Review on the Network Intrusion Detection Systems (NIDS)

Abstract

In today's world, the rapid advancement of Information Technology has resulted in a large number of people accessing the internet globally. The COVID-19 pandemic has further sped up this trend, leading organizations and individuals to move towards online platforms for their daily activities and businesses. Consequently, these online activities have led to various cyber threats for users and networks. The paper analyzed the recent evaluation of Network Intrusion Detection System (NIDS) techniques, such as machine learning models like decision trees, support vector machines, logistic regression, and others, that have been effective in spotting cyber threats, but their effectiveness decreases when dealing with extensive and high-dimensional data. Deep learning models have demonstrated impressive performance in handling extensive and complex datasets. Moreover, ensembles and hybrid models have displayed potential for improved performance compared to stand-alone ML and DL techniques. The paper also included an analysis of commonly utilized datasets for NIDS, such as NSL-KDD, KDD CUP-99, and CICIDS 2017. These datasets are highly important for researchers, organizations, and institutions for further evaluation of NIDS models. Future research efforts could concentrate on addressing existing limitations within NIDS, utilizing advancements in ML, DL, and ensemble techniques to enhance detection capabilities and strengthen network defenses against evolving cyber threats.