What 6G Will Actually Ask of Antenna Hardware

The step change from 5G to 6G — and what it means for antennas

6G is not 5G with higher data rates. It is a fundamentally different set of use cases — sub-millisecond latency for tactile internet, terabit-per-second throughput for holographic communication, sensing and imaging through RF, and AI-native air interfaces that adapt in real time. Each of these requirements pushes antenna hardware into territory that current designs do not reach.

The timeline is 2030 for initial deployments, 2032-2035 for meaningful scale. But the antenna technology is being developed now — and the decisions made in the next three years will determine which companies can supply the hardware when the standards are finalised. Understanding what 6G will demand is not speculative; it is a prerequisite for product planning.

Sub-terahertz spectrum: 100 GHz and above

6G will open spectrum above 100 GHz — the sub-terahertz range — to achieve multi-gigabit and terabit-per-second data rates. The W-band (75-110 GHz) and D-band (110-170 GHz) offer contiguous bandwidths of 20-50 GHz, compared to the 400 MHz-1 GHz available in 5G mmWave.

The physics challenge. At 140 GHz, wavelength is 2.1 mm. Antennas are tiny — a 100-element phased array fits in a 1 cm² package. But the precision required is extraordinary. A surface roughness of 10 micrometres — invisible to the eye — causes significant scattering loss at 140 GHz. Connector contact resistance, which is negligible at 5 GHz, becomes a dominant loss mechanism. Manufacturing tolerances that are acceptable at mmWave become intolerable at sub-THz.

Material evolution. Traditional PCB substrates (FR4, Rogers 4003) are too lossy above 100 GHz. Low-loss laminates (Rogers RT/duoid, Teflon-based) and silicon interposers become necessary. The antenna must be integrated with the transceiver in a module — separate antennas connected by waveguide are impractical due to the loss of any discontinuity.

Integration imperative. At sub-THz frequencies, the antenna cannot be a separate component. It must be co-designed and co-manufactured with the RF front-end, using package-level integration (antenna-on-package) or even on-chip integration (antenna-in-chip). This collapses the traditional antenna supply chain — the antenna becomes part of the semiconductor package, not a standalone product.

Reconfigurable intelligent surfaces (RIS): the antenna as a programmable mirror

One of the most transformative 6G concepts is the reconfigurable intelligent surface — a planar array of sub-wavelength elements that can independently adjust the phase (and sometimes amplitude) of reflected signals. RIS can redirect signals around obstacles, focus energy on specific users, or create programmable propagation environments.

How RIS works. Each element is a simple resonant structure — a patch, slot, or dipole — backed by a tunable load (varactor diode, PIN diode, or MEMS switch). By adjusting the load, the element’s reflection phase is shifted by 0-360°. A surface with hundreds or thousands of elements can create an arbitrary reflected wavefront — steering a beam, focusing at a point, or creating a null in a specific direction.

The antenna design challenge. RIS elements must be sub-wavelength (typically λ/4 to λ/2 spacing) to avoid grating lobes. At 28 GHz, that is 2.7-5.4 mm per element — a surface with 1000 elements measures roughly 10×10 cm. The control network — the wiring that sets each element’s phase — must not disturb the RF performance. The entire surface must be thin (less than λ/10) to conform to walls, ceilings, or building facades.

Manufacturing at scale. RIS requires thousands of tunable elements per surface, each with its own control circuitry. The cost per element must be below $0.10 for the technology to be economically viable — requiring semiconductor-grade manufacturing processes. Traditional PCB fabrication is insufficient; the industry is moving toward wafer-level packaging and integrated circuit techniques borrowed from the semiconductor industry.

Joint communication and sensing (JCAS): the antenna as radar

6G will integrate communication and radar sensing into a single waveform — enabling devices to image their environment, track motion, and detect material properties using the same hardware that transmits data.

Antenna requirements for JCAS. Sensing requires wide bandwidth (for range resolution), low sidelobes (for detecting weak targets near strong reflectors), and MIMO architectures (for angular resolution). The same antenna must simultaneously support communication (optimised for link quality) and sensing (optimised for waveform fidelity). This is a multi-objective optimisation problem with no trivial solution.

Waveform considerations. Sensing waveforms require low autocorrelation sidelobes — typically achieved through frequency-modulated continuous wave (FMCW) or orthogonal frequency-division multiplexing (OFDM) with specific pilot patterns. The antenna must maintain linear phase response across the band to preserve waveform fidelity — a specification that is typically ignored in communication antenna design but is critical for sensing.

Calibration. JCAS requires the antenna array to be calibrated — the relative phase and amplitude of each element must be known to within a fraction of a degree. Manufacturing tolerances, temperature drift, and aging all introduce errors that degrade sensing performance. Self-calibration algorithms can compensate, but they require the antenna to have sufficient observability — typically achieved through coupling networks or dedicated calibration elements.

AI-native antenna design and operation

6G will use AI not just for network optimisation but for real-time antenna control. The antenna will adapt its pattern, polarisation, and beam steering in response to the environment, user mobility, and traffic demands — all on millisecond timescales.

Real-time beam management. Current 5G beam management uses a codebook of predefined beams, selected through a search process that takes 10-100 ms. 6G aims for sub-millisecond beam switching, requiring the antenna to generate arbitrary beams on demand. This demands a fully digital beamforming architecture with one transceiver per element — a significant increase in hardware complexity and power consumption.

Environment-aware optimisation. The antenna system will receive inputs from cameras, LiDAR, and other sensors to build a 3D model of the environment, then compute the optimal beam pattern for each user. This requires the antenna control software to interface with the device’s AI stack — a new level of integration between the RF hardware and the digital processing layer.

Digital twin. Each antenna deployment will have a digital twin — a simulated model that is continuously updated with real-world measurements. The digital twin predicts how the antenna will perform in different configurations, enabling the AI to select the optimal settings without physical trial-and-error. This requires the antenna to be characterised in sufficient detail to populate the digital twin — a significant data acquisition and modelling challenge.

What this means for antenna engineers today

• Learn semiconductor packaging. The antenna is moving onto the chip package. Antenna engineers who understand PCB layout but not IC packaging will find their skills increasingly marginalised.
• Study metasurfaces. RIS and related technologies are rooted in metamaterial physics. Understanding sub-wavelength resonators and their tunable properties is becoming essential.
• Embrace co-design. The antenna can no longer be designed in isolation. It must be co-optimised with the transceiver, the digital processing, and the AI control layer.
• Prepare for new specifications. Waveform fidelity, phase linearity, and calibration stability — previously ignored in communication antenna design — are becoming first-order requirements.

6G is not a distant abstraction. The standards are being written now, the prototypes are being built now, and the supply chain is being assembled now. The antenna technology required — sub-THz integration, reconfigurable surfaces, joint communication-sensing, AI-native control — represents a step change in complexity and capability. The engineers and companies who understand these requirements today will define the products of 2030.