Research

Development of a Platform for Storing LiDAR Sensor Data Using Distributed Ledger Technology

In the development of autonomous vehicles, 3D map information is used for path planning. However, if the terrain changes due to construction or disasters, the map information may become outdated, potentially hindering the operation of autonomous vehicles. To solve this issue, we developed a platform for storing and updating LiDAR sensor data on a distributed ledger in real-time as part of a commissioned research project by the National Institute of Information and Communications Technology (NICT). The project has received the highest evaluation ranking from NICT in 2023.

This platform guarantees the integrity of data by storing the LiDAR data on a distributed ledger, ensuring that the data is not tampered with. Since data for autonomous driving is directly related to human lives, the reliability of the data is crucial. By using distributed ledger technology, we can prevent data tampering and enhance the reliability of the data. Additionally, to further improve data reliability, we developed a system in which nodes autonomously proliferate to run the distributed ledger at each intersection.

One remaining challenge is to see how accurately 3D map information can be reconstructed when real-time LiDAR sensor data from moving vehicles is stored on the distributed ledger.

Research on De-anonymization of Bitcoin Transactions Used for Ransomware Payments Using Llama2

Lab: The University of Queensland, UQ Cyber

Mentor: Prof. Ryan Ko

With the recent rise of cryptocurrency, the damage caused by ransomware has been increasing. Ransomware is a type of malware that infects computers, encrypts data, and demands a decryption key. Victims are required to make payments to obtain the decryption key, and due to its high level of anonymity, Bitcoin is often used for these payments. Existing research has focused on identifying transactions used for ransomware payments by analyzing Bitcoin transactions. However, due to the high level of anonymity in Bitcoin transactions, it is challenging to identify them. In this research, we proposed a method that leverages large language models to aggregate external information and de-anonymize Bitcoin transactions.

The research results were presented at an internal research symposium and nominated for the Best Thesis Award.

Development of an Unauthorized Access Detection System Using Windows Event Logs

To improve the information security of companies, I was responsible for developing an unauthorized access detection system using Windows Event Logs. This system collects information recorded in the Windows Event Logs and detects abnormal access. In this system development, I developed an agent for collecting Windows Event Logs from Windows servers provided on AWS. This system is now being offered to companies and has been adopted by many enterprises.