You are on page 1of 6

2016 IEEE 27th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC): Workshop

Inclusive Radio Communication Networks for 5G and Beyond (IRACON2016)

Compact Dual-band Antenna Array for Massive
Linsheng Li1 , Muhammad Ali1 , and Katsuyuki Haneda1
1 Department of Radio Science and Engineering, Aalto University, Espoo, Finland

Abstract—A compact dual band antenna operating simulta- between the neighbored elements are usually adopted to get a
neously at 5.4 GHz and 15 GHz is presented in this paper. acceptable mutual coupling level.
It is a planar array composed of a grid of 6×8 (48) patch
antennas working at 5.4 GHz and 7×9 (63) patch antennas As the data rate requirement in wireless communication
working at 15 GHz. The whole antenna array occupies an area system is increasing dramatically, microwave and millimeter
of 150 mm × 180 mm which fits the ordinary size requirement wave bands could be adopted due to the large available
of wireless access point. All the antenna patch antennas have a bandwidth and their carrier aggregation will be an essential
reflection coefficient lower than −6 dB across 320 and 2000 MHz
bandwidth at 5 and 15 GHz. The electromagnetic coupling technology to ensure wide bandwidth for data transmission.
between any two antennas is lower than −10 dB within the The 5G trials carried out by NTT DOCOMO and Ericsson
specified bandwidth. Measurement results show good agreement demonstrate ultra-high data rates of 10 Gbps using the 15 GHz
with the simulation of input matching, mutual coupling and frequency band [7]. The 5.4 GHz band and the 15 GHz band
radiation pattern. Compared to the ordinary antenna array has been used widely for WLAN (wireless local area network)
design with half-wavelength spacing between two neighboring
antennas, our proposed dual band array design accommodates and regarded as one of the possible candidate bands for future
1.6 times more antennas at 5.4 GHz and extra 63 antennas 5G communication. Therefore, massive MIMO antenna array
at 15 GHz, while maintaining sufficient radiation efficiency, supporting 5.4 GHz and 15 GHz will be necessary. Several
matching bandwidth and decoupling. massive MIMO testbeds have been built but all of them adopt
Index Terms - antenna array, dual band antenna, massive a single frequency band below 6 GHz and half-wavelength
MIMO. distance between neighbored antenna elements [8]–[12]. To
the best knowledge of the authors, there is still no report on
I. I NTRODUCTION dual frequency band massive MIMO antenna array design [12].
Massive MIMO (Multiple Input Multiple Output) deploying Dual band antenna arrays have been adopted widely in wireless
a large number of antennas at the base station introduces communications and the main challenge is the mutual coupling
a high spectral and energy efficiency [1]–[3]. Theoretically, effect between elements working at different bands when they
massive MIMO can reach Shannon limit through increasing are placed close to each other [13], [14]. High gain and good
the number of its antenna elements. The practical challenge bandwidth are usually difficulty to be achieved in dual band
for implementing massive MIMO system is how to pack the array design. This paper presents a compact dual band antenna
highest number of antenna elements in a defined volume array and the design goals are shown as following.
while maintaining the required performance including the
• A dual bands planar array contains at least 32 antenna
input matching and mutual coupling level. Mutual coupling
elements for both 5.4 GHz and 15 GHz within a space
effects among antenna elements in 1D linear array [4], [5]
of 150 mm × 180 mm which is the typical size of WLAN
and 2D rectangular antenna array [6] on the MIMO system
access point.
performance has been studied. Spectral efficiency will always
• The reflection coefficient of each element is lower than
be improved with the increase of antenna elements within a
−6 dB within at least 320 MHz bandwidth for each
specified area but energy efficiency will be decreased sig-
frequency band.
nificantly due to the high mutual coupling level if antennas
• The mutual coupling between any two ports/elements is
are placed too close to each other [6]. It will be beneficial
less than −10 dB across the required frequency bands.
to put as many antenna elements as possible in a fixed
area while satisfying the mutual coupling requirement to get The rest of the paper is organized as follows: Section II
optimum spectral and energy efficiency. For existing massive introduces the design and simulation of an antenna unit, a 2×2
MIMO array design and analysis, half-wavelength distance sub-arrayand the complete antenna array. Section III gives the
measurement results of the single antenna unit and 2×2 sub-
The research work leading to these results was funded by the Academy of
Finland, under the project Massive MIMO: Advanced Antennas, Systems and array. Section IV summarizes major findings from the present
Signal Processing at mm-Waves (M3MIMO), decision number 288249. contribution and future works.

