betaFeedback

WorksEngineering

Phase-Matched Design of Pinching-Antenna Elements on a 60-GHz Dielectric Rod Waveguide

Abstract

The radiation of pinching antennas, dielectric elements placed on a dielectric waveguide, depends on the element shape. We explain it with a phase-matching, or prism-coupler, picture: the guided rod mode couples into the denser antenna plate at an angle set by the plate-mode effective index, about 31 degrees at 60 GHz, and radiates through the face it reaches first. This yields a simple sizing rule linking element height to contact length. Rectangular, isosceles and right-triangular elements of equal contact length and height are compared from 53 to 67 GHz in full-wave openEMS simulations including the waveguide-to-rod transitions. The rectangle peaks sharply at the rule, reaching 15.9 dBi. The right triangle keeps its gain within 0.7 dB and its side lobes at least 10 dB down over a two-to-one height range. At the rule, both outperform the isosceles element by 1.2 to 2.1 dB, and all remain matched.

Introduction

Pinching antennas (PAs) radiate from a low-loss dielectric waveguide by bringing a small dielectric element into contact with it, so that part of the guided power is coupled out where the element is placed [1]. System-level studies exploit the freedom to position and activate such elements along the waveguide, and mostly model each PA as an isotropic point source [2, 3, 4, 5]; directional element patterns have only recently been included in the channel model [6]. Li et al. [7] showed with full-wave simulations and measurements at 60 GHz that the pattern of a PA depends strongly on its shape: a regular triangle radiates a single, cleaner beam than a square of the same thickness, which was attributed to reduced internal reflections and a tapered polarization current. What is still missing is a design rule: which geometric feature of the element sets its gain, beamwidth and beam direction, and how the element should be dimensioned.

This paper derives such a rule from a phase-matching picture of the PA and tests it with a systematic full-wave comparison of three outlines that share the same contact length and height. The contributions are: (i) we show that the PA, being denser than the guided mode, acts as a prism coupler [8]: power leaves the rod at an angle set by the effective index of the plate mode of the element, not by the bulk permittivity, and this angle predicts both the optimum element height and the beam direction; (ii) we quantify how the contact length LL and the height hh of rectangular, isosceles and right-triangular elements set the S-parameters, gain, beamwidth and beam direction; (iii) we show that elements with a vertical downstream face sized by this rule outperform the isosceles shape, and that a right-triangle outline makes the element tolerant to its height while keeping the side lobes low; and (iv) we provide a model of the complete system in the open-source FDTD solver openEMS [9], including the circular-waveguide (CW) to rod transitions, which reproduces the main-lobe directions reported in [7].

System Model and Numerical Method

The system follows [7] (Figure 1a). A PTFE rod (εr=2.1\varepsilon_{\mathrm{r}} = 2.1, tan⁡δ=2×10−4\tan\delta= 2 \times10^{-4}, diameter 3 mm) runs along xx and carries the HE11\mathrm{HE}_{11} mode, the only guided mode below 72.9 GHz. Each end of the rod is fed by a copper transition in the spirit of [10]: an empty CW of 3.6 mm diameter (TE11\mathrm{TE}_{11} cut-off 48.8 GHz) flares in a 12 mm conical horn to an 8 mm aperture, and the rod enters it with a 10 mm conical taper whose tip lies 2 mm inside the straight section. The transition is axisymmetric, so TE11\mathrm{TE}_{11} couples only to HE11\mathrm{HE}_{11}. Wave ports with analytic TE11\mathrm{TE}_{11} profiles are placed in the straight sections; port 2 also records the cross-polarized TE11\mathrm{TE}_{11} mode. The free rod between the horn apertures is 70 mm long, and the PA is centred on it.

Figure 1. (a) Simulated system (section in the PA plane, z=0z = 0): PTFE rod between two CW-to-rod transitions; P1 is excited, P2 is matched. (b)–(d) PA outlines compared at equal contact length LL and height hh; the dashed line is the in-plate phase-matched ray at θc,p\theta_{c,p} launched from the upstream end of the contact.

