Platform Integrations

Overview

NexAI Platform integrations feature was designed for two primary use cases / deployment options:

• Connect existing NB IoT, LoRaWAN, SigFox and other devices with specific payload formats directly to NexAI platform.
• Stream data from devices connected to existing IoT Platforms to enable real-time interactive dashboards and efficient data processing.

Both use cases have few things in common. There is a server-side component in the deployment topology that prevents direct access to device and provides set of APIs to interact with the device in the field instead. The payload format of the device is not well-defined. Often two devices that have similar sensors have different payload formats depending on a vendor or even software version.

The job of NexAI Integration is to provide secure and reliable API bridge between core platform features (telemetry collection, attributes and RPC calls) and specific third-party platform APIs.

How it works?

At the moment NexAI supports various integration protocols. Most popular are HTTP, MQTT and OPC-UA. Platform also support integration with specific LoRaWAN Network servers, Sigfox backend, various NB IoT devices using raw UDP and TCP integrations. AWS IoT, IBM Watson and Azure Event Hub allows to subscribe to the data feed from devices via MQTT or AMQP.

The list of platform integrations is constantly growing, however, the general integration concepts are the same and explained below.

Once message arrives from External Platform to NexAI it passes validation according to platform specific payload format and security rules. Once message is validated NexAI Integration invokes assigned Uplink Data Converter to extract sub-set of meaningful information out of the incoming message. The message is basically transformed from device and platform specific payload to the format that NexAI uses.

Since TB PE v2.0, Rule Engine is also able to push Downlink messages to the integrations. The example of such message may be:

• notification about shared attribute (configuration) update;
• notification about oneway RPC call to trigger some action on the device;
• any custom message from the rule engine.

The most common use cases are:

• changing data upload frequency based on shared attribute value change
• triggering firmware update procedure based on shared attribute value change
• changing device state based on rpc call;

Once message is pushed by the rule engine, NexAI invokes assigned Downlink Data Converter and transforms the rule engine message to the specific data format that is used by the Integration.

Data Converters

Data Converters is a part of the Platform Integrations feature. There are Uplink and Downlink data converters.

Uplink Data Converter

The main function of Uplink Data Converter is to parse payload of the incoming message and transform it to format that NexAI uses.

To create an Uplink Converter go to Data Converters section and Click Add new data converter —> Create new converter. Enter converter name, select its type, specify a script to parse and transform data. Optional you can turn the Debug mode. Click “Add” to create converter.

Uplink Converter is basically a user defined function with the following signature:

function Decoder(payload, metadata);

Payload

Payload is one of the following content types: JSON, TEXT, Binary(Base64) and is specific to your Integration type.

Default Uplink Converter is dummy, but contains few helper functions to transform incoming payload:

function decodeToString(payload) {
   return String.fromCharCode.apply(String, payload);
}

function decodeToJson(payload) {
   // covert payload to string.
   var str = decodeToString(payload);
   // parse string to JSON
   return JSON.parse(str);
}

There are also btoa and atob functions available to decode Binary(Base64) payload.

Metadata

Metadata is a key-value map with some integration specific fields. You can configure additional metadata for each integration in the integration details. For example, you can put device type as an additional Integration metadata parameter and use it to automatically assign corresponding device type to new devices.

Converter output

Converter output should be a valid JSON document with the following structure:

{
    "deviceName": "Device A",
    "deviceType": "thermostat",
    "customerName": "Company Name",
    "groupName": "Thermostats",
    "attributes": {
        "model": "Model A",
        "serialNumber": "SN-111",
        "integrationName": "Test integration"
    },
    "telemetry": {
        "temperature": 42,
        "humidity": 80
    }
}

NOTE:
The only mandatory parameters in the output JSON are deviceName and deviceType. NexAI also supports assetName and assetType instead of deviceName and deviceType.

NOTE:
NexAI also support optional siteName and groupName. Those parameters will cause NexAI to automatically create site and/or entity group and assign those entities to the site and/or group.

Converter may also output array of device values and/or contain timestamps in the telemetry values. For example:

[
    {
        "deviceName":"SN-111",
        "deviceType":"thermostat",
        "attributes":{
            "model":"Model A"
        },
        "telemetry":[
            {
                "ts":1527863043000,
                "values":{
                    "battery":3.99,
                    "temperature":27.05
                }
            },
            {
                "ts":1527863044000,
                "values":{
                    "battery":3.98,
                    "temperature":27.06
                }
            }
        ]
    },
    {
        "assetName":"OF-123",
        "assetType":"office",
        "attributes":{
            "model":"Model A"
        },
        "telemetry":{
            "ts":1527863041000,
            "values":{
                "battery":3.99,
                "temperature":27.05
            }
        }
    }
]

Update only keys field

To avoid constant updates for telemetry attributes or keys, you can use the “Update only keys list” field.
Any keys provided in this field that exist in the telemetry or attribute arrays in the message after conversion will not be updated if the values associated with those keys have not changed from their previous values.

Please note that in a cluster setup, values, associated with keys specified in the “Update only keys list” field may be updated more than once if a message is received by a different integration executor nodes.

The same behavior is expected if the converter configuration has been updated.

Downlink Data Converter

The main function of Downlink Data Converter is to transform the incoming rule engine message and its metadata to the format that is used by corresponding Integration.

To create a Downlink Converter go to Data Converters section and Click Add new data converter —> Create new converter. Enter converter name, select its type, specify a script to parse and transform data. Optional you can turn the Debug mode. Click “Add” to create converter.

Downlink Converter is basically a user defined function with the following signature:

function Encoder(msg, metadata, msgType, integrationMetadata);

Where

• msg – JSON with rule engine msg
• metadata – list of key-value pairs with additional data about the message (produced by the rule engine)
• msgType – Rule Engine message type. See predefined message types for more details.
• integrationMetadata – key-value map with some integration specific fields. You can configure additional metadata for each integration in the integration details.

Converter output

Converter output should be a valid JSON document with the following structure:

{
    "contentType": "JSON",
    "data": "{\"tempFreq\":60,\"firmwareVersion\":\"1.2.3\"}",
    "metadata": {
        "topic": "temp-sensor/sensorA/upload"
    }
}

Where

• contentType – JSON, TEXT or BINARY (Base64 string) and is specific to your Integration type.
• data – data string according to the content type
• metadata – list of key-value pairs with additional data about the message. For example, topic to use for MQTT integration, etc.

Synchronous vs Asynchronous Downlinks

Most of the integrations are able to process downlink messages to devices asynchronously. For example, each message pushed by the rule engine to MQTT based integration is immediately pushed to the corresponding external MQTT broker.

However, some integrations, like SigFox or generic HTTP integration are not able to push message asynchroniously. These integrations, due to the nature of underlying HTTP protocol, are only able to push downlink information synchronously in reply to uplink message request. In this case, the last downlink message originated by rule engine will be stored in the queue until the new uplink message arrives for particular device.

Debug mode

This feature allows to persis:

• incoming messages from thirdparty system;
• metadata values;
• the results of data converter;
• results of the payload processing.

It enables rapid development of converters and configuration of integrations. This feature allows to validate your configuration setup and should be used only for debug purposes, since it dramatically impacts performance.