MIMO Layer Interference Cancellation With Iterative CONDML Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current MIMO detection methods, such as the maximum likelihood (ML) and maximum a posteriori (MAP) algorithms, face high complexity issues, making them impractical for real-world applications, especially with higher level modulation schemes and multiple data layers, and interference cancellation methods like successive interference cancellation (SIC) require complex ordering decisions.
Innovation Solution
The proposed method employs an iterative conditional maximum likelihood (CONDML) detector that uses feedback from channel decoders for interference cancellation across layers, combining conditional ML detection with the turbo principle, allowing for reduced complexity and efficient decoding through cyclic redundancy checks (CRCs) and log-likelihood ratio calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If maximum likelihood (ML) or maximum a posteriori (MAP) algorithms are used for MIMO detection, then detection performance is improved, but computational complexity increases significantly
Solution Approach 1:
The patent segments the detection process into multiple iterations where each iteration handles a subset of layers. Instead of detecting all L layers simultaneously as in conventional ML/MAP, the patent divides layers into groups and performs iterative detection, reducing the complexity from O(ML) to O(ML−1) per iteration while maintaining near-optimal performance through multiple passes.
Solution Approach 2:
The patent introduces dynamic iteration-based detection where the number of iterations and the grouping of layers can be adapted based on channel conditions and performance requirements. The iterative conditional ML detector dynamically adjusts the detection process across multiple passes, allowing the system to balance between complexity and performance based on operational conditions.
2Device complexity
If successive interference cancellation (SIC) is used to reduce complexity, then computational complexity is reduced, but performance degrades due to error propagation
Solution Approach 1:
The patent implements feedback mechanisms where detection results from previous iterations are fed back to refine subsequent detections. The iterative conditional ML detector uses feedback from partially detected layers to improve the detection of remaining layers, reducing error propagation compared to conventional SIC while maintaining lower complexity than full ML.
Solution Approach 2:
The patent performs preliminary detection of certain layers before final detection of all layers. By detecting some layers first and using their results to aid subsequent detection, the system reduces the search space for remaining layers, thereby reducing overall complexity while maintaining performance through the preliminary information gathered.
3Measurement precision
If iterative detection with feedback is implemented, then detection accuracy is improved, but convergence speed decreases and complexity increases
Solution Approach 1:
The patent applies partial action by performing a limited number of iterations rather than exhaustive iterative detection. The system performs enough iterations to achieve satisfactory performance without continuing until full convergence, thereby balancing accuracy with convergence speed. This partial iteration approach prevents excessive time loss while maintaining adequate detection accuracy.
Data Source
AI summary
A method and apparatus are provided for cancelling interferences of a received MIMO RF signal. The method includes: receiving a signal y; an initial step of obtaining, from the received signal y, at least one data representative of the received signal, called decoded data; computing the decoded data as a function of at least one predetermined criterion, delivering feedback data and a feedback decision; and, as a function of the feedback decision, at least one iteration of the following steps: constructing a set of feedback symbols from the feedback data; obtaining, from the set of feedback symbols, at least one data representative of a feedback decoded signal, called feedback decoded data; computing the at least one feedback decoded data as a function of the at least one predetermined criterion, delivering the feedback data.