Network anomaly detection and network performance status determination

Patent metadata

Field	Value
Publication number	US-10841817-B2
Application number	US-201916715458-A
Country	US
Kind code	B2
Filing date	Dec 16, 2019
Priority date	Jan 26, 2018
Publication date	Nov 17, 2020
Grant date	Nov 17, 2020

Abstract

A system may collect, from a wireless network, first data pertaining to nodes in the wireless network. Each datum of the first data belongs to one of two or more categories/For each of the nodes, for each of the categories, and for each datum belonging to the category, the system may determine if the datum is outside of a first range of values, and if the datum is inside the first range, the system may calculate a first base network performance health (NPH) score that is a function of the nodes, the categories, the data, and time. The system may also apply first deep learning to a first neural network among a plurality of neural networks to update first coefficients for correlating the first base NPH score to a mean opinion score, for each of the categories.

First claim

What is claimed is:

1. A device comprising:
a communication interface;
memory to store instructions;
one or more processors to:
collect, from a wireless network and through the communication interface, first data pertaining to nodes in the wireless network;
perform, for each datum of the first data, learning by:
calculating a first base Network Performance Health (NPH) score; and
applying first deep learning to a first neural network among a plurality of neural networks to update first coefficients for correlating the first base NPH score to a mean opinion scores, wherein each datum of the first data belongs to one of a plurality of categories, and wherein the categories include at least one of service availability, service integrity, service accessibility, mobility, or retainability;

obtain second base NPH scores by using the first coefficients, wherein the second base NPH scores indicate user Quality of Experience (QoE) with the wireless network;
obtain a value of an operating parameter of one of the nodes based the first NPH scores and the second NPH scores; and
send, through the communication interface, the value of the operating parameter to the one of the nodes,

wherein the one of the nodes is configured to set the operating parameter to the value to optimize operation of the wireless network.

2. The device of claim 1 , wherein the first data includes a series of key performance indicator values.

3. The device of claim 1 , wherein the one node includes a base station.

4. The device of claim 1 , wherein when the one or more processors calculate the first base NPH score, the one or more processors are to:
determine if the datum is inside a range where at least about 50% of the first data is found; and
set a first base NPH score, among the first base NPH scores, to a predetermined value if the datum is not in the range.

5. The device of claim 1 , wherein the plurality of neural networks include convolutional neural networks.

6. The device of claim 1 , wherein each datum of the first data is associated with one of a plurality of time intervals, and wherein each of the time intervals is about one hour to 24 hours.

7. The device of claim 1 , wherein when the one or more processors perform learning, the one or more processors are to:
determine a cumulative distribution function (CDF) for the first data;
obtain an area under a curve for the CDF over a first range; and
normalize the area under the curve by dividing the area under the curve by an area under the curve of the CDF computed over a second range.

8. The device of claim 1 , wherein when the one or more processors calculate the first NPH score, the one or more processors are to:
determine whether increasing a value of the datum improves performance of the wireless network or decreasing the value of the datum improves performance of the wireless network.

9. The device of claim 1 , wherein the first data includes a series of key performance indicator values, and wherein the one node includes a base station.

10. The device of claim 1 , wherein the one or more processors are further to:
determine a third base NPH score for a geographical area based on base NPH scores that are functions of the nodes, the data, and time.

11. The device of claim 10 , wherein when the one or more processors determine the third base NPH score, the one or more processors are to:
determine a subset of nodes, among the nodes, that intersect a geographical area, and
sum normalized values of third base NPH scores over the subset of nodes to obtain the third base NPH score.

12. A method comprising:
collecting, from a wireless network, first data pertaining to nodes in the wireless network;
performing learning, for each datum of the first data, by:
calculating a first base Network Performance Health (NPH) score; and
applying first deep learning to a first neural network among a plurality of neural networks to update first coefficients for correlating the first base NPH score to a mean opinion scores, wherein each datum of the first data belongs to one of a plurality of categories, and wherein the categories include at least one of service availability, service integrity, service accessibility, mobility, or retainability;

obtaining second base NPH scores by using the first coefficients, wherein the second base NPH scores indicate user Quality of Experience (QoE) with the wireless network;
obtaining a value of an operating parameter of one of the nodes based the first NPH scores and the second NPH scores; and
sending the value of the operating parameter to the one of the nodes,
wherein the node is configured to set the operating parameter to the value to optimize operation of the wireless network.

13. The method of claim 12 , wherein the first data includes a series of key performance indicator values.

14. The method of claim 12 , wherein the one node includes a base station.

15. The method of claim 12 , wherein calculating the first base NPH scores includes:
determining if the datum is inside a range where at least about 50% of the first data is found;
and setting a first base NPH score, among the first base NPH scores, to a predetermined value if the datum is not in the range.

16. The method claim 12 , wherein the plurality of neural networks include convolutional neural networks.

17. The method of claim 12 , wherein each datum of the first data is associated with one of time intervals, and wherein each of the time intervals is about one hour to 24 hours.

18. The method of claim 12 , wherein the performing includes:
determining a cumulative distribution function (CDF) for the first data;
obtain an area under a curve for the CDF over a first range; and
normalize the area under the curve by dividing the area under the curve by an area under the curve of the CDF computed over a second range.

19. The method of claim 12 , wherein calculating the first NPH score includes:
determining whether increasing a value of the datum improves performance of the wireless network or decreasing the value of the datum improves performance of the wireless network.

20. The method of claim 12 , wherein the first data includes a series of key performance indicator values, and wherein the one node includes a base station.

Assignees

• Verizon Patent & Licensing Inc

Inventors

• Ouyang Ye
• Su Le
• Iyer Krishna Pichumani
• Schmidt Christopher M
• Lu Wenyuan
• Pola Shaun Robert
• Shah Maulik

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• H04W24/02 Primary

  Arrangements for optimising operational condition · CPC title

• H04W84/042

  Public Land Mobile systems, e.g. cellular systems · CPC title