Interference Recognition for Fog Enabled IoT Architecture using a Novel Tree-based Method

Abstract

When connecting and interacting with any device over the internet, the Internet of Things (IoT) holds much potential. Every day, the number of devices increases, and these devices come in a wide variety of shapes, sizes, functions, and levels of complexity. IoT provides a variety of services through applications, but it is plagued by security vulnerabilities and attacks, such as sinkhole attacks, eavesdropping, and denial of service attacks, among others. Also, cyber-attacks are growing more complex, making them harder to identify. These attacks impact the network’s sensitive information because they penetrate the network while behaving normally. This study presents a fog-assisted approach for detecting interference in IoT architecture, including DoS, DDoS, data exfiltration, keylogging, service and OS Scan attacks. In a novel three-phase classification system, we have used tree-based ensembles for this aim. The accuracy of the proposed model has been improved to 95.1 percent (the accuracy is 99% in training phase). This increase in accuracy has been achieved by paying particular attention to the high generality and the absence of over-fitting, which are detailed later in this article.