With increasing growth of wireless communication technologies, high data rates and intelligent services are being continuously demanded. Because of the advantages of large channel capacity, massive connection density and high spectral efficiency, the 5G wireless communication systems have attracted increasing attention in both academic and industrial fields [1]. One of the essential enabling technologies for the 5G communication systems is the Multiple Input Multiple Output (MIMO) technology. By utilizing the multipath property of the wireless communication environment, a MIMO system can improve the spectral efficiency and channel capacity without needing more bandwidth and/or transmitting power [2]. Whenever increasing MIMO system’s antenna elements, larger channel capacity and more link reliability can be obtained. However, because of mobile devices’ limited space, the mutual coupling problem is becoming more pronounced with more penetration of antennas. The MIMO system’s performance is severely affected by the mutual coupling impact [3]. So, various techniques have been employed and provided reasonable solutions to solve this problem. One of these techniques includes the spatial diversity method [4–6]. Good isolation can be obtained by this method but at the expense of decreasing the number of antennas, where the space variation between antennas is used as a key approach to achieve low mutual coupling. Isolation enhancement can be done with more compactness than the spatial diversity method by employing the polarization diversity method [7,8], and the pattern diversity method [9,10]. However the compact size and simple structure design of 5G mobile devices, which have high diversity performance capability with low ECCs, remains a highly requested aim for antenna engineers in designing MIMO antenna systems. Moreover, many other isolation techniques using external decoupling structures have been reported in the literature including parasitic structure [11,12], hybrid elements (neutralization line and parasitic element) [13], neutralization line [14,15], decoupling network [16,17], slot etching [18], hybrid decoupling like ground slot etching and meandered line parasitic element [19] or ground slot and neutralization line [20], and multimode decoupling technique [21]. Although these isolation techniques, which are based on additional decoupling structures, provide good isolation enhancement, the antenna efficiency might be reduced significantly. In other words, the isolation improvement is done at the expense of the antenna efficiency. For example, the antenna efficiency in [15] and [21] is affected notably, decreasing to 30% and 34%, respectively. Good isolation and antenna performance have been reported by utilizing the self-isolated technique [22]; however, the antenna element size is too large. In a self-isolated property, an antenna element operates as a radiated element as well as an isolation one at the same time. So, there will be no need for additional decoupling elements or/and isolation techniques. Nevertheless, a compact-size massive MIMO antenna system with good isolation and high efficiency for 5G mobile devices is still a challenge.

In this paper, a compact self-isolated eight-element antenna system that operates in the (3400–3600 MHz) band is presented for future 5G MIMO mobile handsets. The main feature of this work is that it can effectively solve the isolation problem with keeping high antenna efficiency. Based on the self-isolated property, good isolation is obtained without utilizing decoupling elements and/or isolation techniques. In addition to that, a compact antenna size and simple structure is presented in the proposed system where the antenna element’s overall planar size printed on the side-edge frame is just (9.57 mm × 5.99 mm). Furthermore, a high diversity performance with very low envelope correlation coefficients (ECCs) is attained, where a very good independence antenna’s far-field radiation characteristics can be shown. Besides, good antenna and MIMO performances are achieved. The proposed antenna system was simulated and analyzed using the CST Microwave Studio software (version 2019). A prototype model of the proposed system was fabricated and its performance was measured. Reasonable agreement between the measured and simulated results has been obtained.

