With the increasing prevalence of cyber attacks, it is crucial to develop effective strategies for defending against these threats. In this paper, we investigate the use of Q-learning, a popular reinforcement learning algorithm, for enhancing network security. We explore the development of optimal strategies for defending against cyber attacks by defining the state, action, and reward, and applying Q-learning to learn an optimal policy. Our experiments show that Q-learning can effectively learn an optimal strategy for network defense, and can outperform traditional rule-based approaches in certain scenarios. We also discuss the challenges and limitations of using Q-learning for network security, and suggest future research directions.