Noise Variance Estimation from Training Symbol Subtraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless communication systems face challenges in estimating noise variance without knowledge of the channel, which is crucial for improving signal quality and processing in receivers.
Innovation Solution
The method involves receiving two multi-sample symbols, subtracting one from the other to produce noise samples, and using these samples to estimate noise variance, which can be applied to process data without prior channel knowledge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If noise variance estimation is performed without channel knowledge, then the estimation can be done prior to channel estimation improving processing sequence, but the accuracy of noise variance estimation deteriorates due to lack of channel information
Solution Approach 1:
The patent extracts the signal component from the received signal by utilizing known training symbols. By subtracting the known transmitted training symbols from the received training symbols, the method isolates the noise component, enabling accurate noise variance estimation without requiring channel knowledge. This extraction principle allows separating the noise from the combined signal-noise mixture.
Solution Approach 2:
The method changes the parameter state by using known training symbols to transform the received signal into pure noise samples. By leveraging the known characteristics of training symbols and their relationship with received symbols, the system transforms complex signal processing into a simpler noise estimation problem where the noise variance can be directly calculated from the squared differences.
2Loss of information
If channel estimation is performed before noise variance estimation, then channel information is available for processing, but the noise variance estimation becomes more complex and may require iterative methods
Solution Approach 1:
The patent inverts the conventional processing order by performing noise variance estimation before channel estimation. This inversion is achieved by using known training symbols to directly compute noise samples through subtraction, bypassing the need for channel knowledge. The method proves that noise can be estimated independently of channel information, actually simplifying the overall processing complexity.
3Productivity
If MMSE equalization and LLR computation are performed without accurate noise variance, then processing can proceed, but the performance and accuracy of these operations deteriorate
Solution Approach 1:
The patent performs noise variance estimation as a preliminary action before MMSE equalization and LLR computation. By using known training symbols to estimate noise variance in advance, the system prepares accurate noise statistics that are then utilized in subsequent processing steps. This preliminary estimation ensures that optimal parameters are available for equalization and decoding operations, improving their accuracy without delaying the overall processing timeline.
Data Source
AI summary
A method and system for estimating noise variance. A method for noise variance estimation comprises receiving a first multi-sample symbol and receiving a second multi-sample symbol. The first multi-sample symbol is subtracted from the second multi-sample symbol to produce a set of noise samples. The set of noise samples is used to produce a noise variance estimate. The noise variance estimate is applied in various tasks (e.g. channel estimation, log-likelihood ratio computation, and/or minimum mean squared error equalization) to process data provided to a user.