Multi-Dimensional Channel Estimation for Large Antenna Arrays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large antenna arrays with many closely spaced antennas face a decrease in Signal to Noise Ratio (SNR) and degradation of channel estimates, leading to poor uplink performance, as the number of antenna ports increases, due to high signal correlations.
Innovation Solution
A method involving a network node that obtains matched filter channel estimates, arranges them in a multi-dimensional array reflecting the physical location of antennas, applies an orthogonal transform, de-noises the array, and then applies the inverse transform to provide improved channel estimates, effectively handling high antenna correlations and maintaining performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the number of antenna ports increases to collect more signal energy, then the ability to receive weaker signals improves, but the channel estimation accuracy degrades due to loss in antenna array gain and high signal correlations
Solution Approach 1:
The patent transforms the channel estimation problem from the traditional frequency domain to a multi-dimensional transform domain (combining time/frequency and spatial dimensions). By arranging channel estimates in a multi-dimensional array and applying orthogonal transforms across multiple dimensions, the method exploits correlations in both temporal and spatial domains simultaneously, achieving better noise suppression and maintaining estimation accuracy even with large numbers of antenna ports
Solution Approach 2:
The patent merges the exploitation of temporal correlations and spatial correlations into a unified multi-dimensional transform approach. Instead of processing time/frequency samples and spatial samples separately, the method combines both dimensions in a single transform framework, allowing the system to benefit from both types of correlations working together to improve channel estimation in the presence of high antenna correlations
2Volume of moving object
If closely spaced antennas are used in large antenna arrays, then the physical size of the array can be reduced, but the signal correlations between antennas increase leading to degraded channel estimates
Solution Approach 1:
The patent moves the processing from single-dimensional frequency domain analysis to multi-dimensional transform domain analysis that explicitly incorporates spatial dimension. By organizing channel estimates in a multi-dimensional array where one dimension represents spatial samples and another represents time/frequency samples, the method can exploit spatial correlations caused by close antenna spacing rather than treating them as detrimental effects
Solution Approach 2:
The patent converts the harmful effect of high signal correlations (which normally degrade channel estimation) into a beneficial feature. By applying multi-dimensional orthogonal transforms, the method concentrates the signal energy into fewer transform coefficients while pushing noise and interference to other coefficients, thereby improving signal-to-noise ratio and making the channel estimation more robust to the high correlations inherent in closely spaced antenna arrays
Data Source
AI summary
The disclosure relates to a method (60) performed in a network node (2) for estimating a channel. The network node (2) controls an antenna array (3) comprising a number N of antennas (51, . . . , 5N) in one or more spatial dimensions. The network node (2) comprises a receiver (72) receiving signals from the antenna array (3). The method (60) comprises: obtaining (61) matched filter channel estimates for each sub-carrier and antenna of a signal received by the antennas (51, . . . , 5N), the signal comprising a number K of frequency sub-carriers; arranging (62) the obtained matched filter channel estimates in a first multi-dimensional array, wherein time or frequency domain samples of the matched filter channel estimates are arranged along a first dimension and wherein a second dimension is a first spatial dimension of the number N of antennas, wherein the matched filter channel estimates are ordered in the first multi-dimensional array such as to reflect the physical location in space of the antennas (51, . . . , 5N); applying (63) an orthogonal transform to the first multi-dimensional array, providing a second multi-dimensional array; de-noising (64) the second multi-dimensional array, providing a third multi-dimensional array; applying (65) the inverse of the orthogonal transform to the third multi-dimensional array, providing channel estimates for all branches of the receiver (72). The disclosure also relates to corresponding network node, computer program and computer program products.