Machine learning-based detection of Denial-of-Sleep attacks in wireless sensor networks

The paper comprises detection of Denial-of-Sleep attacks in wireless sensor networks. Such attacks are applicable to IoT devices that operate autonomously and switch between two modes of operation, namely general and power-efficient (sleep) modes, depending on the business rules of the devices. Although being quite effective in general, such attacks are quite stealthy and can significantly reduce and even completely drain the battery life of the device, thereby bringing it into a disabled state. The paper analyzes the source data and applies artificial intelligence methods to detect Denial-of-Sleep attacks. The detection model is built using specific machine learning techniques. The model is validated on real data collected on a test bench with ZigBee wireless modules representing both benign wireless sensor nodes and attackers. The detection quality indicators confirm the effectiveness and applicability of the proposed detection methods in practice.

Authors: V. A. Desnitsky

Direction: Informatics, Computer Technologies And Control

Keywords: wireless sensor network, security, attack detection

---

View full article