Published Paper Details:

In the connected digital world of today, protecting these networked systems is essential. To ensure the integrity of data transmission over the internet and stop fraud or illegal access, it is essential to identify abnormalities in network behaviour. By spotting malicious attacks in network traffic, intrusion detection systems (IDS) play a critical role in network security. There is no thorough study on anomaly detection utilizing four types of ML models under various network environments, despite the fact that ML techniques have been employed in a variety of fields to address security challenges. The majority of surveys only addressed one form of network (supervised learning).Therefore, we outline in this paper how anomaly detection can be accomplished by unsupervised learning (Support vector machine, k-means clustering technique, NDM) and the current solutions in computer networks, cellular networks, SDN, IoT, and cloud networks. The accuracy of this algorithm is between 94 and 98 percent when compared to supervised learning methods. Additionally, we provide the unsupervised detection approach, which is generally suggested without mentioning the type of network. Experimentally corrected datasets are used to confirm these solutions.