In short

To determine the radio coverage, GPS data is sent via TheThingsNetwork, processed with Node-Red and stored in an InfluxDB. The data is read from the database via JS and displayed graphically on a map.

The detailed version

This blog post will be the first of a series related to TheThingsNetwork. For some months now I have been actively involved in the development of the TheThingsNetwork in the city of Cologne. The TheThingsNetwork initiative aims to build an open and free wireless infrastructure for the Internet of Things. The community provides the gateways for use by everyone and the central infrastructure is provided by various data centers worldwide, for Europe by the Netherlands.

When building a radio infrastructure, the most important information is the range and coverage of the city in the area. The TTN Mapper from JP Meijers works very well for this application. If I had a smartphone that would be suitable for this app, this post would probably never have existed.
For my own purposes, however, I wanted to build a webpage that I could adapt flexibly to my own needs. In addition, it was a special attraction for me to familiarize myself with this topic and to master the task.

The Node

The first task was to build a TTN node equipped with a GPS module that sends the GPS data via the TTN. The construction of this node was similar to the node described by Bjoern in his blog. A GPS module is serially connected to an Arduino Pro Mini and an RFM95W is used to send the data. The circuits are powered by a lithium-ion battery and the voltage is stabilized by a low quiescent current LDO and a low dropout voltage type HX7333.  Even if the structure doesn’t look very professional, it still fulfils its function.

The antenna used is an old GSM 900MHz window antenna. The only change to Bjoern’s software is that in my case OTAA activation is used.
The software checks the validity of the GPS data and sends latitude and longitude data encoded in 6 bytes. The relevant lines here:

void get_coords () {
  bool newData = false;
  unsigned long chars;
  unsigned short sentences, failed;
  float flat, flon;
  unsigned long age;

  // For one second we parse GPS data and report some key values
  for (unsigned long start = millis(); millis() - start < 1000;) {
    while (SoftS.available()) {
      char c = SoftS.read();
      Serial.write(c); // uncomment this line if you want to see the GPS data flowing
      if (gps.encode(c)) { // Did a new valid sentence come in?
        newData = true;
      }
    }
  }

  if ( newData ) {
    gps.f_get_position(&flat, &flon, &age);
    flat = (flat == TinyGPS::GPS_INVALID_F_ANGLE ) ? 0.0 : flat;
    flon = (flon == TinyGPS::GPS_INVALID_F_ANGLE ) ? 0.0 : flon;
  }

  gps.stats(&chars, &sentences, &failed);

  int32_t lat = flat * 10000;
  int32_t lon = flon * 10000;

  // Pad 2 int32_t to 6 8uint_t, big endian (24 bit each, having 11 meter precision)
  coords[0] = lat;
  coords[1] = lat >> 8;
  coords[2] = lat >> 16;

  coords[3] = lon;
  coords[4] = lon >> 8;
  coords[5] = lon >> 16;
  }
} ... 

The complete program code can be found in my git. Define an application and device in the TheThingsNetwork console and fill in the keys in the program code.
To decode the data in the TTN cloud define this function in the TTN Application Console.

function Decoder(b, port) {
 var lat = (b[0] | b[1]<<8 | b[2]<<16 | (b[2] & 0x80 ? 0xFF<<24 : 0)) / 10000;
 var lng = (b[3] | b[4]<<8 | b[5]<<16 | (b[5] & 0x80 ? 0xFF<<24 : 0)) / 10000;
return {
 location: {
 lat: lat,
 lng: lng
 },
 love: "TTN payload functions"
 };
}

Data Processing

Node-Red handles the further processing of the data from the TTN cloud. For the first attempts an installation on a RaspberryPi is sufficient or you can use an online version on https://fred.sensetecnic.com/

The gps-logger node is connected to the application in the TTN cloud and receives json fomatted data. The function node decodes the data and prepares the data for the influxdb. The influxdb node finally writes the data to the influxdb. For each gateway receiving the data a record is written in the database. Later when visualizing the data on the map only the record with the highest rssi will be drawn to the map. The record will keep data for the gps location of the node, the gps location and ID of the receiving gateway, rssi, snr, the node’s counter of the record and the device ID of the node. Also the distance between the node and the gateway is calculated and written to the database but not used so far. The structure of the record is defined in the multiple-readings function node.

var lgid = msg.metadata.gateways.length;
var array_aussen = [];
var array_innen = [];
var last_time;
for (i = 0; i < lgid; i++) {
 array_innen = [{
 counter: msg.counter,
 rssi: msg.metadata.gateways[i].rssi,
 snr: msg.metadata.gateways[i].snr,
 lat_gw: msg.metadata.gateways[i].latitude,
 lon_gw: msg.metadata.gateways[i].longitude,
 no_gw: lgid,
 lat_sense: msg.payload.location.lat,
 lon_sense: msg.payload.location.lng,
 dist2gw: distance(msg.metadata.gateways[i].latitude,msg.metadata.gateways[i].longitude,msg.payload.location.lat,msg.payload.location.lng),
 time: new Date(msg.metadata.gateways[i].time).getTime() * 1000000 || 0
 },
 {
 dev_id: msg.dev_id,
 gtw_id: msg.metadata.gateways[i].gtw_id
 }
 ];
 if (array_innen[0].time === last_time) {
 array_innen[0].time += 1000;
 }
 array_aussen.push(array_innen);
}
var msg1 = {};
msg1.payload = array_aussen;
var msg2 = {};
msg2.payload = array_innen[0].time;
return [msg1, msg2];
function distance(gw_lat, gw_lon,sense_lat, sense_lon){
 if(sense_lat<1 || sense_lon < 1){
 return -1;
 }
 else {
 return Math.round(Math.sqrt(Math.pow(70.12*(gw_lon-sense_lon),2) + Math.pow(111.3*(gw_lat-sense_lat),2))*1000);
 }}

The Website

The far more difficult part of the task was the creation of the website. My Javascript knowledge was rather rudimentary in the beginning and I had to learn a lot while programming the website. For those who want to deal with this, the source code is located on the git. The website performs the following functions: The data is fetched by ajax from the influxdb, first the data of the receiving gateways, then the gps data of the node. The data is displayed as colored squares with the JS extension LeafletJS on an Openstreetmap. If you move the mouse over the squares, detailed data such as rssi, snr, number of receiving gateways and ID of the gateway with the highest rssi are displayed. The coding of color of the squares for rssi is the same as in the ttnmapper. If you hover over a gateway, the gateway ID is displayed.

The time period can be limited by passing HTTP GET parameters. In this way, you can also select whether the map should be colored or black and white. To overpass the parameters a simple http form is provided. A separately recorded gpx track can be placed under the data via a control element on the page. In this way, it is possible to illustrate very clearly where there is no radio coverage. The following video shows the possibilities that the website offers perhaps best.

VIDEO – coming soon

last but not least

The workflow described here may not be suitable for processing a lot of mapping data. I provide the members of the TTN community in Cologne with access to this server, so the amount of data is limited. The workflow is very simple and you don’t need to have a very high programming knowledge to understand it. Maybe this will be also interesting for other communities. You are welcome to try out the website, here in direct access to all available timeseries data or via the filter page with input option for the time span and selection of the optional b/w representation.