Enhanced Chicken Swarm Optimization And Improved Convolutional Neural Network Algorithm For Attack Detection Over Iot Based Wireless Sensor Network

Abstract

A useful, adaptable, and interoperable network of electronics, gadgets, and objects has been developed named as the Internet of Things (IoT). IoT has emerged from its early days and is regarded as the most significant technology in changing the Internet into an entirely connected future Internet. Recent developments in computing, networking, communications, software, and hardware technology are the primary factors influencing it Utilizing the potential of IoT in useful applications and services, IoT employs Wireless Sensor Networks (WSN) to remotely gather, exchange, and distribute data. But, the existing system has issues with various and serious security attacks. Also, it has problem with Attack Detection (AD)accuracy for the given dataset. To overcome the abovementioned problems in this research, Enhanced Chicken Swarm Optimization and Improved Convolutional (NN) Neural Network (ECSO-ICNN) algorithm is suggested. Some of the primary stages in this study are the system model (SM), NSL-KDD Data Collection (DC), Cluster Head (CH), Node Selection (NS), data pre-processing, and AD. The amount of Sensor Nodes (SN), sensor devices, SN, destinations, and Multipoint Relays (MPRs) with neighbor and CH nodes that are one-hop and two-hop are all included in the system model. Next, the ECSO method is employed for selecting the CH node. It generates best Fitness Values (FV) by means of higher accuracy and lower Energy Consumption (EC) for the given IoT based WSN. With 42 features and class labels, the NSL-KDD dataset is regarded as a class of attacks. Then, data pre-processing is done by using filtering and Feature Selection (FS) process which is used to handle duplication and redundant features effectively for the given NSL-KDD dataset. The ICNN algorithm, which effectively detects attacks, is the last method utilized for AD. In terms of f-measure, accuracy, recall, and precision, the simulation results show that the suggested ECSO-ICNN strategy performs better than the existing approaches for AD.