Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

한국통신학회논문지

Secure Multi-Satellite Communications in LEO Networks via Multi-Agent Deep Reinforcement Learning

Yongjae Lee, Kyungmin Park, Taehoon Kim, Inkyu Bang

Published Mar 20, 2026• 0 citations
Publisher

Abstract

In this paper, we propose a multi-agent deep reinforcement learning (MADRL) strategy for adaptive beamforming and artificial noise (AN) transmission to enhance physical layer security in low Earth orbit (LEO) satellite networks. Multiple satellites are jointly scheduled to cooperatively transmit data and AN against potential eavesdroppers such as hostile unmanned aerial vehicles. In the proposed scheme, each satellite independently selects its transmission mode (idle, data, or AN) and the corresponding beamforming vector to maximize the secrecy rate within a centralized training decentralized execution (CTDE) framework using the soft actor-critic (SAC) algorithm. The MADRL agents are trained using only statistical channel information of the adversary instead of full instantaneous channel state information. Simulation results demonstrate that the proposed scheme achieves a higher secrecy rate than conventional baseline schemes.

DOI
10.7840/kics.2026.51.3.594

Related papers

Deep Learning-Based Secure Scheduling and Cooperative Artificial Noise Generation in LEO Satellite Networks

Yongjae Lee, Taehoon Kim, Inkyu Bang +2 more

As key enablers of 6G, low Earth orbit (LEO) satellite constellations are expected to offer global and low-latency connectivity. However, the broadcast nature and open-space propagation of satellite communications make it inherently vulnerable to eavesdropping. In dense LEO constellations where multiple satellites simultaneously cover overlapping regions, satellite scheduling can significantly contribute to secrecy performance. In this letter, we investigate a novel deep learning-based secure scheduling problem for physical layer security (PLS) in LEO satellite networks. During the scheduling process, we jointly consider multi-satellite transmission and cooperative artificial noise (AN) generation to defend against eavesdropping. The proposed deep learning-based scheduler, designed using attention-based permutation-invariant neural networks, learns security-oriented scheduling policies based solely on statistical channel information of eavesdroppers. Simulation results demonstrate that the proposed scheduling algorithm significantly enhances secrecy performance in terms of both secrecy rate and secrecy outage probability, outperforming conventional baseline schemes.

IEEE Wireless Communications Letters•Jan 1, 2026•1 citation

Learning-Assisted Secure Transmission Design for Low-Earth Orbit Satellites With High-Altitude Platform Jamming

Yadi Zhai, Zhi Lin, Yong Wang +9 more

Low-earth orbit (LEO) satellite systems provide global coverage and low latency, yet their rapid motion induces severe channel variations, rendering full-service-period physical layer security design prohibitively costly and inefficient. To address this issue, we investigate a high-altitude platform (HAP)-assisted secure transmission scheme, where the HAP transmits artificial noise (AN) to enhance satellite communication security and improve user fairness. Specifically, we formulate a minimum secrecy rate maximization problem with joint power allocation and the HAP beamforming vector as optimization variables. Due to the non-convexity nature of the problem, solving it for a single time slot is computationally intensive, and directly extending the optimization across all time slots incurs prohibitive overhead. Therefore, we propose a learning-based framework that first optimizes key time slots and then trains a kernel ridge regression (KRR) model. In particular, the number of training samples is adaptively adjusted based on the minimum distance between eavesdroppers and legitimate users, while a warm-start strategy is used to accelerate optimization over the sampled slots. The trained KRR model learns the mapping from satellite angular positions to secure transmission parameters and predicts suitable solutions for the remaining time slots. Simulation results show that the proposed method achieves secrecy and user fairness performance comparable to per-slot optimization with much lower computational complexity, offering a scalable and efficient solution for secure LEO satellite communications.

IEEE Open Journal of the Communications Society•Jan 1, 2026•0 citations

Artificial noise aided physical layer security for LEO satellite communications

Can Wang

Low Earth Orbit (LEO) satellite communications are vulnerable to eavesdropping over open wireless links. This paper proposes a simplified artificial noise (AN)-aided physical layer security scheme. A system model with a multi-antenna LEO satellite, a single-antenna ground user, and multiple eavesdroppers is built under shadowed-Rician fading and imperfect channel state information. AN is projected into the null space of the legitimate channel to avoid interference, and fixed power allocation is adopted for low complexity. Simulations show the scheme achieves better secrecy rate and secrecy outage probability than traditional methods, suitable for resource-constrained LEO satellite systems.

