Air-Ground Beam Steering

• Air–ground beam steering is a technology enabling 2D (azimuth and elevation) directive links using advanced techniques like leaky-wave, holographic metasurfaces, and RIS-based eigenmode feeders.
• The approach leverages both hardware innovations (e.g., electronically steered arrays and reconfigurable elements) and algorithmic methods (e.g., manifold optimization and hash-based beam training) to achieve fast, sub-millisecond reconfiguration.
• Integrating sensor data and multidomain optimization improves resilience to multipath, clutter, and mobility, paving the way for real-time air–ground communication and ISAC applications.

Air–ground beam steering refers to the class of electromagnetic architectures, circuit techniques, and array optimization algorithms enabling two-dimensional (azimuth and elevation) directive communication and/or sensing links between aerial platforms (UAVs, HAPs, aircraft) and terrestrial networks. Unlike traditional ground-fixed beamforming, air–ground scenarios demand extreme angular flexibility, sub-millisecond update rates, and resilience to multipath, clutter, and platform dynamics. Contemporary solutions leverage mm-wave leaky-wave arrays, reconfigurable reflectarrays, RIS-based eigenmode feeders, programmable metasurfaces, and hashing-based codebook training, with approaches spanning both hardware (aperture engineering, reconfigurable elements, hybrid feeds) and algorithmic (Riemannian manifold optimization, matching pursuit, hash-based search) domains. The following sections systematically address the dominant physical, circuit, and algorithmic frameworks for air–ground beam steering, referencing current research throughout.

1. Leaky-Wave and Holographic Antenna Architectures

Electronically steered leaky-wave configurations constitute a central approach for compact mm-wave air–ground links. Half-width microstrip leaky-wave antennas (HWMLWA) support a slow-wave TM₀ mode whose phase constant $\beta(\omega)$ enables continuous-angle beam scanning governed by the dispersion law

$$\sin\theta(\omega) = \frac{\beta(\omega)}{k_0(\omega)},$$

with $k_0$ the free-space wavenumber. In electronically loaded geometries, periodic lumped varactors modulate $\beta$ at fixed frequency, allowing for elevation beam steering through voltage control: $$\beta = \beta_0(\omega) + \Delta\beta_{\mathrm{var}}(V_\mathrm{CONT}),\qquad \theta(V_\mathrm{CONT}) = \arcsin\bigg(\frac{\beta_0(\omega_0) + \Delta\beta_{\mathrm{var}}(V_\mathrm{CONT})}{k_0(\omega_0)}\bigg).$$ Azimuth steering is imparted by binary modulation of asymmetric quasi-patch diode loads. For instance, varying the number of ON/OFF diodes on each side ($N_R$/$N_L$) produces a quasi-linear azimuthal steering law: $$\varphi(D_L, D_R) \approx \arctan\big[\alpha \cdot (N_R - N_L)\big],$$ with empirical angular coverage exceeding $90^\circ$. A prototypical realization on Rogers RO4350B demonstrates a peak realized gain of $8$ dBi across $28$–$34$ GHz, radiation efficiency above $80\%$, and $3$ dB E-plane beamwidths of $8$–$12^\circ$. Two-dimensional steering over $\sim$86° in elevation and 90° in azimuth is validated at microsecond reconfiguration speed; platform integration uses IMU and closed-loop bias control for real-time correction of pointing deviations (Alesheikh et al., 2024).

Wideband anisotropic holographic metasurfaces similarly enable 2D steerable high-gain operation. Engineered metasurface phase gradients $\varphi(x,y)$ create leaky-wave apertures whose mainlobe couples to user-defined $(\theta_0,\varphi_0)$, following

$$\varphi_{mn} = -k_0 \big[m\Delta_x\sin\theta_0\cos\varphi_0 + n\Delta_y\sin\theta_0\sin\varphi_0\big],$$

where $(m,n)$ index the cell positions. Frequency tuning and input phase gradients yield elevation and azimuth scanning, respectively; coverage spans $\theta:35^\circ$–$57^\circ$, $\varphi:\pm58^\circ$, gain peaking at $24.5$ dB and $86\%$ measured efficiency. Planar reflectors suppress backward modes, producing robust sectoral and point-to-point air–ground coverage (Mohammadi et al., 15 Jul 2025).

2. Reconfigurable Intelligent Surface and Eigenmode Feeder Systems

RIS-based beamforming couples a low-complexity active multi-antenna feeder (AMAF) to a large passive reflecting aperture, producing far-field radiation patterns through eigenmode matching of the AMAF–RIS near-field channel matrix $H$. The SVD $H = U\Sigma V^H$ furnishes principal feed modes, with the AMAF excitation $\mathbf{w}_A = v_1$ and RIS phase-only weights set by the $u_1$ eigenvector. This technique yields extremely narrow (HPBW $\sim 0.8^\circ$ for $D_R=64\lambda$), ultra-low-sidelobe (PSL $<-50$ dB) pencil beams, or broad flat-topped sectors (mode mixing)—all with only $N_a\ll N_p$ active RF chains.

Key power scaling originates from the compensation of free-space path loss by increasing RIS aperture: $$G_0 \propto \frac{D_A^2 D_R^2}{d^2} \cdot \frac{D_R^2}{\lambda^2},$$ with $d$ the AMAF–RIS separation. If $d$ scales linearly with $D_R$ ($d\propto D_R$), total boresight gain remains nearly independent of RIS size, allowing scaling to arbitrarily high angular resolutions at constant PA DC power. The RIS approach achieves dramatic reductions in system complexity and heat dissipation, crucial for airborne or power-constrained nodes. Phase-only control further allows real-time (sub-millisecond) reprogramming for agile beam/handoff operations (Tiwari et al., 2022).

