Streaming Service Optimization Based on Native Artificial Intelligence Architecture

5G-A network terminals will generate massive data. If the traditional AI training mode is used, the energy consumption of computing and storage will be increased. AI computing needs to evolve from external to endogenous, and mobile communication networks need to be deeply integrated with artificial intelligence. Native Artificial Intelligence will be the inevitable choice of future mobile communication networks. In the 5G-A network, the computing power of the base station is enhanced. A part of the computing power can be moved down to the base station, and the base station executes a service policy that is relatively complex and requires high timeliness. For example, on the NR TDD network, resource limitation and uplink delay increase may occur for large-bandwidth uplink services of live broadcast in the Internet. The intelligent endogenous architecture can be used to implement service identification, user classification, and hierarchical scheduling policies on the base station side, improving user experience.