The PA is a 3 mm-thick PTFE plate standing on the rod in the xyxy plane, with its base sunk 0.2 mm into the rod to model direct contact (a 1.5 mm-wide contact strip). Three outlines are compared at the same base (contact) length LL and height $h ({xref}Fig. 1b–1d`): a rectangle, an isosceles triangle with its apex above the middle of the base, and a right triangle whose vertical leg is at the downstream end. The regular triangle of [7] is the isosceles case with L=20.78L = 20.78 mm, h=18h = 18 mm. The input TE11\mathrm{TE}_{11} mode has its electric field along yy, i.e. in the plane of the plate.

The structure is solved with the FDTD solver openEMS [9] over 52–70 GHz, with cells of 0.25 mm along the rod, 0.15 mm across the rod and plate, and 8-cell PML boundaries (5.5–9 M cells per model). The far field is computed from a Huygens box from which the two CW bores are removed, so that guided power is not counted as radiation; gain is referred to the power accepted at P1. Radiation results are reported over 53–67 GHz, where the unaccounted power Pacc−PP2−PradP_{\mathrm{acc}} - P_{\mathrm{P2}} - P_{\mathrm{rad}}, which includes the dielectric loss, stays within ±3.1\pm3.1 % of PaccP_{\mathrm{acc}} for every element in Table I (and below 6.5 % for all models). Without a PA, the two transitions in series give ∣S11∣<−15.9|S_{11}| < -15.9 dB and ∣S21∣|S_{21}| between −0.43-0.43 dB and −0.17-0.17 dB over this band.

Angles are given as the beam direction ψ\psi in the PA plane (θ=90∘\theta= 90^\circ), measured from the rod axis toward the PA; the azimuth of [7] is ϕ=270∘−ψ\phi= 270^\circ- \psi. For the square and regular-triangle PAs of [7], the model gives main lobes at ϕ=245∘\phi= 245^\circ and 248∘248^\circ against 252∘252^\circ and 240∘240^\circ reported there, the secondary lobes of the square in the same sector (98–134° here, 108–130° in [7]), and the triangle again more directive than the square. Our absolute gains are 1.5–3.2 dB lower; the transition and rod lengths of [7] are not given, so this offset is not interpreted further.

Coupling Mechanism and Sizing Rule

The HE11\mathrm{HE}_{11} field outside the rod decays over about 1.4 mm at 60 GHz, and the mode is slow (ng=βg/k0=1.150n_g = \beta_g/k_0 = 1.150, with βg\beta_g its propagation constant and k0k_0 the free-space wavenumber). A PTFE element in contact is denser than this mode, so the guided wave cannot stay bound under it: as in a prism coupler [8], power tunnels into the element as a wave whose phase velocity along xx matches that of the rod.

phase velocity along xx matches that of the rod. Inside a plate of thickness tt this wave is the fundamental plate mode (TE0\mathrm{TE}_0 for a field in the plate plane) of effective index npn_p [11], and it travels at

cos⁡θc,p=ng/np(1)\cos\theta_{c,p} = n_g/n_p \tag*{(1)}

to the rod axis. For t=3t = 3 mm, np=1.343n_p = 1.343 and θc,p=31.1∘\theta_{c,p} = 31.1^\circ at 60 GHz, whereas the bulk index 2.1\sqrt{2.1} would give 37.5∘37.5^\circ. Over 53–67 GHz, θc,p\theta_{c,p} decreases from 33.4∘33.4^\circ to 28.8∘28.8^\circ, because ngn_g rises faster than npn_p.

