Network Entity Throughput Estimation for Application Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies lack accurate and efficient methods for applications to determine the deliverable throughput of wireless links in radio access networks, leading to suboptimal data rate adjustments and increased network congestion.
Innovation Solution
An apparatus and method that utilize a deep neural network (DNN) to estimate deliverable throughput based on signal quality measurements, including classifications of signal-to-noise ratio (SNR) and feature selections such as the number of competing users, providing accurate and instantaneous or forecasted capacity information to application servers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If applications use default data rates without accurate throughput information, then network overhead is reduced, but application quality and network efficiency deteriorate due to suboptimal data rate adjustments and increased congestion
Solution Approach 1:
The patent introduces an intermediary component (network entity or base station) that calculates deliverable throughput by acting as a mediator between the radio access network and application servers. This intermediary uses signal quality measurements from user devices and applies machine learning models to compute accurate throughput estimates, which are then provided to applications through APIs. This approach delivers precise throughput information without requiring complex modifications to applications themselves, thus maintaining low overhead while improving measurement accuracy.
Solution Approach 2:
The system implements feedback by continuously monitoring signal quality measurements from user devices and using these measurements to calculate and update deliverable throughput information. The throughput calculations are fed back to application servers via APIs, enabling applications to adjust their data rates dynamically based on current network conditions. This closed-loop feedback mechanism ensures accurate throughput information is provided without requiring complex one-time configurations.
2Productivity
If applications adjust data rates without accurate throughput information, then device complexity is reduced, but productivity deteriorates due to suboptimal data rate selection and increased network congestion
Solution Approach 1:
The patent replaces traditional mechanical or rule-based throughput estimation methods with machine learning-based calculations. The system uses trained machine learning models that process signal quality measurements and competing user information to predict deliverable throughput accurately. This substitution of mechanical calculation methods with intelligent algorithms improves productivity by enabling more accurate data rate adjustments while managing calculation complexity through pre-trained models that execute efficiently during operation.
Solution Approach 2:
The system performs preliminary action by pre-training machine learning models offline using historical network data. These pre-trained models capture complex relationships between signal quality, user competition, and actual throughput. During runtime, the models execute efficiently to provide accurate throughput predictions without requiring complex real-time calculations. This preliminary training phase separates the complexity of model development from operational complexity, improving network efficiency while keeping runtime device complexity manageable.
3Measurement precision
If user devices perform throughput estimation locally, then network overhead is reduced, but measurement precision deteriorates due to lack of network-wide information such as competing users and queue sizes
Solution Approach 1:
The patent merges the throughput calculation functionality into the network infrastructure (base station or network entity) rather than distributing it to individual user devices. The base station consolidates all necessary network-wide information including signal quality measurements from multiple users, competing user data, and queue sizes in a single location. This merging enables accurate throughput estimation by having access to complete network state information without requiring each device to independently collect and process fragmented network data, thus improving measurement precision while managing complexity centrally.
Data Source
AI summary
Described are examples for calculating and exposing network capacity and congestion to applications. A network entity such as a radio access network (RAN) intelligent controller (RIC) or virtual base station component receives measurements of a signal quality for a plurality of user devices connected to a RAN. The network entity estimates a deliverable throughput of a wireless link for a user device of the plurality of user devices based on at least the measurements. The network entity can consider other factors such as a number of competing users, queue sizes of the user device and of the competing users, or a scheduling policy. The network entity provides the deliverable throughput to an application server for an application of the user device communicating with the application server via the RAN. The application server can adapt a data rate for the application and the user device based on the deliverable throughput.