Communication

The primary purpose of this framework is to exchange information with other devices. Currently, there are implemented two ways for external connections.

The first way is to plugin an ESP32 microcontroller to the framework's machine. This microcontroller should be able to detect other devices that receive and send information to them. Our framework can establish a UART connection with a plugged ESP32 microcontroller, receive everything sent via serial, and send commands via serial, making the microcontroller responsible for retransmitting the command. In order to accept a connection with an ESP32 microcontroller, it is necessary to insert the correct UART Port and baud rate inside config.ini. The SerialConnection class, from /connections/serial_connector.py, is instantiated when javascript starts a WebSocket connection of this type. The instantiated object keeps trying connection with the UART Port. Once a microcontroller is plugged in, the interface indicates this change, and you can exchange information through the ESP32 microcontroller.

Another way to communicate with our framework is with POST requests. A device, an UAV (drone) per se, wants to send its location to our ground station. This can be achieved with a POST request to the specific ground station URL.

json_tmp = {"id": uav_id, "lat": targetpos.lat, "lng": targetpos.lng, "alt": targetpos.alt, "ip": args.uav_ip + ':' + flask_port}

r = requests.post(path_to_post, data=json_tmp)

Note that the device should attach, on the message, it's own IP and PORT, so our framework can send commands back to it. The specific URL, to receive POSTs, is mapped to a view. So, when the device send it's location on body's request, the post_to_socket view receive the request and extracts the information from it's body. We want to send this information to our interface and also to save it in the log file. Who is responsible for both actions is the PostConsumer, inside /connections/consumers_wrapper/post_consumer.py. This way, the post_to_socket view needs to send the message to PostConsumer, getting an instance of this class and calling this Class function receive_post(message).

post_consumer_instance = get_post_consumer_instance()
await post_consumer_instance.receive_post(new_dict)

Sending a message to an external device is also done by Consumers. When a command button is activated on the interface, the main.js uses the async method socket.send(), to transmit the command direct to the Consumer (back-end). The message received from the main.js, contains which device or group of devices it should be sent. It also contains the ID of the external devices that will receive the command. The first step is to search on the registered device's list for the address (IP) of the devices.

device_to_send_list = get_device_from_list_by_id(device_receiver_id)

There is a list on config.ini mapping the commands code (integer) to a specific endpoint, that should be added to the IP+Port of the external device.

[commands-list]
20 = position_absolute_json,get
22 = position_relative_json,get
24 = auto,get
26 = run_experiment,get
28 = set_auto,get
30 = rtl,get
32 = takeoff_and_hold,get

The list contains the endpoint and the HTTP request type, if it is a GET or POST request.

With the address complete, the command will be sent via HTTP request.

command_path_list = config['commands-list'][command].split(',')
endpoint = command_path_list[0]

if command_path_list[1] == 'get':
  #GET request
  task = asyncio.create_task(self.send_get_specific_device(url, id, device['device']))
else:
  # POST request
  task = asyncio.create_task(self.send_post_specific_device(url, json_to_send))
self.async_tasks.append(task)

Depending on the type of the request, the command will be sent and an asynchronous task will be created.

Data persistence

One of the main features of this project is the data persistence of every event that occurred during the experiments. Log files are generated, when starting the application, and filled in as messages are received, errors are caught, commands are sent, and other events that are of importance to the experiment.

To generate the .log files, the logging package, for Python, is used. Inside /connections/utils/logger.py there is a class Logger, responsible for the persistent logic. It's possible to extend and copy this class to other modules, for example, at the uav_simulator/ module that has this class with different logic.

When the server start, a .log file is created, inside the folder specified by the Logger's path variable, and the file's name is composed by the module name followed by the date created. The example above represents a .log file created inside the uav_simulator module at 25/01/2022 08:18:40.

uav_simulator-2022-01-25-08-18-40.log

To fill this file, it must be inserted in code calls of the methods from the Logger class, according to it's needs. The example above includes the code from the PostConsumer class, inside the method to handle a external message received.

logger.log_info(source=source, data=data, code_origin='receive-info')
try:
  await self.send(json.dumps(data)) # Send to JS via socket
except Exception:
  logger.log_except()

The logger object is global and already instantiated. Two log methods are called, to save the data received and to save the Exception caught when trying to send the message to the front-end via socket.

The .log file format is specified inside the Logger class, using the syntax accepted by the Formatting class, form logging package. For more information on how to format the .log file, https://docs.python.org/3/library/logging.html#logging.Formatter.

