A two-stage Q-learning routing approach for quantum entanglement networks

Abstract

The emerging field of quantum internet offers multiple applications, enabling quantum communication across diverse networks. However, the current entanglement networks exhibit complex processes, characterized by variable entanglement generation rates, limited quantum memory capacity, and susceptibility to decoherence rates. Addressing these issues, we propose a two-stage routing system that harnesses the power of reinforcement learning (RL). The first stage focuses on identifying the most efficient routes for quantum data transmission. The second stage concentrates on establishing these routes and improving how and when to apply entanglement swapping and purification. Our extensive evaluations across various network sizes and configurations reveal that our method not only sustains superior end-to-end route fidelity but also achieves significantly higher request success rates compared to traditional methods. These findings highlight the efficacy of our approach in managing the complex dynamics of quantum networks, ensuring robust and scalable quantum communication. Our method’s adaptability to changing network conditions and its proactive management of quantum resources make an important contribution to quantum network efficiency.