THE ARCHITECTURE DEVELOPMENT OF MACHINE LEARNING-BASED ANOMALY DETECTION DASHBOARD FOR NETWORK MANAGEMENT SYSTEM (NMS)

• Main Author: Putranto Arimurti, Garry
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/55257
• Institution: Institut Teknologi Bandung

The data from Ministry of Communication and Information stated that there are at least 12.548 villages and sub-districts in Indonesia without reliable 4G coverage. Because of that, the collaboration of government and ISP is needed in order to ensure that those regions can receive similar level of services as other regions. To do so, government needs a solution that can be used to monitor the performance of ISP in providing service for the regions which are the focus of this collaboration. Right now, the government uses a NMS solution which only shows the uptime of the device. Hopefully, there will be developed a system that is able to collect the monthly performance metrics of the ISP or their devices and report the collected metrics as an evaluation material. In this final project, the system which is able to detect anomaly by identifying known cases on previous research and clustering new cases by their characteristics using ML is being developed. The results of the identification are then visualized as a dashboard which contains uptime graph, table of cases, graph of cases, and other metrics needed to evaluate the devices’ performance. The dashboard will then be deployed in a cloud computing server so users who want to monitor a specific ISP or device can access it easily on the internet.