Raspberry Pi is the primary development platform for EdgeFlow. Our lightweight Go-based architecture delivers exceptional performance on resource-constrained devices, from Pi Zero to Pi 5.

Supported Models

EdgeFlow supports all Raspberry Pi models with varying capabilities based on hardware specifications.

Model	CPU	RAM	Build Profile	Capabilities
Pi 5	4× Cortex-A76 @ 2.4GHz	2-8 GB	`full`	AI/ML, local LLMs, high-frequency GPIO, NVMe storage
Pi 4	4× Cortex-A72 @ 1.5GHz	2-8 GB	`standard`	Full feature set, database-heavy workloads
Pi 3 B+	4× Cortex-A53 @ 1.4GHz	1 GB	`standard`	Standard flows, basic sensor monitoring
Pi Zero 2 W	4× Cortex-A53 @ 1GHz	512 MB	`minimal`	Basic flows, remote sensors, compact deployment
Pi Zero W	1× ARM11 @ 1GHz	512 MB	`minimal`	Very basic flows only, limited node count

📊 Build Profiles

EdgeFlow offers three build profiles optimized for different hardware:

minimal — 10MB binary, ~50MB RAM — Pi Zero, BeagleBone
standard — 20MB binary, ~200MB RAM — Pi 3/4, Orange Pi (default)
full — 35MB binary, ~400MB RAM — Pi 4/5, Jetson Nano

Quick Installation

The fastest way to get EdgeFlow running on your Raspberry Pi:

One-Line Install

curl -fsSL https://edgx.cloud/install.sh | bash

The install script automatically:

Detects your Pi model and architecture (arm64/armv7)
Downloads the appropriate binary from GitHub releases
Creates a dedicated edgeflow system user
Sets up directory structure (/opt/edgeflow/data, /opt/edgeflow/logs)
Installs and enables the systemd service
Configures security hardening (read-only filesystem, temp protection)

Manual Installation

For more control over the installation process, follow these steps:

Step 1: Enable Hardware Interfaces

Run the Raspberry Pi configuration tool to enable required interfaces:

raspi-config

sudo raspi-config

Navigate to Interface Options and enable:

I2C

Required for I2C sensors: BME280, BH1750, AHT20, SHT3x, ADS1015, LCD displays, RTC modules

SPI

Required for SPI devices: MCP3008 ADC, RC522 RFID, NRF24L01, LoRa modules, MCP2515 CAN

1-Wire

Required for DS18B20 temperature sensors (uses GPIO4 by default)

Serial Port

Required for UART communication: GPS modules (NEO-M8N), serial sensors, Modbus RTU

Step 2: Download EdgeFlow

Download the latest release for your architecture:

All Pi Models (arm64 / armv7 auto-detected)

curl -fsSL https://edgx.cloud/install.sh | bash

Step 3: Create System Service

Install EdgeFlow as a systemd service for automatic startup:

Service Installation

# Install the service
sudo edgeflow service install

# Enable auto-start on boot
sudo systemctl enable edgeflow

# Start the service
sudo systemctl start edgeflow

# Check status
sudo systemctl status edgeflow

Step 4: Configure Permissions

Add your user to the required groups for hardware access:

User Groups

# Add user to hardware groups
sudo usermod -a -G gpio,i2c,spi,dialout $USER

# Apply changes (logout/login or run)
newgrp gpio

Step 5: Verify Installation

Access the EdgeFlow web interface:

Access EdgeFlow

# EdgeFlow runs on port 8080 by default
# Open in browser: http://<your-pi-ip>:8080

# Find your Pi's IP address
hostname -I

Docker Installation

Run EdgeFlow in a Docker container for isolated deployments:

Docker Run

docker run -d \
  --name edgeflow \
  -p 8080:8080 \
  -v edgeflow_data:/opt/edgeflow/data \
  --device /dev/gpiomem \
  --device /dev/i2c-1 \
  edgeflow/edgeflow:latest

Docker Compose

version: '3.8'
services:
  edgeflow:
    image: edgeflow/edgeflow:latest
    container_name: edgeflow
    restart: unless-stopped
    ports:
      - "8080:8080"
    volumes:
      - edgeflow_data:/opt/edgeflow/data
      - edgeflow_logs:/opt/edgeflow/logs
    devices:
      - /dev/gpiomem:/dev/gpiomem
      - /dev/i2c-1:/dev/i2c-1
    # Uncomment for full hardware access
    # privileged: true

volumes:
  edgeflow_data:
  edgeflow_logs:

