Free Simulators & Tools

Level 2 Resources: No Hardware Required

Three-Tier Learning Model

Our labs follow a three-tier approach to accommodate all learning environments:

Level	Environment	Hardware Needed
Level 1	Jupyter Notebooks	None - runs in browser
Level 2	Simulators & Online Tools	None - browser-based
Level 3	Physical Devices	Arduino, Pi, ESP32

Complete Without Hardware

You can complete all labs through Level 2 using only a web browser! The simulators below provide realistic experiences without purchasing any equipment.

Quick Reference: Simulators by Lab

Lab	Topic	Level 2 Simulator	Link
LAB01	Introduction	Colab Notebook	Google Colab
LAB02	ML Foundations	TensorFlow Playground	playground.tensorflow.org
LAB03	Quantization	Netron Model Viewer	netron.app
LAB04	Keyword Spotting	Spectrogram + Colab	Chrome Music Lab
LAB05	Edge Deployment	Netron + TFLite Viewer	netron.app
LAB06	Edge Security	FGSM Colab Demo	TensorFlow FGSM
LAB07	CNNs & Vision	CNN Explainer	poloclub.github.io/cnn-explainer
LAB08	Arduino Sensors	Wokwi Simulator	wokwi.com
LAB09	ESP32 Wireless	Wokwi + HiveMQ	Wokwi / HiveMQ
LAB10	EMG Biomedical	PhysioNet + Colab	physionet.org
LAB11	Profiling	Netron + TF Profiler	netron.app
LAB12	Streaming	MQTT Simulator	mqttlab.iotsim.io
LAB13	Distributed Data	SQLite Online	sqliteonline.com
LAB14	Anomaly Detection	NAB Dataset + Colab	GitHub NAB
LAB15	Energy	Battery Calculator	oregonembedded.com
LAB16	YOLO/CV	Ultralytics HUB	hub.ultralytics.com
LAB17	Federated Learning	Flower Simulation	flower.ai
LAB18	On-Device Learning	Edge Impulse	edgeimpulse.com

Machine Learning Visualization Tools

TensorFlow Playground

Best for: LAB02 (ML Foundations)

Interactive neural network visualization - perfect for understanding gradient descent, activation functions, and overfitting.

URL: playground.tensorflow.org
License: Apache 2.0 (Free)
Features: Real-time training visualization, adjustable hyperparameters

What You Can Do
Lab Exercise

Visualize decision boundaries in real-time
Experiment with different architectures (layers, neurons)
Compare activation functions (ReLU, Tanh, Sigmoid)
Observe overfitting with complex vs simple models

Start with the “Circle” dataset
Try a single hidden layer with 2 neurons - observe underfitting
Add more neurons until it fits
Switch to “Spiral” - what architecture works?

CNN Explainer

Best for: LAB07 (CNNs & Vision)

Interactive visualization of convolutional neural network operations.

URL: poloclub.github.io/cnn-explainer
License: MIT (Free)
Features: Layer-by-layer visualization, filter animations

What You Can Do
Lab Exercise

See how convolution filters extract features
Understand pooling operations visually
Watch activations flow through the network
Explore a trained Tiny VGG network

Upload a digit image
Observe which filters activate
Track how spatial information reduces through pooling
Compare with different input images

Netron Model Viewer

Best for: LAB03 (Quantization), LAB05 (Deployment), LAB11 (Profiling)

Universal neural network model visualizer supporting TFLite, ONNX, and more.

URL: netron.app
License: MIT (Free)
Formats: TFLite, ONNX, Keras, PyTorch, CoreML, and 30+ more

What You Can Do
Lab Exercise

Visualize model architecture from .tflite files
Inspect layer shapes and parameters
Compare float32 vs int8 quantized models
Export architecture diagrams

Export your LAB03 quantized model
Open both float32 and int8 versions in Netron
Compare parameter counts and tensor types
Calculate memory savings

Audio & Signal Processing

Chrome Music Lab - Spectrogram

Best for: LAB04 (Keyword Spotting)

Real-time audio spectrogram visualization using your microphone.