3. Active–Passive Array Integration and Optimization

Active–passive antenna integration, as realized with omni-steering plates, overcomes the limited half-space coverage of standard base-station arrays. By replacing the metallic back-reflector with an array of $M$ passive, varactor-tunable LC resonators, a uniform linear active array achieves 360° air–ground beam steering. The omni-steering plate dynamically toggles between transmission and reflection, re-directing otherwise wasted backward energy to specified aerial or ground sectors.

The joint sum mutual information (MI) maximization problem for simultaneous S&C is formulated as: $$\max_{\alpha,\mathbf{W},\boldsymbol\Theta}\;\; \kappa \sum_{k=1}^K I_{C,k} + (1-\kappa)I_S$$ subject to scheduling, power, and plate-modulus constraints. The optimization decomposes into passive-coefficient adjustment (solved via Riemannian gradient ascent over the complex circle manifold) and sum-weighted MMSE active-beamforming (convex QCQP with soft scheduling variable relaxation and log-barrier penalty). Quantitative evaluations demonstrate full 360° angular beampattern coverage, MI improvements up to $47\%$, and NMSE reductions exceeding $87\%$ over MMSE or random-phase baselines, with robustness to user scheduling and target alignment (Zhu et al., 18 Jan 2026).

4. Fast Beam-Training and Codebook Search Techniques

Large-scale air–ground arrays require operationally tractable beam alignment in the presence of hundreds to thousands of pencil beams covering angular octants. Hashing Multi-Arm Beam (HMB) training provides a scalable solution: by constructing multi-arm codebooks via independent hash functions, the search for the optimal air–ground link reduces from $O(N_C)$ slots (exhaustive) to $Q=LB=O(\ln(1/\epsilon))$ for target mis-identification error $\epsilon$, where $B$ is the per-round bucket size and $L$ the number of independent hash rounds. The technique:

• Constructs a multi-dimensional codebook $\mathcal{C}_1$ covering all steerable pencils.
• Hashes beam indices into multi-beam buckets (arms) using independent bit extraction.
• All APs simultaneously transmit pilots along designated multi-arm beams; users record aggregate received power.
• Voting and soft-decision processes identify the beams eliciting maximal energy, resilient to fast channel variations and multi-AP interference.

Simulation results confirm $>96\%$ alignment accuracy at $Q=24$ (vs $64$ exhaustive) slots for $8\times8$ UPAs, with strong performance under SNR of $0$ dB and dynamically moving UAVs (Xu et al., 2024).

5. Performance, Integration, and Air–Ground-Specific Challenges

Air–ground links impose stringent requirements on scanning range, link budget, and dynamic compensation:

• Steerable apertures must cover elevation from horizon to broadside ($>80^\circ$) and azimuth over at least $\pm50^\circ$, satisfying both ground acquisition and aerial handoff.
• Electronic steering (varactor/PIN-diode/RIS phase) yields sub-microsecond scan times, enabling kHz-rate updates for target tracking and multi-beam hopping.
• Millimeter-wave atmospheric effects (e.g., $0.05$ dB/km clear-air loss at $30$ GHz) and mechanical platform jitter are addressed by integrating sensor data (IMU, INS, GNSS) into active bias feedback.
• Link budgets for mm-wave air–ground links (e.g., $f=18$ GHz, $1$ km range) hinge on high antenna gains ($>24$ dBi), low VSWR, and thermal management via substrate and amplifier selection.

Table: Representative Performances for Air–Ground Beam-Steering Solutions

Approach	Gain (dBi)	Scan Range (θ,ϕ)	Efficiency (%)	Scan Rate
Leaky-wave HWMLWA	7.5–8.2	θ: ±34–+52°, ϕ: ±52–+38°	>80	ns–μs (V, PIN)
Holographic Metasurface	24.5	θ: 35–57°, ϕ: ±58°	86	N/A (fixed/PW)
RIS Eigenmode Feeder	39	(HPBW <1°)	N/A	sub-ms
Omni-Steering Plate	(Array-dependent)	360° az/dec	(Simulation)	(ms-level)

6. Prospects, Limitations, and Open Research Directions

Recent advances in compact, high-efficiency, electronically agile air–ground beam steering have rendered real-time full-space coverage tractable. However, limitations persist in mutual coupling, hardware nonidealities (Q-factor, bandwidth), mechanical integration, and scalability to ultra-large arrays. RIS and hybrid active/passive structures mitigate power draw and thermal constraints but necessitate sophisticated channel modeling, SVD-based precoder recalibration, and robust manifold optimization.

A plausible implication is that future air–ground networks will increasingly rely on hierarchical hybrid multi-aperture designs (combining electronically steerable leaky-wave layers, RIS macro-surfaces, and fast codebook search), tightly coupled with closed-loop sensor data fusion and distributed optimization algorithms to track and compensate aerial mobility in real time. The integration of beamforming, sensing, and communication (ISAC) will become central to both link-level and system-level designs for aerial communication and radar platforms.

References: (Alesheikh et al., 2024, Zhu et al., 18 Jan 2026, Mohammadi et al., 15 Jul 2025, Tiwari et al., 2022, Xu et al., 2024).