Best Practices for MQTT Telemetry on Low-Power Microcontrollers

Introduction to MQTT in IoT

MQTT (Message Queuing Telemetry Transport) has become the de-facto standard protocol for IoT telemetry. Its lightweight design and publish-subscribe model make it perfect for resource-constrained microcontrollers. However, running MQTT inefficiently can quickly drain batteries and overload brokers. Implementing optimization patterns is essential for production-grade scale.

1. Optimize QoS Levels

MQTT offers three Quality of Service (QoS) tiers:

• QoS 0 (At most once): Minimal overhead. Messages are sent without confirmation. Use this for highly frequent, non-critical sensor updates (e.g. ambient temperature).
• QoS 1 (At least once): Demands confirmation handshake. Essential for status updates or alarms where receipt is mandatory.
• QoS 2 (Exactly once): Highest overhead (4-step handshake). Avoid QoS 2 on low-power devices as it keeps the radio active too long.

2. Keep-Alive Frame Optimization

To detect connection losses, MQTT clients send periodic ping requests (PINGREQ). While active devices benefit from short keep-alive intervals (e.g. 60 seconds), battery-powered nodes should increase this threshold to several hours, or connect, transmit telemetry in a single burst, and disconnect immediately before entering Deep Sleep.

3. Payload Compression

Avoid verbose JSON payloads. Instead of:

{ "device_identification_id": "ESP32-S3-009A", "temperature_celsius": 24.52, "humidity_relative": 58.12 }

Use compact keys or binary serialization formats like Protocol Buffers (Protobuf) or MessagePack:

{ "id": "009A", "t": 24.5, "h": 58.1 }

This small change reduces Wi-Fi radio transmission time and bytes written to network sockets significantly.

Working on an IoT network? Let's talk →