The transition from biometric-focused fitness trackers to high-fidelity Spatial AI hardware represents a fundamental paradigm shift in the landscape of ambient computing. As next-generation wearables—ranging from AR glasses to AI-driven smartwatches—evolve to support the low-latency, high-bandwidth demands of 5G-Advanced, Wi-Fi 7, and Satellite NTN, they encounter a formidable convergence of physical and electromagnetic constraints. The traditional ‘antenna keep-out zone’ has effectively vanished, replaced by densely packed sensor suites and premium metallic enclosures that act as near-field suppressors. Navigating the fundamental Chu-Wheeler physical limits requires a departure from legacy design; the antenna can no longer be a discrete peripheral, but must emerge as a system-integrated radiating structure. By leveraging Characteristic Mode Analysis (CMA), Antenna-in-Package (AiP) architectures, and AI-driven geometric optimization, engineers are now forced to synthesize performance within sub-wavelength volumes, ensuring that the wireless link becomes the enabler, rather than the bottleneck, of the burgeoning AI edge-computing era
Introduction: the evolution of wearable connectivity
The evolution of wearable technology from simple BLE-enabled telemetry trackers to high-fidelity Spatial AI nodes marks a fundamental shift toward ambient computing, necessitating a transition to high-bandwidth 5G-Advanced and Wi-Fi 7 protocols
Spatial AI changed the math. AR glasses, AI pins, and flagship smartwatches now want high-bandwidth 5G, Wi-Fi 7, and always-on links to the cloud. The reason is ambient computing: when the device has to run real-time AI, much of the heavy work gets offloaded to the cloud or the edge. That offload only works if the link is fast and steady, which puts direct pressure on antenna efficiency. When the antenna underperforms, the AI stutters instead of just dropping a call. As real-time AI offloading demands deterministic low-latency and high spectral efficiency, the antenna system’s performance becomes the primary determinant of Total Radiated Power (TRP) and battery longevity
However, this requirement for increased wireless throughput collides with a shrinking internal landscape; the proliferation of dense biometric sensor suites (ECG, PPG, IMUs) and premium metallic enclosures has effectively eliminated the traditional electromagnetic “keep-out zone.” Consequently, antenna design must move beyond discrete components to chassis-integrated radiating structures, synthesizing performance within congested, sub-wavelength volumes where the industrial design and the RF system are inextricably linked.
Core physical and engineering challenges
The hard ceiling here is physics. The Chu-Harrington limit says that as an antenna gets smaller relative to the wavelength it has to radiate, its bandwidth and efficiency fall. Shrink the antenna and you lose data rate, range, and battery life.This fundamental bottleneck is compounded by the move toward premium metallic housings—such as titanium and stainless steel—which act as near-field shields, necessitating sophisticated aperture synthesis and chassis-integrated design to prevent the enclosure from functioning as a Faraday cage. Furthermore, the body loading effect from proximal human tissue introduces significant parasitic capacitance and dielectric loss, which detunes resonant frequencies and complicates Specific Absorption Rate (SAR) compliance. Managing multi-radio coexistence within these congested sub-wavelength volumes adds a final layer of complexity, requiring extreme isolation and filtering to manage simultaneous signals across a dense spectral landscape ranging from legacy GPS and LTE to high-frequency Wi-Fi 7 and Non-Terrestrial Network (NTN) bands
Advanced design architectures
To overcome the physical bottlenecks of traditional antenna placement, advanced design architectures are shifting toward a holistic integration of the radiator into the device’s structural and semiconductor DNA. Antenna-in-Package (AiP) technology mitigates the interconnection losses associated with mmWave and emerging 6G frequencies by embedding radiating elements directly within the IC substrate, while Characteristic Mode Analysis (CMA) allows engineers to transcend metallic interference by transforming the device’s own conductive chassis—such as a titanium watch frame—into a primary, controlled radiating structure. This morphological flexibility is further enhanced by Laser Direct Structuring (LDS) and high-density Flexible Printed Circuits (FPC), which enable the conformal mapping of antennas onto the complex, non-planar internal surfaces of wearable hardware. Implementing MIMO architectures within these congested volumes requires decoupling strategies to maintain spatial diversity and minimize the envelope correlation coefficient (ECC), ensuring robust multi-path performance even when radiating elements are separated by sub-wavelength distances.
