Machine Learning Data Rate Control for Wireless Networks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Traditional rate control algorithms in wireless networks result in wastage of critical airtime due to probing packets consuming channel bandwidth and causing packet errors, necessitating a more efficient mechanism for determining optimal data transmission rates to enhance performance and reduce errors.

Innovation Solution

A machine learning-based data rate control method that dynamically generates and adjusts data rate configurations using a deep learning model, incorporating parameters like RSSI, packet length, and channel usage, to optimize packet transmission rates and retransmissions, and updates the model based on feedback to improve accuracy and reduce errors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional rate control algorithms use probing packets to determine optimal transmission rates, then the system can find the best possible rate for data transmission, but probing packets consume channel bandwidth and cause packet errors resulting in wastage of critical airtime

Engineering Contradiction:

Improverate determination accuracyVSAvoidairtime wastage

Core Design Contradiction:

Measurement precisionVSLoss of energy

Solution Approach 1:

The system performs preliminary rate determination through machine learning model training during periods when probing is acceptable, storing the learned rate configurations for future use. This allows the system to have rate determination knowledge ready before critical data transmission occurs, avoiding the need to send probing packets during time-sensitive operations.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system creates a machine learning model that copies and stores the optimal rate determination knowledge learned from probing packets. Instead of repeatedly sending actual probing packets to determine rates, the system uses the stored model predictions which replicate the rate determination function without consuming additional channel bandwidth.

Inventive Principle:

Principle #26Copying

2Adaptability or versatility

If traditional rate control algorithms continuously adjust transmission rates through probing, then the system can adapt to changing channel conditions, but the continuous probing results in packet errors and reduced performance

Engineering Contradiction:

Improverate adaptation capabilityVSAvoidpacket error rate

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system implements feedback mechanisms where transmission outcomes (success/failure) are fed back to the machine learning model. The model uses this feedback to continuously learn and adapt optimal rates for different channel conditions and node characteristics, enabling rate adaptation without requiring continuous probing that would generate packet errors.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The machine learning model enables the system to determine optimal rates autonomously based on learned patterns from historical data. The system serves itself by using the trained model to select rates without external intervention or continuous probing, reducing packet errors while maintaining adaptability to changing conditions.

Inventive Principle:

Principle #25Self-service

3Productivity

If the system uses machine learning models to dynamically determine data rates, then channel usage is enhanced and packet errors are reduced, but the system complexity increases due to model training and maintenance

Engineering Contradiction:

Improvechannel utilization efficiencyVSAvoidmodel training complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The system introduces a machine learning model as an intermediary between channel condition observations and rate selection decisions. This model absorbs the complexity of analyzing multiple parameters (RSSI, packet length, channel usage, node characteristics) and translating them into optimal rate selections, simplifying the overall system architecture while improving channel utilization efficiency.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11470507B2Methods and system for machine learning-based data rate control

Publication Date: 2022.10.11 APRECOMM PTE LTD

AI summary

Embodiments herein provide a method for machine learning-based data rate control in a wireless network system. The method includes detecting a rate change event by a data rate controller for at least one receiving node in the wireless network system. Furthermore, the method includes dynamically generating a data rate configuration comprising a data rate configuration comprising a plurality of rates, and an accuracy level of each of the rates. The method further includes determining by the data rate controller that the accuracy level of at least one rate from the plurality of rates meets an accuracy criteria and causing by the data rate controller to transmit at least one packet using the at least one rate.