A Novel Cellular Handset Design for an Enhanced Antenna Performance and a Reduced SAR in the Human Head

This paper presents a novel cellular handset design with a bottom-mounted short loaded-whip antenna. This new handset design is modeled and simulated using a finite difference time-domain (FDTD)-based platform SEMCAD. The proposed handset is based on a current commercially available bar-phone type with a curvature shape, keypad positioned above the screen, and top-mounted antenna. The specific absorption rates (SARs) are determined computationally in the specific anthropomorphic mannequin (SAM) and anatomically correct model of a human head when exposed to the EM-field radiation of the proposed cellular handset and the handset with top-mounted antenna. The two cellular handsets are simulated to operate at both GSM standards, 900 MHz as well as 1800 MHz, having different antenna dimensions and intput power of 0.6 W and 0.125 W, respectively. The proposed human hand holding the two handset models is a semirealistic hand model consists of three tissues: skin, muscle, and bone. The simulations are conducted with handset positions based on the IEEE standard 1528-2003. The results show that the proposed handset has a significant improvement of antenna efficiency when it is hand-held close to head, as compared with the handset of top-mounted antenna. Also, the results show that a significant reduction of the induced SAR in the human head-tissues can be achieved with the proposed handset.


INTRODUCTION
Due to enormous increase in the number of cellular handset users around the world, many questions are raised about the possible hazard effect of the cellular handset electromagnetic field (EMF) radiation.Thereby, health concerns regarding the use of a cellular handset near the human head have been growing and took a lot of attention by researchers.
The interaction of the cellular handset with the human head has been investigated by many published papers with considering; first, the effect of the human head on the handset antenna performance, including the feed-point impedance, gain, and efficiency [1][2][3][4], second, the impact of the antenna EM radiation on the user's head due to the absorbed power, which is measured by predicting the induced specific absorption rate (SAR) in head tissues [5,6].
The protocol and procedures for the measurement of the peak spatial-average SAR induced inside a simplified head model of the cellular handset users are specified by IEEE Standard-1528 [7] and IEC 62209-1 [8].Both standards specified the specific anthropomorphic mannequin (SAM) as a simplified physical model (phantom) of the human head.This SAM has also been adopted by many committees, associations, and commissions [9][10][11].The SAM has been developed by the IEEE Standards Coordinating Committee 34, Subcommittee 2, Working Group 1 (SCC34/SC2/WG1) as a lossless plastic shell, filled with a homogeneous liquid, and a thin lossless ear spacer, whereas (SCC34/SC2/WG2) has suggested the same SAM but with different plastic shell parameters [5].
Anatomically correct models of a nonhomogeneous human head at different ages were used to evaluate the performance of the handset on a human-head phantom [5,12,13].In this paper, a nonhomogeneous high-resolution numerical correct model of a European female head [14], available with SPEAGE-Schmidt & Partner Engineering AG [15], is used.

International Journal of Antennas and Propagation
Handset models with a keypad positioned above the screen are available commercially.Linux released a more comfortable of such a handset with a top-mounted external antenna and a curvature shape [16].This new design ensures that much of the handset rests in the palm of the hand, thus, improving support and control.In addition to the improved grip, the thumb rests in a comfortable position directly above the buttons of the keypad.The improved angle for the thumb makes it unnecessary to shift the handset around in the hand while typing text [16].
In this paper, the proposed handset design with a bottom-mounted antenna is based on the handset model in [16].An FDTD-based platform SEMCAD [15] is used for simulation.The Antenna performance is evaluated for both handset models in free space, hand-held, and hand-held close to head.A semirealistic hand model consists of three tissues is designed to simulate the human hand.The induced SAR's in head models are evaluated at GSM standards, 900 MHz and 1800 MHz, with antenna intput power of 0.6 W and 0.125 W, respectively.Handset positions, cheek and tilt (15 • ), with respect to head are adopted according to IEEE standard 1528 [7].

Handset structure
The handset model in [16] (will be referred later as model no. 1) is simulated using an FDTD-based platform SEMCAD (Simulation Platform for Electromagnetic Compatibility, Antenna Design and Dosimetry) ver. 12 JUNGFRAU [15].The proposed handset with a bottom-mounted antenna (will be referred later as model no.2) is also designed and simulated, where most handset components, such as PCB, LCD, Battery, and keypad, are considered in the design simulation.These components are not located identically in both handset models due to different antenna positions.Both models are simulated to operate at 900 MHz as well as 1800 MHz. Figure 1(a) shows the physical model of the handset released by Linux [16], whereas Figure 1(b) exhibits the proposed physical model with bottom-mounted antenna.Figure 2 shows the numerical equivalent of both physical models used for the FDTD simulation.The maximum dimensions of both handsets are set to 45 × 16 × 130 mm with a PCB symmetrically embedded inside the housing.The acoustic output position is set according to IEEE standard 1528 [7]. Figure 3 shows the numerical components structure of the handset models.The dielectric parameters of handset materials given in [6] are used.

