Techniques for artificial intelligence capabilities at a network device

Patent metadata

Field	Value
Publication number	US-12445358-B2
Application number	US-202318375934-A
Country	US
Kind code	B2
Filing date	Oct 2, 2023
Priority date	Dec 28, 2018
Publication date	Oct 14, 2025
Grant date	Oct 14, 2025

Abstract

Examples include techniques for artificial intelligence (AI) capabilities at a network switch. These examples include receiving a request to register a neural network for loading to an inference resource located at the network switch and loading the neural network based on information included in the request to support an AI service to be provided by users requesting the AI service.

First claim

What is claimed is:

1. A method comprising:
receiving, through an interface configured to couple with a plurality of ingress links and a plurality of egress links of a switch device, information associated with loading a neural network to an inference resource located at the switch device, the neural network to support an artificial intelligence (AI) service for a network that includes the switch device, the switch device configured to forward data from the plurality of ingress links to the plurality of egress links of the switch device;
causing the neural network to be loaded to the inference resource based on the information; and
receiving, via an ingress link from among the plurality of ingress links, an AI service request, wherein if the AI service request cannot be fulfilled using the neural network loaded to the inference resource, the AI service request is to be forwarded, via an egress link from among the plurality of egress links, to a second inference resource located separate from the switch device;
wherein:
the information comprises registration data; and
the registration data comprises tenant identification data.

2. The method of claim 1 , further comprising:
receiving, via an ingress link from among the plurality of ingress links, an AI service request that includes a data payload;
determining whether the AI service request can be fulfilled using the loaded neural network;
causing the data payload to be inputted to the loaded neural network if the AI service request can be fulfilled using the loaded neural network; and
sending, via an egress link from among the plurality of egress links, a generated result to a requestor of the AI service that is based on the inputted data payload.

3. The method of claim 1 , the inference resource comprising a neural processing unit, a tensor processing unit, a field programmable gate array, an application specific integrated circuit, a graphics processing unit or a central processing unit.

4. The method of claim 1 , the neural network comprising a convoluted neural network, a deep neural network, a recurrent neural network, a convoluted neural network, a multi-task cascaded neural network, a text-to-speech neural network, a Gaussian mixture model neural network, an alternating least square neural network, a gate recurrent unit neural network, an automatic speaker verification neural network, a natural language processing neural network, a compressed sparse row neural network, an inception neural network, a bundle adjustment neural network or a simultaneous localization and mapping/extended Kalman filter neural network.

5. The method of claim 1 , the AI service comprising a vehicle-to-vehicle AI service, an augmented reality AI service, an autonomous driving AI service, a video analytics AI service or a language analysis AI service.

6. The method of claim 1 , wherein causing the neural network to be loaded to the inference resource based on the information comprises causing the neural network to be loaded based on the information indicating a physical location of the switch device in the network is relevant to support the AI service.

7. The method of claim 1 , wherein causing the neural network to be loaded to the inference resource based on the information comprises causing the neural network to be loaded based on the information indicating a physical location of the switch device in the network facilitates meeting a latency requirement for the AI service.

8. The method of claim 1 , wherein the AI service is for a tenant of the network that includes the switch device.

9. The method of claim 1 , wherein:
the registration data also comprises service/performance level data associated with the tenant identification data.

10. A switch device comprising:
an interface configured to couple with a plurality of ingress links and a plurality of egress links, wherein the switch device is configured to forward data from the plurality of ingress links to the plurality of egress links of the switch device;
an inference resource; and
circuitry configured to:
receive, through the interface, information associated with loading a neural network to the inference resource to support an artificial intelligence (AI) service for a network that includes the switch device;
cause the neural network to be loaded to the inference resource based on the information; and
receive, via an ingress link from among the plurality of ingress links, an AI service request, wherein if the AI service request cannot be fulfilled using the neural network loaded to the inference resource, the AI service request is to be forwarded, via an egress link from among the plurality of egress links, to a second inference resource located separate from the switch device;

wherein:
the information comprises registration data; and
the registration data comprises tenant identification data.

11. The switch device of claim 10 , further comprising the circuitry configured to:
receive, via an ingress link from among the plurality of ingress links, an AI service request that includes a data payload;
determine whether the AI service request can be fulfilled using the loaded neural network;
cause the data payload to be inputted to the loaded neural network if the AI service request can be fulfilled using the loaded neural network; and
send, via an egress link from among the plurality of egress links, a generated result to a requestor of the AI service that is based on the inputted data payload.

12. The switch device of claim 10 , the inference resource comprising a neural processing unit, a tensor processing unit, a field programmable gate array, an application specific integrated circuit, a graphics processing unit or a central processing unit.

13. The switch device of claim 10 , the neural network comprising a convoluted neural network, a deep neural network, a recurrent neural network, a convoluted neural network, a multi-task cascaded neural network, a text-to-speech neural network, a Gaussian mixture model neural network, an alternating least square neural network, a gate recurrent unit neural network, an automatic speaker verification neural network, a natural language processing neural network, a compressed sparse row neural network, an inception neural network, a bundle adjustment neural network or a simultaneous localization and mapping/extended Kalman filter neural network.

14. The switch device of claim 10 , the AI service comprising a vehicle-to-vehicle AI service, an augmented reality AI service, an autonomous driving AI service, a video analytics AI service or a language analysis AI service.

15. The switch device of claim 10 , wherein the second inference resource located separate from the switch device comprises the second inference resource located remotely from the switch device.

16. The switch device of claim 10 , wherein the circuitry configured to cause the neural network to be loaded to the inference resource based on the information comprises the circuitry configured to cause the neural network to be loaded based on the information indicating a physical location of the switch device in the network is relevant to support the AI service.

17. The switch device of claim 10 , wherein the circuitry configured to cause the neural network to be loaded to the inference resource based on the information comprises the circuitry configured to cause the neural network to be loaded based on the information indicating a physical location of the switch device in the network facilitates meeting a latency requirement for the AI service.

18. The

Assignees

• Intel Corp

Inventors

• Guim Bernat Francesc
• Prabhakaran Suraj
• Doshi Kshitij A
• Ganesh Brinda
• Verrall Timothy

Classifications

• H04L41/0816

  the condition being an adaptation, e.g. in response to network events · CPC title

• H04L41/5051

  Service on demand, e.g. definition and deployment of services in real time · CPC title

• H04L41/5019

  Ensuring fulfilment of SLA · CPC title

• G06N5/04

  Inference or reasoning models · CPC title

• H04L41/5012

  determining service availability, e.g. which services are available at a certain point in time · CPC title