Equation (1) has two consequences. First, power is collected along the whole contact, so the radiated fraction grows with LL. Second, the launched wave must reach a face through which it can leave. A wave at θc,p\theta_{c,p} meets a horizontal top face 58.9∘58.9^\circ off its normal, beyond the critical angle arcsin⁡(1/np)=48.1∘\arcsin(1/n_p) = 48.1^\circ, and is turned back toward the rod; it meets a vertical downstream face 31.1∘31.1^\circ off its normal and can leave. The wave launched at the upstream end of the contact reaches the top of that face when

h=Ltan⁡θc,p,(2)h = L \tan\theta_{c,p}, \tag*{(2)}

which sets the element height for a given contact length. Figure 2 shows this behaviour for the three outlines at the size of the regular triangle of [7]. In all cases the field leaves the rod along the contact and climbs into the plate at an angle close to θc,p\theta_{c,p}. In the right triangle (Figure 2a) the wave fills the plate under the ramp and exits through the vertical leg as a single beam. In the rectangle (Figure 2b), the region above the ray shows a standing-wave pattern: part of the wave is reflected inside the plate, which broadens the beam. In the isosceles element (Figure 2c) the wave meets the slanted downstream face close to the rod and leaves there, which tilts the beam toward the rod.

Figure 2. ∣E∣|E| at 60 GHz in the PA plane for L=20.78L = 20.78 mm, h=18h = 18 mm (the size of the regular triangle of [7]), normalized to its maximum. Dashed: in-plate ray at θc,p\theta_{c,p} from the upstream end of the contact.

Parametric Study

Figure 3 reports the effect of hh and LL on the three outlines at 60 GHz. All 33 simulated elements have ∣S11∣|S_{11}| between −32.4-32.4 dB and −15.7-15.7 dB at 60 GHz and stay below −10-10 dB over the whole simulated band, so their −10-10 dB fractional bandwidth exceeds 29.5%; it is limited by the simulated band, not by the element. ∣S21∣|S_{21}| falls with LL as more power is radiated: short elements (L≤16L \le16 mm) radiate 75–91 % of the accepted power, and longer elements radiate 98 % at L=30L = 30 mm.

Eight plots showing the effects of height and contact length on gain, beam direction, half-power beamwidth, and scattering parameters

Figure 3. Effect of the height hh at L=20.78L = 20.78 mm (a)–(d) and of the contact length LL at h=18h = 18 mm (e)–(h), all at 60 GHz. In (d) and (h), filled markers show ∣S21∣|S_{21}| and open markers ∣S11∣|S_{11}|.

At fixed L=20.78L = 20.78 mm (Figure 3a–Figure 3d), the rectangle is best for h=9h = 9–12 mm (14.6–14.7 dBi, 17–19° beamwidth) and degrades when made taller; this optimum brackets Ltan⁡θc,p=12.5L \tan\theta_{c,p} = 12.5 mm from (2). The right triangle reaches its maximum for h=12h = 12–18 mm, i.e. for ramp angles from about θc,p\theta_{c,p} to θc,p+10∘\theta_{c,p} + 10^\circ, while the isosceles element stays within 13.0–13.4 dBi for h≥9h \ge9 mm. For the two elements with a vertical downstream face the beam rises from about 10° to 35° as hh grows and saturates at 33.5–35.5°, a few degrees above θc,p\theta_{c,p}, whereas the isosceles beam stays 13–26° from the axis.

At fixed h=18h = 18 mm (Figure 3e–Figure 3h), short elements (h>Ltan⁡θc,ph > L \tan\theta_{c,p}) are poor for every outline. As LL grows, the gain of the two elements with a vertical face rises to 15.1 dBi (right triangle) and 15.9 dBi (rectangle) at L=30L = 30 mm,

GG (dBi)HPBW (°)SLL (dB)ψ\psi (°)η\eta (%)∣S11∣|S_{11}| (dB)∣S21∣|S_{21}| (dB)
L=20.78L = 20.78 mm, h=18h = 18 mm
Right tri.13.920-10.235.593-26.9-11.8
Rectangle13.132-10.033.595-26.9-13.5
Isosceles13.229-8.225.093-21.9-11.4
L=30L = 30 mm, h=18h = 18 mm
Right tri.15.116-10.728.598-26.0-18.2
Rectangle15.910-6.629.598-24.6-18.9
Isosceles13.925-9.618.098-27.7-16.5

Table 1. Comparison at equal LL and hh at 60 GHz

η\eta: radiated/accepted power.

fig:4d shows that, at the rule, the beam of the rectangle and of the right triangle stays within 3° and 6° of θc,p\theta_{c,p} over 53–67 GHz, whereas the isosceles element radiates 12–13° closer to the rod. The phase-matching angle therefore sets both the size of a good element and the direction of its beam.

