Bandwidth Expansion Factor

When a digital predistorter applies an inverse nonlinear function to a signal, it generates intermodulation products that extend beyond the original signal bandwidth. This spectral regrowth is not an impairment but a necessary byproduct of creating the anti-distortion signal. The expansion factor directly quantifies how many times wider the predistorted spectrum becomes relative to the input. For a 100 MHz 5G NR signal, a bandwidth expansion factor of 3× means the DPD must process a 300 MHz effective bandwidth to capture third-order intermodulation products.

The bandwidth expansion factor is directly proportional to the highest nonlinearity order the predistorter must compensate. A predistorter correcting up to the K-th order nonlinearity generates spectral components spanning K times the original bandwidth. Key relationships:

• 3rd-order compensation: 3× bandwidth expansion
• 5th-order compensation: 5× bandwidth expansion
• 7th-order compensation: 7× bandwidth expansion Higher-order correction improves ACLR but demands proportionally faster digital-to-analog converters and higher sampling rates in the observation receiver.

The bandwidth expansion factor directly dictates the minimum sampling rate required throughout the DPD signal chain. Per the Nyquist-Shannon theorem, the predistorter must operate at a sampling rate exceeding twice the expanded bandwidth. For a 200 MHz original signal with a 5× expansion factor, the effective bandwidth reaches 1 GHz, requiring sampling rates above 2 GSPS. This drives ADC/DAC selection, FPGA clock speeds, and overall system cost in wideband 5G and satellite communication systems.

To manage the high sampling rate demands imposed by large expansion factors, multi-rate DPD architectures decouple the predistorter's internal processing rate from the baseband data rate. The predistorter operates at an elevated sampling rate to capture out-of-band distortion products, while the baseband signal remains at its native rate. Interpolation filters upsample the signal before predistortion, and decimation filters reduce the rate in the feedback path. This approach optimizes computational efficiency without sacrificing linearization bandwidth.

If the DPD feedback path sampling rate is insufficient to capture the full expanded bandwidth, aliasing distortion occurs. Out-of-band spectral components fold back into the Nyquist zone, corrupting the observed signal used for coefficient training. This creates a false error signal that degrades predistorter performance rather than improving it. Anti-aliasing filters must be carefully designed to balance rejection of out-of-band energy against phase linearity requirements in the observation path.

In carrier aggregation scenarios, the bandwidth expansion factor applies to the total occupied spectrum spanning multiple component carriers. If two 100 MHz carriers are separated by 300 MHz, the effective signal bandwidth becomes 500 MHz. With a 5× expansion factor, the predistorter must process a 2.5 GHz instantaneous bandwidth. This extreme requirement drives the need for concurrent multi-band DPD architectures that linearize each carrier independently while canceling cross-modulation products between bands.

The physical phenomenon directly responsible for the bandwidth expansion factor. When a signal passes through a nonlinear power amplifier, intermodulation distortion (IMD) generates new frequency components outside the original channel. This spectral regrowth is the unwanted energy that spills into adjacent channels, and the predistortion process must intentionally generate an inverse bandwidth expansion to cancel it out.

The primary regulatory metric that drives bandwidth expansion factor requirements. ACLR quantifies the ratio of transmitted power in the assigned channel to power leaking into adjacent channels. 3GPP specifications mandate ACLR limits (typically -45 dBc for 5G NR base stations). The DPD system must expand the signal bandwidth sufficiently to cancel out-of-band distortion and meet these strict spectral masks.

The mathematical root cause of the bandwidth expansion factor. IMD products appear at frequencies that are sums and differences of integer multiples of the input signal frequencies. Third-order IMD (IMD3) falls closest to the original channel and is the most challenging to filter. The predistorter must generate inverse IMD products across an expanded bandwidth to achieve cancellation at the PA output.

An implementation architecture that directly addresses the bandwidth expansion factor challenge. The predistorter operates at a higher sampling rate than the baseband signal to capture and cancel out-of-band distortion products. For a 100 MHz 5G NR signal with a 5x expansion factor, the DPD path must process at 500 MHz or higher to synthesize the inverse nonlinearity across the full expanded bandwidth.

A critical failure mode when the bandwidth expansion factor exceeds system design limits. If the DPD observation receiver's analog-to-digital converter (ADC) sampling rate is insufficient to capture the full expanded bandwidth, the out-of-band distortion products fold back into the Nyquist zone, corrupting the feedback signal. This makes accurate coefficient extraction impossible and degrades linearization performance.

The regulatory consequence of insufficient bandwidth expansion factor handling. Out-of-band emissions are unwanted RF energy outside the licensed transmission bandwidth, strictly regulated by bodies like the FCC and ETSI. The DPD system's ability to expand the predistorted signal bandwidth and cancel spectral regrowth is the primary mechanism for achieving compliance with these emission limits.