Carrier Aggregation Linearization

When multiple carriers pass through a shared power amplifier, nonlinear mixing generates intermodulation products that fall both in-band and across carrier boundaries. Carrier aggregation DPD must model and cancel these cross-modulation distortion terms, which appear at sum and difference frequencies of the original carriers. This requires a predistorter with sufficient bandwidth to capture third-order, fifth-order, and sometimes seventh-order intermodulation products spanning the full aggregated spectrum.

The aggregated multi-carrier signal is represented as a single wideband complex baseband waveform, even when carriers occupy non-contiguous spectrum blocks. This unified representation allows a single predistorter to process the entire signal, but demands extremely high sampling rates proportional to the total aggregated bandwidth plus guard bands. For example, aggregating three 100 MHz carriers with 50 MHz gaps requires a baseband sampling rate exceeding 600 MHz to avoid aliasing.

Power amplifiers exhibit frequency-dependent behavior across wide aggregated bandwidths, meaning distortion characteristics vary by carrier location. A carrier at the band edge may experience different nonlinear compression and memory effects than a carrier at band center. Advanced DPD architectures incorporate frequency-selective modeling to apply distinct correction profiles to different spectral regions within the same predistorter.

The computational burden of carrier aggregation DPD grows nonlinearly with the number of carriers and total bandwidth. Key scaling factors include:

• Basis function count increases with intermodulation order and memory depth
• Sampling rate must satisfy Nyquist for the full aggregated bandwidth
• Coefficient estimation requires larger matrices for multi-carrier model extraction Efficient implementations often use pruned Volterra series or spline-based models to reduce complexity while maintaining linearization performance.

Unlike single-carrier DPD where a single EVM or ACLR value suffices, carrier aggregation systems must evaluate linearization performance independently per component carrier. A predistorter optimized for aggregate performance may under-correct one carrier while over-correcting another. Validation requires measuring per-carrier EVM, per-carrier ACLR, and aggregate error vector magnitude across the full composite signal.

The observation receiver for carrier aggregation DPD must digitize the full nonlinear bandwidth of the power amplifier output, which extends well beyond the aggregated signal bandwidth due to spectral regrowth. For a system aggregating 200 MHz of carriers, the feedback ADC may require 1 GHz or more of instantaneous bandwidth to capture fifth-order intermodulation products. This drives requirements for ultra-wideband downconverters and high-speed data converters.

A linearization architecture that uses a single predistorter to simultaneously compensate for distortion in two widely separated frequency bands sharing a single power amplifier. This technique exploits the mathematical relationship between baseband signals and their intermodulation products to synthesize a single predistorted signal that linearizes both carriers.

• Eliminates need for separate DPD engines per band
• Reduces hardware complexity and power consumption
• Handles cross-modulation products between carriers
• Critical for intra-band non-contiguous CA scenarios

The ratio of the predistorted signal's bandwidth to the original signal's bandwidth, caused by the spectral regrowth inherent to nonlinear predistortion processing. When a DPD system pre-distorts a signal to cancel PA nonlinearity, it intentionally generates out-of-band intermodulation products that expand the occupied bandwidth.

• Typically 3x to 5x the original signal bandwidth
• Drives ADC/DAC sampling rate requirements
• Must be accommodated in feedback path design
• Increases with higher-order nonlinearity correction

Nonlinear signal distortion generating spurious frequency components at sums and differences of integer multiples of the original input signal frequencies. In carrier aggregation, IMD products from multiple carriers can fall directly into receiver bands, causing desensitization and regulatory compliance failures.

• Second-order: f1 ± f2
• Third-order: 2f1 ± f2, 2f2 ± f1 (most problematic)
• Fifth-order products become significant in wideband systems
• DPD must model and cancel all dominant IMD orders

A metric quantifying the ratio of transmitted power within an assigned channel to the power leaking into an adjacent radio frequency channel. For carrier aggregation systems, ACLR must be measured across all component carriers and their adjacent channels simultaneously.

• 3GPP specifies ACLR limits per band and bandwidth class
• Typical requirement: -45 dBc or better
• CA systems must meet limits on both sides of aggregated spectrum
• Primary figure of merit for DPD linearization performance

A digital predistortion implementation where the predistorter operates at a higher sampling rate than the baseband signal to capture and cancel out-of-band distortion products. This is essential for carrier aggregation because the combined nonlinear distortion bandwidth far exceeds any single component carrier's bandwidth.

• Predistorter rate: typically 3x-5x signal bandwidth
• Requires interpolation and decimation stages
• Enables cancellation of IMD products beyond Nyquist
• Critical for wideband inter-band CA linearization

The unwanted appearance of signal energy in adjacent frequency channels caused by the intermodulation products of a nonlinear power amplifier. In carrier aggregation, spectral regrowth from multiple carriers overlaps and compounds, creating complex interference patterns that span wider frequency ranges than single-carrier regrowth.

• Primary cause of ACLR degradation
• Regrowth bandwidth scales with nonlinearity order
• CA regrowth can span hundreds of MHz
• DPD must model regrowth across entire aggregated bandwidth