We are excited to share our new paper: "Real-time detection of Wi-Fi attacks using hybrid deep learning models on NodeMCU."

Wi-Fi Deauthentication (DA) attacks pose a significant threat. Existing detection systems are often cloud-based, making them unsuitable for resource-constrained, low-power Internet of Things (IoT) environments.

Our Novel Solution

We introduce a unique, cost-effective hybrid deep learning system tailored for the network edge. Our system:

1. Platform: Utilizes the NodeMCU ESP8266 microcontroller for live packet sniffing and feature extraction, making it highly suitable for low-power IoT devices.

2. Model: Integrates powerful sequential deep learning models (LSTM, GRU, RNN) with Logistic Regression (LR) to analyze Wi-Fi traffic in real-time based on key metrics like RSSI, DA packet count, and SNR.

Key Findings & Impact

• Achieved 96% detection accuracy on a dataset of over 7,000 labeled samples.

• Demonstrated superior performance in identifying challenging minority-class threats.

• Delivers a practical, transparent, and explainable AI (XAI) approach that can be readily adapted to diverse IoT and wireless security contexts.

This work addresses the critical need for real-time security at the network edge. We invite researchers, security professionals, and practitioners in edge computing to read our full paper and explore this innovative, embedded security solution.

paper Link: https://www.nature.com/articles/s41598-025-18947-2

Cite this article:
Moharam, M.H., Ashraf, K., Alaa, H. et al. Real-time detection of Wi-Fi attacks using hybrid deep learning models on NodeMCU. Sci Rep 15, 32544 (2025). https://doi.org/10.1038/s41598-025-18947-2