2021-12-12 20:58:32,706; uav-21; receive-info; {'id': 21, 'type': 102, 'seq': 30, 'lat': -15.840081, 'lng': -47.926642, 'alt': -0.03, 'device': 'uav', 'ip': 'http://127.0.0.1:5071/', 'method': 'post', 'time': '2021-12-12T20:58:32.706792', 'status': 'active'}

2022-01-25 20:58:50,365; gs; send-get; http://127.0.0.1:5071/rtl

The example above has two messages, formatted with the date of the event, who triggered the event, where it was triggered and the message itself.

Sequence Diagram

The sequence message diagram below represents the messages flow between external devices and the main modules from this framework.

Project Architecture

Note that the message protocol between the framework and external devices can differ from project to project, changing the way the information is delivered or the commands are handled. But, the messages flow between the back-end and front-end modules should remain similar to this diagram.
Important things to notice are:

  • When a device send information to the framework, the back-end will register this device, if not already on the persistant list, log the information and forward to front-end, to update the interface.

  • There is a Consumer in charge to keep the persistant device list, with the registred devices, inside /connections/consumers_wrapper/update_periodically.py. In this Consumer, there is a task to update the activity status of the devices on the list, every X seconds, specified at config.ini. This is represented on the third group of messages flow in the diagram above.

  • There is the possibility to create checkbox buttons, that will trigger a constant task, while the checkbox is pressed. This is represented on the fourth group of messages flow in the sequence diagram.

Command Buttons

Another important functionality in this framework is the possibility to send commands, through the interface, to available devices. We can register a new button inside the template, create a onClick callback function and send the command via websocket to Django (back-end).

  • Create new button in /templates/index.html
<input class="button" id="new-button" type="button" value="New Command">
  • Register an onclick function, in /static/connections/main.js
var newCommandNumber = 40

document.querySelector('#new-button').onclick = function(e) {
  sendCommand(newCommandNumber);
};
  • Send to the back-end, when button is clicked
function sendCommand(type) {
  jsonToSend = {id: 1, type: type}
  // ...
  if (socket.readyState == WebSocket.OPEN) {
    socket.send(jsonToSend);
  }
}

Notice that the socket object must be instatiated already, and the connection 'OPEN'. The corresponding Consumer will receive the message and handle, acording to it's command type.

A button can contain a different type, as checkbox. This button will have a different logic then a regular button. A specific thread will handle the command, repeating until the checkbox is unmarked. Marking a checkbox will result on the creation of a thread, unmarking it will cancel this thread. In order to create a checkbox button, you need to change the type to type="checkbox", on index.html, and pass the argument "checkbox" when creating a event handler on main.js: sendCommand(30, "checkbox");

The button type will be insert inside the JSON message, sent by main.js to the corresponding Consumer. So the Consumer will know this is not a regular command, and will create or cancel a thread.

Command button logic

You already have a button on interface that sends a command, in this case '40', to a Consumer. This Consumer will be in charge to the command logic.
Inside the 'receive' method of this Consumer's Class, it's up to you to write the command's logic, according to your communication protocol.
When handling with HTTP requests, you can insert the new command to the command's list, inside config.ini file. The Consumer can iterate this list and check the command received, mapping to the right endpoint.

[commands-list]
20 = position_absolute_json,get
22 = position_relative_json,post
32 = takeoff_and_hold,get
...

This list contains a number as the key to the corresponding endpoint address, that will receive the HTTP request. The type of request is represented after the comma, with no spaces. If your communication is using HTTP requests and this list, your new list, with the new command, should look like this:

[commands-list]
20 = position_absolute_json,get
22 = position_relative_json,post
32 = takeoff_and_hold,get
...
40 = new_endpoint,get

Communicating with external devices

The main purpose of this framework is to exchange information with other devices. Currently there is implemented two ways for external connections.

Serial Connection

The first way is plugin a ESP32 microcontroller to the framework's machine. This microcontroller should be able to detect other devices, receive and send information to them. Our framework can stablish an UART connection with a plugged ESP32 microcontroller, receive everything is sent via serial and send commands via serial, making the microcontroller responsible for retransmiting the command. In order to accept a connection with a ESP32 microcontroller, you need to insert the correct UART Port and baudrate, inside config.ini (see below for Changing the code topic and Serial connection subtopic). The SerialConnection class, from /connections/serial_connector.py, is instantiated when javascript starts a websocket connection of this type. The instantiated object keeps trying connection with the UART Port. Once a microcontroller is plugged, the interface indicates this change, and you are able to exchange information through the ESP32 microcontroller.

