OFDM Demapper Using Precomputed Terms to Cut ML Multiplications
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
MIMO OFDM receivers face high complexity, cost, and power consumption due to the large number of multiplication operations required for ML demapping, especially in systems with multiple data streams.
Innovation Solution
The implementation of a per-tone demapper that uses a modified L2-squared-norm cost-function and precomputing/storage of terms to reduce the number of multiplication operations, employing shifting and adding operations to eliminate most or all multiplications during the search process, thereby generating demapped bits with reduced complexity and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ML demapping is used to extract transmitted symbols from multiple data streams, then demapping accuracy is improved, but the number of multiplication operations increases significantly
Solution Approach 1:
The patent precomputes and stores channel matrix terms (H†H) and their products with constellation points before the actual demapping operation. This preliminary computation stores results in lookup tables, eliminating the need for repeated multiplications during the search process, thus reducing the number of multiplication operations while maintaining ML demapping accuracy
Solution Approach 2:
The patent creates lookup tables that store precomputed channel matrix products and their relationships with constellation points. Instead of performing actual multiplications during demapping, the system copies and retrieves precomputed values from these tables, significantly reducing computational complexity while preserving the accuracy of maximum-likelihood demapping
2Measurement precision
If ML demapping with large number of multiplications is performed, then demapping accuracy is improved, but power consumption increases
Solution Approach 1:
The patent performs power-intensive multiplications in advance during system initialization or channel estimation phases, storing results in memory. During actual data demapping, the system retrieves precomputed values without performing additional multiplications, thereby significantly reducing real-time power consumption while maintaining accurate symbol extraction
Solution Approach 2:
The patent replaces real-time multiplication operations with memory retrieval operations by storing precomputed channel matrix products in lookup tables. Since memory access consumes far less power than multiplication operations, this approach dramatically reduces power consumption during the demapping process while preserving demapping accuracy
3Measurement precision
If conventional ML demapping is used for 64-QAM with T streams, then accurate symbol extraction is achieved, but processing complexity increases
Solution Approach 1:
The patent precomputes the channel matrix products H†H and their products with all possible constellation points for 64-QAM modulation. These precomputed values are stored in lookup tables indexed by stream identifiers and constellation point indices, allowing the demapper to retrieve values through simple indexing and addition operations instead of performing complex multiplications during symbol extraction
Solution Approach 2:
The patent creates comprehensive lookup tables that store precomputed channel products for all 64 constellation points across multiple data streams. During demapping, the system copies relevant precomputed values from these tables based on the received signal and channel state, eliminating the need for real-time multiplication operations and significantly reducing processing complexity while maintaining accurate symbol extraction for 64-QAM modulation
Data Source
AI summary
Embodiments of an OFDM receiver and methods for decoding OFDM symbols of two or more data streams with reduced multiplication operations are generally described herein. Other embodiments may be described and claimed. In some embodiments, one or more terms of a modified L2-squared-norm cost function are precomputed and stored for predetermined complex symbol values of one or more tones of OFDM symbols prior to performing a searching process. During the searching process, the cost function is computed using the precomputed terms and received data symbols using shifting and adding operations, rather than multiplication operations. In other embodiments, non-L2-squared-norm cost functions are used.