Machine Learning HARQ Prediction for Adaptive MCS Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing link adaptation techniques in 5G/NR networks suffer from inefficient use of channel capacity due to large step-down values in modulation and coding scheme (MCS) adjustments, leading to prolonged operation below actual capacity, despite the need to maintain block error rate (BLER) targets.
Innovation Solution
Implementing a machine learning (ML) model, such as a Recurrence Neural Network (RNN), to predict the likelihood of decoding success based on historical channel characteristics, allowing for optimized MCS and physical resource block (PRB) selection before transmission, thereby improving BLER, latency, and throughput without requiring standard changes or additional signaling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If large step-down values are used in MCS adjustments to maintain BLER targets, then reliability is improved, but channel capacity utilization deteriorates
Solution Approach 1:
The patent applies dynamics by transitioning from static, fixed step-down values to dynamic, adaptive MCS selection using machine learning predictions. The system continuously learns from historical HARQ feedback and channel conditions to optimize MCS choices in real-time, allowing the step-down value to vary based on actual channel state and prediction accuracy rather than using a fixed large value.
Solution Approach 2:
The patent changes the parameter of MCS selection from conventional fixed-step adjustments to ML-predicted optimal values. By using machine learning models to forecast decoding success probability, the system dynamically adjusts MCS parameters based on predicted channel conditions, replacing the need for large conservative step-down values with more precise, data-driven selections.
2Reliability
If large step-down values are used in MCS adjustments, then BLER target is maintained, but latency increases due to prolonged operation below capacity
Solution Approach 1:
The patent applies preliminary action by using machine learning models to predict decoding success probability before transmission occurs. This allows the system to pre-determine optimal MCS values that are likely to succeed, avoiding the need for large step-down adjustments after failures occur. The prediction is made in advance based on historical patterns, enabling proactive rather than reactive MCS selection.
Solution Approach 2:
The patent implements feedback by continuously monitoring historical HARQ feedback results and using this information to train machine learning models. The models learn from past transmission outcomes and channel conditions to improve future predictions, creating a closed-loop system where actual performance data continuously refines the MCS selection strategy to reduce unnecessary latency.
3Device complexity
If conventional link adaptation techniques are used, then device complexity is low, but productivity is reduced due to inefficient channel capacity use
Solution Approach 1:
The patent introduces an intermediary machine learning model layer between the existing link adaptation framework and the MCS selection process. This intermediary component processes historical channel data and HARQ feedback to generate predictions, sitting atop the conventional system rather than replacing it entirely. The ML model acts as a mediator that enhances productivity while maintaining compatibility with existing network infrastructure.
Data Source
AI summary
Embodiments include methods performed by a radio access network (RAN) node. Such methods include, prior to a first packet being transmitted, predicting likelihood of decoding success by a receiver of the first packet, using a machine learning (ML) model with the following inputs: one or more parameters representative of characteristics of a radio channel over which the first packet will be transmitted during a plurality of regularly spaced time periods (e.g., slots) preceding the transmission of the first packet, and candidates of one or more of the following for the first packet: modulation and coding scheme (MCS), and number of physical resource blocks (PRBs). Such methods include obtaining a first MCS and/or a first number of PRBs to be used for transmitting the first packet, based on the candidate(s) and the predicted likelihood of decoding success. Such methods include transmitting or receiving the first packet using the obtained first MCS and/or first number of PRBs.