IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 65, NO.

4, APRIL 2017 2085

Communication
Dual-Band 4G Eyewear Antenna and SAR Implications
Aykut Cihangir, Chinthana J. Panagamuwa, Will G. Whittow, Gilles Jacquemod,
Frédéric Gianesello, Romain Pilard, and Cyril Luxey

Abstract— This communication presents low-cost and easy-to- The antenna is generally placed behind the user’s ear like in a
manufacture dual-band antenna solution to achieve cellular 4G wireless headset configuration. The specific absorption rate (SAR)
(LTE-Advanced) coverage in smart eyewear devices. Coupling element
is not a concern due to the low output power levels of WLAN and
type antenna has been evaluated with appropriate matching networks to
cover the target bands of 700–960 MHz and 1.7–2.7GHz. To emulate a Bluetooth standards. It is foreseen that these eyewear devices might
realistic device, an ABS plastic dielectric frame has been designed and partially act as smartphones in the future which creates the necessity
manufactured using 3-D printing technology. Simulations for the antenna to make them suitable to communicate with 4G cellular networks.
are carried out in three typical use-case scenarios which are “with Such devices would offer significantly higher data rates compared to
user’s head,” “with head and hand,” and “free space.” The simulations
are validated through S-parameters, efficiency and radiation pattern Bluetooth and pave the way for a new wave of functionality. From
measurements using fabricated frame and antenna prototype in the the antenna design perspective, this requires some dual-band antennas
presence of head and hand phantoms. The specific absorption rate (SAR) to cover 700–960 MHz for the low-band (LB) and 1.7–2.7 GHz for
behavior of the antenna designs is also investigated through simulations the high-band (HB). This is extremely challenging considering the
and measurements. It is demonstrated that SAR values are found to be
above the limitations which might be problematic in practical use if the
very small area that can be allocated to the antenna. As a conse-
transmit power of the eyewear is not reduced. quence, an electrically-small antenna is required especially in the LB.
Index Terms— 4G, antenna, cellular network, coupling element (CE), To maximize the bandwidth potential (BP) from an electrically small
LTE, matching network (MN), specific absorption rate (SAR), smart antenna in a mobile terminal, capacitive coupling elements (CE) have
eyewear. been proposed in the literature. The CEs are nonresonant structures
which are used to excite the proper wave modes on the system
I. I NTRODUCTION ground plane of the device [6], [7]. By exploiting the low quality
The wireless cellular communication standards have undergone factor of the system ground plane, a high BP can be obtained. The
several improvements since the launch of the first analog sys- antenna is then tuned to the desired bands using a proper matching
tems. Recently, with the introduction of smartphones, a continuous network (MN) at the antenna feed. Such antenna solutions with their
necessity to provide higher data rates to the users has arisen. For corresponding MN can be found in [8]–[11]. Previously, the authors
this reason, LTE-Advanced standard was introduced as the fourth have conducted a feasibility study to demonstrate that it was possible
generation of wireless cellular communications. to use CE antennas to cover 4G frequencies in eyewear devices [12].
A parallel avenue of research that has seen an explosion of activity It was shown that the target frequency bands could be covered with
in recent years is body-centric communications that involve trans- a reflection coefficient below −6 dB, with CEs placed in three
mission between on-body and off-body antennas. As an interesting different regions of the printed circuit board (PCB). The feasibility
and hot recent topic, there are a number of smart eyewear devices study included only simulation results of the antennas, without any
already available in the market or being prepared for imminent fabrication or measurement. Taking [12] as the starting point, both
release [1]–[5]. The common properties of all these eyewear systems the CE and the MN are redesigned in this communication, in order
are: a lens-reflector screen in front of the user’s eye, a touchpad on to obtain a real manufactured prototype as realistic as possible.
the side of the eyewear to adjust the settings or use the applications, A realistic ABS plastic eyewear frame was designed and fabricated
a speaker/microphone pair and a Wi-Fi/Bluetooth transceiver. The using 3-D printing technology. Suitable realistic CE type antenna
devices have a wide variety of application areas including navigation, structure with the corresponding MN was optimized and integrated
video capture and streams, accept/reject calls from a paired mobile into this eyewear frame to operate in the 700–960 MHz LB and the
phone. 1.7–2.7 GHz HB. It should be noted that in the 4G operating mode
However, the wireless connectivity of these devices is currently the output power level from the front-end module can be significantly
limited to Wi-Fi 802.11 b/g and Bluetooth where the eyewear higher than Bluetooth devices. So considering an eyewear antenna
connects either to a hotspot, a set-top box or a peripheral smartphone. to operate in 4G cellular networks, SAR becomes an essential
parameter to investigate in order to meet international regulations.
Manuscript received January 1, 2015; revised December 24, 2016; accepted
Section II presents the design and optimization of the manufactured
January 6, 2017. Date of publication February 16, 2017; date of current version
April 5, 2017. antenna design. A comparison of simulations and measurements of
A. Cihangir and G. Jacquemod are with EpOC, Université Nice Sophia the reflection coefficient, total efficiency and radiation patterns taking
Antipolis, Valbonne, France. into account the user’s head is also presented in this section. The
C. Panagamuwa and W. Whittow are with the Wireless Communications effect of the user’s hand and the antenna performance in free space
Research Group, School of Electronic, Electrical and Systems Engineering,
Loughborough University, Loughborough LE11 3TU, U.K. (e-mail: is discussed in Section III. SAR simulations and measurements are
w.g.whittow@lboro.ac.uk; c.j.panagamuwa@lboro.ac.uk). given in Section IV. Finally, conclusions are discussed in Section V.
F. Gianesello and R. Pilard are with STMicroelectronics, Crolles, France.
C. Luxey is with EpOC, Université Nice Sophia Antipolis, Valbonne,
France, and also with the Institut Universitaire de France, Paris, France II. A NTENNA D ESIGN
(e-mail: cyril.luxey@unice.fr).
Color versions of one or more of the figures in this communication are
To validate one of the antenna concept presented in [12], a CE
available online at http://ieeexplore.ieee.org. with a realistic dielectric frame was designed and manufactured. The
Digital Object Identifier 10.1109/TAP.2017.2670562 simulation setup presented in Fig. 1 was prepared for the antenna
0018-926X © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
2086 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 65, NO. 4, APRIL 2017

