Improvements of SERENIOT Based on Machine Learning

The Internet of Things (IoT) is gradually penetrating the lives of citizens with the advancement of the Internet, and people’s privacy is becoming “transparent”, so it is especially important to provide an IoT security detection system. However, most of the current IoT systems and their detection systems are not perfect. The problem occurs in two main areas. While blockchain or other cryptographic algorithms are secure technologies for maintaining the IoT, they inherently have unavoidable vulnerabilities. The second is that they have details that are overlooked early in their design. SERENIoT is a blockchain-based IoT security detection system that has a good overall design, but a vulnerability in the design of “anomaly” detection. To solve this problem, a machine learning-based SERENIoT upgrade system is designed. The system addresses the vulnerability in “anomaly” detection through LSTM, DRL, and DQN techniques to improve the correct detection rate and secure the IoT system.