Enhancing IoT security through emotion recognition and blockchain-driven intrusion prevention

Abstract

As the Internet of Things (IoT) expands, ensuring the security and privacy of interconnected devices poses significant challenges. Traditional intrusion detection and prevention systems (IDPS) for IoT rely primarily on network traffic, anomaly detection, and signature-based approaches. This paper addresses deficiencies in conventional infrastructure security, particularly within Closed-Circuit Television (CCTV) operations, to fortify IoT environments against emerging intrusions and ensure heightened levels of privacy and security. Traditional intrusion detection and prevention systems (IDPSs) for IoT primarily rely on network traffic analysis, anomaly detection, and signature-based approaches. However, there is a promising opportunity to enhance IDPS effectiveness by incorporating CCTV cameras and human-inspired techniques. We present a novel approach to IoT security employing CCTV cameras, Raspberry Pi, and emotion recognition intrusion detection and prevention. Initially, two CCTV cameras are installed and connected to a Raspberry Pi for video recording and preprocessing. Emotions are then detected using a convolutional neural network (CNN). Anomalies are classified according to predefined criteria based on detected emotions: individuals meeting conditions such as fear, multiple failed logins (greater than 2), and activity after 6 PM are classified as intruders, those meeting one or two criteria are labeled suspicious, while others are considered normal (non-intruders). In the event of suspicious activity, an alarm is automatically generated, while for intruders, an internet ban is also applied in addition to an alarm. Our proposed system aims to provide a proactive and context-aware defense mechanism against IoT intrusions by integrating machine learning algorithms and blockchain technology, ensuring the robustness and reliability of IoT security.