The central challenge the team set out to solve is one that has dogged compact MIMO antenna designers for years. Placing several radiating elements close together on a small substrate inevitably causes mutual coupling, where energy intended for one element leaks into its neighbours, degrading isolation, diversity performance, and spectral efficiency. Achieving wide impedance bandwidth and wide axial-ratio bandwidth at the same time, while keeping the elements well isolated, has remained a formidable balancing act. Conventional fixes such as frequency-selective surfaces, electromagnetic bandgap structures, and defected ground planes can improve isolation, but they often enlarge the antenna, complicate fabrication, or narrow the very circular-polarisation bandwidth they are meant to protect.

The new design attacks the problem with an unusual combination of geometric tricks. Each of the four antenna elements is built from an asymmetrical ring radiator paired with an anisotropically defected ground structure and a hexagonally slotted circular patch. These features work together to excite pairs of orthogonal modes with equal amplitude and a ninety-degree phase difference, which is the essential recipe for circular polarisation. The result is a remarkably broad 3-dB axial-ratio bandwidth of roughly 47.8 percent, spanning about 5.74 to 9.34 GHz, alongside a 10-dB impedance bandwidth covering 4.62 to 10 GHz. The team verified the underlying physics using characteristic mode analysis, showing that specific orthogonal mode pairs exhibit the required ninety-degree characteristic angle differences at frequencies near 6.13, 7.5, and 9.2 GHz.

Isolation between the four ports is handled by a cleverly engineered frequency-selective surface. The unit cell, a 10 by 10 mm structure on the same 0.8 mm jeans substrate, is deliberately anisotropic so that it responds differently to orthogonally polarised waves. This is the key detail: a generic decoupling structure would suppress the surface waves that cause coupling but could also wreck the circular polarisation. The proposed band-stop cell, evolved through successive stages from a simple square loop into a composite of square loops, triangular loops, and rectangular partitions, blocks transmission below minus 10 dB from 4.5 to 15 GHz while preserving the antenna’s polarisation purity. Arranged in a cross-shaped layout between adjacent elements, the FSS array pushes inter-element isolation above 25 dB in simulation and above 30 dB in measurement.

The diversity metrics confirm that the array behaves as a proper MIMO system rather than four antennas sharing a board. The envelope correlation coefficient, computed from the scattering parameters, falls below 0.001 across the operating band, indicating that the ports behave nearly independently. Diversity gain exceeds 9.98 dB, the total active reflection coefficient stays below minus 10 dB, the mean effective gains of the ports differ by less than 3 dB, multiplexing efficiency remains above minus 0.1 dB, and channel capacity loss stays under 0.1 bits per second per hertz. Radiation efficiency reaches 98 percent with a peak gain of about 4.5 dBic, and the left-hand circularly polarised waves enjoy cross-polarisation isolation exceeding 22 dB in the broadside direction. Bending tests at radii of 40, 50, 60, and 70 mm confirmed that the flexible substrate tolerates the curvature of a real limb without catastrophic detuning.

The sensing half of the story is where the design becomes genuinely provocative. To model a human knee, the researchers built a cylindrical six-layer phantom replicating skin, fat, muscle, cartilage, bone, and synovial fluid, each layer assigned realistic thickness, density, conductivity, and permittivity. When the antenna rests on the phantom, the fringing near-fields at the patch edges penetrate the tissue stack. Fluid retention raises the effective permittivity of the medium, which capacitively loads the radiating edges, shifts the input impedance, and detunes the resonance. In simulation, the presence of water shifted the resonance from 6.1 GHz, where the reflection coefficient dipped to minus 46.87 dB, to 6.2 GHz with a shallower dip of minus 35.1 dB. The measured sensitivity was one megahertz of frequency shift per millilitre of fluid, with a minimum detectable volume of 20 mL at a depth of 30 mm below the skin layer.

Laboratory validation went beyond simulation. The team prepared a semisolid tissue phantom from agar, water, and corn starch mixtures formulated to mimic the dielectric properties of muscle, fat, skin, and bone, then embedded water-filled polythene packets to represent the oedema condition. The authors are careful to note that this discrete-packet approach models joint effusion, in which fluid accumulates within a confined compartment such as the synovial capsule, rather than diffuse interstitial oedema, and that it provides a controlled, repeatable ground truth for calibration. Measurements were performed in both monostatic configurations, where the same array transmits and receives, and bistatic setups emulating body-to-body wireless networks, with the transmission coefficient falling at roughly 20 dB per decade with distance. Time-domain A-scan analysis further localised the fluid response, with the differential scan highlighting scattering from the fluid layer at approximately 2.4 mm depth in the reflected signal.

Safety, of course, is non-negotiable for any device intended to hug human tissue. Specific absorption rate analysis on the six-layer phantom showed that at 100 mW of input power the 1-gram averaged SAR would reach 4.52 W/kg, exceeding the widely used 1.6 W/kg limit. Because SAR scales linearly with input power, the team calculated a maximum permissible input of 35.4 mW and conducted their compliance analysis at 20 mW with a 5 mm spacing above the skin. Under those conditions the simulated SAR was 0.904 W/kg for 1 gram of tissue and 0.186 W/kg for 10 grams on a normal phantom, dropping to 0.009 and 0.004 W/kg respectively on the hydrated phantom, comfortably within regulatory thresholds.

Compared with prior wearable and circularly polarised MIMO designs, the new antenna stands out on nearly every axis. Its 73.6 percent impedance bandwidth and 47.7 percent axial-ratio bandwidth exceed most reported flexible designs, many of which rely on rigid Rogers substrates unsuitable for conformal wear, suffer from high mutual coupling, or omit SAR analysis entirely. The 98 percent radiation efficiency is the highest among the compared works, and the 10-gram SAR of 0.931 W/kg is lower than several textile competitors. The dual-purpose nature of the design, serving simultaneously as a body-area-network communicator and a microwave fluid sensor, is the novelty the authors emphasise most strongly. If the phantom results translate to clinical settings, a fabric antenna stitched into a knee brace could one day give clinicians a low-cost, radiation-free way to track swelling after injury or surgery, simply by watching a resonance frequency drift.

Subject of Research: A wearable circularly polarised MIMO antenna on jeans substrate for wireless communication and microwave detection of knee oedema

Article Title: A wearable circularly polarised MIMO antenna for detecting oedema in human knee

Article References: Sabarish, J. G., & Das, P. (2026). A wearable circularly polarised MIMO antenna for detecting oedema in human knee. Results in Engineering, 32, Article 113435. https://doi.org/10.1016/j.rineng.2026.113435

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113435

Keywords: wearable antenna, MIMO, circular polarisation, oedema detection, frequency-selective surface, biomedical sensing, jeans substrate, axial ratio bandwidth, SAR, knee phantom, body-centric communication, microwave imaging