Predictive content push-enabled content delivery network

• US 10,592,578 B1
• Filed: 03/07/2018
• Issued: 03/17/2020
• Est. Priority Date: 03/07/2018
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Mokashi, Ronil Sudhir, Verma, Prashant, Farris, Ryan
• Primary Examiner(s): Macilwinen, John M
• Application Number:
  US15/914,823
• Time in Patent Office: 741 Days
• CPC Class Codes: G06F 16/9574 of access to content, e.g. by caching
• CPC Class Codes: G06F 18/295 Markov models or related models, e.g. semi-Markov models; Markov random fields; Networks embedding Markov models
• CPC Class Codes: H04L 47/83 based on usage prediction
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/289 Intermediate processing functionally located close to the data consumer application, e.g. in same machine, in same home or in same sub-network
• CPC Class Codes: H04L 67/55 Push-based network services
• CPC Class Codes: H04L 67/568 Storing data temporarily at an intermediate stage, e.g. caching
• CPC Class Codes: H04L 67/5681 Pre-fetching or pre-delivering data based on network characteristics
• CPC Class Codes: H04L 67/60 Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources admission control or resource allocation H04L47/70
• CPC Class Codes: H04N 21/2225 Local VOD servers
• CPC Class Codes: H04N 21/25 Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies maintenance or administration in data networks H04L41/00
• CPC Class Codes: H04N 21/4331 Caching operations, e.g. of an advertisement for later insertion during playback

Abstract

A content delivery network (“CDN”) is provided herein that predicts content resources (e.g., a data object, such as a video file, an audio file, a script, an image, a document, etc.) that may be requested by a user device in the future and transmits or pushes such resources to the user device prior to receiving a request. The CDN may use artificial intelligence models, such as Markov models, in order to predict which content resources to retrieve and transmit proactively to the user device. The predictive techniques implemented by the CDN may reduce a latency of delivering requested content resources and/or a latency of the user device in rendering and displaying a content page.

Claims

1. A computer-implemented method of proactively transmitting content, the method comprising:

as implemented by one or more computing devices configured with specific computer-executable instructions,

obtaining, from a user device via a network, a request for a content page on a content delivery network (CDN) service, wherein the content page is associated with a first set of data objects, and wherein the CDN service is present on the one or more computing devices at a point of presence (POP);

obtaining the first set of data objects;

transmitting the first set of data objects to the user device;

determining that a first group in a plurality of groups is associated with the request for the content page;

detecting capabilities of the user device and conditions of the network;

determining that the user device can process a first quantity of data objects and the network can transmit the first quantity of data objects using the capabilities of the user device and the conditions of the network;

obtaining one or more Markov models corresponding to the first group;

applying a first Markov model in the one or more Markov models, wherein the first Markov model corresponds to the requested content page;

identifying a second set of data objects to obtain based on the application of the first Markov model, wherein the second set of data objects are associated with a second content page different than the requested content page, and wherein the user device has not yet requested the second content page after requesting the requested content page;

determining that a quantity of the first set of data objects and the second set of data objects is less than the first quantity of data objects;

obtaining the second set of data objects; and

transmitting the second set of data objects to the user device.

2. The computer-implemented method of claim 1, wherein the first Markov model comprises at least a first node representing the requested content page, a second node coupled to the first node via a first arc that represents the second content page, and a third node coupled to the first node via a second arc that represents a third content page, and wherein a probability associated with the first arc is greater than a probability associated with the second arc.

3. The computer-implemented method of claim 2, further comprising:

identifying a third set of data objects to obtain based on the application of the first Markov model, wherein the third set of data objects are associated with the third content page, and wherein the user device has not yet requested the third content page after requesting the requested content page;

determining that a quantity of the first set of data objects, the second set of data objects, and the third set of data objects is less than the first quantity of data objects;

obtaining the third set of data objects; and

transmitting the third set of data objects to the user device.

4. The computer-implemented method of claim 2, wherein the first Markov model further comprises at least a fourth node preceding the first node that represents a fourth content page, and wherein the user device requested the fourth content page prior to requesting the requested content page.

5. The computer-implemented method of claim 1, wherein a second computing device external to the one or more computing devices trained the first Markov model using access logs collected by the CDN service corresponding to the user device and other user devices associated with the first group.

6. A system comprising:

a first computing device comprising a first processor configured with first computer-executable instructions that, when executed by the first processor, cause the first computing device to train a first stochastic model; and

a second computing device comprising a second processor in communication with the first computing device and configured with second computer-executable instructions that, when executed by the second processor, cause the second computing device to;

obtain, from a user device via a network, a request for a content page, wherein the content page is associated with a first set of data objects;

obtain the first set of data objects;

cause transmission of the first set of data objects to the user device;

determine that a first group in a plurality of groups is associated with the request for the content page;

determine that the user device can process a first quantity of data objects based on detected capabilities of the user device;

obtain the first stochastic model from the first computing device, wherein the first stochastic model corresponds to the first group;

identify a second set of data objects to obtain using the first stochastic model, wherein the second set of data objects are associated with a second content page different than the requested content page, and wherein the user device has not yet requested the second content page after requesting the requested content page;

determine that a quantity of the first set of data objects and the second set of data objects is less than the first quantity of data objects;

obtain the second set of data objects; and

cause transmission of the second set of data objects to the user device.