•Jan 1, 2026•0 citations

Secrecy Outage Probability of Secure Transmission with Artificial Noise in Low Earth Orbit Satellite Networks

Yongjae Lee, Taehoon Kim, Inkyu Bang

Satellite communication has become essential for achieving seamless global connectivity in the 5 G and the future 6G networks. However, preventing eavesdropping attacks on data transmission remains a challenge due to the broadcasting nature of radio signals. Physical layer security (PLS) is a promising approach that can offer a lightweight solution suitable for satellites with hardware constraints such as limited computing and processing power. Recent studies considering PLS techniques have investigated secrecy performance in various satellite communication environments such as non-orthogonal multiple access but multi-satellite scheduling problems have been less highlighted. Particularly, in this paper, we investigate secrecy outage probability in low Earth orbit satellite networks by jointly considering multi-satellite scheduling for data and artificial noise transmissions against eavesdropping attacks. We propose a novel multi-satellite scheduling scheme and derive a closed-form of secrecy outage probability over Nakagami-$m$fading channels. We verify our analysis and evaluate our proposed scheme compared to several baseline schemes through extensive simulations.

International Conference on Communications•Jun 8, 2025•3 citations

Artificial noise aided physical layer security for low earth orbit satellite communications

Can Wang

Low Earth Orbit (LEO) satellite communications are vulnerable to eavesdropping due to open wireless links. To address this issue, a simplified artificial noise (AN)-aided physical layer security scheme is proposed. A practical system model including a multi-antenna LEO satellite, a single-antenna ground user, and multiple eavesdroppers is established, considering shadowed Rician fading and imperfect channel state information (CSI). The AN is projected into the null space of the legitimate user’s channel to avoid interference, while suppressing eavesdroppers. A fixed power allocation strategy is adopted for easy implementation. Simulation results show that the proposed scheme outperforms traditional methods in secrecy rate and secrecy outage probability (SOP) with low computational complexity, making it suitable for resource-constrained LEO satellite systems.

•Jan 1, 2026•0 citations

Eavesdropping Game Based on Multi-Agent Deep Reinforcement Learning

Delin Guo, Lan Tang, Lüxi Yang +1 more

This paper considers an adversarial scenario between a legitimate eavesdropper and a suspicious communication pair. All three nodes are equipped with multiple antennas. The eavesdropper, which operates in a full-duplex model, aims to wiretap the dubious communication pair via proactive jamming. On the other hand, the suspicious transmitter, which can send artificial noise (AN) to disturb the wiretap channel, aims to guarantee secrecy. More specifically, the eavesdropper adjusts jamming power to enhance the wiretap rate, while the suspicious transmitter jointly adapts the transmit power and noise power against the eavesdropping. Considering the partial observation and complicated interactions between the eavesdropper and the suspicious pair in unknown system dynamics, we model the problem as an imperfect-information stochastic game. To approach the Nash equilibrium solution of the eavesdropping game, we develop a multi-agent reinforcement learning (MARL) algorithm, termed neural fictitious self-play with soft actor-critic (NFSP-SAC), by combining the fictitious self-play (FSP) with a deep reinforcement learning algorithm, SAC. The introduction of SAC enables FSP to handle the problems with continuous and high dimension observation and action space. The simulation results demonstrate that the power allocation policies learned by our method empirically converge to a Nash equilibrium, while the compared reinforcement learning algorithms suffer from severe fluctuations during the learning process.

IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC)•Jul 4, 2022•0 citations
In this paper, we propose a multi-agent deep reinforcement learning (MADRL) strategy for adaptive beamforming and artificial noise (AN) transmission to enhance physical layer security in low Earth orbit (LEO) satellite networks. Multiple satellites are jointly scheduled to cooperatively transmit data and AN against potential eavesdroppers such as hostile unmanned aerial vehicles. In the proposed scheme, each satellite independently selects its transmission mode (idle, data, or AN) and the corresponding beamforming vector to maximize the secrecy rate within a centralized training decentralized execution (CTDE) framework using the soft actor-critic (SAC) algorithm. The MADRL agents are trained using only statistical channel information of the adversary instead of full instantaneous channel state information. Simulation results demonstrate that the proposed scheme achieves a higher secrecy rate than conventional baseline schemes.
As key enablers of 6G, low Earth orbit (LEO) satellite constellations are expected to offer global and low-latency connectivity. However, the broadcast nature and open-space propagation of satellite communications make it inherently vulnerable to eavesdropping. In dense LEO constellations where multiple satellites simultaneously cover overlapping regions, satellite scheduling can significantly contribute to secrecy performance. In this letter, we investigate a novel deep learning-based secure scheduling problem for physical layer security (PLS) in LEO satellite networks. During the scheduling process, we jointly consider multi-satellite transmission and cooperative artificial noise (AN) generation to defend against eavesdropping. The proposed deep learning-based scheduler, designed using attention-based permutation-invariant neural networks, learns security-oriented scheduling policies based solely on statistical channel information of eavesdroppers. Simulation results demonstrate that the proposed scheduling algorithm significantly enhances secrecy performance in terms of both secrecy rate and secrecy outage probability, outperforming conventional baseline schemes.
Low-earth orbit (LEO) satellite systems provide global coverage and low latency, yet their rapid motion induces severe channel variations, rendering full-service-period physical layer security design prohibitively costly and inefficient. To address this issue, we investigate a high-altitude platform (HAP)-assisted secure transmission scheme, where the HAP transmits artificial noise (AN) to enhance satellite communication security and improve user fairness. Specifically, we formulate a minimum secrecy rate maximization problem with joint power allocation and the HAP beamforming vector as optimization variables. Due to the non-convexity nature of the problem, solving it for a single time slot is computationally intensive, and directly extending the optimization across all time slots incurs prohibitive overhead. Therefore, we propose a learning-based framework that first optimizes key time slots and then trains a kernel ridge regression (KRR) model. In particular, the number of training samples is adaptively adjusted based on the minimum distance between eavesdroppers and legitimate users, while a warm-start strategy is used to accelerate optimization over the sampled slots. The trained KRR model learns the mapping from satellite angular positions to secure transmission parameters and predicts suitable solutions for the remaining time slots. Simulation results show that the proposed method achieves secrecy and user fairness performance comparable to per-slot optimization with much lower computational complexity, offering a scalable and efficient solution for secure LEO satellite communications.
Low Earth Orbit (LEO) satellite communications are vulnerable to eavesdropping over open wireless links. This paper proposes a simplified artificial noise (AN)-aided physical layer security scheme. A system model with a multi-antenna LEO satellite, a single-antenna ground user, and multiple eavesdroppers is built under shadowed-Rician fading and imperfect channel state information. AN is projected into the null space of the legitimate channel to avoid interference, and fixed power allocation is adopted for low complexity. Simulations show the scheme achieves better secrecy rate and secrecy outage probability than traditional methods, suitable for resource-constrained LEO satellite systems.
Satellite communication has become essential for achieving seamless global connectivity in the 5 G and the future 6G networks. However, preventing eavesdropping attacks on data transmission remains a challenge due to the broadcasting nature of radio signals. Physical layer security (PLS) is a promising approach that can offer a lightweight solution suitable for satellites with hardware constraints such as limited computing and processing power. Recent studies considering PLS techniques have investigated secrecy performance in various satellite communication environments such as non-orthogonal multiple access but multi-satellite scheduling problems have been less highlighted. Particularly, in this paper, we investigate secrecy outage probability in low Earth orbit satellite networks by jointly considering multi-satellite scheduling for data and artificial noise transmissions against eavesdropping attacks. We propose a novel multi-satellite scheduling scheme and derive a closed-form of secrecy outage probability over Nakagami-mmmfading channels. We verify our analysis and evaluate our proposed scheme compared to several baseline schemes through extensive simulations.
Low Earth Orbit (LEO) satellite communications are vulnerable to eavesdropping due to open wireless links. To address this issue, a simplified artificial noise (AN)-aided physical layer security scheme is proposed. A practical system model including a multi-antenna LEO satellite, a single-antenna ground user, and multiple eavesdroppers is established, considering shadowed Rician fading and imperfect channel state information (CSI). The AN is projected into the null space of the legitimate user’s channel to avoid interference, while suppressing eavesdroppers. A fixed power allocation strategy is adopted for easy implementation. Simulation results show that the proposed scheme outperforms traditional methods in secrecy rate and secrecy outage probability (SOP) with low computational complexity, making it suitable for resource-constrained LEO satellite systems.
This paper considers an adversarial scenario between a legitimate eavesdropper and a suspicious communication pair. All three nodes are equipped with multiple antennas. The eavesdropper, which operates in a full-duplex model, aims to wiretap the dubious communication pair via proactive jamming. On the other hand, the suspicious transmitter, which can send artificial noise (AN) to disturb the wiretap channel, aims to guarantee secrecy. More specifically, the eavesdropper adjusts jamming power to enhance the wiretap rate, while the suspicious transmitter jointly adapts the transmit power and noise power against the eavesdropping. Considering the partial observation and complicated interactions between the eavesdropper and the suspicious pair in unknown system dynamics, we model the problem as an imperfect-information stochastic game. To approach the Nash equilibrium solution of the eavesdropping game, we develop a multi-agent reinforcement learning (MARL) algorithm, termed neural fictitious self-play with soft actor-critic (NFSP-SAC), by combining the fictitious self-play (FSP) with a deep reinforcement learning algorithm, SAC. The introduction of SAC enables FSP to handle the problems with continuous and high dimension observation and action space. The simulation results demonstrate that the power allocation policies learned by our method empirically converge to a Nash equilibrium, while the compared reinforcement learning algorithms suffer from severe fluctuations during the learning process.