Device-to-device proximity service method based on reinforcement learning

The invention discloses a device-to-device proximity service method based on reinforcement learning. The method comprises steps that service request equipment transmits a service request signal, and the service request signal is transmitted to a mobile communication base station which cannot achieve the direct communication through proximity service providing equipment; the mobile communication base station scores and ranks the service providing devices with the service conditions according to the connection communication history of the service request device and all the service providing devices; the mobile communication base station selects a preset number of alternative service providing devices for the service request device according to the sorting result; when the service request device and the alternative service providing device move to be within the communication distance, communication is started, and services are provided; and if the two move beyond the communication distance, the service request e