978-1-5090-3254-9/16/$31.00 ©2016 IEEE

The unit element is based on a cross patch model [15] unit adopted for the 5. Fig.6 4. Simulated reflection coefficient for 5. E-shaped patch [17] and The unit element operates at dual bands of 5.4 GHz could be placed input matching simulation result and Fig. matching bandwidth was ≥ 4 GHz.. ports 2/4 and 3/5 are array design has as many as 1.2 15.4 5.4 14.4 GHz with extra 63 patch elements for 15 GHz compared coupling level between each element is shown in Fig.8 6 Frequency (GHz) Frequency (GHz) Fig.58 mm is used for the design.2 14. The performance of the lower frequency patch assure a low surface wave loss for the high frequency band will be affected by the small square patches and retuning (fhigh = 15 GHz). 3 shows the input within the 150 mm× 180 mm area.4 GHz while patch sub-array based on [15] as this type of antenna is the other four square patch antennas at each corner operates suitable for the large frequency ratio.4 GHz only 5×6 (30) patch antennas for 5.2 4. Unit element frequency bands antenna including coplanar multi-resonators antennas [15]. Our proposed antenna matching of the 15 GHz patch antenna. The thickness of the procedure. 4 6 8 10 12 14 16 4 4.. For an ordinary half-wavelength antenna array design.4 GHz) as well as to each corner.6 times antenna elements for exactly the same due to the symmetric design. 4. Different techniques could be used to design a dual/multi A. 1. 2 shows the 5.6 14. The design fits the goals in terms of the and optimization was made to get a required performance required input matching bandwidth (≥ 320 MHz) as well as including the input matching.4 15. 4.4 GHz and 15 GHz array is a cross and the center patch is optimized to operate at 5. Simulated reflection coefficient for 15 GHz square patches Fig. Stacked patch scheme at 15 GHz as shown in Fig.. a cross patch was first designed for 5.33 and a thickness defines the feeding positions of the probes. Due to the ordinal design with half-wavelength spacing. Duroid RT5870 with r = 2. The measurement.2 5.4 GHz and substrate is selected to guarantee an adequate bandwidth for then four smaller patches working at 15 GHz are added to the lower frequency band (flow = 5. 2.8 15 15. 1 shows the design of the unit element. Simulated coupling level between all elements in the antenna unit (the desired bands are highlighted) II. In the design of 1.8 5 5. The antenna 15 GHz. 1. the coupling level between the elements (the design goal is more than 32 elements for each frequency at the required band is ≤ −10 dB.4 GHz patch can achieve an input matching band- has 48 elements for 5.4 GHz and U-slot cut in a broadband antenna [18] and so on.4 GHz cross patch Fig.6 5. stacked patches [16].4 GHz and 63 for 15 GHz elements width of 380 MHz. 3. and Mobile Radio Communications (PIMRC): Workshop: Inclusive Radio Communication Networks for 5G and Beyond (IRACON2016) 2 3 0 S 22 S 33 S 44 Input matching (dB) -10 S 55 1 -20 4 5 -30 -40 14 14. The mutual 5. All the to the symmetric design some of the curves coincide with each . design is optimized to fit the specific dual frequency band the numbers represent different feeding ports and the circles requirement. mutual coupling and gain level. Indoor. For the 15 GHz the input band). Dual band antenna unit 0 S 12 0 S 13 -10 S 14 Coupling level (dB) -4 S 15 380 MHz -20 S 21 |S11| (dB) S 23 -8 -30 S 24 -40 S 25 -12 S 31 -50 S 32 -16 S 34 -60 . D ESIGN OF ANTENNA UNIT AND ARRAY simulations are carried out in CST microwave studio.4 4.8 16 Frequency (GHz) Fig. the number of the elements for the predefined space and it The 5.6 15. 2016 IEEE 27th Annual International Symposium on Personal. Probe feeding technique is is not adopted here due to the internal feeding issue and adopted for this design and SMA connectors are used for the high cost of fabrication compared with a planar structure. Fig.

