Under the large-scale application of 5G-A and AI models, transmission networks face challenges such as bandwidth surges, latency sensitivity, and rising energy consumption, making traditional optimization methods difficult to adapt to dynamic environments. This paper focuses on the practice of AI in transmission optimization, analyzes the adaptation logic of machine learning, deep reinforcement learning and transmission network, and studies congestion control, resource scheduling, and collaborative reasoning from the perspective of congestion control, resource scheduling, and collaborative reasoning in combination with communication network, data center, and edge computing scenarios. Through case studies such as Singapore's M1 microwave transmission network and Shanghai Unicom's 5G-A intelligent scheduling system, verify the effectiveness of AI in improving bandwidth utilization, reducing latency, and optimizing energy consumption. Research has shown that AI dynamic optimization can increase bandwidth utilization by more than 40%, reduce end-to-end latency by about 30%, and save energy consumption by 18% -35%. It provides a feasible path for the intelligent upgrade of transmission
📖 افتح في inklap 🔗 DOI 📮 اطلب بحثاً