Could not download file: This paper is available to authorised users only.


L. Gotsev, B. Jekov, E. Kovatcheva , R. Nikolov, I. Barzev, E. Shoikova

University of Library Studies and Information Technologies (BULGARIA)
Technological advances have resulted in new paradigms and increasingly powerful tools for exploring cybersecurity data science, but much less attention has been directed at methods and strategies used to teach. The process of acquiring knowledge and skills within learning processes should result in applicable knowledge and integrated skills and abilities in a real-world context. Cyber threats of the Internet of Things (IoT) are growing at an explosive pace making the existing security measures inadequate. Machine Learning (ML) algorithms can be employed to improve IoT security. Based on the existing knowledge of cyber-threats, ML algorithms can analyse network traffic, update threat knowledge databases, and keep the underlying systems protected from new attacks.
The paper's primary aim is developing a case study to demonstrate the integration of ML algorithms for IoT security analytics as-a-process. This technological demo supports an active project-based learning through experimental lab sessions which occur over weeks in the infrastructure of the University Data Science CoE. The virtual lab is open for students, researchers, and practitioners to learn, test, refine, upgrade and apply models on real-world data-intensive projects in an interactive environment. The project-based learning is associated with a range of learner outcomes including conceptual knowledge, problem solving skills, and motivation.
The case study addresses and illustrates the challenges of security processes implementation to the existing IoT network infrastructure and excel skill-building in cybersecurity for Master's students. There is an in-depth investigation of real-world IoT network traffic and manipulation of authentic attacks. Machine Learning algorithms are used to produce accurate outputs from large complex databases, where the generated outputs can be used to predict and detect vulnerabilities in IoT-based systems. The learning progress is achieved through questioning experts, making observations, making connections, and demonstrating learning to others.
In Section 1, а summary of research efforts, addressing innovative learning approaches and security issues using ML algorithms in the IoT domain are discussed. Overview of several IoT security threats that affect both data integrity and network availability and their categorization are introduced, such as Denial of Service, Man-in-the-middle and Malware. In the following Section 2, we discuss the usage of ML algorithms as a solution. In Section 3, a comparative analysis of the accuracy of the several ML algorithms that have proven extremely helpful in mitigating security in IoT domains such as Random Forest, Naive Bayes, Multi-Layer Perceptron and Deep Learning is shown. ML is accomplished by conducting experiments with the public available IoT-23 dataset, which contains labelled information of both malicious and benign IoT network traffic. The Zeek as an open-source tool is used to capture the traffic for the IoT-23 dataset creation. This dataset is pre-processed to be suitable for experiments done in a Python environment while using the Scikit-learn library of Python.
In conclusion, as education is currently undergoing significant change brought about by emerging reform in pedagogy and technology, efforts have sought to close the gap between technologies as educational additive to effective integration as a means to promote and cultivate student centred, inquiry based and project based learning.