Predictive networking architecture for next-generation multiservice, multicarrier WANs

Patent metadata

Field	Value
Publication number	US-9749188-B2
Application number	US-201414276431-A
Country	US
Kind code	B2
Filing date	May 13, 2014
Priority date	May 13, 2014
Publication date	Aug 29, 2017
Grant date	Aug 29, 2017

Abstract

In one embodiment, network traffic data is received regarding traffic flowing through one or more routers in a network. A future traffic profile through the one or more routers is predicted by modeling the network traffic data. Network condition data for the network is received and future network performance is predicted by modeling the network condition data. A behavior of the network is adjusted based on the predicted future traffic profile and on the predicted network performance.

First claim

What is claimed is:

1. A method, comprising:
receiving, at a device, network traffic data regarding traffic flowing through one or more routers in a network;
predicting, by a first learning machine executing on the device, a future traffic profile through the one or more routers by modeling the network traffic data;
receiving, at the device, network condition data for the network, wherein the network condition data and the network traffic data are different;
predicting, by a second learning machine executing on the device, future network performance by modeling the network condition data;
inputting into a third learning machine of a Predictive Control Manager (PCM) the network traffic modeling data from the first learning machine and the network condition modeling data from the second learning machine; and
adjusting, by a closed-loop controller on the PCM, a behavior of the network based on an output from the third learning machine, wherein the output from the third learning machine is based on the predicted future traffic profile output by the first learning machine and the predicted network performance output by the second learning machine,
wherein the closed-loop controller allows updated predictions from the first learning machine and the second learning machine to be used by the PCM when adjusting the behavior of the network.

2. The method as in claim 1 , wherein the network traffic data includes one or more of: an observed bandwidth consumed by the traffic, an observed application type of the traffic, flow characteristics of the traffic, or statistical measurements of the traffic.

3. The method as in claim 1 , wherein the network condition data comprises one or more of: delay measurements, bandwidth measurements, jitter measurements, packet loss measurements, or routing measurements.

4. The method as in claim 3 , wherein the network condition data is received from multiple different sources, and wherein the network condition data is modeled by merging the network condition data using a Kalman filter.

5. The method as in claim 1 , wherein the network traffic data is modeled using a time series model.

6. The method as in claim 1 , wherein the behavior of the network is adjusted by:
changing a quality of service (QoS) parameter.

7. The method as in claim 6 , wherein the QoS parameter is one or more of: a queue length, an allocated bandwidth percentage, or a class of service for an application.

8. The method as in claim 1 , wherein the behavior of the network is adjusted by:
changing a call-admission control policy.

9. The method as in claim 1 , wherein the behavior of the network is adjusted by:
changing a routing path based on an application type associated with a flow of traffic.

10. An apparatus, comprising:
one or more network interfaces to communicate with a network;
a processor coupled to the one or more network interfaces and adapted to execute one or more processes; and
a memory configured to store a process executable by the processor, the process when executed operable to:
receive network traffic data regarding traffic flowing through one or more routers in the network;
predict, by executing a first learning machine, a future traffic profile through the one or more routers by modeling the network traffic data;
receive network condition data for the network, wherein the network condition data and the network traffic data are different;
predict, by a second learning machine, future network performance by modeling the network condition data;
input into a third learning machine of a Predictive Control Manager (PCM) the network traffic modeling data from the first learning machine and the network condition modeling data from the second learning machine; and
adjust, by a closed-loop controller on the PCM, a behavior of the network based on an output from the third learning machine, wherein the output from the third learning machine is based on the predicted future traffic profile output by the first learning machine and the predicted network performance output by the second learning machine,
wherein the closed-loop controller allows updated predictions from the first learning machine and the second learning machine to be used by the PCM when adjusting the behavior of the network.

11. The apparatus as in claim 10 , wherein the network traffic data includes one or more of: an observed bandwidth consumed by the traffic, an observed application type of the traffic, flow characteristics of the traffic, or statistical measurements of the traffic.

12. The apparatus as in claim 10 , wherein the network condition data comprises one or more of: delay measurements, bandwidth measurements, jitter measurements, packet loss measurements, or routing measurements.

13. The apparatus as in claim 12 , wherein the network condition data is received from multiple different sources, and wherein the network condition data is modeled by merging the network condition data using a Kalman filter.

14. The apparatus as in claim 10 , wherein the network traffic data is modeled using a time series model.

15. The apparatus as in claim 10 , wherein the behavior of the network is adjusted by:
changing a quality of service (QoS) parameter, wherein the QoS parameter is one or more of: a queue length, an allocated bandwidth percentage, or a class of service for an application.

16. The apparatus as in claim 10 , wherein the behavior of the network is adjusted by:
changing a call-admission control policy.

17. The apparatus as in claim 10 , wherein the behavior of the network is adjusted by:
changing a routing path based on an application type associated with a flow of traffic.

18. A tangible, non-transitory, computer-readable media having software encoded thereon, the software when executed by a processor operable to:
receive network traffic data regarding traffic flowing through one or more routers in a network;
predict, by executing a first learning machine, a future traffic profile through the one or more routers by modeling the network traffic data;
receive network condition data for the network, wherein the network condition data and the network traffic data are different;
predict, by a second learning machine, future network performance by modeling the network condition data;
input into a third learning machine of a Predictive Control Manager (PCM) the network traffic modeling data from the first learning machine and the network condition modeling data from the second learning machine; and
adjust, by a closed-loop controller on the PCM, a behavior of the network based on an output from the third learning machine, wherein the output from the third learning machine is based on the predicted future traffic profile output by the first learning machine and the predicted network performance output by the second learning machine,
wherein the closed-loop controller allows updated predictions from the first learning machine and the second learning machine to be used by the PCM when adjusting the behavior of the network.

Assignees

• Cisco Tech Inc

Inventors

• Vasseur Jean-Philippe
• Dasgupta Sukrit
• Mermoud Grégory

Classifications

• H04L41/5025

  by proactively reacting to service quality change, e.g. by reconfiguration after service quality degradation or upgrade · CPC title

• H04L41/16 Primary

  using machine learning or artificial intelligence · CPC title

• H04L41/5003

  Managing SLA; Interaction between SLA and QoS · CPC title

• H04L43/062

  related to network traffic · CPC title

• H04L43/0876

  Network utilisation, e.g. volume of load or congestion level · CPC title