Millimeter wave MIMO base station cooperation beam selection method based on wide learning

The invention provides a millimeter wave MIMO base station cooperation wave beam selection method based on wide learning, and aims at a downlink wave beam selection problem of a multipoint cooperation millimeter wave MIMO scene, a longitudinal federated learning framework is used for reference, and a mapping problem of an original centralized multi-base station uplink wide wave beam response and an optimal transmission narrow wave beam is solved through vertical cutting of a data feature space. The problem is converted into a distributed learning problem, and a specific base station cooperation distributed beam selection framework is designed. And the communication overhead of the forward link is reduced by mining the sparsity of the intermediate parameters in the training process. And an incremental updating mode of the local network of the base station in the cooperative mode is designed, so that the updating complexity of the network is effectively reduced. According to the method, the capability of mining