Employing Deep Reinforcement Learning to Cyber-Attack Simulation for Enhancing Cybersecurity

• DOI: 10.3390/electronics13030555
• Publication Date: 2024-01-30T12:49:47Z
• Authors: Sang Ho Oh; Jeongyoon Kim; Jae Hoon Nah; Jongyoul Park
• Keywords: Cyber-attack

In the current landscape where cybersecurity threats are escalating in complexity and frequency, traditional defense mechanisms like rule-based firewalls signature-based detection proving inadequate. The dynamism sophistication of modern cyber-attacks necessitate advanced solutions that can evolve adapt real-time. Enter field deep reinforcement learning (DRL), a branch artificial intelligence has been effectively tackling complex decision-making problems across various domains, including cybersecurity. this study, we advance by implementing DRL framework to simulate cyber-attacks, drawing on authentic scenarios enhance realism applicability simulations. By meticulously adapting algorithms nuanced requirements contexts—such as custom reward structures actions, adversarial training, dynamic environments—we provide tailored approach significantly improves upon methods. Our research undertakes thorough comparative analysis three sophisticated algorithms—deep Q-network (DQN), actor–critic, proximal policy optimization (PPO)—against RL algorithm Q-learning, within controlled simulation environment reflective real-world cyber threats. findings striking: actor–critic not only outperformed its counterparts with success rate 0.78 but also demonstrated superior efficiency, requiring fewest iterations (171) complete an episode achieving highest average 4.8. comparison, DQN, PPO, Q-learning lagged slightly behind. These results underscore critical impact selecting most fitting for simulations, right choice leads more effective strategies. impressive performance study marks significant stride towards development adaptive, intelligent systems capable countering increasingly contributes robust model simulating provides scalable be adapted challenges.