Creating a OPC-UA server for a virtual weather station.

update on 26/06/2016 for node-opcua 0.0.55.


In my quest of exploring the “Internet of Things” world, I decided to create a simple weather station with 3 sensors mounted on my Raspberry computer. I needed to buy some equipment to build the prototype. After studying different type of sensors (1-Wire,Analog,I2C), I finally opted for I2C sensors. (I used I2C chips a long time ago, in a Junior Enterprise Project). I ordered a breadboard and I2C temperature and humidity sensor.

While waiting for the equipment to be delivered, I though it was time to start coding the Server Application.

I got the idea of using a free Web-Service to get some realtime temperature and pressure information that I need to simulate the data.

The server is written in Javascript, using NodeJS.

retrieving Weather data using a REST API.

The virtual weather station need to extract the weather data from a web service. worldweatheronline provides a free API. The Free API comes with some restrictions, though, and only allows each user to make up to a maximum of 250 requests per day, a.k.a 1 request every 6 minutes)

getting a key at worldweatheronline

You will need to register to worldweatheronline to obtain your API key. Store you API key in a file name worldweatheronline.key, in your project folder.

testing the API

The API is documented here.

For example, typing the following URL in the address bar of your web browser.<YOURAPIKEY>

will return the following JSON data

     "data": {
         "current_condition": [{
             "cloudcover": "75",
             "humidity": "67",
             "observation_time": "11:54 AM",
             "precipMM": "0.3",
             "pressure": "1003",
             "temp_C": "12",
             "temp_F": "54",
             "visibility": "10",
             "weatherCode": "116",
             "weatherDesc": [{
                 "value": "Partly Cloudy"
             "weatherIconUrl": [{
                 "value": "http:\/\/\/images\/wsymbols01_png_64\/wsymbol_0002_sunny_intervals.png"
             "winddir16Point": "W",
             "winddirDegree": "260",
             "windspeedKmph": "31",
             "windspeedMiles": "19"
         "request": [{
             "query": "London, United Kingdom",
             "type": "City"
         "weather": [{
             "date": "2014-05-11",
             "precipMM": "2.5",
             "tempMaxC": "13",
             "tempMaxF": "56",
             "tempMinC": "5",
             "tempMinF": "41",
             "weatherCode": "266",
             "weatherDesc": [{
                 "value": "Light drizzle"
             "weatherIconUrl": [{
                 "value": "http:\/\/\/images\/wsymbols01_png_64\/wsymbol_0017_cloudy_with_light_rain.png"
             "winddir16Point": "W",
             "winddirDegree": "267",
             "winddirection": "W",
             "windspeedKmph": "33",
             "windspeedMiles": "20"

Reading weather data

It is now time to code a function to extract the temperate, the pressure, and the humidity using nodeJs.

preparing the project

First of all, let create the ‘'’package.json’’’ file for our project.

mkdir myweatherstation
cd myweatherstation
npm init

While we are here, let’s install some of the npm modules that we need.

npm install request --save
npm install node-opcua --save

accessing the worldweatheronline API key

Our application will need to access our API developer key. Let’s put it in a file named worldweatheronline.key in our project folder. The key value can be easily read in nodejs using this code.

// read the World Weather Online API key.
var fs = require("fs");
var key = fs.readFileSync("worldweatheronline.key");

Lets write a small worldweather_demo.js, to experiment the api.

testing the rest api

Our purpose is to create a getCityWeather asynchronous function that pass to a callback function an object containing the temperature and pressure of a city. This function will be used this way:

/*global require,console */
/*jshint evil:true */
_"get city weather"
var city = "London";
getCityWeather(city,function(err,data) {
    if (!err) {
    console.log("data = data",data);
        console.log(" city =",city);
        console.log(" time =",data.observation_time);
        console.log(" temperature =",    data.temperature);
        console.log(" pressure    =",    data.pressure);

get city weather

Let’s write the method that reads the weather of a city.

