LETTER IEICE Electronics Express, Vol.10, No.11, 1 8 GA optimization of transparent MIMO antenna for smartphone Sindhuja Patchaikani and Yoshihiko Kuwahara a) Graduate School of Engineering, Shizuoka University, 3 5 1 Jokoku, Naka-ku, Hamamatsu 432 8561, Japan a) tykuwab@ipc.shizuoka.ac.jp Abstract: A shared aperture transparent MIMO antenna for smartphone is proposed in this study. Genetic algorithm (GA) is used to optimize the antenna geometry by interfacing Matlab tool and HFSS- EM simulator using visual basic (VB) scripting. The performance of transparent conducting film as an antenna element is enhanced through numerical simulation. An optimal antenna design is obtained through this optimization algorithm. The evaluated antenna s electrical performance like VSWR, spatial correlation coefficient (SCC), and peak gain (db) are desirable in the cellular frequency bands of 0.850 GHz and 2.2 GHz. Keywords: transparent antenna, MIMO, diversity, GA optimization Classification: Microwave and millimeter wave devices, circuits, and systems References [1] N. Guan, H. Furuta, K. Himeno, K. Goto and K. Ito: IEICE Trans. Commun. 90-B [9] (2007) 2219. [2] T. Peter and R. Nilavalan: PIERS Proc. (2012) 836. [3] N. Abdullah and Y. Kuwahara: IEEE Trans. Antennas Propag. 60 [3] (2012) 1228. [4] N. Abdulla and Y. Kuwahara: Proc. EUCAP 2012, CA06-10 (2012). [5] P. Sindhuja and Y. Kuwahara: IEICE Technical Report, AP2012-108 (2012) 87. [6] C.-F. Huang and L. Chen: Electron. Lett. 38 [20] (2002) 1162. [7] S. K. Ahmed, Z. A. A. Al-Hussein and M. K. Hussein: Thi Qar University Journal for Engineering Sciences 2 [4] (2011). [8] Y. Rahmat-Samii and E. Michielssen (Editors): Electromagnetic Optimization by Genetic Algorithms (Wiley-Interscience, New York, 1999) 36 49, 417 418. 1 Introduction Both the forthcoming 4G and WiMAX technologies require high data rates and longer range to provide quality services to end users. In the mobile 1
device, the antenna is the only device that touches the network frequency. Optimizing the performance of the antenna in smaller mobile devices to adapt 4G technology is becoming highly challenging. Antenna size, mutual coupling between multiple antennas and antenna material are the prime factors need to be considered during the process of antenna design. Conventional mobile antennas are mounted in the body of the mobile phone which is made up of non-metallic substance whereas the body of the future smartphone may consider to be made up of metallic substance for its less weight and thinness. Hence, installation of antennas over the metal body seems complex due to unwanted electrical influences. The design of realistic diversity antennas on mobile terminals for MIMO systems remains a challenging issue. The main challenge in designing array antennas on a small mobile terminal is to achieve a high SCC in addition with low VSWR and high gain. There is a need to design an appropriate and realistic diversity antenna array on mobile terminals for MIMO systems and also to mitigate multipath the fading effects. Thin profile, less weight, and reduced size of upcoming smart phone model urges antenna design engineers to bring up a more adaptable antenna pattern in the near future. Glass display makes up a significant proportion of the mobile phone structure. If an antenna can be made from a transparent conducting film and also be integrated into the display [1], the restriction of design space for MIMO antennas can be easily solved. This antenna pattern overcomes the difficulties in the existing half wavelength dipole antenna system. Currently, antennas with soft visual impact became an attractive solution to optimize their integration in display touch panel [2]. In this study, we have suggested a technique in the designing of array antenna for Smartphone (iphone 4S model).with reference to vehicular rear defogger antenna we proposed our antenna pattern [3, 4]. In the first stage, we studied only the performance of transparent conducting film when attached to the display panel along with the hollow metal body [5]. Optimization of antenna pattern is carried out with the help of genetic algorithm [6, 7, 8]. Two ports are set for simplicity. The antenna performance is observed at various cellular frequency bands such as 0.850 GHz, and 2.2 GHz. We have achieved and confirmed the proposed array antenna s electromagnetic characteristics, numerically using HFSS-EM simulator. 2 Concept of shared aperture transparent MIMO antenna The main concept of our antenna design is the transformation of separately placed antenna element around the body of the mobile device into single antenna aperture with multiple antenna ports integrated with the display panel of the mobile device. In reference [5], we have modified the multi-band shared aperture array antenna using the rear defogger for the Smartphone. The heating wire is replaced to the transparent sheet. Figure 1 and 2 gives details of the antenna. The display panel (Glass) of mobile device with relative permittivity (ε r = 2
Fig. 1. Non-optimized referenced Smartphone antenna model with proposed antenna design (Front view). Fig. 2. Non-optimized Smartphone antenna model (Side view) 7), loss tangent (0.00036), and thickness 1 mm acts as a substrate. The rear side of the mobile phone (metal case) with conductivity, σ =5 10 7 S/m act as a ground plane. Then the proposed antenna, integrate along with the top of glass surface, made of transparent conductive film of conductivity (σ =5 10 5 S/m) and thickness (0.175 mm), acts as a radiating element. Antenna ports are set, where one end is connected to the edge of horizontal strips and other end to the metal body. The performance of the antenna pattern shown by the figure 1 is not so desirable at lower frequency [5]. Hence an auto-optimization method has to be carried out, to enhance the proposed antenna s electrical characteristics like VSWR, peak gain at each port and also reduce the SCC value for all combination of antenna ports used in the design at the desired frequency bands of 0.85 GHz, and 2.2 GHz. 3 Numerical analysis The main objective of this simulation is to obtain the best aperture shape, at which the suggesting antenna can produce more desirable performance at required frequency bands. 3