The proposed eight-element MIMO antenna system for 5G mobile phones is presented in Fig. 1. The eight antenna elements of the proposed MIMO system are located over a mobile device’s two long side edges. The main board size is (150 × 75) mm². This is a typical size of a 5.5-inch phone device. A side edge frame of 7 mm height is selected for a slim smartphone device requirement. Both the system circuit board and the frame are designed by employing a double-sided FR4 substrate of 0.8 mm height. It has an approximate relative permittivity of 4.3 and a dielectric loss tangent of 0.02. At the bottom and the top of the main system board, two clearance areas are left for the current 4G antennas and other different antennas. The size of each clearance area is (8 × 75) mm². As can be observed from Fig. 1, the eight antenna elements are printed on the inner and outer surfaces of the two long side-edges of the frame substrate. They are located perpendicularly on the mobile device’s main substrate, with each side having four antenna elements. All antennas have the same structure and dimensions. A 50 Ω SMA connector is employed for feeding antenna elements via holes from the backside of the main system ground. As illustrated in Fig. 1, the distance between the front sides of the antennas which are printed on the inner frame surfaces is descried by D, and the distance between the back sides of the antenna elements is denoted by d, while the distance between the antenna feeding lines is indicated by DL. For example, D15 means the distance between the front sides of Ant1 and Ant5 and so on. In addition to that, the four antennas located at the mobile frame’s corners are symbolized as Ant1, Ant2, Ant3 and Ant4, while the four antennas located at the middle of the mobile frame are denoted by Ant5, Ant6, Ant7 and Ant8. The following parameters are optimized for the proposed MIMO antenna system (D15 = D47 = D26 = D38 = 30.43 mm, D56 = D78 = 34.13 mm), (d15 = d47 = d26 = d38 = 33 mm, d56 = d78 = 39.27 mm), and (DL15 = DL47 = DL26 = DL38 = 39.62 mm, DL56 = DL78 = 34.13 mm). One can observe from the distance parameters the similarity of the frame long side-edges.

Fig. 2 shows the perspective view of a single self-isolated compact antenna element. As seen from this figure, the antenna element can be divided into two parts, a front part and a back part. Fig. 3 depicts the detailed structure of the two parts. The front part of the antenna, as observed in Fig. 3a, consists of an I-shaped feeding line and a modified Hilbert fractal monopole antenna. The second iteration of the Hilbert fractal geometry is considered and its modification is done by extending its long arm length. The symbol FL is used to describe the I-shaped feeding line which is fine-tuned for matching purposes. The long line length of the Hilbert’s arm is represented by LL and the short one is described by LS. Very good antenna miniaturization is obtained by utilizing the Hilbert space-filling property, where the antenna element overall planar size is just (9.57 mm × 5.99 mm) which operates at the desired (3.4–3.6) GHz band. Due to its small-size characteristic of 0.111 λ0 × 0.069 λ0 at 3.5 GHz and its printed design on the slim mobile frame, the proposed antenna element is appropriate for 5G massive MIMO antenna systems with slim and full-screen smartphone requirements. The smaller the antennas size the greater the distance between antennas. So, the probability of increasing the isolation between antennas is greater. The back part of the antenna element which serves as a virtual ground is demonstrated in Fig. 3b. It has an L-shape which is shorted to the system ground by a 0.5 mm short stub directly under its vertical part. The length and width of the L-shape vertical stub are denoted by VL and VW respectively, while the horizontal stub dimensions are denoted by HL and HW, respectively. The coupling capacitance generated by the L-shaped shorted-stub is used to provide the desirable impedance matching. It is also employed to direct the antenna radiation for better diversity characteristics and radiation performance. For obtaining the desired antenna and MIMO performances of the proposed MIMO antenna system, the following parameters are optimized as follows: FL = 9.97 mm, LL = 7.70 mm, LS = 2.25 mm, VL = 5.84 mm, VW = 1 mm, HL = 6 mm, HW = 0.8 mm.

