ttn-osem-integration

Microservice for openSenseMap that provides a direct integration with TheThingsNetwork to allow straightforward measurement upload from LoRa-WAN devices.

It decodes measurements from an uplink payload from the TTN HTTP Integrations API for a configured senseBox, and adds the decoded measurements to the database.

There are multiple decoding options provided via profiles, which may be easily extended to support other sensor configurations or value transformations.

configuring a box

To associate a device on TTN with a box on the openSenseMap, there is some configuration required on the openSenseMap. The box has to contain a field box.integrations.ttn with the following structure:

ttn: {
  // the app_id & dev_id you recieved when registering on TTN
  app_id: 'abcd',
  dev_id: '1234',
  // decode the messages according to this profile format, see below
  profile: 'lora-serialization',
  // optional. some profiles require additional configuration
  decodeOptions: [],
  // optional. if specified, only messages recieved on this LoRa-port are stored
  port: 3,
}

decoding profiles

sensebox/home

Decodes messages of all sensors of the senseBox:home. Takes registered sensors into account and decodes payload accordingly. Therefore, the senseBox:home lora sketch should not be changed except TTN IDs. The correct sensorIds are matched via their titles. Decoding fits the senseBox:home lora sketch.

lora-serialization

Allows decoding of messages that were encoded with the lora-serialization library. The decoders temperature, humidity, uint8, uint16, unixtime and latLng are supported. Each encoded value is matched to a sensor via it's _id, sensorType, unit, or title properties. There may be one or more property defined for each value via sensor_id, sensor_title, sensor_type, sensor_unit. If one property matches a sensor, the other properties are discarded.

The following example config allows decoding of measurements of 3 sensors:

"ttn": {
  "profile": "lora-serialization",
  "decodeOptions": [
    { "decoder": "temperature", "sensor_unit": "°C" },
    { "decoder": "humidity", "sensor_id": "588876b67dd004f79259bd8b" },
    { "decoder": "uint16", "sensor_type": "TSL45315", "sensor_title": "Beleuchtungsstärke" }
  ]
}

special decoders unixtime & latLng

These decoders do not generate a measurement for a sensor of the box, but will be used as timestamp or location for the measurements. The unixtime and latLng decoders must be defined before the measurements, and are applied to all following measurements, until the next latLng or unixtime decoder is specified. This means that it is possible to send measurements of several timestamps at once:

"ttn": {
  "profile": "lora-serialization",
  "decodeOptions": [
    // first measurement, will have time of transmission as timestamp
    { "decoder": "temperature", "sensor_unit": "°C" },
    // 2nd measurement for same sensor, will have custom timestamp
    { "decoder": "unixtime" }, // no sensor properties required for special decoders
    { "decoder": "temperature", "sensor_unit": "°C" },
    // 3rd measurement, another timestamp
    { "decoder": "unixtime" },
    { "decoder": "temperature", "sensor_unit": "°C" }
  ]
}

debug

Simple decoder, which decodes a given number of bytes to integer values. Requires a config like the following, where the measurements are applied to the sensors in the order of box.sensors.

ttn: {
  profile: 'lora-serialization',
  decodeOptions: [3, 1, 2] // specifies the number of bytes to consume for each measurement
}

json

It's also possible to add measurements which already have been decoded by a TTN payload function. The property payload_fields has to contain JSON in the format accepted by the openSenseMap-API. This is the case, if the TTN application has a Payload Function defined.

cayenne-lpp

Allows decoding of messages that were encoded with the Cayene LPP format. The decoders temperature, relative_humidity, barometric_pressure, luminosity, and analog_in are supported. Each encoded value is matched to a sensor via it's _id, sensorType, unit, or title properties. There may be one or more property defined for each value via sensor_id, sensor_title, sensor_type, sensor_unit. If one property matches a sensor, the other properties are discarded.

If GPS measurements are provided, they will automatically be attached to the measurement in the API. No need to add GPS to the decodeOptions. Mobile boxes will then update their location on the openSenseMap accordingly.

The following example config allows decoding of measurements of 3 sensors:

"ttn": {
  "profile": "cayenne-lpp",
  "decodeOptions": [
    { "decoder": "temperature", "channel": 1, "sensor_unit": "°C" },
    { "decoder": "relative_humidity", "channel": 1, "sensor_id": "588876b67dd004f79259bd8b" },
    { "decoder": "illuminance", "channel": 1, "sensor_type": "TSL45315", "sensor_title": "Beleuchtungsstärke" }
  ]
}

deployment

There is a Dockerfile, as well as an docker-compose.yml which includes a mongodb instance. If you want to run the application directly, you need to have the dependencies listed below installed. For configuration, see below. Once configured, run

yarn install
npm start

dependencies

• node.js >= 6.x
• yarn
• mongodb >= 3.x

configuration

Configuration is handled by node-config. See config/default.json.

{
  "port": 3000,
  "loglevel": "trace",
  "openSenseMap-API-models": {
    "db": {
      // See example config json of @sensebox/opensensemap-api-models
      "mongo_uri"
    }
  }
}

development

• JSDoc documentation can be found under ./docs/ or sensebox.github.io/ttn-osem-integration. To update it, run npm run docs.
• To run the test suite, either run npm run test while the application is running, or ./run_tests.sh (requires bash & docker).
• Please follow the existing code style. Double check by running npm run lint.

license

MIT, see LICENSE