Over-the-Air DPD

Uses a spatially located observation antenna in the far-field to sample the combined radiated waveform. This captures the composite nonlinear behavior of the entire array, including mutual coupling and beamforming-dependent distortion, rather than isolating individual power amplifier outputs. The feedback path inherently includes over-the-air channel effects, requiring de-embedding techniques to isolate PA nonlinearity.

OTA DPD directly addresses the dynamic impedance modulation that occurs when beamforming weights change. As the array steers, each PA experiences a varying load impedance, altering its nonlinear characteristics. OTA DPD learns a beam-indexed or weight-aware predistorter that adapts linearization parameters based on the active beam configuration, ensuring consistent ACLR across all steering angles.

Unlike per-element DPD that linearizes individual PAs, OTA DPD targets the spatially combined E-field at the receiver location. This approach inherently compensates for:

• Antenna mutual coupling effects between adjacent elements
• Cross-coupling and crosstalk in the RF front-end
• Array manifold phase misalignment
• Beam-squint across wideband signals The result is a cleaner constellation and lower EVM at the intended receiver.

OTA DPD dramatically reduces hardware complexity by using a single observation receiver to capture the combined array output. This eliminates the need for per-branch feedback paths, couplers, and ADCs, which become prohibitive in massive MIMO systems with 64+ elements. The trade-off is increased algorithmic complexity to de-embed individual PA contributions from the composite feedback signal.

OTA DPD commonly employs an indirect learning architecture where the post-distorter is trained by swapping input and output signals. The over-the-air feedback is used to identify the inverse of the composite array nonlinearity. This approach avoids the need for an explicit PA model and directly computes predistorter coefficients that minimize far-field distortion, though convergence can be sensitive to measurement noise in the OTA channel.

In wideband massive MIMO systems, beam-squint causes the beam direction to shift with frequency across the signal bandwidth. OTA DPD can jointly compensate for this frequency-dependent spatial effect and PA nonlinearity by incorporating frequency-selective predistortion that accounts for the varying array factor at different subcarriers, maintaining linearity across the entire occupied bandwidth.

A linearization strategy that accounts for dynamic beamforming weight changes in phased arrays. As the beam is steered, the effective load impedance seen by each power amplifier shifts, altering its nonlinear signature. This technique adapts the predistorter coefficients in real-time based on the current beamforming configuration to maintain linearity across all steering angles.

The variation in load impedance presented to an individual power amplifier due to electromagnetic coupling with neighboring elements during beam steering. This mismatch causes the PA's AM-AM and AM-PM characteristics to change dynamically. Over-the-air DPD must compensate for this effect, as the far-field combined signal reflects the aggregate of all elements operating under varying impedance conditions.

A signal processing method to mitigate unintended electromagnetic interaction between adjacent antenna elements in a MIMO array. Energy radiated by one element induces currents in nearby elements, creating a coupled distortion path. Over-the-air DPD captures this combined effect in the far-field and computes an inverse model that jointly cancels both PA nonlinearity and inter-element crosstalk.

A cost-effective array linearization architecture using one observation receiver to sequentially sample outputs from multiple PAs. Rather than dedicating a feedback path per element, a switch network routes each PA's attenuated output to a shared receiver. The over-the-air variant captures the combined radiated field with a single probe antenna, dramatically reducing hardware complexity for massive MIMO systems.

A predistortion technique incorporating the array's spatial signature to jointly optimize linearization across all angles of departure. By modeling how nonlinear distortion projects into the far-field through the array manifold, this method ensures that spectral regrowth is suppressed not just at boresight but across the entire spatial sector. Critical for maintaining regulatory ACLR compliance in all directions.

A linearization method that explicitly models the S-parameter coupling network between antenna elements. The coupling matrix captures both mutual coupling and crosstalk paths. By inverting this matrix within the predistortion computation, the technique decouples the array's radiated field and linearizes each effective spatial channel independently, enabling cleaner far-field beam patterns.