Implementation of Naive Bayes Classifier for Reducing DDoS Attacks in IoT Networks

Abstract

One of the objectives of intelligent devices is to enhance human well-being in terms of convenience and efficiency. Using the Internet of Things (IoT) paradigm, smart environments may now be created. Privacy and security are major concerns in any IoT-based smart real-world setting. There are security dangers to smart environment applications due to IoT-based systems' vulnerabilities. IoT-specific intrusion detection systems (IDSs) are urgently needed to protect against attacks that leverage some of these security flaws. Intrusion Detection Systems (IDS) have played a significant role in network and information system security for more than two decades. Because of its unique properties, such as low-resource devices, specialised protocol stacks, and standardized communication protocols, it is difficult to apply classic IDS techniques to IoT. For the Internet of Things, standard intrusion detection systems suffer from a number of limitations that can be mitigated by combining machine learning technology with them. The purpose of this work is to explain how classic statistical approaches may be used to examine scope distribution diversity in order to choose and optimize features. The Correlation Coefficient approach is used to select the best features for the development of the classifier in this work's "Distributed Diverse Technique of Feature Optimization to Prevent Intrusion Activities on IoT Networks." To train our Naive Bayes classifier, we employed proposal-selected feature sets. There was a decrease in false alarms as well as an increase in classification accuracy.