US 11,929,912 B2
Method and system for generating network configurations using graph neural network
Mahmoud Mohamed Bahnasy, Shenzhen (CN); Fenglin Li, Shenzhen (CN); Shihan Xiao, Shenzhen (CN); Xiangle Cheng, Shenzhen (CN); and Majid Ghaderi, Shenzhen (CN)
Assigned to HUAWEI TECHNOLOGIES CO., LTD., Shenzhen (CN)
Filed by Huawei Technologies Co., Ltd., Shenzhen (CN)
Filed on Dec. 2, 2022, as Appl. No. 18/074,319.
Application 18/074,319 is a continuation of application No. PCT/CN2020/094058, filed on Jun. 3, 2020.
Prior Publication US 2023/0124663 A1, Apr. 20, 2023
Int. Cl. H04L 45/02 (2022.01); H04L 45/00 (2022.01)
CPC H04L 45/08 (2013.01) [H04L 45/14 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A system comprising:
a processor; and
a non-transient memory storing instructions which, when executed by the processor, cause the system to:
receive a first matrix M representative of a network having a plurality of routers, wherein each element Mij in M indicates a connectivity state for data traffic between two routers from the plurality of routers based on a set of network requirements;
store in the non-transient memory, a graph neural network (GNN) comprising a plurality of nodes v and a plurality of edges, wherein each of the plurality of nodes corresponds to a respective router from the plurality of routers, and each of the plurality of edges corresponds to a physical link connection between two or more of the plurality of routers; and
train the GNN, based on the first matrix M and a second matrix Xv having a plurality of elements, to generate an output matrix Ov, wherein Ov includes a plurality of nodes labelled based on the first matrix M, wherein training the GNN comprises:
initializing the GNN based on the second matrix Xv;
aggregating neighbourhood information of each node from the plurality of nodes v of the GNN;
calculating a hidden state of each node from the plurality of nodes v of the GNN based on the respective neighbourhood information; and
generating the Ov based on the hidden state of each node from the plurality of nodes v of the GNN.