Turbo Loop Receiver QR Decomposition Reuse
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Turbo-looped multiuser detection architectures in wireless communication networks face significant computational complexity due to the need for matrix inversion in linear MMSE equalization, especially in large MIMO systems, which is impractical for real-world applications.
Innovation Solution
The method involves reusing components of QR decomposition from previous turbo loops to reduce computational overhead, specifically by estimating interference plus noise covariance matrices and using partial results of QR decomposition to simplify equalizer calculations in subsequent loops.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If linear MMSE equalization is performed in successive turbo loops, then near-optimal detection performance is achieved, but computational complexity increases at a rate more than linear due to increasing dimensions
Solution Approach 1:
The patent performs QR decomposition in advance before the turbo loops begin, storing the Q and R components for later reuse. This preliminary action eliminates the need to perform full matrix inversion during each turbo loop iteration, significantly reducing computational complexity while maintaining detection performance.
Solution Approach 2:
The patent discards the computationally expensive matrix inversion operation during turbo loops and recovers the necessary computational results by reusing the pre-computed QR decomposition components. This allows the system to achieve the same equalization effect without repeating the heavy computational burden.
2Measurement precision
If QR decomposition is used to compute the inverse of the covariance matrix, then high accuracy solution is achieved without actual matrix inversion, but significant computational overhead remains for moderately large MIMO systems
Solution Approach 1:
The QR decomposition is performed once in advance and the results are stored for reuse across multiple turbo loops. This preliminary computation avoids repeating the same heavy mathematical operations in each iteration, reducing overall computational overhead while maintaining solution accuracy.
Solution Approach 2:
The pre-computed QR decomposition components serve multiple purposes: they enable efficient equalization in each turbo loop, support the computation of interference plus noise covariance matrices, and facilitate the reuse strategy across different iterations. This multi-functionality maximizes the value of the initial computational investment.
3Reliability
If the linear equalizer is updated on each turbo loop for near optimality, then detection performance is improved, but computational overhead increases due to repeated matrix inversion
Solution Approach 1:
Instead of performing full matrix inversion in each turbo loop, the patent discards this expensive operation and recovers the necessary equalization results by reusing the pre-computed QR decomposition. This maintains the ability to update the equalizer effectively while avoiding the computational burden of repeated matrix inversion.
Solution Approach 2:
The patent creates and reuses copies of the QR decomposition results across multiple turbo loops. Rather than recomputing from scratch each time, the system copies and applies the pre-computed Q and R matrices to different stages of the detection process, significantly reducing computational overhead.
Data Source
AI summary
An improved receiver design implements a method for modeling users in SIC turbo loop multiuser detection architectures that reduces the number of implementation cycles, and thereby reduces the computational overhead associated with computing the inverse of the received signal covariance matrix, by efficiently reusing components of a QR decomposition. By reusing some of the computational results from the previous turbo loop's equalizer calculation, the disclosed receiver significantly reduces the computational burden of updating the linear equalizer on each turbo loop. Depending on the embodiment, this reduction can be accomplished in at least two different ways, depending on the dimensionality and other aspects of the implementation.