Table I. Simulation conditions Fig. 3. Optimizing antenna parameters in GA optimization 3.1 Genetic algorithm (GA) optimization Smaller the size of the antenna lower the electrical performance at lower frequency band. Manual optimization of such smaller dimension antenna is not much easier and it needs more complex steps. Hence we adapted genetic algorithm technique to do auto-optimization of antenna pattern by interlinking MATLAB tool and HFSS simulating software. This algorithm works for the best positioning of both horizontal (x1, x2) and vertical conducting strips (y1, y2, y3, y4, y5, y6, y7) over the limited surface area of touch panel (See Fig. 3). The spacing between the horizontal and vertical conducting strips is determined by GA calculation. For convenience, we assumed that the antenna pattern is symmetry along x-direction, hence the value for antenna parameter x3 is (115-x2) and x4 is (115-x1). VSWR, SCC, and 1/peak gain (db) are set as objective functions which are exported from HFSS simulator. In general GA works for minimization of the objective function. Thus our objectives are processed towards minimum value. Figure 4 explains how the antenna parameters involved in the optimization 4
Fig. 4. Flowchart for genetic optimization process in proposed antenna design process. 3.2 Pareto ranking This algorithm outputs several sets of Pareto-optimal solutions rather than single optimal solution. Each optimal solution is a set of values of three objectives [VSWR, SCC, 1/gain]. Pareto ranking of multi-objective optimization is carried out as below to find the final optimal solution [8]. For explanation here, we consider only three sets of solution. In set1 [VSWR, SCC, 1/gain]={1.4970, 0.6052, 0.3082}, though the VSWR is least the SCC values goes high. In set2 {6.7450, 0.1768, 0.4239}, SCC is least but the VSWR values goes high. In set3 {1.9306, 0.3855, 0.3825}, which has good compromising values corresponds to each objective. Hence set3 is considered as the best optimal solution. 4 Simulation results The algorithm outputs the best optimal antenna dimensions [X] and also its corresponding objective function s values [FVAL]. X is a matrix with columns equals to number of optimizing variable and with rows as same as the number of Pareto solutions, whereas the matrix FVAL has same number of rows as X, and columns equals to the number of objective functions. Figure 5 shows the Pareto relationship among each obtained output data and the arrow locates the best data. Among the 9 optimal solutions generated by GA, one is decided as the best optimized value and the corresponding gene combination [X] is given below, [X] = [8.10, 16.10, 32.10, 51.10, 34.10, 48.10, 4.10, 45.10, 51.10] 5
Fig. 5. Pareto Plots between VSWR and SCC and 1/GAIN (db) obtained at each best gene combinations (antenna dimensions) at 0.850 GHz and 2.2 GHz frequency and the corresponding aperture shape to the above [X] value is shown by figure 6. Based on the results of reference [5], we set the antenna parameters like number of vertical conducting strips, width of conducting strips as 9 and 2 mm respectively. The proposed algorithm mainly works to achieve good electrical performance by adjusting the spacing between the strips. In this case, 6 strips are located separately and three of them get overlapped and merged with each other by GA optimization. The VSWR frequency characteristics and 3D-radiation pattern of each port of optimized antenna pattern at 0.85 GHz and 2.2 GHz is shown by the figure 7 and figure 8 respectively. Genetic algorithm makes this optimization process simple and it took 4hours 15minutes for entire calculation. Lower SCC values at two bands seems that the antenna ports are highly isolated, which is the prime factor for array antennas. By observing the antenna pattern in the figure 6, most of the transparent conducting strips are distributed along the borders of the touch panel, and central area is left undisturbed. Hence it is predicted that the influence of this conducting transparent film over touch sensor in the middle part will be quite low. Table II clearly shows that the antenna performance is improved through this optimization particularly at lower frequency. This optimization methodc IEICE 2013 6
Fig. 6. 3D simulated Smartphone model with optimized antenna pattern Fig. 7. VSWR characteristics of Optimized Transparent Smartphone antenna ology can be used further to enhance the multiband capability of proposed antenna pattern. 5 Conclusion In this study, we have proposed and developed a numerical procedure combining, GA and HFSS to optimize the shared aperture array antenna pattern for miniaturized Smartphone model. We also suggest that a commercially available transparent conductive film (ITO, AgHT-4, FTO) has been implemented as an antenna material to avoid antenna space limitation problem. The electrical performance of proposed antenna has been confirmed at the most accessing cellular frequency bands of 0.850 GHz and 2.2 GHz. We are 7
Fig. 8. 3D Radiation Pattern of Optimized Transparent Smartphone antenna Table II. Simulated results also working on our study further, to improve the performance for 4G and Wi-Fi accessing frequency bands while considering heating loss at transmitting and compatibility with the touch panel. 8