Digital Pre-Distorter Structure for Low-Complexity Linearization

The patent segments the non-linear compensation function into multiple additive components (memoryless polynomial terms and memory effect terms with different delay orders). This segmentation allows each term to be computed independently and added together, avoiding the exponential complexity of full Volterra series while maintaining linearization accuracy through the systematic inclusion of dominant non-linear and memory effects.

The patent applies partial action by selectively including only the most significant terms from the Volterra series expansion - specifically memoryless polynomial terms up to a certain order and memory terms with specific delay orders. This partial inclusion captures the dominant non-linear and memory effects sufficient for accurate linearization while avoiding computation of less significant higher-order terms that would exponentially increase complexity.

The patent uses partial action by including only the most significant memoryless and memory terms that capture the dominant non-linear effects. This selective inclusion reduces the number of parameters that need to be accurately determined and maintained, thereby reducing sensitivity to device variations such as process, temperature, and supply voltage changes while still achieving sufficient linearization accuracy.

The patent performs preliminary determination of compensator coefficients during manufacturing or initial calibration, storing them for use during normal operation. This preliminary action reduces the need for continual updating of parameters during operation, making the system more robust to device variations by establishing a stable baseline compensation that is less sensitive to subsequent environmental changes.

The patent performs preliminary determination and storage of compensator coefficients during manufacturing or initial calibration. These pre-determined coefficients are then used during normal operation without requiring continual updating, significantly reducing power consumption and computation while maintaining adequate linearization accuracy through the robust structure of the selected polynomial and memory terms.

The patent uses a simplified compensator structure with a limited number of polynomial and memory terms that can be determined once and used for an extended period. This approach treats the coefficient determination as a one-time or infrequent operation rather than a continuous process, reducing ongoing power consumption and computational resources while maintaining sufficient performance through the carefully selected terms.