US 11,907,765 B2
Fog computing systems and methods
Juan Luís Pérez Rico, Barcelona (ES); Alberto Gutiérrez Torre, Barcelona (ES); Josep Lluís Berral García, Barcelona (ES); and David Carrera Perez, Barcelona (ES)
Assigned to BARCELONA SUPERCOMPUTING CENTER—CENTRO NACIONAL DE SUPERCOMPUTACIÓN, Barcelona (ES); and UNIVERSITAT POLITÈCNICA DE CATALUNYA, Barcelona (ES)
Appl. No. 17/050,368
Filed by BARCELONA SUPERCOMPUTING CENTER—CENTRO NACIONAL DE SUPERCOMPUTACIÓN, Barcelona (ES); and UNIVERSITAT POLITÈCNICA DE CATALUNYA, Barcelona (ES)
PCT Filed Jul. 6, 2018, PCT No. PCT/EP2018/068367
§ 371(c)(1), (2) Date Oct. 23, 2020,
PCT Pub. No. WO2019/206436, PCT Pub. Date Oct. 31, 2019.
Claims priority of application No. 18382285 (EP), filed on Apr. 26, 2018.
Prior Publication US 2021/0117758 A1, Apr. 22, 2021
Int. Cl. G06N 3/08 (2023.01); G06F 9/50 (2006.01); G06N 3/063 (2023.01); H04L 67/1087 (2022.01); H04L 67/125 (2022.01); G06F 18/214 (2023.01); G06N 3/045 (2023.01); G06N 3/044 (2023.01)
CPC G06F 9/5072 (2013.01) [G06F 18/214 (2023.01); G06N 3/044 (2023.01); G06N 3/045 (2023.01); G06N 3/063 (2013.01); G06N 3/08 (2013.01); H04L 67/1089 (2013.01); H04L 67/125 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A fog computing system comprising a plurality of fog nodes including edge nodes at an edge or bottom level, and middle nodes at one or more middle levels intermediating between the edge nodes and a cloud node at a cloud or top level, wherein
the cloud and middle nodes are non-bottom level nodes, the middle and edge nodes are non-top level nodes, and said non-bottom and non-top nodes form a hierarchical structure with each of the non-bottom nodes having children nodes at a next lower level, and each of the non-top nodes having a parent node at a next higher level; wherein
each of the edge nodes is configured to perform the following functions: receiving sensor data produced by sensors, assigning a reception timestamp to each of said sensor data indicating a time of receipt to produce a corresponding series of timestamp-ordered sensor data, training a first local model through a machine learning based on said series of timestamp-ordered sensor data, and sending said series of timestamp-ordered sensor data to the parent node of the edge node; wherein
each of the middle nodes is configured to perform the following functions: collecting a series of timestamp-ordered sensor data from the children nodes of the middle node to obtain a collected series of timestamp-ordered sensor data, training a first supra-local model through the machine learning based on the collected series of timestamp-ordered sensor data, and sending the collected series of timestamp-ordered sensor data to the parent node of the middle node.