Enhancing Cloud Server Resilience: Honeypots and Intrusion Detection Systems in Browser Hijacking Defense

Thus, the necessity of an efficient anomaly detection in the context of modern cloud environment is undeniable when it comes to combating complex browser hijacking threats. This paper aims at unraveling the real application of machine learning for cloud server to improve its reliability where data from honeypots and Intrusion Detection Systems (IDS) are used to train and test three Isolation Forest, One-Class SVM, and DBSCAN machine learning algorithms. It also achieved a high level of Precision, Recall, and Accuracy, which is why Isolation Forest can be considered a well-posed algorithm for the detection of anomalies when data is both massive and has a large number of features. One-Class SVM performed notably well in the Recall where it was able to identify a large proportion of the anomalies at the expense of the Specificity. Hence, DBSCAN had poor Precision @ all and Recall @all values proving that it had a tendency to miss some degree of detection performance results in exchange for high accuracy. The study indicates that each algorithm has benefits and weaknesses, and depend on what the security system needs and what kind of information is to be protected, the detection method should be chosen. Annually, cloud breaches continue to be reported, and therefore, this research sheds light on the right choice of and approach towards anomaly detection to protect cloud infrastructure.