8. (b) H-plane are separated with a distance of 16.11 o -20 o -20 o -20 o -60 60 -60 60 S 1. Radiation pattern of the 5. The required bands are highlighted and in a coupling level lower than −20 dB within more than coupling level is below −10 dB for the desired bands. and Mobile Radio Communications (PIMRC): Workshop: Inclusive Radio Communication Networks for 5G and Beyond (IRACON2016) o o 0 0 dB 00 dB o -30 o 30 o -30 o 30 11 12 13 -10 -10 o -20 o -20 o -60 -30 60 -60 o -30 60 3 4 -40 -40 o o o o -90 90 -90 90 10 5 9 -40 -40 -30 -30 o -20 120 o -20 120 o 1 2 o -120 -120 -10 -10 o o o 0 dB o -150 180 o 150 -150 0 dB 180 o 150 6 7 8 (a) (b) Fig. Based on the simulation results. As shown in Fig. to design compact MIMO antennas with a high isolation The highlighted boxes show the interested bandwidth and the and there are several methods to improve the isolation level mutual coupling level is below −10 dB in this design. 2×2 Sub-array array design. S43 = S52 . 5. 2016 IEEE 27th Annual International Symposium on Personal. 1 and it has a gain of 7.7 mm which will result other.4 GHz.8 mm requirement without any extra decoupling network.8 mm. The prototype is predefined space and also fulfill the mutual coupling level shown in Fig.4 GHz and 15 GHz and after that a 2×2 sub-array is patch antennas.12 -30 -30 -30 S 1. 7. Fig. The main constraint for the design array are fabricated. 5 and 6.13 -40 -40 -40 S 21 o o o o -90 90 -90 90 -50 S 23 -40 S 24 -40 -60 S 25 -30 -30 o -70 S 26 o -20 -20 120 120 o -120 o -80 S 27 -120 -10 .10 Coupling level (dB) S 1. The reflection coefficient . e. 5 shows the radiation pattern of the 2. a 2×2 sub-array and a complete antenna for the 2×2 sub-array. Fig. simulated to verify the mutual coupling level. the final space between The radiation patterns for E-plane and H-plane are shown the cross patches is optimized and the horizontal spacing is in Figs. between antenna elements [20]–[22]. The simulation work includes optimization of the cross. 8 shows the coupling level between the elements.4 GHz cross patch antenna by simulation. III. The planar array is composed of 6×8 = 48 separations between the antenna elements are studied for 5.0 dBi. 2 GHz bandwidth.4 GHz cross patch antenna with a gain of 7. The square patches for 15 GHz the 2×2 sub-array are measured. Indoor.4 GHz cross patch antennas and 7×9 = 63 15 GHz square 5. It is difficult to the symmetric design most of the coupling curves coincide.. -10 o o -90 o 0 dB o -150 0 dB 150 4 6 8 10 12 14 16 -150 150 o o 180 Frequency (GHz) 180 (a) (b) Fig. 6. These are the minimum distance we shows the radiation pattern of the 15 GHz antenna of patch can use to keep the mutual coupling level below −10 dB.4 mm to acquire the 5. Radiation pattern of the 15 GHz square patch antenna by simulation. (a) E-plane. 9 and the PCB occupies an area of 150. Coupling between all antenna array elements in the sub-array Fig. Antenna array of them only work for a single frequency band. 10. M EASUREMENT RESULTS patch antennas to keep the coupling level below −10 dB A single unit. Due between each element is an important factor [19]. The complete antenna array is designed based on the we will try to place as many antennas as possible in the distance given by previous simulation results. 2×2 antenna sub-array (a) E-plane. One of the mothods is decoupling networks but this will occupy extra space and most C. For the compact MIMO antenna design... 7. 6 required coupling level. mutual coupling Fig.6 dBi.g. The fabricated prototypes are shown in comes from the mutual coupling level between the cross Fig. Fig. number 2 as referred in Fig.4 mm while the vertical space is 0. In our design. Proper × 178. (b) H-plane o o 00 dB o 00 dB 0 S 19 o o o -30 30 -30 -10 -10 30 -10 S 1. The reflection coefficient and mutual coupling in patches working at 5. mutual coupling level below −10 dB will not alter the gain and efficiency in the compact B.

