BIER Multi-fault Recovery Mechanism For Multimedia Multicast In Edge Networks

In dynamic edge network multimedia scenarios, BIER (Bit Indexed Explicit Replication) multicast can effectively save bandwidth and improve network throughput. However, due to the high failure rate and highly dynamic nature of dynamic edge networks, it is necessary to consider the failure recovery mechanism of BIER to effectively cope with concurrent failures of multiple links and nodes, as well as to consider cascading failures and network congestion that may be caused by improper rerouting. Existing solutions are mainly based on ex ante backups, which cannot effectively cope with dynamic edge network multi-fault scenarios. In this paper, we propose an innovative solution based on Deep Reinforcement Learning (DRL) approach for rerouting the multi-failure problem, aiming at fast recovery from failures and optimizing network load balancing. The method employs the deep Q-learning algorithm, which incorporates the real-time load profile of the network and the multi-fault correlation. Simulation results validate the effectiveness of the scheme, demonstrating significant computational efficiency and performance advantages over Dijkstra’s algorithm, the ant colony algorithm, etc.