DQL-based Wide-area Path Scheduling for Deterministic Time-delay Broadband Services in Power Communication Data Networks

A wide-area path scheduling algorithm based on deep reinforcement learning is proposed for the problem that the existing wide-area path scheduling methods lack delay sensitivity and reliability in the current deterministic delay broadband services of power communication data networks. The algorithm firstly integrates service delay, node bandwidth and link load balance degree to establish a quantitative evaluation model; secondly finds all alternative routing schemes satisfying multiple constraints for the source and destination nodes of each service by using a depth-first search algorithm; and finally solves the optimal primary and alternate routing schemes using the DQN algorithm. Comparative analysis through simulation experiments shows that when the algorithm converges and stabilizes, the algorithm can use less time to solve reasonable primary and alternate routing schemes in the face of multi-service scenarios.