Table I and Fig. 4e–4h compare the three outlines at the size of [7] and at L=30L = 30 mm, h=18h = 18 mm, which satisfies (2). At the size of [7] (h/(Ltan⁡θc,p)=1.43h/(L\tan\theta_{c,p}) = 1.43, above the rule) the right triangle already has the highest gain (13.9 dBi against 13.1–13.2 dBi) and the narrowest beam (20° against 29–32°), and, like the rectangle, a lower ∣S11∣|S_{11}| than the isosceles element (−26.9 dB against −21.9 dB). At the rule, the rectangle reaches 15.9 dBi and a 10° beam, but with −6.6 dB side lobes; the right triangle gives 15.1 dBi, a 16° beam and −10.7 dB side lobes with half the plate area. Both exceed the isosceles element (13.9 dBi, 25°) by 1.2–2.1 dB, and point their beam 10.5–11.5° further from the rod. The isosceles element does better when it is lower (14.6 dBi at h=12h = 12 mm for L=30L = 30 mm), but the right triangle and the rectangle at the rule still exceed it by 0.4 dB and 1.3 dB.

where (2) is met, while the isosceles element saturates at 13.9 dBi.

Fig. 4 tests (2) with the two height sweeps at fixed contact length, L=20.78L = 20.78 mm and L=30L = 30 mm, plotted against the normalized height h/(Ltan⁡θc,p)h/(L\tan\theta_{c,p}). For the rectangle (Fig. 4a) both sweeps peak at a normalized height of 1: 14.7 dBi at 0.96 for L=20.78L = 20.78 mm and 15.9 dBi at 1.00 for L=30L = 30 mm, with 1.6–2.3 dB less at 1.33–1.44. The right triangle (Fig. 4b) also needs h≳Ltan⁡θc,ph \gtrsim L\tan\theta_{c,p}, losing 1.8 dB at 0.72 (L=20.78L = 20.78 mm) and 1.5 dB at 0.66 (L=30L = 30 mm), but above the rule it is tolerant: its gain varies by only 0.7 dB from 0.96 to 1.92 at L=20.78L = 20.78 mm and drops by 0.7 dB at 1.33 for L=30L = 30 mm, against 2.3 dB for the rectangle, and its side lobes stay at or below −10 dB at all these heights. The ramp removes the plate region above the phase-matched wave in which the rectangle traps reflected power. The isosceles element (Fig. 4c) has no maximum at the rule: its gain stays within 12.9–13.4 dBi at L=20.78L = 20.78 mm and decreases with height at L=30L = 30 mm, from 14.6 dBi at 0.66 to 13.6 dBi at 1.33, while its beam stays 13–26° from the axis.

Conclusion

A pinching-antenna element on a dielectric rod behaves as a prism coupler: the guided wave enters the denser plate at the phase-matching angle θc,p\theta_{c,p} of the plate mode, 31° for a 3 mm PTFE plate at 60 GHz, and radiates through the face it reaches first. For elements with a vertical downstream face this gives the sizing rule h≈Ltan⁡θc,ph \approx L\tan\theta_{c,p}, which locates the gain maximum of the rectangular element and the onset of the gain plateau of the right-triangular one at both simulated contact lengths, while the beam points within 6° of θc,p\theta_{c,p} across 53–67 GHz. A right-triangle element, whose ramp removes the plate region above the phase-matched wave, trades 0.9 dB of gain against the rectangle for tolerance to its height and side lobes at or below −10 dB, and both shapes outperform the isosceles element of equal size by 1.2–2.1 dB. These results give a simple, physically grounded starting point for designing PAs with a prescribed gain and beam direction, and the fabrication and measurement of the proposed elements will follow.

