Four Tiers of IoT Hardware: What to Choose and When

So, you want to build an IoT device. Maybe you have a specific idea in mind - a smart sensor for your home, a machine monitoring system for your workshop, or a device that sends data to the cloud. Whatever your goal is, one of the first and most important questions you'll face is: what base hardware should I use?

In this article, we’ll break down the IoT device landscape into four practical tiers. Each tier strikes a different balance between power, flexibility, and complexity. For each one, we’ll highlight the most popular device in that category.

1. Arduino UNO + Comm Modules: The Classic Starting Point

The Arduino UNO remains a timeless entry point for hardware projects. Based on the ATmega328P, it runs at 16 MHz with 2 KB of RAM - enough for basic control tasks like reading sensors, toggling relays, or blinking LEDs. While it lacks built-in wireless connectivity, you can add Wi-Fi or Bluetooth via modules such as ESP-01, HC-05, or LoRa transceivers.

Upside: One major advantage is compatibility - nearly every Arduino library and sensor breakout board supports the UNO natively. It also includes internal EEPROM, which is useful for storing configuration data without extra components.

Downside: Wireless modules need to be connected and configured separately, which adds wiring complexity and consumes serial ports or GPIOs. Also, performance is limited - the UNO can’t handle modern encryption, multitasking, or advanced protocols.

Still, for educational purposes or simple connected devices, it’s a great low-cost and low-barrier way to start building.

2. ESP32: The DIY Champion

The ESP32 is a widely-used microcontroller with built-in Wi-Fi and BLE, making it ideal for wireless IoT projects. BLE can help with provisioning before connecting to Wi-Fi, simplifying setup for end users. It is possible to write software in Arduino IDE which really speeds up development time.

Upside: It’s affordable, runs at up to 240 MHz, and has enough RAM and flash to handle common tasks like sending sensor data or running simple logic. It's also great for prototyping - thanks to its simple pinout and compact size, you can hand-solder it or design a basic custom PCB for enclosures or signal integrity.

Downside: It lacks some security features - for example, it doesn't support WPA2-Enterprise Wi-Fi - which limits its use in corporate or industrial networks.

3. STM32 + Zephyr: Industrial Flexibility

Moving up to STM32 microcontrollers with the Zephyr RTOS brings more power, control, and professional-grade features - but also higher cost.

Downside: Not only are the chips themselves typically more expensive than ESP32, but development time can increase as well. Unlike the Arduino-based workflow, working with Zephyr requires proper toolchains, device trees, and deeper embedded knowledge.

Upside: In return, you get a mature RTOS, real-time performance, better power management, and a wide range of peripherals. Zephyr also offers modern development practices like device abstraction, secure boot, and native support for industrial protocols and connectivity beyond Wi-Fi, such as CAN, LoRa, and BLE.

4. Raspberry Pi: Full Power at the Edge

Raspberry Pi and similar Linux-based boards offer unmatched flexibility. With full operating systems like Raspberry Pi OS or Ubuntu, you get access to advanced networking stacks, containerization (e.g., Docker), and virtually all modern security protocols - including WPA2-Enterprise, TLS, and secure VPNs.

Upside: This makes it perfect for edge computing, where data processing happens locally instead of in the cloud - saving bandwidth and reducing cloud service costs. You can run Python scripts, machine learning models, or even lightweight web servers directly on the device.

Downside: This power comes at a cost. The board itself is more expensive, and designing a custom PCB with a Pi-class processor is complex and expensive. The chip packaging (e.g., BGA) makes DIY soldering or home assembly nearly impossible without specialized tools and manufacturing.

Costs

When selecting the right IoT platform, cost is a crucial factor, especially when scaling up your project. The choice between different tiers can significantly impact the overall expense, not just in terms of initial development, but also long-term maintenance and scaling. Here’s a quick comparison of costs across the tiers:

Software development

When choosing an IoT platform, it's important to consider not only the cost of the hardware but also the time required to write the code. The development process varies significantly across the tiers.

External Hardware

When it comes to external sensors, almost any sensor can be connected to any of these platforms, as they all support common communication protocols like SPI and I2C. However, if you need to interface with sensors via USB, you'll need to use Tier as lower-tier platforms typically don’t have native USB host support.

Summary

For simple projects like a Wi-Fi temperature sensor, Tier 1 (Arduino UNO) and Tier 2 (ESP32) is ideal due to its low cost and fast prototyping. For more complex tasks requiring real-time performance or advanced security, consider Tier 3 (STM32 + Zephyr) or Tier 4 (Raspberry Pi).

However, for prototypes, speed is crucial. Start with Tier 1 or Tier 2 to quickly validate the core idea, then transition to more powerful platforms as needed to expand functionality and refine the design. This allows you to move quickly, test your idea, and gather feedback.

‍