URL: musiclab.chromeexperiments.com/Spectrogram
License: Free (Google)
Features: Live microphone input, frequency visualization

What You Can Do
Lab Exercise

Visualize your voice as frequency patterns
See how different words create different spectrograms
Understand why MFCCs are useful features
Compare sustained sounds vs transients

Say “yes” and “no” repeatedly
Observe the different spectrogram patterns
Try other wake words (“hey”, “okay”, “stop”)
Notice which frequencies are most distinctive

Academo Spectrum Analyzer

Best for: LAB04 (Keyword Spotting), LAB10 (EMG)

Upload audio files to visualize their frequency content.

URL: academo.org/demos/spectrum-analyzer
License: Free
Features: File upload, real-time analysis, presets (whale, siren, birds)

Hardware Simulators

Wokwi - Arduino & ESP32 Simulator

Best for: LAB08 (Sensors), LAB09 (Wireless)

Full-featured embedded systems simulator running in your browser.

URL: wokwi.com
License: Free tier available
Supported: Arduino Uno/Nano/Mega, ESP32, Raspberry Pi Pico

Write and run Arduino/ESP32 code
Simulate sensors (temperature, motion, light)
Test I2C/SPI communication
WiFi simulation with real HTTP requests!

DHT22 temperature/humidity sensor
PIR motion sensor
Analog potentiometers
LEDs, buttons, displays
Serial monitor

HiveMQ Public MQTT Broker

Best for: LAB09 (MQTT), LAB12 (Streaming)

Free public MQTT broker for testing IoT communication.

URL: hivemq.com/mqtt/public-mqtt-broker
Broker: broker.hivemq.com:1883
WebSocket Client: Built-in browser client

What You Can Do
Lab Exercise

Publish/subscribe to topics in browser
Test MQTT communication without setup
Connect Wokwi ESP32 to real broker
Visualize message flow

Open HiveMQ WebSocket client
Subscribe to edge-analytics/sensor/#
In Wokwi, publish sensor data to same topic
Watch messages flow in real-time

MQTT Lab (IoTSim.io)

Best for: LAB12 (Streaming), LAB13 (Distributed)

Simulate thousands of IoT devices for testing.

URL: mqttlab.iotsim.io
License: Free tier available
Features: Device simulation, traffic generation

ML Platforms with Free Tiers

Edge Impulse

Best for: LAB04-LAB18 (All ML Labs)

Complete ML platform for edge devices with browser-based training.

URL: edgeimpulse.com
License: Free for individuals
Features: Data collection, training, deployment

Free Tier Includes

Unlimited projects
4 hours compute/month
Export to Arduino, ESP32, Raspberry Pi
Community support

What You Can Do
Recommended Projects

Upload sensor data via browser
Train models without coding
Export optimized C++ for microcontrollers
Test in browser before deployment

Keyword Spotting (LAB04): Use Edge Impulse audio classification
Motion Detection (LAB10): Accelerometer gesture recognition
Anomaly Detection (LAB14): Unsupervised anomaly detection block

Ultralytics HUB

Best for: LAB16 (YOLO/CV)

Train and deploy YOLO models in the cloud.

URL: hub.ultralytics.com
License: Free tier available
Features: Dataset management, training, export

Google Colab

Best for: All Labs

Free Jupyter notebooks with GPU access.

URL: colab.research.google.com
License: Free (Google account required)
Features: GPU/TPU access, pre-installed ML libraries

Open Any Notebook in Colab

Replace GitHub URL prefix:

github.com/user/repo/blob/main/notebook.ipynb
     ↓
colab.research.google.com/github/user/repo/blob/main/notebook.ipynb

Security & Adversarial ML

FGSM Attack Demo (Colab)

Best for: LAB06 (Edge Security)

Interactive adversarial attack demonstration.

URL: TensorFlow FGSM Tutorial
License: Apache 2.0 (Free)
Features: Step-by-step FGSM implementation

Data Tools

SQLite Online

Best for: LAB13 (Distributed Data)

Browser-based SQL database for testing queries.

URL: sqliteonline.com
License: Free
Features: SQL editor, import/export, visualization

PhysioNet

Best for: LAB10 (EMG/Biomedical)

Free physiological signal databases.

URL: physionet.org
License: Various open licenses
Datasets: ECG, EMG, EEG, and more

Energy & Power Tools

Battery Life Calculator

Best for: LAB15 (Energy Optimization)

Estimate device battery life based on power profile.

