LoRa / LoRaWAN Integration with ASI Biont: AI-Driven IoT Monitoring Without the Internet

Why LoRa and LoRaWAN Need an AI Agent

LoRa (Long Range) and LoRaWAN are the backbone of low-power wide-area networks (LPWAN) for IoT. A single LoRa module, like the Ebyte E32-900T20D, can transmit sensor data over 10–15 km in rural areas or 2–5 km in urban environments using license-free sub-GHz bands (868 MHz in Europe, 915 MHz in the US). LoRaWAN adds a standardized protocol stack with a network server, application server, and over-the-air activation (OTAA). These systems are ideal for remote monitoring where cellular coverage is weak or WiFi is unavailable—think agricultural fields, pipeline corridors, or off-grid weather stations.

But raw data from a temperature/humidity sensor or a pulse counter is just numbers. Without an intelligent agent to analyze trends, detect anomalies, and trigger actions, you’re left staring at a dashboard. ASI Biont’s AI agent changes that: it connects directly to your LoRa gateway or serial LoRa module via COM port (RS-232/RS-485), parses the data, and automates responses—no cloud dependency, no custom server code.

How ASI Biont Connects to LoRa Hardware

ASI Biont supports two primary integration paths for LoRa devices:

1. LoRa Module via COM Port (Hardware Bridge)

For a direct LoRa module like the Ebyte E32-900T20D or a Dragino LoRa/GPS HAT, you connect it to a PC or single-board computer (Raspberry Pi, Orange Pi) over UART. The user runs the bridge.py application on that machine, specifying the COM port (e.g., /dev/ttyS0 on Linux or COM4 on Windows) and baud rate (typically 9600 or 115200). bridge.py establishes a long-polling connection to ASI Biont’s cloud server. When the AI agent needs to send a command (e.g., AT+SETCONFIG=... or a raw payload), it uses the industrial_command() tool with protocol='serial://' and command: 'write' (or 'read'). The bridge reads from the module and returns responses back to the chat.

2. LoRaWAN Gateway via MQTT or Modbus TCP

If you have a LoRaWAN gateway (e.g., a RAK7249 or a Kerlink iStation) that forwards uplinks to an MQTT broker or exposes a Modbus TCP interface, ASI Biont connects via execute_python using paho-mqtt (MQTT) or pymodbus (Modbus TCP). The AI writes a Python script that subscribes to the gateway’s uplink topic, decodes the payload (typically in base64 or JSON), and processes it. For downlinks (commands), the script publishes to the gateway’s downlink topic.

Real-World Use Case: Remote Soil Moisture Monitoring and Irrigation Control

Scenario: A farm deploys 10 soil moisture sensors (e.g., Dragino LSE01) connected to a LoRaWAN gateway. Each sensor sends a payload with moisture percentage and temperature every 30 minutes. The goal: automatically turn on a solenoid valve (via a LoRaWAN actuator like the Dragino LAQ4) when soil moisture drops below 30%, and send a Telegram alert if the temperature exceeds 40°C.

Step 1: User Describes the Setup in Chat

The user writes to the ASI Biont agent:

“Connect to my LoRaWAN gateway at mqtt://192.168.1.100:1883, topic uplink sensor/+/data (payload is JSON with fields moisture, temp, sensor_id). Downlink topic is actuator/+/cmd. Subscribe to all sensors, log data to a PostgreSQL database, and if any sensor’s moisture < 30, publish {"valve":"on"} to topic actuator/${sensor_id}/cmd. If temp > 40, send Telegram alert to chat ID 123456789.”

Step 2: AI Generates and Runs the Python Script

ASI Biont’s sandbox (with paho-mqtt, psycopg2, and requests installed) writes a script like this:

import paho.mqtt.client as mqtt
import psycopg2
import json
import requests

# Database connection
conn = psycopg2.connect(
    host="localhost",
    dbname="farm_iot",
    user="admin",
    password="secret"
)
cur = conn.cursor()

# Telegram bot token and chat ID
TELEGRAM_TOKEN = "YOUR_BOT_TOKEN"
CHAT_ID = "123456789"

def on_message(client, userdata, msg):
    topic = msg.topic
    payload = json.loads(msg.payload.decode())
    sensor_id = payload.get("sensor_id")
    moisture = payload.get("moisture")
    temp = payload.get("temp")

    # Log to database
    cur.execute(
        "INSERT INTO sensor_data (sensor_id, moisture, temp, recorded_at) VALUES (%s, %s, %s, NOW())",
        (sensor_id, moisture, temp)
    )
    conn.commit()

    # Check thresholds
    if moisture < 30:
        actuator_topic = f"actuator/{sensor_id}/cmd"
        client.publish(actuator_topic, json.dumps({"valve": "on"}))
        print(f"Opened valve for sensor {sensor_id}")

    if temp > 40:
        message = f"ALERT: Sensor {sensor_id} temperature {temp}°C exceeds 40°C!"
        requests.post(
            f"https://api.telegram.org/bot{TELEGRAM_TOKEN}/sendMessage",
            json={"chat_id": CHAT_ID, "text": message}
        )

client = mqtt.Client()
client.on_message = on_message
client.connect("192.168.1.100", 1883, 60)
client.subscribe("sensor/+/data")
client.loop_forever()

Note: This script runs in the cloud sandbox with a 30-second timeout for single executions. For long-running MQTT listeners, ASI Biont uses the industrial_command() tool with MQTT protocol, which keeps the connection alive and routes messages through the chat interface.

Step 3: AI Monitors and Takes Action

The AI agent now continuously receives sensor data through the MQTT subscription. When anomalies occur, it sends commands back through the same channel. The farmer can ask: “What was the average moisture last night?” and the AI queries the database and responds in chat.

Wiring Diagram for Direct LoRa Module (Ebyte E32-900T20D)

On the Raspberry Pi, you enable the serial interface (sudo raspi-config → Interface Options → Serial Port → “login shell over serial” → No, “serial port hardware” → Yes). Then run bridge.py --token=YOUR_TOKEN --ports=/dev/ttyS0 --default-baud=9600.

Why This Beats Traditional Dashboards

• No code to write: The AI generates the entire integration script from a natural-language description. No need to learn MQTT, Modbus, or serial protocols.
• No server-side setup: The bridge handles COM port access locally; the sandbox runs in the cloud. You only need a PC or SBC connected to the LoRa module.
• Real-time anomaly detection: The AI can be configured to alert you via Telegram, email, or even call an API to trigger a physical siren.
• Offline-first: For remote areas without internet, the local bridge caches commands and syncs when connectivity returns.

Conclusion

LoRa and LoRaWAN are perfect for long-range, low-power IoT—but they’re only half the story. Pairing them with ASI Biont’s AI agent turns raw sensor data into actionable intelligence. Whether you’re monitoring soil moisture, tracking livestock, or securing a pipeline, you can connect any LoRa module via COM port, MQTT, or Modbus TCP in minutes. Describe your hardware in the chat, and let the AI handle the rest.

Try it now at asibiont.com – no dashboard, no buttons. Just a conversation with an AI that connects to your devices.