Beam-Squint Compensation

In a phased array, traditional phase shifters apply a frequency-flat phase shift. For wideband signals, this causes different frequency components to constructively interfere at slightly different angles, creating beam squint. Compensation applies a frequency-dependent phase progression—often implemented via true time delay (TTD) units or digital FIR filters—to ensure all subcarriers point in the identical spatial direction. This is mathematically equivalent to removing the linear phase slope error across the array aperture.

Beam-squint becomes critically intertwined with digital predistortion (DPD) in massive MIMO transmitters. When squint is uncompensated, different frequencies illuminate different PA elements with varying effective isotropic radiated power (EIRP). This creates a frequency-selective nonlinear distortion profile across the band. A beam-squint-aware DPD must linearize not just per-element, but must account for the spatial-frequency coupling where a PA's nonlinear memory effects are excited differently depending on the instantaneous beam angle at each subcarrier.

The fundamental hardware solution to beam squint replaces or augments narrowband phase shifters with true time delay (TTD) elements:

• Phase shifters: Apply a constant phase φ, causing beam angle to vary as θ = arcsin(φ·c / 2π·f·d). Squint is proportional to fractional bandwidth.
• True time delay: Applies a constant time delay τ, making beam angle θ = arcsin(τ·c / d) frequency-independent.
• Hybrid architectures: Use TTD at sub-array level for coarse squint correction and phase shifters at element level for fine steering, balancing complexity and performance.

Modern systems implement squint compensation entirely in digital baseband before the DAC. Each antenna element's signal is convolved with a fractional delay filter—typically a Farrow structure or Lagrange interpolator—that applies the precise time delay required for coherent combining at the target angle. This approach eliminates bulky analog TTD networks and enables per-subcarrier beamsteering in OFDM systems. The filter tap resolution must be sufficient to achieve sub-picosecond delay accuracy for mmWave arrays with large fractional bandwidths.

When beam squint is left uncompensated in a linearized array, the adjacent channel leakage ratio (ACLR) becomes spatially anisotropic. The main beam at the carrier frequency may be perfectly linearized, but the spectral regrowth at band edges is radiated in a slightly offset direction. This creates a scenario where a user at the intended angle experiences higher out-of-band interference than predicted by boresight measurements. Compensation ensures the DPD's linearization benefit is spatially uniform across the entire occupied bandwidth.

The severity of beam squint is quantified by the bandwidth-angle product. For an array with element spacing d and number of elements N, the squint angle Δθ is approximately:

• Δθ ≈ (Δf/f_c) · tan(θ_0) where Δf is the signal bandwidth, f_c is the center frequency, and θ_0 is the scan angle. The 3 dB beamwidth is approximately λ/(N·d·cos(θ_0)). Squint becomes problematic when Δθ exceeds a fraction of this beamwidth. For a 64-element array at 28 GHz with 400 MHz bandwidth scanned to 45°, squint can exceed 2°, significantly degrading array gain.

The physical-layer alternative to phase-shifter-based beamforming that fundamentally eliminates beam squint. Instead of applying a frequency-independent phase shift, true time delay (TTD) elements introduce a variable time delay to each antenna element. This creates a beam angle that is constant across all frequencies because the time delay, not the phase, determines the direction. TTD is implemented using switched transmission lines, photonic delay lines, or tunable delay circuits. While TTD solves squint at the hardware level, it is often bulky and lossy at mmWave frequencies, motivating hybrid approaches that combine coarse TTD units with digital squint compensation.

A key metric defining when beam squint becomes significant. The fractional bandwidth is the ratio of signal bandwidth to center frequency (B/f_c). Squint becomes problematic when the fractional bandwidth exceeds approximately 5-10% in large arrays. For a 28 GHz mmWave system with a 400 MHz bandwidth, the fractional bandwidth is ~1.4%, yielding negligible squint. However, at 6 GHz with 400 MHz bandwidth, the ratio jumps to ~6.7%, causing noticeable beam deflection. This threshold guides system architects in deciding whether to implement squint compensation or accept the performance degradation.

A digital compensation technique that applies a frequency-selective phase correction to the baseband signal before digital-to-analog conversion. The correction term is derived from the array geometry and the desired steering angle. For a uniform linear array with element spacing d, the phase shift required at frequency f for angle θ is (2πf/c)·d·sin(θ). The squint compensation pre-calculates the phase error at each subcarrier relative to the center frequency and applies an inverse rotation. This method is effective for moderate fractional bandwidths but introduces computational overhead in OFDM systems with large FFT sizes.

The dual of beam squint in the spatial domain, also called spatial-wideband effect or beam broadening. While beam squint describes the frequency-dependent angle shift, the spatial-wideband effect describes how the array's spatial signature changes across subcarriers. This causes the channel's angle of arrival to appear frequency-dependent, degrading channel estimation and precoding in massive MIMO-OFDM systems. Compensation requires frequency-dependent beam codebooks or beamspace processing that treats each subcarrier's spatial channel independently. This effect is particularly severe in terahertz communications where fractional bandwidths can exceed 20%.

An integrated compensation architecture that simultaneously corrects for power amplifier nonlinearity and beam-squint-induced distortion across the array bandwidth. In a massive MIMO transmitter, each PA in the array experiences a different frequency-dependent load impedance due to squint, causing the nonlinear behavior to vary per element and per subcarrier. Joint linearization models this coupling by extending the DPD basis functions with frequency-dependent terms that capture the interaction between beam angle deviation and PA compression. This approach prevents the DPD from overfitting to the center-frequency beam pattern while leaving edge subcarriers under-corrected.

A passive microwave lens structure that provides true time delay beamforming without active electronics. The Rotman lens uses a parallel-plate region with carefully designed contour shapes to create linear time-delay progressions across its output ports. When fed from different input ports, the lens produces different beam angles with inherently low squint over wide bandwidths. Rotman lenses are commonly used in satellite communications and radar systems where bandwidths exceed 20%. Their primary limitation is insertion loss and physical size, making them impractical for very large arrays but valuable as a benchmark for hybrid digital-analog squint compensation schemes.