7. The system of claim 6, wherein the first stochastic model comprises a first Markov model, wherein the first Markov model comprises at least a first node representing the requested content page, a second node coupled to the first node via a first arc that represents the second content page, and a third node coupled to the first node via a second arc that represents a third content page, and wherein a probability associated with the first arc is greater than a probability associated with the second arc.

8. The system of claim 7, wherein the second computer-executable instructions further cause the second computing device to:

identify a third set of data objects to obtain using the first Markov model, wherein the third set of data objects are associated with the third content page, and wherein the user device has not yet requested the third content page after requesting the requested content page;

determine that a quantity of the first set of data objects, the second set of data objects, and the third set of data objects is less than the first quantity of data objects;

obtain the third set of data objects; and

cause transmission of the third set of data objects to the user device.

9. The system of claim 7, wherein the second computer-executable instructions further cause the second computing device to:

identify a third set of data objects to obtain using the first Markov model, wherein the third set of data objects are associated with the third content page, and wherein the user device has not yet requested the third content page after requesting the requested content page;

determine that a quantity of the first set of data objects, the second set of data objects, and the third set of data objects exceeds the first quantity of data objects; and

determine not to obtain and cause transmission of the third set of data objects to the user device, thereby reducing a likelihood that at least one of the user device or the network becomes congested.

10. The system of claim 7, wherein the first Markov model further comprises at least a fourth node preceding the first node that represents a fourth content page, and wherein the user device requested the fourth content page prior to requesting the requested content page.

11. The system of claim 6, wherein the first computer-executable instructions further cause the first computing device to define the first group using access logs corresponding to the user device and other user devices that exhibited browsing behavior common with the user device.

12. The system of claim 6, wherein the second computer-executable instructions further cause the second computing device to:

obtain a link header in response to transmission of a request for the first set of data objects, wherein the link header comprises a reference to a third set of data objects to obtain;

compare the second set of data objects and the third set of data objects; and

store any differences between the second set of data objects and the third set of data objects in access logs.

13. The system of claim 12, wherein the first computer-executable instructions further cause the first computing device to:

re-train or update the first stochastic model using at least the differences; and

cause transmission of the re-trained or updated first stochastic model to the second computing device for use when future content page requests are obtained.

14. The system of claim 6, wherein the second computer-executable instructions further cause the second computing device to:

determine a first format of data objects to transmit to the user device using the capabilities of the user device;

obtain a version of the first set of data objects in the first format and a version of the second set of data objects in the first format; and

cause transmission of the version of the first set of data objects in the first format and the version of the second set of data objects in the first format to the user device.

15. Non-transitory, computer-readable storage media comprising computer-executable instructions, wherein the computer-executable instructions, when executed by a computer system, cause the computer system to:

obtain, from a user device via a network, a request for a content page, wherein the content page is associated with a first set of data objects;

obtain the first set of data objects;

cause transmission of the first set of data objects to the user device;

determine that a first group in a plurality of groups is associated with the request for the content page;

obtain a first artificial intelligence model, wherein the first artificial intelligence model corresponds to the first group;

identify a second set of data objects to obtain using the first artificial intelligence model, wherein the second set of data objects are associated with a second content page different than the requested content page, and wherein the user device has not yet requested the second content page after requesting the requested content page;

obtain the second set of data objects; and

cause transmission of the second set of data objects to the user device in response to a determination that a quantity of the first set of data objects and the second set of data objects can be processed by the user device.

16. The non-transitory, computer-readable storage media of claim 15, wherein the first artificial intelligence model comprises at least one of a first Markov model or a neural network.

17. The non-transitory, computer-readable storage media of claim 15, wherein the first Markov model comprises at least a first node representing the requested content page, a second node coupled to the first node via a first arc that represents the second content page, and a third node coupled to the first node via a second arc that represents a third content page, and wherein a probability associated with the first arc is greater than a probability associated with the second arc.

18. The non-transitory, computer-readable storage media of claim 15, wherein the computer-executable instructions further cause the computer system to:

identify a third set of data objects to obtain using the first Markov model, wherein the third set of data objects are associated with a third content page, and wherein the user device has not yet requested the third content page after requesting the requested content page;

determine that transmission of the first set of data objects, the second set of data objects, and the third set of data objects to the user device will not cause an available bandwidth of the network as a percentage of a total bandwidth of the network to drop below a threshold value;

obtain the third set of data objects; and

cause transmission of the third set of data objects to the user device.

19. The non-transitory, computer-readable storage media of claim 15, wherein the first Markov model further comprises at least a fourth node preceding the first node that represents a fourth content page, and wherein the user device requested the fourth content page prior to requesting the requested content page.

20. The non-transitory, computer-readable storage media of claim 15, wherein the computer-executable instructions further cause the computer system to:

obtain a link header in response to transmission of a request for the first set of data objects, wherein the link header comprises a reference to a third set of data objects to obtain;

compare the second set of data objects and the third set of data objects; and

store any differences between the second set of data objects and the third set of data objects in access logs such that the first artificial intelligence model can be re-trained or updated using at least the differences to improve predictions when future content page requests are obtained.