US Patent No. 10,716,089

DEPLOYMENT OF TRAINED NEURAL NETWORK BASED RSS FINGERPRINT DATASET


Patent No. 10,716,089
Issue Date July 14, 2020
Title Deployment Of Trained Neural Network Based Rss Fingerprint Dataset
Inventorship Sean Huberman, Guelph (CA)
Joshua Karon, Toronto (CA)
Henry L. Ohab, Toronto (CA)
Assignee MAPSTED CORP., Mississauga, Ontario (CA)

Claim of US Patent No. 10,716,089

1. A method, executed in a processor of a server computing device, of deploying received signal strength (RSS) fingerprint dataset, based on a trained neural network for mobile device indoor navigation, the method comprising:receiving RSS parameters from a plurality of mobile devices, the RSS parameters acquired for a set of positions within an indoor area, the RSS parameters being determined by the plurality of mobile devices using a respective RSS sensor device;
training a neural network implemented in the processor at lease in part based on the RSS parameters, the neural network comprising a first neural network layer corresponding to a set of RSS input parameters for a wireless signal in accordance with a first wireless communication protocol, and at least a second neural network layer corresponding to the set of RSS input parameters for the wireless signal in accordance with at least a second wireless communication protocol, an RSS input parameter being based on a postulated RSS model;
when a density of points represented by the set of positions having the RSS parameters exceeds a deployment threshold density, deploying the RSS fingerprint dataset within a fingerprint map, based on the trained neural network, the fingerprint map encompassing the set of positions; and
navigating another mobile device in the indoor area using the deployed RSS fingerprint dataset within the fingerprint map.