Systems and methods for network traffic shaping for streaming and downloadable video
Disclosed are systems and methods that provide a computerized framework for network traffic shaping and optimization using a novel, trained machine learning (ML) algorithm. The disclosure provides for the designed framework to enhance the efficiency and quality of data transmission over a network by dynamically determining the appropriate delivery method—streaming or downloading—for various types of content. The disclosed framework can execute a trained ML algorithm that analyzes real-time network conditions and content characteristics, and the like, to make such determinations as to how to deliver content to users, for which content is then provided to requesting content provider applications executing on user devices.
1 . A method comprising:
receiving a request for a user device to access content over a network, the content accessible via an application associated with the user device;
analyzing the request, and selecting a machine learning (ML) model from a set of ML models, the ML model being a trained ML model for handling requests related to at least a type of the application, the ML model being trained using a set of operations, the set of operations comprising:
collecting, over a time period, session data for a set of devices executing the application,
analyzing the session data to determine a labeled training data set, the labeled training data set comprising information related to online and offline content, and
training the ML model based on the labeled training data set, the training being performed in accordance with at least an efficiency threshold that corresponds to a speed at which the ML model can be executed at runtime;
executing the ML model, the execution causing a network traffic shaping operation to be executed that corresponds to a type of the request; and
providing, via the network traffic shaping operation, the user device access to the content in response to the request.
2 . The method of claim 1 , further comprising the request type being an online request or offline request, the online request comprising the request corresponding to streaming of the content by the user device, the offline request comprising the request corresponding to a download of the content by the user device.
3 . The method of claim 1 , further comprising:
performing the network traffic shaping operation by modifying network traffic associated with the provided access to the content to correspond to the type of the request.
4 . The method of claim 1 , further comprising the session data comprising information related to at least one of a user device identifier (ID), user device type, user device version, application ID, application type, application version, content type and content ID.
5 . The method of claim 1 , further comprising:
identifying network capabilities related to the user device;
identifying a network component specific to the network capabilities; and
performing the network traffic shaping operation via the network component.
6 . The method of claim 5 , further comprising the network capabilities relating to at least one of a fourth generation (4G) network and a fifth generation (5G) network, wherein the network component is a packet gateway (PGW) or user plane function (UPF).
7 . The method of claim 1 , further comprising:
further training the ML model based on the access provided to the user device.
8 . The method of claim 1 , further comprising the ML model being a convolutional neural network (CNN).
9 . A system comprising:
a processor configured to:
receive a request for a user device to access content over a network, the content accessible via an application associated with the user device;
analyze the request, and select a machine learning (ML) model from a set of ML models, the ML model being a trained ML model for handling requests related to at least a type of the application, the processor being configured to train the ML model using a set of operations, the set of operations comprising:
collecting, over a time period, session data for a set of devices executing the application,
analyzing the session data to determine a labeled training data set, the labeled training data set comprising information related to online and offline content, and
training the ML model based on the labeled training data set, the training being performed in accordance with at least an efficiency threshold that corresponds to a speed at which the ML model can be executed at runtime;
execute the ML model, the execution causing a network traffic shaping operation to be executed that corresponds to a type of the request; and
provide, via the network traffic shaping operation, the user device access to the content in response to the request.
10 . The system of claim 9 , wherein the processor is further configured such that the request type is an online request or offline request, the online request comprising the request corresponding to streaming of the content by the user device, the offline request comprising the request corresponding to a download of the content by the user device.
11 . The system of claim 9 , wherein the processor is further configured to:
perform the network traffic shaping operation by modifying network traffic associated with the provided access to the content to correspond to the type of the request.
12 . The system of claim 9 , wherein the processor is further configured such that the session data comprises information related to at least one of a user device identifier (ID), user device type, user device version, application ID, application type, application version, content type and content ID.
13 . The system of claim 9 , wherein the processor is further configured to:
identify network capabilities related to the user device;
identify a network component specific to the network capabilities; and
perform the network traffic shaping operation via the network component.
14 . The system of claim 13 , wherein the processor is further configured such that the network capabilities relate to at least one of a fourth generation (4G) network and a fifth generation (5G) network, wherein the network component is a packet gateway (PGW) or user plane function (UPF).
15 . A non-transitory computer-readable storage medium tangibly encoded with computer-executable instructions, that when executed by a processor, perform a method comprising:
receiving a request for a user device to access content over a network, the content accessible via an application associated with the user device;
analyzing the request, and selecting a machine learning (ML) model from a set of ML models, the ML model being a trained ML model for handling requests related to at least a type of the application, the ML model being trained using a set of operations, the set of operations comprising:
collecting, over a time period, session data for a set of devices executing the application,
analyzing the session data to determine a labeled training data set, the labeled training data set comprising information related to online and offline content, and
training the ML model based on the labeled training data set, the training being performed in accordance with at least an efficiency threshold that corresponds to a speed at which the ML model can be executed at runtime;
executing the ML model, the execution causing a network traffic shaping operation to be executed that corresponds to a type of the request; and
providing, via the network traffic shaping operation, the user device access to the content in response to the request.
16 . The non-transitory computer-readable storage medium of claim 15 , further comprising the request type being an online request or offline request, the online request comprising the request corresponding to streaming of the content by the user device, the offline request comprising the request corresponding to a download of the content by the user device.