Multimodal Multi-Objective Network Routing Optimization

This paper proposes a routing optimization approach that integrates genetic algorithms (GA) for multi-modal multi-objective optimization problems (MMOPs) with reinforcement learning (RL) to enhance network routing efficiency and throughput while reducing latency and packet loss rates. Our method employs genetic algorithms to generate multiple high-quality paths for each node pair, forming an action space for reinforcement learning. This accelerates the RL learning process and increases the flexibility of actions. Through the adaptability of RL, the routing strategy is further optimized to improve routing efficiency and guarantee network service quality.