Cross-Band Predistorter

Unlike conventional DPD that linearizes in-band distortion, the cross-band predistorter selectively targets intermodulation distortion (IMD) products that fall into neighboring frequency bands. It synthesizes a cancellation signal equal in amplitude but opposite in phase to the predicted cross-band distortion.

• Operates on baseband representations of multiple concurrent carriers
• Requires accurate cross-band behavioral models to predict IMD landing zones
• Cancellation depth of 15-20 dB is typically achievable for cross-band products

The predistorter uses a two-dimensional indexing structure based on the instantaneous magnitudes of two concurrent baseband signals. This 2D-LUT or 2D polynomial approach captures the envelope-dependent nature of cross-band distortion.

• Address space: |x₁(n)| and |x₂(n)| form the 2D index
• 2D Memory Polynomial (2D-MMP) extends this with temporal cross-terms
• Enables hardware-efficient FPGA implementation with dual-port memory blocks

Cross-band predistorters must account for cross-band memory effects where the nonlinear behavior in one frequency band is influenced by the past envelope history of a signal in a different band. These arise from:

• Thermal coupling between amplifier stages handling different carriers
• Bias circuit modulation where envelope power from one band sags the shared supply
• Trapping effects in GaN HEMT devices with multi-carrier excitation
• Modeled via cross-band envelope lag terms in Volterra-derived structures

Coefficients for the cross-band predistorter are extracted using joint estimation techniques that simultaneously solve for in-band and cross-band terms. The Multi-Band Indirect Learning Architecture (MB-ILA) is the dominant approach:

• A post-distorter is trained on the attenuated PA output
• Coefficients are copied to the forward-path predistorter
• Least squares (LS) or recursive least squares (RLS) solvers handle the augmented parameter vector
• Cross-band terms typically require 30-50% additional coefficients versus single-band DPD

Cross-band predistortion is critical for 3GPP carrier aggregation scenarios where multiple component carriers are transmitted through a common power amplifier. Without it, cross-band IMD violates spectral emission masks.

• Supports intra-band contiguous and non-contiguous CA configurations
• Essential for dual-band uplink carrier aggregation in LTE-Advanced and 5G NR
• Enables multi-standard radios transmitting LTE and NR simultaneously through one PA chain

Cross-band predistorters face a quadratic increase in complexity versus single-band DPD due to the 2D basis function space. Key optimization strategies include:

• Pruned Volterra models that eliminate statistically insignificant cross-terms
• 2D-LUT with interpolation to reduce real-time multiply-accumulate operations
• Subband decomposition to process wideband signals at reduced sample rates
• Typical implementation: 40-60% more DSP slices than equivalent single-band DPD on FPGA

Nonlinear interference products generated by the interaction of multiple carrier signals within a power amplifier. These products fall on top of or near the desired transmit bands, making them particularly destructive to signal integrity.

• Third-order intermodulation (IM3) products are typically the strongest and closest to the carriers
• Cross-band distortion cannot be filtered without also removing desired signal content
• Requires active cancellation via dedicated predistorter blocks

The generation of unwanted frequency components resulting from the nonlinear mixing of two or more signals within an active device. IMD products appear at sum and difference frequencies of the input carriers and their harmonics.

• IM3 products at 2f₁-f₂ and 2f₂-f₁ are the primary targets of cross-band predistorters
• IMD power increases 3 dB for every 1 dB increase in input power in the nonlinear region
• Characterized by the third-order intercept point (IP3)

A behavioral model that extends the memory polynomial to two dimensions by including cross-terms dependent on the envelope magnitudes of both concurrent bands. This captures cross-band memory effects essential for accurate predistortion.

• Indexes predistortion coefficients using |x₁(n)| and |x₂(n)| simultaneously
• Includes memory depth terms for both self-distortion and cross-distortion paths
• Balances modeling accuracy with computational tractability for FPGA implementation

A long-term memory effect in multi-band amplifiers where the nonlinear behavior in one frequency band is influenced by the past envelope history of a signal in a different band.

• Caused by shared bias networks, thermal coupling, and trapping effects in semiconductor devices
• Cannot be compensated by memoryless or single-band memory models
• Requires cross-band envelope tracking terms in the predistorter model

A closed-loop DPD adaptation method where a post-distorter model is identified from the attenuated PA output and then copied to the predistorter in the forward path.

• Avoids the need to solve for the PA inverse directly
• Post-distorter training uses the PA output as input and the desired signal as target
• Cross-band predistorter coefficients are extracted jointly with in-band terms

A parameter identification technique that simultaneously estimates all coefficients of a multi-band predistorter model, including cross-band terms, in a single optimization step.

• Uses least squares or recursive least squares algorithms on the composite error signal
• Ensures cross-band cancellation terms are optimally aligned with in-band linearization
• Critical for maintaining stability when cross-band coupling is strong