TNDRM: Data Normalization Using Uncovering Complex Data Structures With Topological Nonlinear Dimensionality Reduction Using Manifold Learning

Abstract

With the increasing complexity of Distributed Denial of Service (DDOS) attacks in Wireless Sensor Networks (WSNs), the accurate detection of these threats has become imperative. This research presents a robust preprocessing technique for DDOS attack detection, focusing on data normalization through the integration of topological nonlinear dimensionality reduction via manifold learning (TNDRM). Our methodology revolves around transforming the intricate high-dimensional feature space of WSN data into a lower-dimensional representation, all while preserving the intrinsic topology and geometry of the original data. Achieved through manifold learning techniques, this process enables a more meaningful understanding of complex data structures, essential for effective analysis. A pivotal step in our approach involves the normalization of the data within the reduced dimensional space. Leveraging statistical techniques, specifically z-score normalization and min-max scaling, we mitigate the impact of varying scales and outliers in the data. The accuracy of machine learning algorithms is improved with normalization because it guarantees uniformity and consistency by removing aberrations.