MIMO Soft-Decision Detection Using Parallel Candidate Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multiple antenna systems using spatial multiplexing face high computational complexity in generating soft-decision values, with detection times varying based on channel states and noise magnitudes, making them inefficient compared to Maximum Likelihood Detection (MLD).
Innovation Solution
A method and apparatus for generating soft-decision information in a multiple antenna system using parallel detection to determine candidate symbol vectors and calculate Log Likelihood Ratios (LLRs), which reduces complexity and maintains performance comparable to MLD, with constant detection time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If Maximum Likelihood Detection (MLD) is used to detect candidate symbol vectors, then detection performance is optimized, but computational complexity becomes very high
Solution Approach 1:
The patent segments the detection process into two stages: first identifying a small set of candidate symbol vectors using simplified criteria, then performing refined detection only on these candidates. This segmentation reduces the overall computational complexity while maintaining detection performance close to MLD.
Solution Approach 2:
Instead of performing full MLD on all possible symbol vectors, the patent applies partial detection by identifying and processing only the most promising candidate vectors. This partial action approach achieves near-MLD performance with significantly reduced computational effort.
2Device complexity
If tree search algorithms like LSD or QRD-M are used to reduce complexity, then computational complexity decreases, but detection time varies depending on channel state or noise magnitude
Solution Approach 1:
The patent performs preliminary identification of candidate symbol vectors using simple distance criteria before the actual detection process. This preliminary action ensures that the subsequent detection stage operates on a fixed, small set of candidates, making detection time constant and independent of channel state or noise magnitude.
Solution Approach 2:
The patent changes the detection approach by using fixed threshold-based candidate selection instead of adaptive tree search parameters. This parameter change eliminates the variability in detection time that occurs with channel-state-dependent algorithms like LSD and QRD-M.
3Device complexity
If max-log approximation is used to generate soft-decision values, then design complexity is simplified, but performance degradation occurs compared to optimal soft-decision detection
Solution Approach 1:
The patent uses hard-decision candidate identification as a simplified copy or approximation of the optimal soft-decision process. By copying the essential function of MLD (identifying the best candidate) without implementing the full complex soft-decision calculation, the system achieves low design complexity while maintaining good performance.
Data Source
AI summary
A method and an apparatus for generating soft-decision information in a multiple antenna system are provided. The method includes determining Q candidate symbol vectors for a first transmission symbol of a received signal vector by performing parallel detection on a received signal vector and a channel matrix, determining a candidate symbol vector having a shortest Euclidean distance to the received signal vector from among the Q candidate symbol vector, as an approximate Maximum Likelihood (ML) symbol vector, determining (Q−1) candidate symbol vectors for each of the remaining transmission symbols of the received signal vector by performing partial parallel detection on the received signal vector and the channel matrix using the approximate ML symbol vector, and calculating Log Likelihood Ratios (LLRs) of bits of the first transmission symbol using the candidate symbol vectors, wherein Q represents a modulation order.