URL: oregonembedded.com/batterycalc.htm
License: Free

Getting Started Checklist

To complete all labs at Level 2, bookmark these essential tools:

Additional Resources

Deeper Playground - Extended TF Playground
ConvNetJS - JS neural networks
3D CNN Visualization - Interactive 3D view
MQTTX - Desktop MQTT client
Flower - Federated learning framework

Contributing

Know of a great free simulator we missed? Open an issue to suggest additions!

--- title: "Free Simulators & Tools" subtitle: "Level 2 Resources: No Hardware Required" --- ## Three-Tier Learning Model Our labs follow a three-tier approach to accommodate all learning environments: | Level | Environment | Hardware Needed | |-------|-------------|-----------------| | **Level 1** | Jupyter Notebooks | None - runs in browser | | **Level 2** | Simulators & Online Tools | None - browser-based | | **Level 3** | Physical Devices | Arduino, Pi, ESP32 | ::: {.callout-tip} ## Complete Without Hardware You can complete **all labs through Level 2** using only a web browser! The simulators below provide realistic experiences without purchasing any equipment. ::: --- ## Quick Reference: Simulators by Lab | Lab | Topic | Level 2 Simulator | Link | |-----|-------|-------------------|------| | LAB01 | Introduction | Colab Notebook | [Google Colab](https://colab.research.google.com) | | LAB02 | ML Foundations | TensorFlow Playground | [playground.tensorflow.org](https://playground.tensorflow.org) | | LAB03 | Quantization | Netron Model Viewer | [netron.app](https://netron.app) | | LAB04 | Keyword Spotting | Spectrogram + Colab | [Chrome Music Lab](https://musiclab.chromeexperiments.com/Spectrogram) | | LAB05 | Edge Deployment | Netron + TFLite Viewer | [netron.app](https://netron.app) | | LAB06 | Edge Security | FGSM Colab Demo | [TensorFlow FGSM](https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/tutorials/generative/adversarial_fgsm.ipynb) | | LAB07 | CNNs & Vision | CNN Explainer | [poloclub.github.io/cnn-explainer](https://poloclub.github.io/cnn-explainer/) | | LAB08 | Arduino Sensors | Wokwi Simulator | [wokwi.com](https://wokwi.com) | | LAB09 | ESP32 Wireless | Wokwi + HiveMQ | [Wokwi](https://wokwi.com) / [HiveMQ](https://www.hivemq.com/mqtt/public-mqtt-broker/) | | LAB10 | EMG Biomedical | PhysioNet + Colab | [physionet.org](https://physionet.org) | | LAB11 | Profiling | Netron + TF Profiler | [netron.app](https://netron.app) | | LAB12 | Streaming | MQTT Simulator | [mqttlab.iotsim.io](https://mqttlab.iotsim.io) | | LAB13 | Distributed Data | SQLite Online | [sqliteonline.com](https://sqliteonline.com) | | LAB14 | Anomaly Detection | NAB Dataset + Colab | [GitHub NAB](https://github.com/numenta/NAB) | | LAB15 | Energy | Battery Calculator | [oregonembedded.com](https://oregonembedded.com/batterycalc.htm) | | LAB16 | YOLO/CV | Ultralytics HUB | [hub.ultralytics.com](https://hub.ultralytics.com) | | LAB17 | Federated Learning | Flower Simulation | [flower.ai](https://flower.ai) | | LAB18 | On-Device Learning | Edge Impulse | [edgeimpulse.com](https://edgeimpulse.com) | --- ## Machine Learning Visualization Tools ### TensorFlow Playground **Best for:** LAB02 (ML Foundations) Interactive neural network visualization - perfect for understanding gradient descent, activation functions, and overfitting. - **URL:** [playground.tensorflow.org](https://playground.tensorflow.org) - **License:** Apache 2.0 (Free) - **Features:** Real-time training visualization, adjustable hyperparameters ::: {.panel-tabset} ### What You Can Do - Visualize decision boundaries in real-time - Experiment with different architectures (layers, neurons) - Compare activation functions (ReLU, Tanh, Sigmoid) - Observe overfitting with complex vs simple models ### Lab Exercise 1. Start with the "Circle" dataset 2. Try a single hidden layer with 2 neurons - observe underfitting 3. Add more neurons until it fits 4. Switch to "Spiral" - what architecture works? ::: --- ### CNN Explainer **Best for:** LAB07 (CNNs & Vision) Interactive visualization of