Wireless Receiver Multiuser Detection Residual Error Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multiuser detection methods in wireless communication networks suffer from the accumulation of estimation errors in successive turbo loops due to faulty assumptions about previously modeled users, leading to suboptimal performance in non-linear successive interference cancellation (SIC) receiver architectures.
Innovation Solution
The method improves user modeling in SIC turbo loop multiuser detection by utilizing 'soft' information and confidence levels from preceding turbo loops to scale error covariance matrices, effectively representing residual interference and minimizing estimation errors in successive loops.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-linear successive interference cancellation (SIC) receiver architectures with turbo loops are used to approach optimal MLD detection performance, then detection accuracy is improved, but estimation errors accumulate in successive loops due to faulty assumptions about previously modeled users
Solution Approach 1:
The patent implements feedback by calculating residual errors from the difference between received signals and modeled signals, then feeding these residual errors back into the equalization process. This allows the system to correct estimation errors from previous turbo loops by adjusting the error covariance matrix based on actual performance, preventing error accumulation while maintaining high detection accuracy
Solution Approach 2:
The patent performs preliminary equalization and error calculation before the main detection process. By pre-calculating residual errors and using them to scale error covariance matrices in advance, the system prepares corrected estimation parameters that prevent error accumulation from the outset, rather than correcting errors after they have propagated through multiple loops
2Device complexity
If linear equalization methods such as multiuser MMSE with interference rejection are used, then implementation complexity is reduced, but performance does not approach optimal MLD detection performance
Solution Approach 1:
The patent introduces dynamic adaptation by iteratively updating error covariance matrices based on residual errors from each turbo loop. This dynamic adjustment allows the system to achieve near-MLD performance through adaptive refinement, while maintaining the lower implementation complexity of linear equalization methods by avoiding the exhaustive search required by pure MLD approaches
3Productivity
If more users are modeled in successive turbo loops, then spectral efficiency increases, but the faulty assumption that previously modeled users are accurately known leads to error accumulation
Solution Approach 1:
The patent calculates residual errors for each modeled user and feeds this information back to adjust the modeling of subsequent users. This feedback mechanism ensures that errors from modeling earlier users do not propagate to later users, allowing the system to model more users (increasing spectral efficiency) without suffering from cumulative estimation errors
Data Source
AI summary
An improved receiver design implements a practical method for modeling users in SIC turbo loop multiuser detection architectures, wherein in each loop unsubtracted estimation errors from previous loops are used to appropriately scale the error covariance matrix for each user, thereby accurately representing the remaining residual interference in the data stream for each desired user. The effect of estimation errors in previous interference cancellation operations is thereby minimized, and symbol estimations in successive turbo loops are improved, for example during multiuser MMSE, multiuser MMSE with interference rejection combining (MMSE-IRC), sample matrix inversion (SMI), or any of their adaptive variants (least mean-square, recursive least square, Kalman filter etc.). The estimated residual symbol energy can be computed per symbol, and then applied to entire data streams, to groups of symbols, or to each symbol separately.