A Hybrid Intrusion Detection System for Smart Home Security Based on Machine Learning and User Behavior

Abstract

Given the growth of internet technology, lives are increasingly being led into the virtual world. Virtual functions such as working, shopping, or monitoring properties have been made possible by network interconnectivity. The Internet of Things (IoT), a quickly rising digital technology, is considered to be the third revolution in information technology after computers and internet. IoT is a network that has the ability to connect an object to the internet through sensors, smart devices, and smart equipment. There are many applications in IoT and the smart home is one of the most important. Smart homes have become part of people’s daily lives. Many people attempt to monitor and control their smart homes using smartphones, tablets, or computers because smart home systems make residential living areas more comfortable and convenient. Many devices in homes are now being connected to the internet; these devices can easily become targets for attacks and can cause serious problems that affect users’ lives. Some of these attacks are difficult to detect because attackers can be intelligent or use the same protocols that are employed by users to submit legitimate requests. Another issue is that anomalous requests can locate the existence of the smart home network environment and do not exhibit any malicious behavior. The security system in a smart home should have the capability to detect any expected threats that might come from the network, devices, sensors, or users. An intrusion detection system (IDS) is designed to detect and mitigate attacks on the network. However, various constraints on smart home sensors and device manufacturers do not ensure the security and privacy of the wireless sensor networks. This is because they use a one tier standard intrusion detection and most IDSs have crippling limitations that cannot provide accurate results. Therefore, this dissertation proposes an efficient framework called a Hybrid Intrusion Detection (HID-SMART) system. HID-SMART is a model that detects and examines smart home requests. The model utilizes analysis techniques on the network tier and employs static analysis techniques by matching the user profile. The first tier contained a machine learning technique. This technique is studied by the smart home’s network traffic. The second tier examines all requests sent to the system based on patterns of user behavior profiles. The reason for having a two-tiered intrusion detection system is to increase the system’s security and restrain the error rate since typically more than one user can control and monitor a smart home. HID-SMART employs several machine learning algorithms for the network behavior tier and a misuse detection technique for the user behavior tier. The model obtains approximately 95% accuracy and achieves less than a 0.1% false positive and false negative.