Hardware Support

EdgeFlow provides native support for 52+ hardware nodes across multiple categories:

Supported Sensors

Category	Protocol	Supported Devices
Temperature	1-Wire, I2C	DS18B20, DHT22, BME280, MAX31865 (PT100/PT1000)
Humidity	GPIO, I2C	DHT22, DHT11, BME280, AHT20, SHT3x, HDC1080
Pressure	I2C	BME280, BMP280, BMP180, MS5611
Light	I2C	BH1750, TSL2561, VEML7700, MAX44009
Motion	GPIO	PIR (HC-SR501), RCWL0516 (microwave)
Distance	GPIO	HC-SR04, JSN-SR04T (waterproof), VL53L0X (laser)
Air Quality	I2C	CCS811 (eCO2/TVOC), SGP30, BME680
ADC/DAC	I2C, SPI	ADS1015, ADS1115, MCP3008, MCP3208, PCF8591

Supported Actuators & Modules

Category	Protocol	Supported Devices
Motor Control	GPIO/PWM	L298N, L293D, TB6612, Servo motors, Stepper (A4988)
Relay Control	GPIO	1/2/4/8-channel relay modules, solid-state relays
Displays	I2C	LCD 16x2/20x4 (PCF8574), OLED SSD1306, SH1106
RTC	I2C	DS1307, DS3231, PCF8523
RFID	SPI	RC522 (MIFARE)
Wireless	SPI/GPIO	NRF24L01, LoRa SX1276/SX1278, RF433 transmitter/receiver
GPS	UART	NEO-M8N, NEO-6M, BN-880
CAN Bus	SPI	MCP2515

GPIO Pin Reference

EdgeFlow uses BCM (Broadcom) pin numbering. Here's a quick reference for common connections:

Power Pins

3.3V — Pins 1, 17
5V — Pins 2, 4
GND — Pins 6, 9, 14, 20, 25, 30, 34, 39

I2C Bus

SDA — GPIO2 (Pin 3)
SCL — GPIO3 (Pin 5)

SPI Bus

MOSI — GPIO10 (Pin 19)
MISO — GPIO9 (Pin 21)
SCLK — GPIO11 (Pin 23)
CE0 — GPIO8 (Pin 24)
CE1 — GPIO7 (Pin 26)

UART

TXD — GPIO14 (Pin 8)
RXD — GPIO15 (Pin 10)

1-Wire (Default)

DATA — GPIO4 (Pin 7)

PWM Hardware

PWM0 — GPIO12, GPIO18
PWM1 — GPIO13, GPIO19

Configuration

EdgeFlow configuration is stored in /opt/edgeflow/config.yaml:

config.yaml

server:
  host: "0.0.0.0"
  port: 8080

database:
  driver: sqlite
  path: "./data/edgeflow.db"

hardware:
  gpio_enabled: true      # Enable GPIO access
  i2c_enabled: true       # Enable I2C bus
  spi_enabled: false      # Enable SPI bus
  i2c_bus: 1              # Default I2C bus (1 for Pi 2+)

engine:
  max_concurrent_executions: 100
  execution_timeout: 5m
  retry_attempts: 3

scheduler:
  timezone: "UTC"         # Your timezone

logging:
  level: "info"           # debug, info, warn, error
  format: "json"

metrics:
  enabled: false
  prometheus_port: 9090

Performance Optimization

Optimize your Raspberry Pi for better EdgeFlow performance:

Reduce GPU Memory

For headless setups, minimize GPU memory allocation:

/boot/config.txt

# Reduce GPU memory to minimum (16MB for Pi 4/5, 64MB for Pi Zero)
gpu_mem=16

# For Pi 5 with 8GB RAM running AI workloads
gpu_mem=128

Disable Unused Services

Free up resources by disabling services you don't need:

Disable Services

# Disable Bluetooth (if not using BLE nodes)
sudo systemctl disable bluetooth
sudo systemctl disable hciuart

# Disable Avahi/mDNS (if not using .local discovery)
sudo systemctl disable avahi-daemon

# Disable triggerhappy (keyboard shortcuts daemon)
sudo systemctl disable triggerhappy

# Disable ModemManager (if not using cellular modems)
sudo systemctl disable ModemManager

Overclock Pi 5 (Optional)

For maximum performance on Pi 5 with adequate cooling:

