In recent years, mobile services have developed rapidly and traditional satellite-terrestrial networks have been unable to support them. We are faced with the problems of how to locate mobile terminals accurately and process the data we collected quickly to reduce communication pressure. In order to solve this problem, this paper studies a pointing and tracking method based on artificial intelligence for mobile stations and terminals in satellite-terrestrial network, to make sure that our mobile stations and terminals can access best antenna signal and suffer minimal communication interference from other stations or terminals. An AI-based self learning (ASL) network framework is designed to support filtering and correct original sampling data, mobile tracking of mobile stations and terminals, and unsupervised satellite selection and antenna adjustment scheme. Deep learning of historical information data of stations and terminals to achieve real-time pointing and tracking, and predict the distribution of stations and terminals at some time in the future. Finally, the ASL is compared with existing systems to measure their functionality and usability. INDEX TERMS Satellite-terrestrial network, artificial intelligence, mobile tracking, deep learning, unsupervised self learning.

No abstract