Eight plots showing normalized height sweeps, beam direction, gain pattern, and three-dimensional normalized electric-field patterns

Figure 4. (a)–(c) Gain at 60 GHz versus the normalized height h/(Ltan⁡θc,p)h/(L \tan\theta_{c,p}) for the two height sweeps at fixed contact length; the dash-dotted line marks (2). (d) Beam direction versus frequency, (e) gain in the PA plane (θ=90∘\theta= 90^\circ) at 60 GHz, and (f)–(h) 3D normalized ∣E∣|E| at 60 GHz, all for L=30L = 30 mm, h=18h = 18 mm; the dash-dotted lines mark (1), and in (f)–(h) xx is the rod axis (propagation direction), the PA lies in the xyxy plane toward +y+y, and the pattern in that plane is projected below each beam as filled contours. Colours and markers as in Figure 3.

Acknowledgment

Generative AI tools were used solely for language editing, grammar correction, and improving the clarity of the text. All scientific content, methodology, analysis, results, and conclusions were developed, verified, and approved by the authors.

References

  1. [1]A. Fukuda, H. Yamamoto, H. Okazaki, Y. Suzuki, and K. Kawai, “Pinching antenna—Using a dielectric waveguide as an antenna,” NTT DOCOMO Tech. J., vol. 23, no. 3, pp. 5–12, Jan. 2022.
  2. [2]Z. Ding, R. Schober, and H. V. Poor, “Flexible-antenna systems: A pinching-antenna perspective,” IEEE Trans. Commun., vol. 73, no. 10, pp. 9236–9253, Oct. 2025.arxiv.org/abs/2412.02376
  3. [3]Z. Wang, C. Ouyang, X. Mu, Y. Liu, and Z. Ding, “Modeling and beamforming optimization for pinching-antenna systems,” IEEE Trans. Commun., vol. 73, no. 12, pp. 13904–13919, Dec. 2025.DOI
  4. [4]D. Tyrovolas, S. Tegos, Y. Xiao, P. Diamantoulakis, S. Ioannidis, C. Liaskos, G. Karagiannidis, and S. D. Assimoniis, “Ergodic rate analysis of two-state pinching-antenna systems,” in Proc. IEEE Int. Conf. Commun. Workshops (ICC Workshops), 2026, pp. 1–6.arxiv.org/abs/2511.01798
  5. [5]D. Tyrovolas, S. A. Tegos, Y. Xiao, P. Diamantoulakis, S. Ioannidis, C. K. Liaskos, G. K. Karagiannidis, and S. D. Asimonis, “How many pinching antennas are enough?” IEEE Internet Things J., vol. 13, no. 10, pp. 21 994–22 006, 2026.
  6. [6]R. Zhang, Y. Shao, and Y. Liu, “Directional pinching-antenna systems,” arXiv:2511.19133, 2025.arxiv.org/abs/2511.19133
  7. [7]H. Li, W. Liu, Z. Chen, C. Song, and G. Chen, “Geometry-dependent radiation of pinching antennas: Theory, simulation, and measurement,” IEEE Commun. Lett., vol. 30, pp. 2093–2097, 2026.
  8. [8]P. K. Tien and R. Ulrich, “Theory of prism-film coupler and thin-film light guides,” J. Opt. Soc. Am., vol. 60, no. 10, pp. 1325–1337, Oct. 1970.DOI
  9. [9]T. Liebig, “openEMS – open electromagnetic field solver.” General and Theoretical Electrical Engineering (ATE), University of Duisburg-Essen. [Online]. Available: https://www.openEMS.de.
  10. [10]T.-L. Vu, S. Barlier, Y. Stricot, R. Sauleau, M. Ettorre, and D. González-Ojevero, “Experimental characterization of bending effects for solid and hollow dielectric waveguides at V-band,” Sci. Rep., vol. 11, no. 1, p. 20679, Oct. 2021.DOI
  11. [11]K. Okamoto, Fundamentals of Optical Waveguides, 2nd ed. Amsterdam, The Netherlands: Elsevier, 2006.

Paper details

vlnc:W_vra3bt85
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 14

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Engineering

Generative AI tools

View statement
Looking for related papers…
Contents