MIMO Detector Triangularizes Channel Matrix for Low Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multiple-input multiple-output (MIMO) communication systems face challenges in detecting transmitted vectors in noisy fading channels, particularly with high spectral efficiency, as existing detectors like Zero-Forcing (ZF) and Minimum Mean Square Error (MMSE) suffer from sub-optimal performance, high complexity, and limited spatial diversity, while list detectors and lattice decoding algorithms struggle with computational complexity and parallelization.
Innovation Solution
A method that triangularizes the channel matrix, decouples the minimization problem by selecting a reference antenna, and uses successive layer detection with spatial decision-feedback equalization to compute soft-output information, enabling near-optimal performance with reduced complexity and parallelization suitable for VLSI implementations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Maximum-Likelihood (ML) detection is used to achieve high-performance detection in MIMO systems, then detection accuracy is improved, but computational complexity becomes increasingly unfeasible with growth of spectral efficiency
Solution Approach 1:
The patent segments the detection problem into two stages: first performing linear detection (ZF or MMSE) to obtain initial symbol estimates, then applying nonlinear processing through spatially ordered decision-feedback equalization. This segmentation allows the system to achieve near-ML performance without the exhaustive search complexity of pure ML detection, as the nonlinear detector processes already-refined estimates rather than raw received signals.
Solution Approach 2:
The patent applies linear detection (ZF or MMSE) as a preliminary step before nonlinear detection. This preliminary action refines the received signals and reduces interference, making the subsequent nonlinear processing more effective and efficient. The linear detector prepares the data by providing initial estimates that the nonlinear detector can then refine further, avoiding the need for complete ML exhaustive search.
2Device complexity
If sub-optimal linear detection algorithms like Zero-Forcing (ZF) or Minimum Mean Square Error (MMSE) are used to reduce complexity, then computational complexity is reduced, but spatial diversity order and detection performance deteriorate
Solution Approach 1:
The patent merges linear detection and nonlinear detection into a hybrid detector that combines the advantages of both approaches. The linear detector provides computational efficiency and initial refinement, while the nonlinear detector recovers spatial diversity order and improves detection accuracy. This merging allows the system to achieve performance close to ML detection with significantly reduced complexity compared to pure nonlinear approaches.
Solution Approach 2:
The patent implements decision-feedback equalization where decisions from previously detected symbols are fed back to cancel interference in subsequent detections. This feedback mechanism allows the nonlinear detector to exploit spatial diversity more effectively by using detected symbols to improve the detection of remaining symbols, thereby recovering the spatial diversity order that linear detectors lack.
3Measurement precision
If list detectors and lattice decoding algorithms are used to improve detection performance, then detection accuracy is improved, but computational complexity and difficulty of parallelization increase
Solution Approach 1:
The patent segments the detection process into linear preprocessing and nonlinear postprocessing stages, where the linear stage handles computational intensive operations in a parallelizable manner, and the nonlinear stage performs simpler operations on already-refined data. This segmentation reduces the overall computational complexity compared to list detectors and lattice decoding algorithms that require exhaustive or near-exhaustive searches.
Data Source
AI summary
Embodiments of a method and an apparatus for detecting multiple complex-valued symbols belonging to discrete constellations. The method and apparatus is a detector that finds a closest vector, or a close approximation of it, to a received vector. The disclosure also gets (optimally, in case of two transmit sources) or closely approximates (for more than two transmit sources) the most likely sequences required for an optimal bit or symbol a-posteriori probability computation. Also part of the present disclosure is represented by Also embodiments of a method and an apparatus to determine a near-optimal ordering algorithms for the aforementioned purpose. The method and apparatus achieves optimal performance for two transmit antennas and achieves near-optimal performance for a higher number of antennas, with a lower complexity as compared to a maximum-likelihood detection method and apparatus. The method and apparatus are suitable for highly parallel hardware architectures.