公开(公告)号：US20180278311A1

公开(公告)日：2018-09-27

申请号：US15991672

申请日：2018-05-29

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Jing QIU ,  Kin Nang LAU ,  Xiaona WANG

IPC: H04B7/06

CPC classification number: H04B7/0617 ,  H04B7/0626 ,  H04B7/0628 ,  H04B7/063 ,  H04B7/0695

Abstract: A communication beam determining method and a corresponding apparatus are disclosed. The method includes respectively sending, by a network side device, downlink sounding signals by using M beams with a first width, where main lobe directions of any two of the M beams are different; receiving, by the network side device, sounding results returned by user equipment (UE), and determining N beams with a second width based on the sounding results, where the second width is less than the first width, a coverage area of a set of the N beams is smaller than a coverage area of a set of the M beams, and M and N are integers and not less than 2; and respectively sending, by the network side device, downlink scanning signals by using the N beams, and determining, based on scanning results returned by the UE, a first beam for data transmission with the UE.

公开(公告)号：US20240185087A1

公开(公告)日：2024-06-06

申请号：US18404069

申请日：2024-01-04

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Mengyao MA ,  Kin Nang LAU ,  Liqun SU

IPC: G06N3/098

CPC classification number: G06N3/098

Abstract: An intelligent model training method and apparatus. A plurality of participating nodes jointly train an intelligent model. This method is performed by one of the plurality of participating nodes, and the method includes: performing a Kth time of model training on the intelligent model to obtain first gradient information; and sending first synthetic gradient information to a central node, where the first synthetic gradient information includes synthetic information of the first gradient information and residual gradient information, and the residual gradient information represents a residual estimate of synthetic gradient information that is not transmitted to the central node before the Kth time of model training, where K is a positive integer.