Beam Broadening Approaches

• Beam Broadening Approaches are techniques that intentionally expand the beam’s mainlobe to mitigate beam squint in wideband/sub-THz systems.
• They leverage methods such as quadratic-phase profiles, frequency-averaged eigenmode synthesis, and beamspace windowing to maintain spectral efficiency.
• These strategies enable tiled hybrid architectures to achieve robust performance while reducing hardware complexity and power consumption.

Beam broadening approaches encompass a set of algorithmic and hardware strategies in wideband hybrid and massive MIMO systems that intentionally expand the spatial mainlobe of a beam, typically to address the beam squint phenomenon in wideband or sub-THz multiuser settings, reduce sensitivity to array imperfections, or design tile-level architectures with reduced RF complexity. These approaches span phase-only broad beams, quadratic-phase profiles, frequency-averaged eigenmode synthesis, partitioned beams, spatial windowing in beamspace, and even physical array reconfigurations. The motivation and implementation of beam broadening are fundamentally linked to the rise of tiled (subarray-based) hybrid beamforming architectures in modern millimeter-wave and sub-THz systems.

1. Physical Origins and System-Level Motivation

Wideband analog (RF) beamforming in massive MIMO and tiled hybrid arrays gives rise to the beam squint effect—a frequency-dependent shift of the beam's spatial angle that causes mainlobe misalignment and pronounced gain loss outside of narrow carrier-matched scenarios. This effect is quantified by the spatial frequency dispersion $\Delta\Omega = |\Omega_c|\,(B/f_c)$, where $\Omega_c$ is the carrier-frequency spatial frequency and $B/f_c$ is the fractional bandwidth. Narrowband phase-only beamformers achieve peak gain at $f=0$ but experience rapid gain roll-off across the band, particularly at larger steering angles (Haddad et al., 6 Dec 2025).

Beam broadening addresses this challenge by expanding the angular/spatial support of the beam pattern on each tile or subarray, ensuring that a broader mainlobe covers the spatial spread induced by the frequency-dependent channel structure—without resorting to cost-prohibitive true time delay (TTD) circuits or amplitude-phase controlled hardware. Practical tiled architectures enable the hardware to realize such broadening at reduced complexity, power, and insertion loss (Zhang et al., 2019, Haddad et al., 6 Dec 2025, Chung et al., 2020).

2. Quadratic-Phase and Frequency-Averaged Beam Designs

A principal method for beam broadening leverages phase-only quadratic phase profiles across tile antennas. The RF beamforming vector per tile is set according to

$$\phi[n] = \Omega_c n + \frac{\Delta\Omega}{2N_a} n^2,\quad n = -\frac{N_a-1}{2},\ldots, \frac{N_a-1}{2}$$

which yields a mainlobe spanning $\Omega_c \pm \Delta\Omega/2$, ensuring that the beam retains array gain across the target frequency band (Haddad et al., 6 Dec 2025). This quadratic phase profile can be efficiently realized with phase shifters and does not require per-antenna amplitude control.

Another approach involves frequency-averaged dominant eigenmode synthesis: an approximation of the dominant eigenvector of $$\frac{1}{B} \int \mathbf{h}_{mk}(f)\mathbf{h}_{mk}^H(f)df$$, implemented by phase-only matching to the principal eigenvector. Partitioned broad beams may also be employed, where tiles are assigned to distinct sub-bands of spatial frequencies.

Empirical findings indicate that quadratic-phase broadening can recover up to 80% of the ideal spectral efficiency at large steering angles, and, for fractional bandwidths up to 20%, nearly bridge the gap between phase-only and full TTD-equipped tiles (Haddad et al., 6 Dec 2025).

3. Beamspace and Windowed Processing in Tiled Architectures

In beamspace-based hybrid architectures, each tile executes a spatial DFT across its aperture, and a windowing (selection) operation in the DFT domain isolates the mainlobe and its frequency spread. AoA-dependent beamspace windows are chosen to match the mainlobe width over frequency, enabling reduced-dimensional representation that encapsulates the dominant wideband spectral content (Noroozi et al., 6 Dec 2025). Window size is chosen precisely to match the mainlobe width, with tradeoffs between mainlobe fidelity and sidelobe control.

The table below summarizes the main methods:

Broadening Strategy	Implementation Domain	Typical Use Case
Quadratic phase profile	RF, tile-level, phase-only	Wideband squint mitigation
Partitioned broad beams	RF, tile assignment	Angular sub-band multiplexing
Frequency-averaged eigenmode	RF, phase-only	Robust tile-level design
Beamspace windowing (DFT)	Digital, per-tile beamspace	Radar/MIMO, dimensionality reduction

Beamspace approaches are particularly key for scalable radar arrays, where reduction in processing and training complexity is essential (Noroozi et al., 6 Dec 2025).

