AI/ML Modulation Constellations With SNR-Adaptive Bit Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing modulation schemes in 6G/5G communication systems suffer from shaping losses and inefficiencies in signal-to-noise ratio (SNR) requirements, limiting the effective use of spectrum and channel capacity.
Innovation Solution
Employing AI/ML empowered modulation schemes with non-uniform constellations and bit-to-symbol mapping, constrained by quadrant symmetry Lagrangian (QSL), quadrant symmetry constraint (QSC), or rectangular structure constraint (RSC), to optimize bitwise mutual information based on SNR conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional uniform QAM modulation schemes are used, then implementation simplicity is maintained, but shaping losses occur and bitwise mutual information is suboptimal
Solution Approach 1:
The patent applies local quality by transitioning from uniform QAM constellation points to non-uniform constellation points where each point's position is locally optimized based on its probability of transmission. The constellation design assigns different spatial positions to different symbol groups, with inner points having higher probability and outer points having lower probability, thereby locally optimizing the information transmission efficiency at each constellation point rather than using a uniform distribution.
Solution Approach 2:
The patent employs parameter changes by modifying the constellation point parameters from uniform to non-uniform distribution. Specifically, the in-phase and quadrature components are independently optimized to create non-uniform spacing between constellation points, with the parameter optimization guided by bitwise mutual information metrics. This allows the system to adapt constellation parameters to achieve optimal performance for different code rates and channel conditions.
2Productivity
If non-uniform constellations are used to improve bitwise mutual information, then channel capacity increases, but SNR requirements become more stringent
Solution Approach 1:
The patent applies dynamics by making the constellation design adaptive to different operating conditions. The system dynamically selects and optimizes constellation parameters based on the target code rate and channel SNR conditions. Different non-uniform constellation configurations are designed for different code rates (e.g., 1/3, 2/3, 3/4, 4/5), allowing the system to adapt its constellation structure to match the required reliability and capacity trade-offs for each scenario.
Solution Approach 2:
The patent employs preliminary action by pre-optimizing constellation parameters for specific code rates and SNR conditions before actual transmission. The constellation design process involves preliminary optimization of in-phase and quadrature components to maximize bitwise mutual information for predetermined code rates. This pre-optimization ensures that when the system operates at a specific code rate, the constellation is already configured for optimal performance without requiring real-time adaptation during transmission.
3Ease of manufacture
If quadrant symmetry constraints are applied to simplify constellation design, then implementation ease increases, but optimization flexibility is reduced
Solution Approach 1:
The patent applies asymmetry by deliberately breaking the traditional quadrant symmetry constraint to achieve better optimization precision. While conventional designs enforce identical constellation structures in all four quadrants, this patent allows asymmetric optimization within quadrants by independently optimizing in-phase and quadrature components. This controlled asymmetry enables finer-grained optimization of bitwise mutual information while maintaining sufficient structural regularity for practical implementation.
Data Source
AI summary
A two-dimensional constellation for data signals having improved bitwise mutual information of data points is based on a signal-to-noise ratio (SNR) and a code rate where, based on the SNR, data bits are mapped to pre-defined in-phase and quadrature values. The in-phase and quadrature values denote points in the two-dimensional space such that the efficiency of bitwise mutual information is adapted based on the SNR. The mapping is preferably subject to a constraint selected from one of quadrant symmetry Lagrangian (QSL), quadrant symmetry constraint (QSC), or rectangular structure constraint (RSC).