4 GHz patch (a) single element (b) 2×2 sub-array 0 S44 Sim S 44 Input Matching (dB) -10 S55 Sim S 55 -20 -30 -40 14 14. 9. Mutual coupling for 5.11 Sim S 12.2 4. 13.8 16 Frequency (GHz) Fig.4 5. Antenna array -10 S 21 Coupling level (dB) -20 Sim S21 S 31 -30 Sim S 31 -40 S 41 Sim S -50 41 -60 -70 -80 4 4.6 5.2 5.2 4. Indoor.6 14. Input matching for 5.6 14. sub-array and complete -50 antenna array 14 14.8 6 Frequency (GHz) Fig.6 5.8 5 5.2 14.2 15.8 6 Frequency (GHz) Fig.4 GHz cross patch are shown in Fig.8 5 5.4 GHz is shown in 12 mutual coupling measurement.4 GHz patch 0 Fig. 10 (b) are measured. For the 5. Mutual coupling for 15 GHz patch measurement results for 5. the four cross patches in the which indicates a coupling level lower than −10 dB within 2×2 sub-array as shown in Fig.6 4.2 14.4 15. 5 Sim S Coupling level (dB) -10 11.6 4.4 GHz the mutual coupling performance for 5.8 15 15.8 15 15.5 S 12.6 15. Input matching for 15 GHz patch 0 S 11. 14.11 -30 (c) complete antenna array -40 Fig.4 4. 11.4 15.2 5. 10. 12. .10 Sim S 12. 11 along with the simulation results. A the required bandwidth. and Mobile Radio Communications (PIMRC): Workshop: Inclusive Radio Communication Networks for 5G and Beyond (IRACON2016) 0 Measured Simulated -4 300 MHz |S11 | (dB) -8 -12 4 4.4 14.6 15. 2016 IEEE 27th Annual International Symposium on Personal.4 14. fabricated prototypes of the antenna unit. The measurement result agrees with the simulation result except a little narrower comparison between the simulated and measured results of bandwidth for the S11 lower than −6 dB.10 -20 S 12.8 16 Frequency (GHz) Fig.2 15.4 5.4 4.