Fig. 1. Optimized eyewear frame on the SAM head.

Fig. 2. Top view of the eyewear frame (left) and placement of the PCB Fig. 3. Model of Antenna Design and its matching network.
inside the frame. Position of user’s hand and fingers (right).

design, in the EMPIRE XCcel electromagnetic software [13]. The

specific anthropomorphic mannequin (SAM) head [14] was used
for simulations and measurements. 3-D-printing fast prototyping
technology was selected for the fabrication of the eyewear frame
using ABS-M30 material. This material is modeled in simulations
with an εr = 2.97 and tan δ = 0.029. The parts of the frame on the
right side of the head acts as a dielectric casing for the eyewear PCB
(casing length is 168 mm, height is 19 mm, and width is 6.8 mm).
There is a hollow region inside the casing for the placement of the
antenna PCB, having a width of 4.8 mm, leaving dielectric walls Fig. 4. Simulated and measured reflection coefficient of Antenna Design
with frame and 3-element MN on SAM head.
of 1 mm thickness on each side (Fig. 2). Although the antenna was
optimized for the “with head” condition, two additional use-cases
were also evaluated both in simulations and measurements, one being
“antenna and frame in free space condition” and the other being “with
the user’s hand” (Fig. 2) emulating the user using the touchpad.
For the electrical properties of the homogeneous tissue, the liquid
properties at 1450 MHz were used in simulations, to be a midvalue
of the target band. The corresponding properties are: εr = 40.5 and
tan δ = 0.367.
The CE is printed on one side of a 0.8 mm thick FR4
PCB (εr = 4.4 and tan δ = 0.02). The other side of the PCB is
used as the ground plane. The location of the CE is toward the end
Fig. 5. Manufactured PCBs with Antenna.
of the PCB, close to the eye. It was hypothesized that the efficiency
will be higher in this configuration since the CE is situated further
away from the lossy head tissues due to the curved geometry of good except above 2.2 GHz. This is thought to be due to the frequency
the head and the straight geometry of the frame. The T-shaped CE dependent characteristics of the head liquid, which was used as a
geometry can be observed in Fig. 3. constant value through the whole simulated frequencies. The antenna
The ground plane is printed on the inner side of the FR4 PCB. can indeed cover the target-band with dual-band behavior with a
The CE and the MN are situated on the outer side, together with the reflection coefficient below −6 dB.
WFL connector that will be used to feed the antenna. The simulated Fig. 5 shows the manufactured PCBs used in the measurements.
and measured reflection coefficients of this antenna are presented The simulated and measured total efficiency for the general use-case
in Fig. 4. The agreement between simulation and measurement is (“with head” configuration) can be seen in Fig. 6. The measured total
IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 65, NO. 4, APRIL 2017 2087

