FamilyGuard: A Security Architecture for Anomaly Detection in Home Networks

Area of Science:

• Cybersecurity
• Internet of Things (IoT)
• Machine Learning

Background:

• The modern residential environment features interconnected smart devices, creating convenience but also introducing significant security vulnerabilities.
• The heterogeneity of applications and protocols in smart homes complicates security management and threat detection.
• Existing security measures may not adequately address the unique challenges posed by the evolving smart home landscape.

Purpose of the Study:

• To propose the FamilyGuard architecture, a novel security layer for smart homes.
• To simplify the management of smart home environments through intelligent security solutions.
• To enhance user security by detecting network traffic anomalies.

Main Methods:

• Development of the FamilyGuard architecture incorporating a new security layer.
• Implementation of an anomaly detection module utilizing machine learning algorithms.
• Training one-class classifiers on network flow data for anomaly identification.

Main Results:

• Experimental validation confirmed the efficacy of the FamilyGuard architecture's core components.
• The developed anomaly detection module successfully identified network traffic irregularities.
• The proposed solution demonstrated the capability to provide personalized security features.

Conclusions:

• The FamilyGuard architecture offers a viable solution for improving smart home security.
• Machine learning-based anomaly detection is effective in identifying threats in heterogeneous smart home networks.
• The system provides enhanced and personalized security for smart home users utilizing affordable hardware.