LTE Channel State Information Estimation via Covariance Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating channel state information (CSI) in LTE wireless networks are computationally intensive, particularly when dealing with large system bandwidths, due to the need for joint estimation of rank indicator (RI) and precoding matrix index (PMI), which requires extensive matrix operations and can lead to significant complexity and performance loss.
Innovation Solution
A computationally efficient method is introduced that decouples RI and PMI estimation, using channel covariance estimation and Taylor series approximation of the inverse, allowing for simplified CQI calculation with only one matrix inversion per PMI trial and linearizing the effective SINR mapping process, reducing hardware and software complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If joint estimation of RI and PMI is performed to achieve optimum performance metrics, then measurement precision is improved, but device complexity increases significantly
Solution Approach 1:
The patent segments the joint RI-PMI estimation problem into separate independent estimation processes. RI is estimated first using channel covariance matrix, then PMI is estimated separately using the decoded RI value and channel state information. This segmentation maintains estimation accuracy while dramatically reducing computational complexity by avoiding exhaustive search over all RI-PMI combinations.
Solution Approach 2:
The patent performs preliminary RI estimation using channel covariance matrix before PMI estimation. By determining RI first and using it to guide subsequent PMI estimation, the system avoids the need for joint exhaustive search, achieving the same performance metrics with reduced computational burden through preliminary action.
2Measurement precision
If matrix inversion is calculated for each selected RE in bandwidth to achieve accurate metrics, then measurement precision is improved, but productivity decreases due to prohibitively large computation effort
Solution Approach 1:
The patent merges multiple RE measurements into a single channel covariance matrix estimation. Instead of performing matrix inversion at each individual RE, the system combines channel measurements across multiple REs to form a covariance matrix, then performs a single matrix inversion operation that captures the essential channel characteristics, significantly improving computational efficiency while maintaining accuracy.
Solution Approach 2:
The patent uses channel covariance matrix as a representative copy that captures the essential characteristics of multiple individual RE channel measurements. By working with this compact covariance representation rather than individual RE measurements, the system achieves the same information content with much reduced computational complexity for matrix operations.
3Measurement precision
If subset size is increased to reduce performance loss from channel averaging, then measurement precision is improved, but device complexity increases due to more extensive matrix operations
Solution Approach 1:
The patent replaces the mechanical approach of averaging individual channel matrices with a statistical approach using channel covariance matrix. Instead of directly averaging channel matrices which requires handling large-dimensional matrices, the system computes covariance statistics that capture essential channel properties in a compact form, reducing matrix operation complexity while maintaining or improving estimation accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method, computer program, device and system are provided for determining channel state information for use in a wireless communications network. The channel state information includes a rank indicator (RI), precoding matrix index (PMI) and channel quality indicator (CQI). The RI, PMI or CQI can be determined based on channel covariance estimation and the Taylor series approximation of its inverse. Further, the RI and PMI can be determined separately.