Gao. [12] A. 10 (a). pp. 11–22. Puglielli. and L. L.” IEEE Signal Process. “Dual-polarized turning torso antenna array for massive mimo systems. G. a compact dual band antenna array with a Fig. “Performance analysis compared with the traditional half-wavelength scheme. 2014. Wright.4 GHz [6] S. K. The reflection coefficient and mutual -40 -40 coupling level between any two ports could be measured but o -90 90 o -90 o 90 o -40 -40 it will be too time consuming to measure each feeding port.4 GHz patch transmitters in fixed physical spaces: The effect of transmit correlation and mutual coupling. Y. Artiga. 30. and L. Dec. K. Miers. The antenna array contains curve for co-polarization and red curve for cross-polarization part 48 cross patch antennas at 5. Chen. and F.” in vol.4 GHz patch. According to the measurement vol.. Marzetta. B. Larsson. The H-plane measurement result for 5. Ma. Ratnarajah. Marzetta. Kundargi. Radiation pattern for 15 GHz patch.6 times antenna elements for 5. Biswas. and . “Massive mimo for next generation wireless systems.8 mm.” www. no. Lau. 2013. along of large multi-user mimo systems with space-constrained 2d antenna with extra 63 patch elements for 15 52. G. “Miniaturized [1] T.” Bell Labs Tech. Marzetta. Tufvesson.. USA. O. 15.14 -10 where mutual coupling level is lower than −20 dB between o 150 o -150 0 dB 14 and 16 GHz. Masouros. no. P. 1803–1804. N. Perruisseau-Carrier. Jul. the result is shown in Fig. Commun. area of 150. 2014. the mutual coupling between any two ports is lower [4] C. pp.. Lu. Ying. N. Parini. IEEE APSURSI. “Mutual coupling coefficient lower than −6 dB. Liu. Ratnarajah. IEEE EuCAP. D. M. Shepard. B. no. Masouros. Indoor. Yu. IEEE APSURSI. 10 (b) with the final result shown in Fig.” IEEE Commun. 180 o The radiation pattern for antenna elements working at both frequency bands are measured in the anechoic chamber. Zhong. and T. MobiCom. Rusek. pp. T. This -60 60 -30 could be due to the fabrication error in the PCB and differences -40 between the connector model used for simulation and final o -45 o -90 90 product. Devillers. Y. results. pp. 2015.” in Proc. 2014. 2016 IEEE 27th Annual International Symposium on Personal. Courtade. V. Conf.” IEEE Trans. -30 -30 o It is still an open question on how to measure the antenna -120 o -20 120 -120 o -20 120 o performance of massive MIMO array with more than 100 -10 150 o -10 o o o 0 dB 150 0 dB feeding ports. T.4 GHz patches but it is difficult to implement with [8] J. [9] R. Edfors. “Scaling up mimo: Opportunities and research platform. and Mobile Radio Communications (PIMRC): Workshop: Inclusive Radio Communication Networks for 5G and Beyond (IRACON2016) o The reflection coefficient of the 15 GHz square patch 00 dB o o antenna is measured based on a single element prototype -30 -10 30 as shown in Fig. 19th Annu. Zhang. Our proposed antenna array effects in multi-user massive mimo base stations. I. B. E. Proc. and T. Mag. Vieira. (a) E-plane. for the 5. 20. [7] Ericsson.4 GHz and 63 square patch antennas for 15 GHz configured in a uniform planer array. The mutual -30 o coupling measurement was measured based on the 2×2 sub.” IEEE Trans. Cuthbert.4 GHz is unavailable 0 0 dB o o o o 00 dB o -30 30 -30 30 due to the limited measurement facility. 15 and Fig. Tufvesson. pp. O. L. This array is arrays. and F. C. 16 and the polarization and red curve for cross-polarization part measured radiation patterns agree with the simulation results. [11] C. H. 13. wall. The fabricated prototype of the complete array occupies an [3] E. SMA connectors in one patch. challenges with very large arrays. blue curve for co- measurement results are shown in Fig. Edfors. vol. 2794– elements give a matching bandwidth of 300 MHz while the 2804. [2] F. Jan. bandwidth for 15 GHz is more than 2 GHz for a reflection [5] X.” in Proc. 2013. 61. the measurement results -45 -40 are quite sensitive to the dimension tolerance. Feb. 7. pp. J. Malkowsky. Mar. design has as many as 1. 2012. 40–60. Larsson. “A flexible 100-antenna the current available SMA connectors as the distance between testbed for massive mimo. “Docomo and ericsson succeed in worlds first trial to achieve suitable for small base stations for 5G and future wireless a cumulative 20Gbps with two simultaneously connected mobile devices communication. “Massive mimo: An introduction. As for the 15 GHz design. -150 180 o -150 180 o IV. and T..” in Proc. May R EFERENCES 2015. Sellathurai. 2013. Dual polarized feeding would be preferred in 5G outdoor trial. The Fig. (b) H-plane. Parini. G. Wong. antenna array with co2z hexaferrite substrate for massive mimo. -10 -10 -20 The measurement results for the complete antenna array -60 o -20 60 o -60 o o 60 -30 -30 is still under going. F.8 mm × 178.” in Proc. L. o 120 -120 -20 array shown in Fig. Nikolic. NY. Nieman. Narevsky. Gao. A o -20 o frequency shift happens for the input matching results. Wireless Commun. E-plane radiation pattern for 5. 186–195. Jul. [10] Q. 287– the two feeding ports is not wide enough to accommodate two 293. Persson. vol. and J.. L. pp. New York. 1. C. 16. Mag. Jul. Tufvesson. 2. “Large-scale mimo than −10 dB within the required bands and the 5. C. to be published. blue total of 111 elements is presented. “Argosv2: A flexible many-antenna O. and Z. Z. S. GC Wkshps. Edfors. C ONCLUSION (a) (b) In this paper. 163–166. Z.