TABLE I
S IMULATED AND M EASURED M AXIMUM A NTENNA G AIN

Fig. 6. Simulated and measured total efficiency of Antenna with frame on

SAM head.

Fig. 8. Measurement configurations for the three targeted use-cases: “free

space” using foam spacer (left), “with SAM head” (center), and “with SAM
head and phantom hand” (right).

by as much as 6 dB in the LB and 5 dB in the HB, compared to

vertical polarization (because of the main horizontal currents flowing
along the length of the PCB).
Table I summarizes the simulated and measured maximum gain
values. A fair agreement can be observed between simulations and
measurements.

III. M EASUREMENT IN D IFFERENT U SE C ASE

As previously mentioned in Section II, three realistic use-cases
were targeted for the smart eyewear devices. This section presents
the measurements of those three use-cases. The pictures of the
measurement setup of different configurations are shown in Fig. 8.
The left figure shows the setup used for “free space” condition
where the eyewear is not worn by the user but might need to
be operational for incoming calls. The ABS frame with integrated
antenna was placed horizontally on a foam support, to keep the
antenna leveled in the horizontal plane. The center figure presents
the general use-case, “with SAM head” configuration, which was
used in the previous measurements in Section II. The right figure
shows the set-up for “with SAM head and phantom hand” use-case
which emulates the case in which the user uses the touchpad on the
side of the device. A positioner placed below the hand was used
to keep the hand phantom stable and the finger positioned at the
center of the ABS frame. A real human male head configuration was
Fig. 7. Measured and simulated gain patterns of Antenna with frame on also tested for S-Parameter measurements. The reflection coefficient
SAM head (2-D cuts).
and total efficiency measurements for the antenna can be found
in Figs. 9 and 10, respectively.
efficiency ranges from 5% to 9% in the LB and between 18% and As expected, the reflection coefficient performance is slightly
31% in the HB. The simulated and measured 2-D gain patterns of worse in free space condition when compared to the “with head”
the antenna are compared in Fig. 7, and show reasonable agreement case, because of the lossy nature of the head and also since the MN
despite the presence of some ripples. was fine tuned with the presence of this SAM head. In “with head
The radiation is weak in the direction of the head due to losses. and hand” case, a slight detuning in the HB can be seen due to the
The radiation is always maximum in the direction away from the existence of the finger close to the CE. The reflection coefficient
head (in horizontal plane) and splits in several beams as the frequency measured with a real head is similar compared to the SAM head.
increases, however, without the creation of any strong deep nulls. The In free space, the efficiency in LB, ranges from 30% to 58%, with a
dominant polarization is in the horizontal direction and is superior minimum seen around 700 MHz mainly because of a poor reflection
2088 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 65, NO. 4, APRIL 2017

TABLE II
S IMULATED AND M EASURED 1 AND 10 g SAR (W/kG ) FOR A NTENNA .
P OWER N ORMALIZED TO 0.25 W A CCEPTED P OWER

Fig. 9. Measured reflection coefficient for different use case.