To further demonstrate the working principle of the proposed compact self-isolated antenna element, the effects of LL and VL on the input reflection coefficient have been demonstrated as depicted in Figs. 4 and 5, respectively. Fig. 6 illustrates the surface current distribution at 3.5 GHz. One can see from Fig. 4 that four values have been chosen for LL: 7.5 mm, 7.7 mm (the proposed one), 8 mm and 8.5 mm. As expected, the antenna resonance frequency shifts according to the antenna length. So in this case study, the antenna resonance frequency shifts to a lower value as the effective value of LL increases. As can be observed form Fig. 5 that four values have been introduced for the VL parameter which are 2.85 mm, 3.85 mm, 5.85 mm (the proposed one) and 6.85 mm. So the optimized VL parameter value, which is an effective parameter of the coupling capacitance generated by the virtual antenna ground, plays a vital role for obtaining a very good impedance matching. In addition to that, the virtual antenna ground (L-shaped shorted-stub) has been employed to direct the maximum antenna gain direction to a direction far away as much as possible from other antennas maximum gain directions. Thus, the isolation among antennas in the proposed MIMO system is improved. It is worth noting that these case studies have been taken on a single antenna (Ant1 alone) before extending the system to the proposed eight antenna elements which will be discussed in the next section. Fig. 6a shows the surface current distribution of Ant1 front side. As depicted in this figure, the current mainly concentrates on the path (ABCDEFGH) and of course on the feeding line. The length of this current flowing path (the current direction is denoted by arrows) is about 22.85 mm, which is 0.26 λ0 (where λ0 is the free space wavelength of 3.5 GHz). So this current path represents a 0.25 λ0 monopole-like mode operating at 3.5 GHz resonance frequency. One can also see that the direction of the current flow is inverted at the IJ section and the current intensity is very weak, so the points M and N are considered as the current null points. The current intensity at the KL section is also weak, and point O is accounted as the current null point because of the capacitance in the IJ section (the lower section) which blocks the current flow at this point. Fig. 6b illustrates the current distribution of the antenna virtual ground (L-shaped shorted-stub) with the main system’s ground. As observed, the current intensity is mainly focused on the region confined under the antenna virtual ground at the main system’s ground plane, which is the reason behind the self-isolated property of the proposed antenna element. The advantages of the proposed antenna structure with its self-isolated property in the proposed eight antenna elements will be discussed in the next section.

The proposed 8-element MIMO smartphone antenna was fabricated and experimentally tested. The assembled prototype model is depicted in Fig. 16. Through the measurement process, each antenna under test (AUT) is connected to the vector network analyzer (VNA) and other antenna elements are connected to 50 Ω matched terminal loads (MTL). Due to the similarity and for the brevity, only the necessary simulated and measured S-parameters have been plotted in Figs. 17a and 17b, respectively. As observed, acceptable impedance matching and good isolation performance (better than 15 dB) are obtained by the proposed MIMO antenna system. The measured and simulated results are in good agreement. However, there are some deviations between them which may be due to fabrication tolerances, soldering effects, and/or measurement errors.

Tab. 1, presents a summary for the performance comparison of the proposed eight-element MIMO antenna system with some other recent MIMO antenna systems reported for the 5G mobile phone devices. Compared with other MIMO antenna systems, the proposed one has got a smaller size and lower ECC. Good system isolation and eight antenna total efficiencies are obtained by adopting the self-isolation property where there is no additional antenna efficiency loss by other decoupling elements and/or isolation techniques. So it can be seen from the Tab. 1, that the proposed eight-element MIMO system is able to provide a massive MIMO system with a very comparable antenna and MIMO performance.

In this study, an eight-port MIMO antenna array is introduced and examined for 5G massive MIMO smartphone devices. Very good antenna miniaturization is achieved by utilizing the Hilbert fractal space-filling property. Due to the self-isolated property of the proposed antenna structure, good isolation is achieved (better than 15 dB) without using decoupling elements and/or isolation techniques. The main trait of the self-isolated property is that a single antenna can operate as a radiated and an isolated element simultaneously so that the antenna efficiencies do not decrease. The simulated and measured results of the antenna and MIMO performances obtained, as well as the structure simplicity, compactness, self-isolated property and position independency, show that the proposed eight element MIMO antenna system is a convincing candidate for the future 5G mobile phone devices.