강화학습을 이용한 무인기 기만 경로 유도 시뮬레이션
A UAV Spoofing Path Guided Simulation Using Reinforcement Learning
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
In preparation for cases where Unmanned Aerial Vehicles (UAVs) are abused for surveillance or terrorism, this study proposes a technique to guide a UAV to a target point using a spoofing signal that interferes with the Global Navigation Satellite System (GNSS). However, the waypoint estimation-based approach used in spoofing requires repetitive computations, making realtime processing challenging and reducing its responsiveness to target point changes. This paper proposes a technique that uses reinforcement learning to guide UAV spoofing paths in real time by dynamically learning and adapting to changes in flight states without the need for waypoint estimation. To effectively learn real-time flight state change data, the Advantage Actor-Critic (A2C) reinforcement learning model is utilized. In the simulation, a spoofing path guided simulation that controls flight in real time through reinforcement learning was developed. The proposed reinforcement learning model was applied, and the reinforcement learning model was verified through a simulation experiment where the target point of guided spoofing was changed.
Abstract
In preparation for cases where Unmanned Aerial Vehicles (UAVs) are abused for surveillance or terrorism, this study proposes a technique to guide a UAV to a target point using a spoofing signal that interferes with the Global Navigation Satellite System (GNSS). However, the waypoint estimation-based approach used in spoofing requires repetitive computations, making realtime processing challenging and reducing its responsiveness to target point changes. This paper proposes a technique that uses reinforcement learning to guide UAV spoofing paths in real time by dynamically learning and adapting to changes in flight states without the need for waypoint estimation. To effectively learn real-time flight state change data, the Advantage Actor-Critic (A2C) reinforcement learning model is utilized. In the simulation, a spoofing path guided simulation that controls flight in real time through reinforcement learning was developed. The proposed reinforcement learning model was applied, and the reinforcement learning model was verified through a simulation experiment where the target point of guided spoofing was changed.
참고문헌 (9)
- Chae, M. -H., "Commercial Fixed-Wing Drone Redirection System Using GNSS Deception", IEEE Transactions on Aerospace and Electronic Systems, 59, pp.5699~5713 (2023)[DOI]
- Choi, S. -H., "Recent R&D Trends of Anti-Drone Technologies", Electronics and Telecommunications Trends, 33, pp.78~88 (2018)[DOI]
- Duan, Y., "Benchmarking Deep Reinforcement Learning for Continuous Control" (2016)[DOI]
- Haarnoja, T., "Soft Actor-critic: Off-policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor" (2018)[DOI]
- Lee, J. -S., "Simulation of GPS spoofing technique for estimating the destination and guiding of the drone", Proceedings of Symposium of the Korean Institute of Communications and Information Sciences, pp.239~240 (2019)
- Mnih, V., "Playing Atari with Deep Reinforcement Learning", NIPS Deep Learning Workshop (2013)
- Moon, I. -C., "Modeling and Simulation on One-vs-One Air Combat with Deep Reinforcement Learning", JKSS, 29, pp.39~46 (2020)[DOI]
- Ruder, S, "An overview of gradient descent optimization algorithms" (2017)[DOI]
- Sutton, R. S., "Reinforcement Learning:An Introduction", pp.328~333 (2018), The MIT Press