result comparisons are shown in Table II. All results are normalized
to 0.25 W accepted power after accounting for the port mismatch
losses (S11 ). In all cases, the measured results are lower than their
simulation counterparts. Although ferrite beads were used around the
RF cable in order to choke any external cable currents, there is still
radiation from the cable, as highlighted in the ellipse in Fig. 11(b).
As the antenna is placed in close proximity to the head, the SAR
levels are sensitive to manufacturing tolerances of the frame. This
affects the positioning of the antenna inside the frame, the thickness
of the frame as well its dielectric properties. This was verified by
carrying out additional measurements with the PCB (without the
frame) placed directly against the flat section of the DASY4 twin
Fig. 10. Measured total efficiency for different use case. phantom; these results showed better agreement with simulations.
As with all measurement equipments, the DASY4 has a number
of parts where measurement uncertainty can be introduced. The
three main areas are identified as Measurement equipment (probe),
Mechanical Constrains (shell and probe positioning), and Physical
parameters (liquid properties). As a result, the worse-case expanded
uncertainty budget for the DASY4 quoted in the manufacturer’s
documentation is ±21.7% [18].
In all cases, simulated and measured SAR values exceed the
international limits. This is partially due to the fact that the total
antenna efficiency of the eyewears is double the typical values
with generic smartphones with head and hand of the user. In this
communication, the antenna is also much closer to the phantom head
compared to a smartphone that normally rests against the pinna and
has the antenna placed on the side of the case away from the head.
Fig. 11. (a) Antenna positioned against the right head of the DASY4’s twin This is indeed an important finding of this communication which is
phantom. (b) Visualization of measured SAR area scan at 1900 MHz. highly problematic for the safety of the users. Due to the possible
placements of an antenna to be integrated in a 4G eyewear, it is
believed that it will lead to relatively high efficiency and therefore
coefficient at this frequency. In the HB, the free space efficiency
it will be extremely difficult to comply with the US and European
varies between 40% and 63%. The total efficiency decrease due to
SAR standards. As the total efficiency of the antenna should be large
the hand is on average around 5%, with a difference up to 10% in the
enough for 4G cellular communications, the SAR levels could be
HB. The total efficiency with the SAM head and hand is ranging from
reduced by decreasing the transmit power of the front-end module of
6% to 8% in the LB and 12% to 21% at the HB. This antenna has
the eyewear making it operate in a different class mode.
a higher efficiency than antennas integrated in typical smartphones
(with user’s head and hand). Therefore, it demonstrates the suitability
of our eyewear device for 4G communications. V. C ONCLUSION
An antenna design to cover the 4G cellular communication bands
IV. SAR S IMULATIONS AND M EASUREMENTS for a smart eyewear device was presented in this communication.
RF Power absorbed by a unit mass of tissue (W/kg) is referred to A realistic ABS frame which fits the SAM head was designed and
as SAR and is the internationally recognized dosimetric parameter manufactured using 3-D printing technology. The CE was placed at
for testing wireless devices. In Europe, the SAR limit in the head is the end of the PCB, close to the eye and as far as possible from the
2 W/kg averaged over 10 g for 6 min [17]. In the USA, it is 1.6 W/kg lossy head tissue. It was shown that this design can achieve dual-
averaged over 1 g for 30 min [14]. The 1 and 10 g spatially averaged band coverage of 700–960 MHz and 1.7–2.7 GHz with a −6 dB
SAR was simulated in EMPIRE XCcel using the SAM head phantom. reflection coefficient. Considering three possible realistic use-case,
Measurements were carried out using a Dosimetric and near-field S-Parameters and total efficiency measurements were repeated for the
Assessment System (DASY4) [18]. The experimental setup is shown antenna design in “free space” and “with head and hand” conditions,
in Fig. 11(a). SAR measurements were taken with continuous wave showing no significant detuning. The reflection coefficient was also
RF input signals at 900 and 1900 MHz. Simulation and measurement measured in the presence of a real human head. Slight and acceptable
IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 65, NO. 4, APRIL 2017 2089