convolutional neural network operations. - **URL:** [poloclub.github.io/cnn-explainer](https://poloclub.github.io/cnn-explainer/) - **License:** MIT (Free) - **Features:** Layer-by-layer visualization, filter animations ::: {.panel-tabset} ### What You Can Do - See how convolution filters extract features - Understand pooling operations visually - Watch activations flow through the network - Explore a trained Tiny VGG network ### Lab Exercise 1. Upload a digit image 2. Observe which filters activate 3. Track how spatial information reduces through pooling 4. Compare with different input images ::: --- ### Netron Model Viewer **Best for:** LAB03 (Quantization), LAB05 (Deployment), LAB11 (Profiling) Universal neural network model visualizer supporting TFLite, ONNX, and more. - **URL:** [netron.app](https://netron.app) - **License:** MIT (Free) - **Formats:** TFLite, ONNX, Keras, PyTorch, CoreML, and 30+ more ::: {.panel-tabset} ### What You Can Do - Visualize model architecture from .tflite files - Inspect layer shapes and parameters - Compare float32 vs int8 quantized models - Export architecture diagrams ### Lab Exercise 1. Export your LAB03 quantized model 2. Open both float32 and int8 versions in Netron 3. Compare parameter counts and tensor types 4. Calculate memory savings ::: --- ## Audio & Signal Processing ### Chrome Music Lab - Spectrogram **Best for:** LAB04 (Keyword Spotting) Real-time audio spectrogram visualization using your microphone. - **URL:** [musiclab.chromeexperiments.com/Spectrogram](https://musiclab.chromeexperiments.com/Spectrogram) - **License:** Free (Google) - **Features:** Live microphone input, frequency visualization ::: {.panel-tabset} ### What You Can Do - Visualize your voice as frequency patterns - See how different words create different spectrograms - Understand why MFCCs are useful features - Compare sustained sounds vs transients ### Lab Exercise 1. Say "yes" and "no" repeatedly 2. Observe the different spectrogram patterns 3. Try other wake words ("hey", "okay", "stop") 4. Notice which frequencies are most distinctive ::: --- ### Academo Spectrum Analyzer **Best for:** LAB04 (Keyword Spotting), LAB10 (EMG) Upload audio files to visualize their frequency content. - **URL:** [academo.org/demos/spectrum-analyzer](https://academo.org/demos/spectrum-analyzer/) - **License:** Free - **Features:** File upload, real-time analysis, presets (whale, siren, birds) --- ## Hardware Simulators ### Wokwi - Arduino & ESP32 Simulator **Best for:** LAB08 (Sensors), LAB09 (Wireless) Full-featured embedded systems simulator running in your browser. - **URL:** [wokwi.com](https://wokwi.com) - **License:** Free tier available - **Supported:** Arduino Uno/Nano/Mega, ESP32, Raspberry Pi Pico ::: {.panel-tabset} ### What You Can Do - Write and run Arduino/ESP32 code - Simulate sensors (temperature, motion, light) - Test I2C/SPI communication - WiFi simulation with real HTTP requests! ### Built-in Components - DHT22 temperature/humidity sensor - PIR motion sensor - Analog potentiometers - LEDs, buttons, displays - Serial monitor ### Lab Exercises - [LAB08: Multi-Sensor Arduino](../simulations/wokwi-arduino-sensors.qmd) - [LAB09: ESP32 WiFi IoT](../simulations/wokwi-esp32-wifi.qmd) ::: --- ### HiveMQ Public MQTT Broker **Best for:** LAB09 (MQTT), LAB12 (Streaming) Free public MQTT broker for testing IoT communication. - **URL:** [hivemq.com/mqtt/public-mqtt-broker](https://www.hivemq.com/mqtt/public-mqtt-broker/) - **Broker:** `broker.hivemq.com:1883` - **WebSocket Client:** Built-in browser client ::: {.panel-tabset} ### What You Can Do - Publish/subscribe to topics in browser - Test MQTT communication without setup - Connect Wokwi ESP32 to real broker - Visualize message flow ### Lab Exercise 1. Open HiveMQ WebSocket client 2. Subscribe to `edge-analytics/sensor/#` 3. In Wokwi, publish sensor data to same topic 4. Watch messages flow in real-time ::: --- ### MQTT Lab (IoTSim.io) **Best for:** LAB12 (Streaming), LAB13 (Distributed) Simulate thousands of