Lehtovuori. Apr. Alon. [20] C. Luxey. and T. 1183– 1192.. M. IEEE ICCW. Ma. 4341–4344. pp.. vol. [22] S. K. Li. and S.” in Proc. A. no. vol. 1995.. no. 20.” IEEE Trans.” IEEE Antennas Wireless Propag. 100–103. 2014. pp. 8. no. Chiu. pp. Dioum.. Antennas Propag. Fu. Soltani and R. [19] S. 1234–1241. H. and S. “A novel compact dual- band lte antenna-system for mimo operation. Lett. May 2015. 1140–1149. Lee. “T-shaped decoupling network for wideband isolation improvement between two strongly coupled antennas.. 63. “On the constrains to isolation improvement in multi-antenna systems. H. 4. 61. Mar. Maci. [13] Y. Jun. S. Wu. 64. S. 2291–2296. Nov. vol. 2013. M. C. and R. Cui. 62. S. 1706–1707. pp. S. Lehtovuori. A.” in Proc. vol. Rahmat-Samii. 15. Apr. Venkatasubramanian. . 2016. L. Lu. vol. vol. pp. “Dual-band probe-fed stacked patch antenna for gnss applications. pp. Antennas Propag. S. no. Luk. Indoor.. N. Noguchi. Fang. “Design of wideband/dual-band e-shaped patch antennas with the transmission line mode theory. C. D. M. no. Haneda. L. 2015. “Single-layer single- patch dual-band and triple-band patch antennas. Mok. IEEE EuCAP. F. and Y. vol. A. “A scalable massive mimo array architecture based on common modules. G. [17] K. and K.” in Proc.” IEEE Trans. Icheln. Murch.” Electron. Rajagopalan. Salvador. A.. and K. Borselli. 4. Falciani. H. 4. 158–161. 64.” IEEE Trans. Wang. “Very compact fully lumped decoupling network for a coupled two-element array. 2016. IEEE LAPC. 2009. [15] C. no. and Mobile Radio Communications (PIMRC): Workshop: Inclusive Radio Communication Networks for 5G and Beyond (IRACON2016) E. Diallo. 2015. Li. “Investigation of a triple-band multibeam mimo antenna for wireless access points. Venkatasubramanian. “A compact planar printed mimo antenna design.” IEEE Antennas Wireless Propag. Aug. Y. vol. Wong. 3. Sep. 2016. Pan. pp. Farssi. Icheln. [14] I. 8.” IEEE Trans. Heino. 2015. 31. and K. Haneda.” IEEE Trans. pp. [21] L. Antennas Propag. Lett. Apr. “Dual frequency planar antenna at s and x bands. C. Antennas Propag. 2016 IEEE 27th Annual International Symposium on Personal. Lett. Antennas Propag.. [18] W. 1310–1315. and C. C. pp. [16] Z.