Antenna design and specifications
Instead of using a helical antenna, a short-whip antenna top loaded with a small cylinder [17] is suggested for both designs of models as depicted in Figure 4 .Table 1 shows the physical and electrical antenna specifications that optimized at both GSM standards for both handset models.

Cellular handset in free-space
To align the simulated handset components to the FDTD grid accurately, a minimum spatial resolution of 0.

Cellular handset in hand
A semirealistic human-hand model consists of three tissues: skin, muscle, and bone, are designed using SEMCAD [15] to simulate both handset models in hand, as shown in Figure 5.The FDTD grid has a minimum spatial resolution of 0.5 × 0.5 × 0.5 mm 3 and maximum spatial resolution of 10 × 10 × 10 mm 3 in the x, y, and z directions, with grading ratio of 1.2.For the hand-held of model no.

Cellular handset in hand close to head
As defined in IEEE standard 1528-2003 [7], two handset positions are considered in presence of human-head, cheek and tilt (15 • ).The head is simulated using both, homogeneous and nonhomogeneous phantoms.The homogeneous head model is a SAM phantom available with [15] and consists of two dielectric materials, shell and liquid.The material parameters are defined in [7, 8], with shell and ear spacer defined in [5], at 900 MHz and 1800 MHz.
The nonhomogeneous head phantom is a highresolution European 40-year female head (HR-EFH), derived from MRI scan [15], and is imported to the SEMCAD platform.This CAD phantom consists of 121 different slices, with slice thicknesses of 1 mm (ear region) and 3 mm, and a transverse spatial resolution of 0.2 mm.The following different 25 tissues are recognized: air, blood vessel, bones, brain/grey matter, brain/white matter, cerebellum, cerebrospinal fluid, ear (cartilage), eye-cornea, eye-lens, eyevitreous body, fat, jaw, mastoid cells (bones), mid-brain, Head and hand tissues properties are set according to the material properties data-base in [15] and to that given in [18], where both are based on [19].
The FDTD-grid for each handset in hand close to head has a minimum spatial resolution of 0.5 × 0.5 × 0.5 mm 3 and maximum resolution of 10 × 10 × 10 mm 3 in the x, y, and z directions with grading ratio of 1.2.The absorbing boundary conditions (ABCs) are set as a perfectly matched layer (PML) mode with a very high-strength thickness [15].
Table 2 lists the amounts of mesh cells according to FDTD-grid setting for both handset models in hand close SAM and HR-EFH, at 900 MHz and 1800 MHz.
The simulations (in all cases) assume a steady-state voltage at the 900 and 1800 MHz, with a feed point of a 50-Ohm voltage source of 1-mm gap.A transient excitation of 12 periods is set as guarantee to achieving a steady state.The absorbing boundary conditions (ABCs) are set as a perfectly matched layer (PML) mode with a very high strength thickness [15].
In case of the handset close to head (both SAM and HR-EFH), the acoustic output referenced to earpiece is set according to IEEE standard 1528 [7].Due to different antenna positions in both handset models, the distances between the antennas feed points and the nearest tissue voxel are different too.For the handset model no. 1 the acoustic output position is set at the origin, whereas for the handset model no. 2 the acoustic output position is set at (x = −15, y = 0 and z = −104 mm).Figures 6(a

EM INTERACTION BETWEEN THE HANDSET ANATENNA AND HUMAN HEAD
The EM interaction between the handset antenna and human head is evaluated by; first, evaluating the effect of human head and hand on the handset antenna performance through computing the antenna parameters, including input return loss, gain, radiation efficiency, and total efficiency, second, evaluating the impact of antenna EM radiation on the head through computing the induced SAR and power absorption.

Antenna performance
Table 3 demonstrates the antenna parameters including; input return loss, gain, radiation efficiency, and total efficiency, for both handset models in all cases at 900 MHz.Table 4 lists the antenna parameters at 1800 MHz. Figure 7 shows the radiation beam pattern in (V/m) for both handset models in hand close to HR-EFH at cheek position and for both 900 and 1800 MHz frequencies, whereas Figure 8 shows the radiation beam pattern at tilt position.

