A Comparative Study of Distributed Denial of Service Attacks On The Internet Of Things By Using Shallow Neural Network

Abstract

The Internet of Things (IoT) is a term used to indicate a world in which objects are linked to the Internet. in some way, but not in the way that most people imagine. However, for the Internet of Things to be a success, computing must go beyond standard scenarios involving laptops and smartphones to include the networking of common intelligent Intelligence integration with the environment”, ”Smart homes, cities, and other wearable devices are examples. As a result, there will be new computing problems and features. Because of its variety, the Internet of Things has a difficult time guaranteeing universal privacy in areas like smart homes, smart hospitals, and so forth. Vulnerability can appear in a variety of forms. The internet of things has grown in popularity during the previous era. The internet of things (IoT), which may be characterized as a network of networked gadgets, has exploded in popularity during the last decade. Many elements of our lives have been fast-devoured by the Internet of Things (IoT). Smart homes, savvy cities, and other wearable devices are examples. IoT devices work to achieve their objectives, which include the building of a contemporary city. At the same time, there are a lot of security flaws in IoT devices that attackers could exploit. Distributed Denial of Service (DDoS) is the most common hazard to IoT security. The main goal of these assaults is to knock down victim computers and prevent legitimate people from accessing them using malicious software. The goal of this research is to provide compression of two algorithms 1. Scaled Conjugate Gradient (SCG) and 2. Levenberg–Marquardt algorithms (LMA) by training a Shallow neural network look into and assesses security vulnerabilities linked to DDoS attacks, as well as solutions like layered IoT device protection. In this research, it is discovered that the conjugant gradient algorithm has better accuracy as compared with Levenberg–Marquardt algorithm.