IoT devices for testing. - **URL:** [mqttlab.iotsim.io](https://mqttlab.iotsim.io) - **License:** Free tier available - **Features:** Device simulation, traffic generation --- ## ML Platforms with Free Tiers ### Edge Impulse **Best for:** LAB04-LAB18 (All ML Labs) Complete ML platform for edge devices with browser-based training. - **URL:** [edgeimpulse.com](https://edgeimpulse.com) - **License:** Free for individuals - **Features:** Data collection, training, deployment ::: {.callout-note} ## Free Tier Includes - Unlimited projects - 4 hours compute/month - Export to Arduino, ESP32, Raspberry Pi - Community support ::: ::: {.panel-tabset} ### What You Can Do - Upload sensor data via browser - Train models without coding - Export optimized C++ for microcontrollers - Test in browser before deployment ### Recommended Projects 1. **Keyword Spotting** (LAB04): Use Edge Impulse audio classification 2. **Motion Detection** (LAB10): Accelerometer gesture recognition 3. **Anomaly Detection** (LAB14): Unsupervised anomaly detection block ::: --- ### Ultralytics HUB **Best for:** LAB16 (YOLO/CV) Train and deploy YOLO models in the cloud. - **URL:** [hub.ultralytics.com](https://hub.ultralytics.com) - **License:** Free tier available - **Features:** Dataset management, training, export --- ### Google Colab **Best for:** All Labs Free Jupyter notebooks with GPU access. - **URL:** [colab.research.google.com](https://colab.research.google.com) - **License:** Free (Google account required) - **Features:** GPU/TPU access, pre-installed ML libraries ::: {.callout-tip} ## Open Any Notebook in Colab Replace GitHub URL prefix: ``` github.com/user/repo/blob/main/notebook.ipynb ↓ colab.research.google.com/github/user/repo/blob/main/notebook.ipynb ``` ::: --- ## Security & Adversarial ML ### FGSM Attack Demo (Colab) **Best for:** LAB06 (Edge Security) Interactive adversarial attack demonstration. - **URL:** [TensorFlow FGSM Tutorial](https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/tutorials/generative/adversarial_fgsm.ipynb) - **License:** Apache 2.0 (Free) - **Features:** Step-by-step FGSM implementation --- ## Data Tools ### SQLite Online **Best for:** LAB13 (Distributed Data) Browser-based SQL database for testing queries. - **URL:** [sqliteonline.com](https://sqliteonline.com) - **License:** Free - **Features:** SQL editor, import/export, visualization --- ### PhysioNet **Best for:** LAB10 (EMG/Biomedical) Free physiological signal databases. - **URL:** [physionet.org](https://physionet.org) - **License:** Various open licenses - **Datasets:** ECG, EMG, EEG, and more --- ## Energy & Power Tools ### Battery Life Calculator **Best for:** LAB15 (Energy Optimization) Estimate device battery life based on power profile. - **URL:** [oregonembedded.com/batterycalc.htm](http://oregonembedded.com/batterycalc.htm) - **License:** Free --- ## Getting Started Checklist To complete all labs at Level 2, bookmark these essential tools: - [ ] [Google Colab](https://colab.research.google.com) - All notebooks - [ ] [TensorFlow Playground](https://playground.tensorflow.org) - ML basics - [ ] [CNN Explainer](https://poloclub.github.io/cnn-explainer/) - Vision - [ ] [Wokwi](https://wokwi.com) - Hardware simulation - [ ] [Netron](https://netron.app) - Model inspection - [ ] [Edge Impulse](https://edgeimpulse.com) - Complete ML pipeline - [ ] [HiveMQ Broker](https://www.hivemq.com/mqtt/public-mqtt-broker/) - MQTT testing --- ## Additional Resources - [Deeper Playground](https://deeperplayground.org/) - Extended TF Playground - [ConvNetJS](https://cs.stanford.edu/people/karpathy/convnetjs/) - JS neural networks - [3D CNN Visualization](https://adamharley.com/nn_vis/cnn/3d.html) - Interactive 3D view - [MQTTX](https://mqttx.app/) - Desktop MQTT client - [Flower](https://flower.ai) - Federated learning framework --- ::: {.callout-note} ## Contributing Know of a great free simulator we missed? [Open an issue](https://github.com/ngcharithperera/edge-analytics-lab-book/issues) to suggest additions! :::