Securing IoT-Cloud Applications with AQ-KGMO-DMG Enhanced SVM for Intrusion Detection

Konduru Siva Naga Narasimharao, P. V. Lakshmi — Fri, 28 Feb 2025 00:00:00 +0700

In contemporary society, the Internet has evolved into an indispensable facet of daily life, serving myriad functions across various domains. Intrusion detection, as a cornerstone of information security, plays a pivotal role in fortifying networks against potential threats, emphasizing the necessity for robust and reliable methods capable of discerning and mitigating network vulnerabilities effectively. In this work, a pioneering network intrusion detection model is introduced, leveraging Adaptive Quantum-Inspired KGMO with Dynamic Molecular Grouping (AQ-KGMO-DMG) for feature selection and employing Simplified Support Vector Machines (SVM) for the classification of intrusion data. The utilization of the UNSW-NB15 dataset serves as the litmus test for evaluating the efficacy of the developed intrusion detection model. Notably, this approach enhances the accuracy in categorizing classes with minimal instances while concurrently mitigating the false alarm rate (FAR). A notable innovation in this methodology involves the transformation of raw traffic vector data into a visual representation, thereby reducing computational costs significantly. To reduce the computation cost further, the raw traffic vector data is converted into picture format. The experimental findings showed that the proposed model performed better than conventional techniques in terms of FAR, accuracy, and computation cost.

Enhancing Security of Databases through Anomaly Detection in Structured Workloads

Fri, 28 Feb 2025 00:00:00 +0700

In today’s world, the protection of databases in any global organization has become paramount due to the rapid growth of data and the new generations of cyber threats. This highlights the need for more enhanced security precautions to secure these databases containing sensitive information. One of the most advanced ways of enhancing database security is using an anomaly detection system, especially for structured workloads. Structured workloads typically exhibit predictable patterns of data access and usage, making them susceptible to displaying anomalies that may indicate unauthorized access, data manipulation, or other security breaches. Anomaly detection methods can identify patterns that are unusual, an indication of malicious activity, or a data security breach. The present research utilized the Isolation Forest algorithm to detect outliers in high-dimensional data sets. The main contribution and novelty of this research lies in leveraging the Isolation Forest algorithm for structured database workloads to proactively identify and mitigate potential security threats. Our study showed that the proposed model, with an accuracy of 85%, outperformed various state-of-the-art methods. Furthermore, anomaly detection systems powered by advanced algorithms and machine learning enable real-time database activities analysis, addressing challenges like preprocessing, model training and scalability.

Journal of ICT Research and Applications

Securing IoT-Cloud Applications with AQ-KGMO-DMG Enhanced SVM for Intrusion Detection

Enhancing Security of Databases through Anomaly Detection in Structured Workloads