DEEP FUSION REASONING ENGINE (DFRE) FOR DYNAMIC AND EXPLAINABLE WIRELESS NETWORK QoE METRICS

Patent metadata

Field	Value
Publication number	US-2020022016-A1
Application number	US-201916365096-A
Country	US
Kind code	A1
Filing date	Mar 26, 2019
Priority date	Mar 27, 2018
Publication date	Jan 16, 2020
Grant date	—

Abstract

In one embodiment, a network quality assessment service that monitors a network obtains multimodal data indicative of a plurality of measurements from the network and subjective perceptions of the network by users of the network. The network quality assessment service uses the obtained multimodal data as input to one or more neural network-based models. The network quality assessment service maps, using a conceptual space, outputs of the one or more neural network-based models to symbols. The network quality assessment service applies a symbolic reasoning engine to the symbols, to generate a conclusion regarding the monitored network. The network quality assessment service provides an indication of the conclusion to a user interface.

First claim

What is claimed is:

1 . A method comprising:
obtaining, by a network quality assessment service that monitors a network, multimodal data indicative of a plurality of measurements from the network and subjective perceptions of the network by users of the network; using, by the network quality assessment service, the obtained multimodal data as input to one or more neural network-based models; mapping, by the network quality assessment service and using a conceptual space, outputs of the one or more neural network-based models to symbols; applying, by the network quality assessment service, a symbolic reasoning engine to the symbols, to generate a conclusion regarding the monitored network; and providing, by the network quality assessment service, an indication of the conclusion to a user interface.

2 . The method as in claim 1 , wherein the plurality of measurements from the network comprises one or more of: packet delays, packet drops, jitter, received signal strength indicator (RSSI), or Peak Signal to Noise Ratio (PSNR), and wherein the multimodal data indicative of the subjective perceptions of the network by the users of the network comprises one or more of: demographics information, subscriber churn rate, or user privacy information.

3 . The method as in claim 1 , wherein one or more neural network-based models comprise a deep learning model.

4 . The method as in claim 1 , wherein the conceptual space represents quality of experience (QoE) in the network as a concept comprising a plurality of domains.

5 . The method as in claim 1 , further comprising:
using a surprise detector to detect unexpected data in the obtained multimodal data; and, in response, automatically adjusting the one or more neural network-based models.

6 . The method as in claim 5 , wherein automatically adjusting the one or more neural network-based models comprises:
using the symbolic reasoning engine to determine that the one or more neural network-based models should be adjusted, based on the detected unexpected data in the obtained multimodal data.

7 . The method as in claim 5 , wherein automatically adjusting the one or more neural network-based models comprises:
adding a model for the unexpected data to the one or more neural network-based models.

8 . The method as in claim 1 , wherein the symbolic reasoning engine is non-axiomatic.

9 . An apparatus, comprising:
one or more network interfaces to communicate with a network; a processor coupled to the network interfaces and configured to execute one or more processes; and a memory configured to store a process executable by the processor, the process when executed configured to:
obtain multimodal data indicative of a plurality of measurements from the network and subjective perceptions of the network by users of the network;
use the obtained multimodal data as input to one or more neural network-based models;
map, using a conceptual space, outputs of the one or more neural network-based models to symbols;
apply a symbolic reasoning engine to the symbols, to generate a conclusion regarding the monitored network; and
is provide an indication of the conclusion to a user interface.

10 . The apparatus as in claim 9 , wherein the plurality of measurements comprises one or more of: packet delays, packet drops, jitter, received signal strength indicator (RSSI), or Peak Signal to Noise Ratio (PSNR).

11 . The apparatus as in claim 9 , wherein one or more neural network-based models comprise a deep learning model.

12 . The apparatus as in claim 9 , wherein the conceptual space represents quality of experience (QoE) in the network as a concept comprising a plurality of domains.

13 . The apparatus as in claim 9 , wherein the process when executed is further configured to:
use a surprise detector to detect unexpected data in the obtained multimodal data; and, in response, automatically adjust the one or more neural network-based models.

14 . The apparatus as in claim 13 , wherein the apparatus automatically adjusting the one or more neural network-based models by:
using the symbolic reasoning engine to determine that the one or more neural network-based models should be adjusted, based on the detected unexpected data in the obtained multimodal data.

15 . The apparatus as in claim 13 , wherein the apparatus automatically adjusting the one or more neural network-based models by:
adding a model for the unexpected data to the one or more neural network-based models.

16 . The apparatus as in claim 9 , wherein the symbolic reasoning engine is non-axiomatic.

17 . A tangible, non-transitory, computer-readable medium storing program instructions that cause a network quality assessment service that monitors a network to execute a process comprising:
obtaining, by the network quality assessment service, multimodal data indicative of a plurality of measurements from the network and subjective perceptions of the network by users of the network; using, by the network quality assessment service, the obtained multimodal data as input to one or more neural network-based models; mapping, by the network quality assessment service and using a conceptual space, outputs of the one or more neural network-based models to symbols; applying, by the network quality assessment service, a symbolic reasoning engine to the symbols, to generate a conclusion regarding the monitored network; and providing, by the network quality assessment service, an indication of the conclusion to a user interface.

18 . The computer-readable medium as in claim 17 , wherein one or more neural network-based models comprise a deep learning model.

19 . The computer-readable medium as in claim 17 , wherein the symbolic reasoning engine is non-axiomatic.

20 . The computer-readable medium as in claim 17 , wherein the plurality of measurements comprises one or more of: packet delays, packet drops, jitter, received signal strength indicator (RSSI), or Peak Signal to Noise Ratio (PSNR).

Assignees

• Cisco Tech Inc

Inventors

• Fenoglio Enzo
• Latapie Hugo M
• Gray Kenneth
• Rezig Sawsen
• Ward David Delano

Classifications

• H04L41/5009 Primary

  Determining service level performance parameters or violations of service level contracts, e.g. violations of agreed response time or mean time between failures [MTBF] · CPC title

• H04L41/22

  comprising specially adapted graphical user interfaces [GUI] · CPC title

• H04W28/0268 Primary

  using specific QoS parameters for wireless networks, e.g. QoS class identifier [QCI] or guaranteed bit rate [GBR] (negotiating SLA or negotiating QoS H04W28/24) · CPC title

• H04L41/16

  using machine learning or artificial intelligence · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title