/boot/config.txt (Pi 5 only)

# Overclock CPU to 3GHz (requires active cooling)
arm_freq=3000

# Overclock GPU
gpu_freq=1000

# Note: Monitor temperature with: vcgencmd measure_temp

Use SSD/NVMe Storage (Pi 5)

For database-heavy workloads, use NVMe storage:

NVMe Setup

# Install NVMe SSD via PCIe HAT
# Update config to use NVMe for data directory

database:
  driver: sqlite
  path: "/mnt/nvme/edgeflow/edgeflow.db"

Raspberry Pi 5 Specifics

Pi 5 offers significant performance improvements and new capabilities:

Feature	Pi 4	Pi 5	Improvement
CPU Performance	1.5GHz Cortex-A72	2.4GHz Cortex-A76	2-3× faster
GPIO Speed	Standard	Direct chip interface	~10× faster
I2C Speed	400 kHz	1 MHz	2.5× faster
Memory Bandwidth	~4 GB/s	~5.6 GB/s	40% faster
Storage	SD/USB only	NVMe via PCIe	10× faster I/O
Local AI/LLM	Limited	7B models (8GB)	✅ New capability

🔧 Pi 5 GPIO Note

Raspberry Pi 5 uses a different GPIO chip (gpiochip4) compared to Pi 4 (gpiochip0). EdgeFlow automatically detects your Pi model and configures the correct GPIO interface. If you experience GPIO issues on Pi 5, ensure you're running the latest EdgeFlow version.

Optimal Pi 5 Use Cases

Local AI/LLM — Run 7B parameter models with Ollama on 8GB variant
High-Frequency Sensor Polling — GPIO speeds support >1kHz sampling
Time-Series Databases — InfluxDB with NVMe storage
Multi-Protocol Hubs — MQTT broker + WebSocket dashboard + database
Computer Vision — Object detection at edge with USB accelerators

Troubleshooting

❌ Permission denied on GPIO

Cause: User not in gpio group

sudo usermod -a -G gpio $USER
# Logout and login again

❌ I2C device not found

Cause: I2C not enabled or wrong address

# Enable I2C
sudo raspi-config
# Interface Options → I2C → Enable

# Scan for devices
sudo i2cdetect -y 1

❌ SPI device not working

Cause: SPI not enabled or kernel module not loaded

# Enable SPI
sudo raspi-config
# Interface Options → SPI → Enable

# Verify module loaded
lsmod | grep spi

❌ High CPU usage

Cause: Polling frequency too high

# Reduce sensor polling intervals
# In your flow, change inject node interval
# from 100ms to 1000ms or higher

❌ DS18B20 not detected

Cause: 1-Wire not enabled or wrong pin

# Enable 1-Wire
sudo raspi-config
# Interface Options → 1-Wire → Enable

# Check for sensors
ls /sys/bus/w1/devices/

❌ Service won't start

Cause: Configuration error or port in use

# Check service logs
sudo journalctl -u edgeflow -f

# Check if port is in use
sudo lsof -i :8080

Common I2C Addresses

Reference table for common I2C device addresses:

Address	Device	Description
`0x20-0x27`	PCF8574	I/O Expander, LCD backpack
`0x23`	BH1750	Light sensor
`0x38`	AHT20	Temperature/Humidity
`0x3C-0x3D`	SSD1306	OLED Display
`0x40`	HDC1080	Temperature/Humidity
`0x44`	SHT3x	Temperature/Humidity
`0x48-0x4B`	ADS1015/1115	ADC
`0x50-0x57`	AT24Cxx	EEPROM
`0x5A`	CCS811	Air Quality (eCO2/TVOC)
`0x58`	SGP30	Air Quality
`0x68`	DS1307/DS3231	RTC
`0x76-0x77`	BME280/BMP280	Temp/Humidity/Pressure

Service Management

Common commands for managing the EdgeFlow service:

Service Commands

# Start EdgeFlow
sudo systemctl start edgeflow

# Stop EdgeFlow
sudo systemctl stop edgeflow

# Restart EdgeFlow
sudo systemctl restart edgeflow

# View status
sudo systemctl status edgeflow

# View logs (live)
sudo journalctl -u edgeflow -f

# View last 100 log lines
sudo journalctl -u edgeflow -n 100

# Disable auto-start
sudo systemctl disable edgeflow

# Enable auto-start
sudo systemctl enable edgeflow