DESIGN OF JAMMING ATTACK DETECTION AND CLASSIFICATION MODEL ON WIRELESS NETWORK BASED ON MACHINE LEARNING

• Main Author: Rizki Utami, Citra
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/77945
• Institution: Institut Teknologi Bandung

Wireless network plays an important role in providing stable and uninterrupted service for data transmission process. The importance of network operation security, especially the data transmission process, really needs to be considered so that daily activities continue to run smoothly, for example business activities, education, banking, health, and so on. One of the disturbances in the data transmission process on a wireless network is a jamming attack. This disturbance/attack resulted in the loss of communication connection of telecommunications equipment, most of which were connected to the internet network and used in the data transmission process. This jamming attack blocks radio wave transmission signals that have the same frequency as the jammer. Almost all areas of life already depend on this wireless network, therefore, jamming attacks can cause information transmission failure which can be fatal. This attack also makes an area affected by jamming lose service to the internet network, Denial of Service (DOS). A lot of research has been done in developing jamming attack detection techniques and utilizing Machine Learning (ML) as the solution to support accurate and low latency detection. In this thesis, the development of jamming attack detection techniques will be carried out using wireless network metric values , PDR and RSS, and supervised learning algorithms such as K-Nearest Neighbors (KNN), decision tree, random forest, gradient boosting, adaptive boosting, naïve bayes, and Support Vector Machine (SVM). The result shows that hybrid model was selected as the best model with respective accuracy, precision and recall achievements of 96.8% on 30% of the testing dataset.