| CPC H04W 16/22 (2013.01) [G06N 3/02 (2013.01); G06N 3/08 (2013.01); H04B 17/318 (2015.01); H04B 17/3913 (2015.01); H04L 41/147 (2013.01); H04L 41/16 (2013.01); H04L 43/04 (2013.01); H04L 43/0888 (2013.01); H04W 8/22 (2013.01); H04W 24/02 (2013.01); H04W 24/10 (2013.01); H04W 84/02 (2013.01); H04W 88/18 (2013.01); H04W 92/02 (2013.01); H04W 92/04 (2013.01)] | 20 Claims |
|
1. An information handling system executing an intelligent throughput performance analysis and issue detection system, comprising:
a network interface device to establish a wireless link with a wireless network;
a processor to execute code instructions of a neural network of the intelligent throughput performance analysis and issue detection system trained to predict throughput values at the information handling system based on controlled connectivity testing metrics gathered in a controlled laboratory from a plurality of tested information handling systems that includes the information handling system;
the processor to execute code instructions of the intelligent throughput performance analysis and issue detection system to:
gather a measured throughput of the wireless link, and operational connectivity metrics for the information handling system that describe antenna positional information, antenna type, antenna gains, antenna adaptation controller parameters, system noise profiles, signal strength measurements, and wireless link performance metrics;
output from the trained neural network a first predicted throughput value based on the gathered operational connectivity metrics;
determine a difference exceeding a maximum tolerance between the first predicted throughput value and the measured throughput; and
the network interface device to transmit a notification of erroneously predicted throughput, values output from the neural network, and the operational connectivity metrics for the information handling system to a remote neural network training platform.
|
|
8. A method of training a neural network for an intelligent throughput performance analysis and issue detection system, comprising:
gathering, via a plurality of information handling systems transceiving data via wireless links in a controlled laboratory, and for each wireless link, a test measured throughput value and controlled connectivity testing metrics describing antenna positional information, antenna adaptation controller parameters, signal strength measurements, and wireless link performance metrics;
modeling a relationship between the controlled connectivity testing metrics and a test predicted throughput value for wireless links transceiving data according to the controlled connectivity testing metrics through a multi-layer, feed-forward neural network;
training the neural network by inputting the controlled connectivity testing metrics for each of the plurality of information handling systems and back-propagating to minimize a first error signal measuring a difference between the test predicted throughput value and the test measured throughput value; and
loading the trained neural network onto a first information handling system within the plurality of information handling systems for predicting throughput of a future-established wireless link between the first information handling system and a remote wireless network, based on input values gathered during transceiving of data via the future-established wireless link.
|