SAR and power loss computation in head
The impact of the electromagnetic (EM) wave irradiation on the living body is measured by evaluating the SAR which is defined as the amount of EM energy absorption in the unit mass as follow [20]: where σ E (S/M) is the conductivity, E (V/m) is the the induced electric field vector, and ρ (kg/m 3 ) is the material density.Using SEMCAD platform, an algorithm based on SCC34/SC2/WG2 computational dosimetry, IEEE-1529 [21], the spatial peak SAR can be computed over any required mass.
The spatial-peak SAR should be evaluated in a cubical volume of the body tissues that is within 5% of the required mass [15].The averaged peak-SAR (Spatial-peak SAR [IEEE-1529]) can be specified over a cube of 1g and 10g mass, and normalized to a certain source power.Referred to the IEEE standard C95.1b-2004 [22] (for low-power devices, uncontrolled environment), the antenna input power is set to 0.6 W at 900 MHz and 0.125 W at 1800 MHz, respectively, in all cases.Table 5 lists the computed peak SAR averaged over 1g and 10g, and the absorbed power in tissues, for both handset models at both positions and at 900 MHz.Table 6 lists the computed parameters at 1800 MHz.
Figure 9 shows the sliced-distribution of the averaged peak SAR 1g in the HR-EFH phantom exposed to EM radiation of both model no. 1 and model no. 2 antennas at cheek position and at different frequencies, whereas Figure 10 shows the sliced-distribution of the averaged peak SAR 1g in the HR-EFH phantom exposed to EM radiation at tilt position.

TOTAL ISOTROPIC SENSITIVITY
The total isotropic sensitivity (TIS) [15] is a measure of the handset receiving performance.The TIS and TRP (total radiated power) together determine effectiveness of the handset as a piece of radio equipment, in particular the maximum range at which the handset can operate from the base station with some given level of performance [23].The computed TIS for both handset models at 900 MHz and 1800 MHz are given in Tables 5 and 6.

COMPUTATION ERROR
The computation error is defined as [24] Computation error = P in − P rad + P abs + P Loss /P in , where P in is the input power, P rad is the radiation power, P abs is the absorbed power in tissues, and P Loss is the total power loss.P Loss includes the dielectric loss (P d ) and the metallic ohmic loss (P c ).

DISCUSSION
The results in Tables 3 and 4 reveal that presence of a head close to the handheld set of model no. 1 degrades the handset performance, significantly reducing the handset total efficiency to about (8%-28%) of the total efficiency of the handset in free space.Adopting a bottom-mounted antenna, model no.2, the total efficiency of the handset model no. 1 can be improved by (3.3%-45.5%),whereas the   gain is reduced by (0.19-2.15 dBi).The antennas of both handset models were matched well for all the cases.Since the proposed handset model has an antenna in a low-noise area of the handset and well separated from the potentially noisy components, it has the potential to achieve better TIS.According to the results obtained in Tables 5 and   6, the different cases of the handset model no. 2 in hand close to head do show better TIS values, as compared with model no. 1, due to the improved total efficiency.Moreover, Tables 5 and 6 show that the averaged peak-SAR 1g induced in head close to hand-held of model no. 1 can be reduced by (28%-92.2%)using the proposed handset  Table 5: The computed averaged peak-SAR (over 1g and 10g) and power absorption in tissues, radiated power, total loss, total isotropic sensitivity, and computation error for both handset models in hand close to head at different positions and at 900 MHz.The differences in the induced SAR and absorption power values in both SAM and HR-EFH phantoms are due to their different masses, volumes, and densities distribution.According to simulation results, HR-EFH mass is approximately 4.71 kg and the volume is approximately 4118 cm 3 , while the SAM mass is approximately 6.024 kg (considering a homogeneous density of 1000 kg/m 3 ) and the volume is approximately 6043 cm 3 .The proposed human-hand model mass is approximately 0.248 kg and its volume is approximately 186 cm 3 .
All computations are performed on a 2.0-GHz Intel centrino Laptop machine (Dell, inspiron-630 m) with 2 GB memory (dual-channel technology), and operating under MS Windows-vista.The runtime and memory requirements depend on the simulation space.Less memory and runtime are required for the handset simulation in free space, whereas, more memory and runtime are required for the handset in hand close to head.The machine-memory is enough to achieve all simulations with the mesh cells amounts listed in Table 2.The runtimes are about 1-10 hours.

