Efficient and Secure IoT Based Smart Home Automation Using Multi-Model Learning and Blockchain Technology

Abstract

The concept of smart houses has grown in prominence in recent years. Major challenges linked to smart homes are identification theft, data safety, automated decision-making for IoT-based devices, and the security of the device itself. Current home automation systems try to address these issues but there is still an urgent need for a dependable and secure smart home solution that includes automatic decision-making systems and methodical features. This paper proposes a smart home system based on ensemble learning of random forest (RF) and convolutional neural networks (CNN) for programmed decision-making tasks, such as categorizing gadgets as “OFF” or “ON” based on their normal routine in homes. We have integrated emerging blockchain technology to provide secure, decentralized, and trustworthy authentication and recognition of IoT devices. Our system consists of a 5V relay circuit, various sensors, and a Raspberry Pi server and database for managing devices. We have also developed an Android app that communicates with the server interface through an HTTP web interface and an Apache server. The feasibility and efficacy of the proposed smart home automation system have been evaluated in both laboratory and real-time settings. It is essential to use inexpensive, scalable, and readily available components and technologies in smart home automation systems. Additionally, we must incorporate a comprehensive security and privacy-centric design that emphasizes risk assessments, such as cyberattacks, hardware security, and other cyber threats. The trial results support the proposed system and demonstrate its potential for use in everyday life.