Multilevel MIMO Receiver Decoding for Lower LLR Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
MIMO-OFDM systems face high computational complexity in extracting Log-Likelihood Ratios (LLR) and A-Posteriori Probabilities, especially with multiple antennas, leading to the use of suboptimal techniques due to increased complexity with the number of constellation points.
Innovation Solution
A multilevel receiver with a hierarchical architecture is introduced to reduce complexity by staged LLR computation and channel decoding, using a multilevel coding scheme that divides constellation bits into least significant, most significant, and middle bits, with each level having different coding rates and decoding stages to select candidate sequences and calculate LLRs efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If more antennas are introduced at the receive side to achieve high data rate transmission, then the data rate and transmission performance are improved, but the computational complexity of extracting Log-Likelihood Ratios (LLR) and A-Posteriori Probabilities (APPs) increases significantly
Solution Approach 1:
The patent applies segmentation by dividing the multilevel decoding process into multiple stages, where each stage processes a subset of the data. The receiver performs staged LLR computation and channel decoding in successive stages, breaking down the complex single-step decoding into manageable segments that reduce the computational burden at each step while maintaining overall decoding performance.
2Device complexity
If suboptimal techniques are used to reduce computational complexity in multilevel decoding, then the computation burden is reduced, but the decoding performance and reliability deteriorate
Solution Approach 1:
The patent implements feedback through an iterative decoding process where the receiver performs multiple stages of LLR computation and channel decoding. Each stage uses the results from previous stages to refine the decoding, with feedback loops that allow the system to progressively improve decoding accuracy while maintaining reduced computational complexity compared to optimal single-step decoding.
3Measurement precision
If optimal multilevel decoding is performed for large constellation sizes, then the decoding accuracy is maximized, but the number of correlations and computational resources required increase dramatically
Solution Approach 1:
The patent segments the multilevel decoding process into multiple computational stages, where each stage handles a portion of the constellation decoding. This segmentation reduces the number of correlations required at each individual stage while maintaining overall decoding accuracy, as the computational burden is distributed across multiple simpler steps rather than requiring one complex correlation operation.
Solution Approach 2:
The patent applies preliminary action by performing staged LLR computation and preliminary channel decoding before final decision-making. The receiver computes LLRs and performs intermediate decoding steps in advance, preparing processed data that simplifies the final decoding decision and reduces the number of full correlations needed while preserving decoding accuracy.
Data Source
AI summary
A method of detecting sequences of multi-level encoded symbols. The multi-level encoded symbols are mapped and modulated with a modulation scheme having a number of constellation points identified by a sequence of bits arranged in at least a first and a second group. The first group is encoded with a first encoding scheme, and the second group is encoded with a second coding scheme, and the multi-level encoded symbols are transmitted by multiple transmitting sources and received as a received vector by multiple receiving elements.A first set of candidate sequences is selected and a first set of probability information is calculated for the first set of candidate sequences. Then the first group of bits of the symbols are decoded. The decoded bits of the first group are re-encoded and used to select a sub-set of constellation points.A second set of candidate sequences is selected based on this sub-set of constellation points and a second set of probability information is calculated for the second set of candidate sequences. Finally, the second group of bits of the symbols are decoded.