differences were encountered between simulations and SAM head [6] J. Villanen, J. Ollikainen, O. Kivekas, and P. Vainikainen, “Coupling ele-
phantom measurements. ment based mobile terminal antenna structures,” IEEE Trans. Antennas
Propag., vol. 54, no. 7, pp. 2142–2153, Jul. 2006.
Finally, SAR simulations and measurements were conducted
[7] J. Holopainen, R. Valkonen, O. Kivekas, J. Ilvonen, and P. Vainikainen,
at 0.9 and 1.9 GHz demonstrating a very important issue about user- “Broadband equivalent circuit model for capacitive coupling elemen-
antenna interactions. Measured 1g SAR values higher than the US t–based mobile terminal antenna,” IEEE Antennas Wireless Propag. Lett.,
limits for the general public were found at 0.9 and 1.9 GHz. The vol. 9, pp. 716–719, 2010.
measured 10 g SAR was below the European limits for the general [8] J. Villanen, C. Icheln, and P. Vainikainen, “A coupling element-
based quad-band antenna structure for mobile terminals,” Microw. Opt.
public at 0.9 GHz but above the limit at 1.9 GHz. Note the measured Technol. Lett., vol. 49, no. 6, pp. 1277–1282, Jun. 2007.
values would be below the limits for all cases for military personnel [9] A. Andujar, J. Anguera, and C. Puente, “Ground plane boosters as a
or trained users in controlled environments where five times higher compact antenna technology for wireless handheld devices,” IEEE Trans.
SAR values are permissible. The high-SAR values are largely due Antennas Propag., vol. 59, no. 5, pp. 1668–1677, May 2011.
[10] R. Valkonen, J. Ilvonen, and P. Vainikainen, “Naturally non-selective
to the fact that the total efficiency is double the typical values with handset antennas with good robustness against impedance mistun-
generic smartphones with the head and hand. As the total efficiency ing,” in Proc. 6th Eur. Conf. Antennas Propag. (EuCAP), Mar. 2012,
of the antenna is large enough for cellular communications with pp. 796–800.
4G networks, the SAR levels could be reduced by decreasing the [11] R. Valkonen, J. Ilvonen, C. Icheln, and P. Vainikainen, “Inherently non-
transmit power of the eyewear devices; making them operate in a resonant multi-band mobile terminal antenna,” Electron. Lett., vol. 49,
no. 1, pp. 11–13, Jan. 2013.
different class mode; or adding a series resistor within the RF transmit [12] A. Cihangir et al., “Feasibility study of 4G cellular antennas for eyewear
chain. communicating devices,” IEEE Antennas Wireless Propag. Lett., vol. 12,
pp. 1704–1707, 2013.
ACKNOWLEDGMENT [13] EMPIRE XCcel, accessed on Mar. 5, 2017. [Online]. Available:
The authors would like to thank Orange Labs La Turbie and http://www.empire.de/
the CIM-PACA design platform. They would also like to thank [14] IEEE Recommended Practice for Determining the Peak Spatial-
Average Specific Absorption Rate (SAR) in the Human Head From
F. Devillers from Orange Labs, for his valuable help in the mechan-
Wireless Communications Devices: Measurement Techniques, IEEE
ical adjustments of the eyewear, and phantom head and hand for Standard 1528-2003, Sep. 2013.
measurements. [15] Optenni Lab, accessed on Mar. 5, 2017. [Online]. Available:
https://www.optenni.com/optenni-lab/overview
R EFERENCES [16] Satimo StarLab, accessed on Mar. 5, 2017. [Online]. Available:
[1] Google Glass, accessed on Mar. 5, 2017. [Online]. Available: http://www.mvg-world.com/fr/products/field_product_family/antenna-
http://www.google.com/glass/start/ measurement-2/starlab
[2] M100 Smart Glass, accessed on Mar. 5, 2017. [Online]. Available: [17] Human Exposure to Radio Frequency Fields From Hand-Held and Body-
http://www.vuzix.com/UKSITE/consumer/products_m100.html Mounted Wireless Communication Devices. Human Models, Instrumen-
[3] RECON Jet, accessed on Mar. 5, 2017. [Online]. Available: tation, and Procedures. Procedure to Determine the Specific Absorption
http://www.reconinstruments.com/products/jet/ Rate (SAR) for Hand-Held Devices Used in Close Proximity to the Ear
[4] Olympus MEG4.0, accessed on Mar. 5, 2017. [Online]. Available: (Frequency Range of 300 MHz to 3 GHz), document IEC 62209-1:2005,
http://www.olympus.co.jp/jp/news/2012b/nr120705meg40j.jsp 2005.
[5] Optinvent ORA, accessed on Mar. 5, 2017. [Online]. Available: [18] DASY4 Manual: DASY4 Manual V4.1, Schmid Partner Eng. AG, Zürich,
http://www.optinvent.com/ Switzerland, Mar. 2003.