Channel-Aware Wireless Waveform Generation for Link Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless communication systems face challenges in efficiently supporting a diverse range of devices with varying data traffic profiles and requirements, such as high data rate, low latency, and high reliability, due to sub-optimal link adaptation methods that are limited to a finite set of configurations and lack granular assessment of radio propagation conditions.
Innovation Solution
Implementing machine learning-based neural networks at base stations and user equipment to dynamically adapt transmission and reception waveforms to prevailing channel conditions, using specialized reference signals for training, enabling fine-tuned parameter selection for optimal performance across varying geographical environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional link adaptation methods with finite configurations are used, then device complexity is reduced, but system efficiency and adaptability to diverse traffic profiles deteriorate
Solution Approach 1:
The patent changes the fundamental parameter of link adaptation from selecting among finite pre-defined configurations to continuously optimizing transmission parameters based on machine learning models. The neural networks dynamically adjust modulation order, coding rate, and other transmission parameters to match actual channel conditions, transforming the system from discrete configuration selection to continuous parameter optimization, thereby achieving superior system efficiency while managing complexity through algorithmic approaches.
Solution Approach 2:
The patent replaces traditional mechanical/link-layer configuration selection mechanisms with machine learning-based neural networks. Instead of relying on predefined tables and rule-based selection, the system uses trained neural network models to predict optimal transmission parameters, substituting the mechanical configuration selection process with intelligent computational models that adapt to diverse traffic profiles and channel conditions.
2Reliability
If granular assessment of radio propagation conditions is implemented, then transmission reliability is improved, but measurement and detection difficulty increases
Solution Approach 1:
The patent introduces machine learning models as intermediary components between raw channel measurements and transmission parameter selection. The neural networks process complex channel state information, interference patterns, and traffic characteristics, transforming difficult-to-interpret raw measurements into reliable transmission decisions. This intermediary processing layer simplifies the overall measurement and detection task while enhancing transmission reliability through intelligent pattern recognition.
Solution Approach 2:
The patent implements universal machine learning models that can assess multiple aspects of radio propagation conditions simultaneously. The neural networks are trained to evaluate various channel parameters, interference scenarios, and traffic profiles using a unified approach, making the measurement and detection process more manageable while providing comprehensive assessment for reliable transmission across diverse conditions.
3Adaptability or versatility
If dynamic waveform adaptation is implemented, then adaptability to different traffic profiles is improved, but device complexity and computational requirements worsen
Solution Approach 1:
The patent applies preliminary action by pre-training machine learning models offline using extensive datasets representing diverse traffic profiles and channel conditions. The neural networks are trained beforehand to recognize patterns and optimize parameters for various scenarios. During actual operation, the pre-trained models quickly infer optimal transmission parameters without requiring complex real-time computations, thereby achieving high adaptability while keeping online computational complexity manageable.
Solution Approach 2:
The patent uses copying by deploying trained neural network models that have learned optimal transmission strategies from extensive training data. Instead of performing complex optimization calculations in real-time, the system copies the knowledge gained during offline training into compact neural network models that can be executed efficiently on devices with limited computational resources, achieving high adaptability with reduced runtime complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of transmitting data by a transmitting entity in a wireless communications network to a receiving entity via a communications channel between the transmitting entity and the receiving entity is provided. The method comprises receiving data for transmission to the receiving entity via the communications channel, dividing the data into portions for transmission, receiving, for each of the portions of data, an indication of channel information from the receiving entity for use by the transmitting entity in determining values for one or more communications parameters with which the portion of data should be transmitted, determining the values for the one or more communications parameters based on the received channel information, dynamically generating, for each of the portions of data, a waveform representative of the portion of data, the waveform representations being generated in accordance with the values of the one or more communications parameters, and transmitting each of the portions of data, using the generated waveform representations and in accordance with the values of the one or more communications parameters, to the receiving entity.