Attack detection and secure IOT using adaptive neuro fuzzy inference scheme

Abstract

Nowadays, the wide adoption of the modern Internet of Things newline(IoT) paradigm has brought about the tremendous development of smart newlinecities. Smart cities operate in real-time world to promote ease and also the newlinehuman life quality with regard to efficiency and comfort. A security concern newlinealong with privacy is considered as a foremost issue in several smart cities. newlineThe security vulnerability in IoT-centered systems creates security threats newlinewhich affect smart surroundings applications. Therefore, there is basically a newlinerequirement for Intrusion Detection Systems (IDS) for mitigating the newlineIoT-related security outbreaks which took the entire benefits of security newlineliabilities. In existing works, the accuracy in the procedure of detection and newlinesecurity are the main challenge. To trounce these drawbacks, in the first stage newlineof the work the IDS are proposed intended for the identification of attacks in newlineIoT of city depending on the Deep Learning Modified Neural Networks newline(DLMNN) classification approach. Initially, the values of sensor from smart newlinecities are provided to the training phase of IDS system which is then newlineemployed for the purpose of testing values. Then, the step of preprocessing is newlinecarried out, followed by feature selection using entropy-Hummingbird newlineOptimization Algorithm (HOA). The classified outcomes are analyzed newlineafterwards and is employed for the prediction of outcome. Then, the task of newlinesecured data sharing is carried with the use of Krill Heard Advanced newlineEncryption Standard (KH-AES) approach. Finally, the resultant outcome is newlineforecasted. Experimental result of the suggested method employed in newlineclassification, feature selection, and secured data sharing are then compared newlinewith the traditional methods. newline newline