CONCLUSION
A cellular handset with a keypad over the screen and a bottom-mounted antenna has been proposed and numerically modeled, with the most handset components, using an FDTD-based SEMCAD platform.The proposed handset model is based on the commercially available model with a top-mounted external antenna.Both homogeneous and nonhomogeneous head phantoms have been used with a semirealistic hand design to simulate the handset in hand close to head.The simulation results showed a significant improvement in the antenna performance with the proposed handset model in hand close to head, as compared with the handset of top-mounted antenna.Also, using this proposed handset, a significant reduction in the induced SAR and power absorbed in head has been achieved.

Figure 1 :
Figure 1: The physical model of (a) the handset released by Linux, and (b) the proposed handset with bottom-mounted antenna.

Figure 2 :
Figure 2: The CAD representation of both handset models.

Figure 3 :
Figure 3: Numerical components structure of (a) the handset model no. 1, and (b) the proposed handset model no. 2.

Figure 4 :
Figure 4: The proposed loaded short-whip antenna with dimensions.

Figure 5 :
Figure 5: The CAD representation of the proposed semirealistic hand model holding the proposed handset; (a) all hand tissues, (b) hand-bones only.
) and 6(b) show both handset models close to head (SAM) at cheek position indicating the coordinate system, whereas Figure6(c)shows the handset model no. 2 in hand close to head.

Figure 6 :
Figure 6: Coordinate system; (a) the handset model no. 1 referenced as seen from the right side of the SAM, at cheek position, (b) the handset model no. 2 referenced as seen from the right side of the SAM, at cheek position, and (c) the handset model no. 2 in hand close to SAM at cheek position.
(a) Handset model no. 1 at cheek position and operating at 900 MHz (b) Handset model no. 2 at cheek position and operating at 900 Handset model no. 1 at cheek position and operating at 1800 MHz (d) Handset model no. 2 at cheek position and operating at 1800 MHz

Figure 7 :
Figure 7: The three-dimensional radiation pattern in (V/m) of both handset models in hand close to HR-EFH at cheek position and operating at different frequencies.
Handset model no. 1 at tilt position and operating at 900 MHz (b) Handset model no. 2 at tilt position and operating at 900 Handset model no. 1 at tilt position and operating at 1800 MHz (d) Handset model no. 2 at tilt position and operating at 1800 MHz

Figure 8 :
Figure 8: The three-dimensional radiation pattern in (V/m) of both handset models in hand close to HR-EFH at tilt position and operating at different frequencies.

Figure 9 :
Figure 9: Sliced-distribution of the averaged peak SAR 1g in the yz-plane of the HR-EFH phantom in cases of handset models at cheek position.The antenna input powers are 0.6 W and 0.125 W for the frequencies 900 MHz and 1800 MHz, respectively.(a) Model no. 1 at 900 MHz, (b) Model no. 2 at 900 MHz, (c) Model no. 1 at 1800 MHz, (d) Model no. 2 at 1800 MHz.

Figure 10 :
Figure 10: Sliced-distribution of the averaged peak SAR 1g in the yz-plane of the HR-EFH phantom in cases of handset models at tilt position.The antenna input powers are 0.6 W and 0.125 W for the frequencies 900 MHz and 1800 MHz, respectively.(a) Model no. 1 at 900 MHz, (b) Model no. 2 at 900 MHz, (c) Model no. 1 at 1800 MHz, (d) Model no. 2 at 1800 MHz.

Table 1 :
The proposed antenna dimensions and specifications for both handset design models at different frequencies.

Table 2 :
The generated FDTD-grid properties of both handset models in hand close to head, SAM and HR-EFH.

Table 3 :
Computational results of the antenna performance parameters of both handset models at 900 MHz in all cases.

Table 4 :
Computational results of the antenna performance parameters of both handset models at 1800 MHz in all cases.

Table 6 :
The computed averaged peak-SAR (over 1g and 10g) and power absorption in tissues, radiated power, total loss, total isotropic sensitivity, and computation error for both handset models in hand close to head at different positions and at 1800 MHz.1800 MHz-Cheek 1800 MHz-Tilt SAM HR-EFH (Adult) SAM HR-EFH (Adult) Handset model Model no. 1 Model no. 2 Model no. 1 Model no. 2 Model no. 1 Model no. 2 Model no. 1 Model no. 2 model no.2, and the power absorbed in head can also be reduced by (27.9%-77.5%).The computation errors are less than 2% for all cases in presence of HR-EFH, whereas for the cases of SAM presence they are (1.4%-4.4%).