POST Requests

Another way to communicate with our framework is with POST requests. A device, let's say an UAV (drone), wants to send it's location to our ground station. This can be achieved with a POST request to a specific URL, registered in config.ini file (see below for Changing the code topic and HTTP Requests subtopic). Note that the device should attach, on the message, it's own IP and PORT, so our framework can send commands back to it. The specific URL, to receive POSTs, is mapped to a view. So, when the device send it's location on body's request, the post_to_socket view receive the request and extracts the information from it's body. We want to send this information to our interface and also to save it in the log file. Who is responsible for both actions is the PostConsumer, inside /connections/consumers_wrapper/post_consumer.py. This way, the post_to_socket view needs to send the message to PostConsumer, getting an instance of this class and calling this Class function receive_post(message).

Changing the code

Some of the framework's informations are initialized by the config.ini file. Below are listed the parameters that can be changed.

Serial connection

One way this framework can comunicate with a network is with a dedicated ESP32, using UART Protocol. The ESP device connected via serial has a specific PORT and Baudrate, that can be changed inside config.ini with the [serial] tag:

[serial]
# The serial PORT the ESP32 is connected
port = COM4

# Rate of information transferred in the serial port
# Needs to be the same in ESP32 connection
baudrate = 115200

# If this Protocol is used
serial_available = false

HTTP Requests

Another way to comunicate with nodes of the network is receiving/sending information via POST/GET Requests. Django provides a routing system that acessibles URLs trigger methods, or Views. A device can send a POST request to http://127.0.0.1:8000/update-info/ (or IP/PORT running the application). Notice that a device should send inside the message it's own IP/PORT, so the application can send commands via HTTP requests. This structure is described with more details below, on the Project Struct topic.

Inside the config.ini file, below the [post] tag, you can change some of the protocol's variables:

[post]
# Default ip/port of django's server.
# If started on a different configuration, you need to change it here.
ip = http://127.0.0.1:8000/

# Endpoint that'll receive POST requests with device's information
path_receive_info = update-info/

List of devices updater

The application saves the latest messages of unique devices in a list, inside the update_periodically_consumer.py, for each execution. From time to time, it's sent to the front-end, via web-socket, with the activity status of each device. A device can be active, on hold and inactive, depending on the interval of it's last message. These variables can be adjusted in the config.ini file, below the [list-updater] tag:

[post]
# The amount of seconds to a device be considered 'inactive'
seconds_to_device_be_inactive = 50

# The amount of seconds to a device be considered 'on hold'
seconds_to_device_be_on_hold = 25

# The amount of seconds to update the list of devices in Front-End
update_delay = 20

Folders structure

.
├── config              # Contains Django's configurations files
├── connections         # Main app folder
├── static              # Static js, css, images files
├── templates           # Files with template language (html)
├── uav_simulator       # Logic to run Ardupilot SITL simulator 
├── config.ini          # Contains project's adjustable parameters
├── manage.py           # Django’s command-line utility for administrative tasks
├── requeriments.txt    # Contains all packages and versions required for Windows
├── ´´_linux.txt        # Contains all packages and versions required for Linux
└── README.md

We will open the folders that require more attention:

Connections This is the main app's folder, with the necessary tools to allow connections and information exchange with other devices

.
├── ...             
├── connections
|   ├── consumers_wrapper   # Folder with all websocket consumers
|   ├── LOGS                # Stores all .log files generated
|   ├── utils               # Auxiliary tools
|   ├── ...
|   ├── routing.py          # Paths for websocket connections
|   ├── serial_connector.py # Auxiliary class to stablish/handle connection with esp32 device
|   ├── urls.py             # Paths for views
|   ├── views.py            # Methods called when specific url is accessed
└── ...

Templates This folder contains the files written with template language. These files can be associated with traditional HTML files, but with tags, interpreted by Django.

.
├── ...             
├── templates
|   ├── index.html   # Home page, runs main.js
└── ...

Static Contains static files, like Images, CSS and Javascript. Django provides django.contrib.staticfiles to help you manage them. These files can be configured inside config/settings.py:

STATIC_URL = '/static/'

In the templates files, can be used the static template tag to build the URL for the given relative path using the configured STATICFILES_STORAGE:

{% load static %}
<img src="{% static 'my_app/example.jpg' %}" alt="My image">
.
├── ...             
├── static
|   ├── connections   # Folder to separate the main app's static files
|   |    ├── css      # All css used inside connections
|   |    ├── images   # All images used in the project
|   |    ├── js       # All javascript files
    └──
└── ...

Releases

No releases published

Packages

No packages published

Languages

  • Python 83.7%
  • JavaScript 9.2%
  • CSS 3.7%
  • HTML 3.0%
  • Shell 0.4%