ArduNeun

Using Neun in Arduino and ESP32 projects: benchmarks and basic C++ scripts for simulating neuron models on microcontrollers, plus Python utilities to visualize and analyze the results.

Developed by the GNB-UAM.

Features

• Minimal examples of neuron simulation with Neun on microcontrollers.
• Real-time and periodic real-time versions.
• Fixed-point and float variants.
• Performance benchmarks for ESP32-S3 and ESP8266, including tests with different network configurations (single neuron, one connection, full network, and empty).
• Python scripts for live plotting and benchmark comparison.
• Build and upload with PlatformIO.

Requirements

• PlatformIO (CLI or VS Code extension).
• Neun, which must be available at include/lib/neun.
• Python 3 (only for the plotting scripts).
• A supported board:
  • ESP32-S3 (configured for waveshare_esp32_s3_zero)
  • ESP8266 (configured for d1_mini_pro)

Installation

# 1. Clone the repository
git clone https://github.com/GNB-UAM/ArduNeun.git
cd ArduNeun

# 2. Add Neun at include/lib/neun
git clone https://github.com/GNB-UAM/Neun.git include/lib/neun

If Neun has its own directory layout, adjust the path so the project's #include directives can find it.

Repository structure

ArduNeun/
├── benchmark/             # Benchmark code and data
├── include/               # Headers (lib/neun goes here)
├── lib/                   # Local libraries
├── src/                   # Main programs (one per PlatformIO environment)
├── test/                  # Tests
├── platformio.ini         # Build environments
├── flash_benchmark.sh     # Runs the flash benchmark
├── fast_plot.py           # Plotting utilities
├── mini_plot.py
├── plot_realtime.py       # Real-time plot
├── plot_realtime_periodic.py
├── plot_esp_benchmark.py  # Benchmark plots
└── plot_flash_benchmark.py

PlatformIO environments

Each environment in platformio.ini builds a different program from src/.

Environment	Board	Source
esp32-s3-basic	ESP32-S3	basic.cpp
esp32-s3	ESP32-S3	main.cpp
esp32-s3-float	ESP32-S3	main-float.cpp
esp32-s3-realtime	ESP32-S3	main_realtime.cpp
esp32-s3-realtime-periodic	ESP32-S3	real-time-periodic.cpp
esp32-s3-flash-single	ESP32-S3	main_flash_single_neuron.cpp
esp32-s3-flash-one-connection	ESP32-S3	main_flash_one_connection.cpp
esp32-s3-flash-full	ESP32-S3	main_flash_full.cpp
esp32-s3-flash-empty	ESP32-S3	main_flash_empty.cpp
esp8266-basic	ESP8266	basic.cpp
esp8266_flash_single	ESP8266	main_flash_single_neuron.cpp
esp8266_flash-one-connection	ESP8266	main_flash_one_connection.cpp
esp8266_flash-full	ESP8266	main_flash_full.cpp
esp8266_flash-empty	ESP8266	main_flash_empty.cpp

Configuration notes:

• ESP32-S3: C++20 (-std=gnu++20) using the pioarduino platform.
• ESP8266: C++2a (-std=gnu++2a) using the development branch of the ESP8266 Arduino framework.
• Serial monitor at 115200 baud.

Usage

Build and upload a program

# Build
pio run -e esp32-s3-basic

# Build and upload to the board
pio run -e esp32-s3-basic -t upload

# Open the serial monitor
pio device monitor -b 115200

Replace esp32-s3-basic with any environment from the table above.

Benchmarks

The *-flash-* environments let you measure behavior under different network configurations. To run them you can use:

./flash_benchmark.sh

Then generate the plots with:

python plot_flash_benchmark.py
python plot_esp_benchmark.py

Real-time visualization

With a board flashed with esp32-s3-realtime (or esp32-s3-realtime-periodic) and connected over USB:

python plot_realtime.py            # real-time mode
python plot_realtime_periodic.py   # periodic real-time mode

You may need to adjust the serial port inside the scripts (for example /dev/ttyACM0 on Linux or COMx on Windows).

Contributing

Contributions are welcome. Open an issue to discuss major changes, or send a pull request directly.

License

This project is distributed under the GPL-3.0 license.