Application of Artificial Intelligence and Deep Learning in Operating System Fingerprinting Prevention Using Neural Networks

—Information gathered from an operating system (OS) is a powerful weapon for cybercriminals. Existing solutions rely on a monitoring system across transport and application layers. This study developed an Operating System fingerprinting preventive model using deep learning techniques like CNN, DBN, and RNN on the UNSW-NB-15 dataset. Principal component analysis was used to reduce 43 features to 10. Bayesian optimized RNN performed best with a high accuracy of 99.99% for preventing OS fingerprinting. The study recommends the application of other dimensionality reduction (DR) techniques such as linear discriminant analysis and non-negative matrix factorization for optimal feature analysis. The study suggests for further studies; the implementation of other parameter tuning with different hyperparameter optimization techniques for optimal model-building time.