_"accessing the worldweatheronline API key"
function getCityWeather(city,callback) {
    var api_url=""+city+"+&format=json&key="+ key;
    var options = {
        url: api_url,
        "content-type": "application-json",
        json: ""
    var request = require("request");
    request(options, function (error, response, body) {
      if (!error && response.statusCode === 200) {
        var data  = perform_read(city,body);
      } else {
_"extract useful data"

extract useful data

The perform_read function convert the raw json data retrieved from the API,into a simpler Javascript object for our application.

function perform_read(city,body) {
    var obj = JSON.parse(body);
    var current_condition =[0];
    var request =[0];
    return  {
        city:               request.query,
        date:               new Date(),
        observation_time:   current_condition.observation_time,
        temperature:        parseFloat(current_condition.temp_C),
        humidity:           parseFloat(current_condition.humidity),
        pressure:           parseFloat(current_condition.pressure),
        weather:            current_condition.weatherDesc.value

reading data periodically

The Weather Station Server will have to query the weather data of a city on a regular basis. However, we have to be careful not to send too many queries to the web-server, as any exceeding requests will get rejected and will lead to a error:

<h1>Developer Over Rate</h1>

In NodeJs, the setInterval function can be used to perform a action periodically.

var london_data = {}
setInterval(function() {
   getCityWeather("London",function(err,data) {
       if (!err) {
            london_data = data;
}, 60*1000);

Lets edit worldweather_demo2.js to experiment this.

making a round robin read

Why not make our server expose the weather variables of more than one city ? Supposing we hold an array containing the city we want to monitor, our periodic call to the REST API will have to query the data for each city in turn. We will store the most up to date weather data inside a map city_data_map .

I chose 10 cities spread in different continents and hemisphere.

Just to make it fun, I added Longyearbyen, the northenmost city in the world and ‘Ushuaia’ one of the southernmost city.

/*global require,setInterval,console */
var cities = [ 'London','Paris','New York','Moscow','Ho chi min','Benjing','Reykjavik' ,'Nouakchott','Ushuaia' ,'Longyearbyen'];
_"get city weather"
var city_data_map = { };
// a infinite round-robin iterator over the city array
var next_city = function(arr) {
   var counter = arr.length;
   return function() {
      counter += 1;
      if (counter>=arr.length) {
        counter = 0;
      return arr[counter];
function update_city_data(city) {
    getCityWeather(city,function(err,data) {
         if (!err) {
            city_data_map[city] = data;
            console.log(city,JSON.stringify(data, null," "));
         }  else {
            console.log("error city",city , err);
// make a API call every 10 seconds
var interval = 10* 1000;
setInterval(function() {
     var city = next_city();
}, interval);

Weather Server Skeleton

It is now time to create the skeleton of our weather server.


/*global require,console,setInterval */
Error.stackTraceLimit = Infinity;

_"making a round robin read"

var opcua = require("node-opcua");

var server = new opcua.OPCUAServer({
   port: 4334 // the port of the listening socket of the server

server.buildInfo.productName = "WeatherStation";
server.buildInfo.buildNumber = "7658";
server.buildInfo.buildDate = new Date(2014,5,2);
function post_initialize() {
    function construct_my_address_space(server) {
       _"construct the address space"
    server.start(function() {
        console.log("Server is now listening ... ( press CTRL+C to stop)");
        console.log("port ", server.endpoints[0].port);
        var endpointUrl = server.endpoints[0].endpointDescriptions()[0].endpointUrl;
        console.log(" the primary server endpoint url is ", endpointUrl );        

construct the address space

The server address space will be made of a Cities folder containing one folder for each city.

// declare some folders
var cityFolder = server.engine.addFolder("ObjectsFolder",{ browseName: "Cities"});
function create_CityNode(city_name) {
    // declare the city node
    var cityNode = server.engine.addFolder(cityFolder,{ browseName: city_name });
    _"construct city weather variables"
cities.forEach(function(city) {
_"extracting a DataValue"

extracting a DataValue

Let’s write a helper function (extract_value) to extract a city weather variable as DataValue. Since the city weather data are read asynchronously at a very low rate, it is possible that the data doesn’t exist yet when the client will send its request. We have to be careful to handle this case appropriately. In the absence of city data, I have chose to send a BadUncertainInitalValue status code.

function extract_value(city_name,property) {
    var city = city_data_map[city_name];
    if (!city) {
        return opcua.StatusCodes.BadDataUnavailable
    var value = city[property];
    return new opcua.Variant({dataType: opcua.DataType.Double, value: value });

construct city weather variables

Each city node exposes 3 read-only variables that can be instantiated this way:

    componentOf: cityNode,
    browseName: "Temperature",
    dataType: "Double",
    value: {  get: function () { return extract_value(city_name,"temperature"); } }
    componentOf: cityNode,
    browseName: "Humidity",
    dataType: "Double",
    value: {  get: function () { return extract_value(city_name,"humidity"); } }
    componentOf: cityNode,
    browseName: "Pressure",
    dataType: "Double",
    value: {  get: function () { return extract_value(city_name,"pressure"); } }

testing the server

It is now time to start the server for testing.

node weather.js

Putting everything together, the weather.js script looks like this