Materials and emerging technologies
New materials are opening doors that geometry alone cannot. Liquid metal alloys and stretchable conductive polymers enables the development of elastomeric radiators capable of maintaining stable electromagnetic resonance under high mechanical strain, facilitating seamless integration into ergonomic straps and flexible substrates Metamaterials and high impedance surfaces attack a specific enemy: surface waves and interference from the metallic backplane. Engineered surfaces can suppress those effects and recover efficiency that the enclosure would otherwise eat. High-permittivity dielectrics do something neat. Specialized ceramics or polymers electrically lengthen the antenna, so it behaves like a bigger radiator while staying physically small.
The role of AI in antenna design and optimization
Antenna design has always been part math, part craft. AI is starting to change the craft part. Genetic algorithms and machine learning can explore shapes no human would draw, finding geometries that fit irregular spaces and still radiate well. The results sometimes look strange. They work because the optimizer doesn’t care what an antenna is supposed to look like. AI-driven dynamic impedance matching enables real-time adaptation to fluctuating “body loading” effects, utilizing predictive models to adjust tunable matching networks as the user’s physical interaction with the device changes.
Neural beamforming extends the idea to spatial hardware. For AR glasses, AI-based predictive steering can maintain high-throughput links while significantly reducing the power consumption and computational overhead associated with continuous beam-tracking. This keeps the link pointed the right way, which improves connectivity and saves battery at the same time.
Case studies: specific hardware applications
The practical implementation of these antenna strategies is best exemplified by the divergent electromagnetic constraints of emerging form factors, ranging from high-bandwidth spatial hardware to ultra-miniaturized wearables Next-generation smartwatches must navigate the transition from plastic to premium metallic and sapphire-backed enclosures, necessitating the use of the chassis itself as a radiating element to maintain efficiency for optimal RF performance. Spatial AI glasses distribute the problem. Antennas spread across the temples and bridge give 360-degree coverage, but now heat from the compute has to be managed right next to the user’s head, and the frame has to stay light enough to wear. Smart rings are the extreme case. Multi-band antennas in sub-10 mm radii, with almost no internal air space, force every technique above into a volume smaller than a coin.
Future outlook and conclusion
6G will push toward terahertz frequencies. Shorter wavelengths mean smaller antenna elements, which sounds like relief, until you remember that THz links are fragile and demand line of sight. The size problem gets traded for a pointing problem. Non-terrestrial networks are the other path. Satellite links built into wearables could fill the gaps where cellular coverage dies, and NTN-capable antennas are already moving from concept to product. Sustainability deserves a mention too. Integrated LDS plastics and mixed metals are hard to separate at end of life, and the industry will have to reckon with recycling as these devices multiply.
The thread through all of it is simple. The antenna used to be a component. In AI-driven hardware, it is a system. In the era of Spatial AI and ambient computing, where real-time intelligence is tethered to the quality of the electromagnetic link, the antenna has become the central architectural bottleneck and the primary enabler of device performance. As physical keep-out zones vanish, the successful integration of advanced materials, AI-driven optimization, and chassis-integrated architectures will distinguish the next generation of hardware. Ultimately, the antenna is no longer just a component—it is the critical interface between the physical user and the global digital intelligence layer, and its evolution will dictate the boundaries of what is possible in the post-smartphone era .
Juhi Godhwani is an antenna engineering leader specializing in compact multiband antenna systems for advanced wireless devices. Her work spans LTE, Wi-Fi/Bluetooth, GNSS, NFC, and emerging NB-NTN technologies, from research and prototyping through regulatory certification and high-yield mass production. She is a named inventor on multiple U.S. antenna patents and a member of IEEE.
