ML-Aided Channel Estimation Using Dual Pilot Sequences

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing channel estimation methods in wireless communication networks, particularly in massive MIMO scenarios, face performance degradation due to high DMRS signaling overhead, leading to sub-optimal estimation results and reduced cellular system capacity.

Innovation Solution

The implementation of a Machine Learning (ML)-aided channel estimation technique using two pilot sequences, one designed for frequency domain and another for time domain changes, allowing for accurate full channel estimation without the need for additional interpolation, thereby reducing the number of pilot symbols and enhancing spectral efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If Linear Minimum Mean Squared Error (LMMSE) estimation algorithms are used for channel estimation, then the implementation is simple, but the estimation accuracy degrades significantly in complex wireless communication channels

Engineering Contradiction:
Improveimplementation simplicityVSAvoidchannel estimation accuracy
Core Design Contradiction:
Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent replaces traditional signal processing algorithms (LMMSE) with a Machine Learning model (Convolutional Neural Network) to perform channel estimation. This substitution enables the system to achieve superior estimation accuracy in complex channels while maintaining practical implementability through the trained ML model.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If the frequency of DMRS symbols per layer is increased to support high-mobility scenarios, then the channel tracking capability improves, but the signaling overhead increases and cellular system capacity is reduced

Engineering Contradiction:
Improvechannel tracking capabilityVSAvoidcellular system capacity
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent changes the approach from increasing the quantity of pilot symbols to using a Machine Learning model that can achieve accurate channel estimation with fewer pilot symbols. This parameter change in the estimation methodology allows the system to maintain reliable channel tracking in high-mobility scenarios while reducing DMRS signaling overhead and preserving cellular system capacity.

Inventive Principle:
Principle #35Parameter changes

3Adaptability or versatility

If multiple orthogonal DMRSs are allocated for MIMO transmission, then the support for multiple transmission layers is enabled, but the pilot sequence length increases and spectral efficiency is reduced

Engineering Contradiction:
ImproveMIMO transmission supportVSAvoidspectral efficiency
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent replaces traditional pilot-based channel estimation methods with a Machine Learning-based approach that can efficiently handle MIMO transmission with multiple layers. This substitution enables the system to maintain adaptability for MIMO scenarios while reducing the number of required pilot symbols, thereby improving spectral efficiency.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4462740A1Machine learning-aided channel estimation in wireless communication network
Publication Date: 2024.11.13 NOKIA SOLUTIONS & NETWORKS OY
  • EP4462740A1 patent drawingFigure 1
  • EP4462740A1 patent drawingFigure 2
  • EP4462740A1 patent drawingFigure 3

AI summary

A technical solution is provided, which enhances the performance of Machine Learning (ML)-based channel estimation by using two different pilot sequences. More specifically, a first pilot sequence (e.g., a sequence of Demodulation Reference Signals (DMRSs)) is designed to track a change in a wireless communication channel in a frequency domain, while a second pilot sequence (e.g., a sequence of Phase-Tracking RSs (PTRSs)) is designed to track a change in the wireless communication channel in a time domain. By using the first and second pilot sequences thus designed, raw channel estimates in the frequency and time domains, respectively, are obtained on a receiving side. Then, the raw channel estimates are fed to a ML model that is configured to predict a full (time-frequency) channel estimate based thereon. The full channel estimate may be subsequently used in a data decoding algorithm executed on the receiving side.