4. Adaptive and Reconfigurable Array Techniques

Emerging reconfigurable architectures such as movable antenna (MA)-aided hierarchical sub-connected hybrid beamforming (HSC-HBF) physically shift tile centers to provide additional geometric degrees of freedom for broadband focusing. By optimizing the tile positions $\{\Delta_t\}$, these arrays emulate the focusing effect of TTD, actively compensating for the frequency-dependent propagation paths in the near field and achieving robust squint suppression (Zhang et al., 11 Nov 2025). Simulations indicate that the MA-HSC-HBF architecture holds $\geq95\%$ of the center-frequency gain across the entire frequency band and offers up to $144\%$ improvement in sum-rate over conventional fixed-point arrays.

This approach reduces the need for hardware- and calibration-intensive TTD networks and remains algorithmically tractable via block-coordinate and convex subproblem optimization.

5. System Trade-Offs: Performance, Complexity, and Hardware

Beam broadening methods are critically intertwined with the hardware and architectural choices of contemporary tiled (subarray) hybrid arrays:

• Tile size ($N_a$) vs. number of tiles ($N_d$): Larger tile size (fewer tiles) improves analog gain and reduces front-end power, but exacerbates beam squint due to narrower per-tile beamwidth. Smaller $N_a$ increases multiuser flexibility and spatial DoF, at higher power consumption (Haddad et al., 6 Dec 2025).
• Phase-only vs. amplitude-phase control: Phase-only broadening, particularly via quadratic-phase designs, is sufficient for up to 20% fractional bandwidth. Full amplitude-phase or TTD hardware is not required in many regimes, reducing complexity.
• Shared vs. disjoint tile assignment: Disjoint allocation with broad beams and per-subcarrier digital processing matches or outperforms full sharing for most practical user separations, with less hardware complexity.
• Power and efficiency: Tiled architectures with beam broadening cut RF-chain power by a factor of 3–4 while preserving near-optimal spectral efficiency for fixed wireless access and small cell deployments (Haddad et al., 6 Dec 2025, Zhang et al., 2019).

A plausible implication is that system design should simultaneously optimize tile size, tile beamwidth, and allocation, targeting the regime where phase-only broadening suffices for the operational bandwidth.

6. Algorithmic and Experimental Validation

Numerical and anechoic-chamber experiments have validated the feasibility and performance boundaries of beam broadening:

• Quadratic-phase broadening: For $\theta=55^\circ$ in a wideband sub-THz system, standard narrowband beamforming incurs over 2 b/s/Hz spectral efficiency loss due to beam squint. The quadratic-phase broad beam recovers approximately 80% of the ideal spectral efficiency (Haddad et al., 6 Dec 2025).
• Beamspace tiled MVDR: In massive MIMO radar, tiled windowed-beamspace architectures achieve detection and interference rejection comparable to full-dimensional MVDR, with $>10^4\times$ reduction in covariance matrix inversion cost (Noroozi et al., 6 Dec 2025).
• MA-aided HSC-HBF: This approach emulates TTD focusing and achieves sum-rate gains up to $144\%$ over fixed-point arrays (Zhang et al., 11 Nov 2025).

7. Design Guidelines and Practical Recommendations

The current literature provides several key design takeaways for beam broadening in tiled and hybrid systems:

• Phase-only RF broadening (quadratic-phase) eliminates the need for TTD or amplitude control in fractional bandwidths up to 20%.
• Tile sizes of $N_a\approx32$ (for $N=256$ element ULA) achieve a balance of power, analog gain, and loss, with 8–12 W front-end power typical in sub-THz systems (Haddad et al., 6 Dec 2025).
• For well-separated users, single-user RF beamforming with broad beams achieves $>95\%$ performance of hybrid MU-MIMO.
• Beamspace windowing and partitioned broadening can be tuned to dimensionality, power, and scalability constraints, with mainlobe-matching windows optimizing spatial-spectral trade-offs.
• Newer spatially reconfigurable architectures (MA-HSC-HBF) further extend robust broadband beamforming to near-field and ultra-wideband scenarios.

Collectively, beam broadening approaches represent a critical enabler for scalable, energy-efficient, and spectrally robust hybrid architectures in next-generation MIMO, radar, and sub-THz wireless networks (Haddad et al., 6 Dec 2025, Noroozi et al., 6 Dec 2025, Zhang et al., 11 Nov 2025, Zhang et al., 2019).