Compact Substrate Integrated Waveguide Filtering Antennas: A Review

In recent years, with the rapid development of integrated circuits, the filtering antenna (FA) circuit of the substrate integrated waveguide (SIW) has attracted widespread attention in wireless communications. Numerous structures have been proposed to improve the performance of SIW FA. Hence, this paper presents the development and miniaturization trend of the SIW FA, and briefly introduces the structure and historical development of the SIW FA. In terms of the mode, multi-mode cavities and incomplete mode cavities are constructed. Multi-layer structures, special structures and the slow-wave technology are constructed in physical structure. Several methods and design examples of the SIW FA miniaturization are introduced from these two aspects. Based on the combination of the filter and antenna, the FA remains the advantages of the SIW structure, mainly including small size, low insertion loss, high quality-factors, and easy integration in planar radio frequency components. Therefore, it can further reduce the circuit, allowing SIW FAs to be applied in many aspects such as the 5G (5th generation mobile communication technology) and wireless local area network communications.


I. INTRODUCTION
Radio frequency (RF) front-end system is an evolving technical solution to support high-performance wireless communication systems. In traditional RF front-end systems, separate components are usually employed, such as filters and antennas. Nevertheless, these separate devices would make the system bulky and costly. Besides, there are also additional losses at the connection of the two devices. The development of modern requires small size, low cost, low loss. Therefore, to meet the above requirements, the filtering antenna (FA) that integrates the filter and antenna is proposed. In addition, to further optimize the transition part between the FA and planar circuit, the substrate integrated waveguide (SIW) The associate editor coordinating the review of this manuscript and approving it for publication was Mohammad Tariqul Islam . structure is proposed in [1] and [2]. As shown in Fig.1, its specific structure mainly consists of two parallel metal plates, a dielectric substrate, and a series of metallic vias. The influence of bandgap loss can be eliminated by adjusting the distance between metallic vias. On the one hand, the SIW structure retains the advantages of the traditional waveguide, such as the high quality-factor and the high-power capacity, while reducing the circuit size to a large extent. Furthermore, as it is easier to integrate with other devices, there is no need for transitions between components, making the overall system less lossy. Therefore, the SIW technology has advantages in small size, low cost, low loss systems and reduces the total requirement for transition. After a series of improvements, the SIW can be applied to filters [3], [4], antennas [5], [6], power amplifiers [7], [8], power divider [9], oscillators [10] and polarizers [11]. It is apparent that the SIW structure can be  utilized in the RF front-end as a standard waveguide. Hence, the SIW has excellent development prospects in the design of different devices.
In traditional wireless systems, filters and antennas are usually designed independently and cascaded directly. Matching networks are used between filters and antennas which would lead to an increase in size and insertion loss (IL). Hence, a great configuration is proposed to integrate the filter and antenna, which can realize the functions of filtering and radiation within a compact structure. The integration of the filter and antenna is shown in Fig. 2(a), and the approximate curves of gain and S 11 of the FA are shown in Fig. 2(b). In order to further intuitively compare the performance and size differences between filter and FA, as shown in Fig. 3(a), the output end of the filter was connected with 50 impedance, and as shown in Fig. 3(c), the output of the FA is etched with a long slot to make energy diffused outward. Since the S 21 parameters of the filter and the gain of the FA have similar port characteristics, a comparison between them can be obtained. As shown in Fig. 3(b) and (d), the S 21 parameters of the filter are 0dB at the center frequency of 5GHz, while the gain of the FA is 3.27dB under the same configuration. In S 11 curve, both FA and filter are third-order curves. It can be seen that FA has a gain, and the filter will inevitably lose part of its energy during transmission. By comparing the model diagrams of the two devices simultaneously, it can be seen intuitively that the filter size is slightly larger than that of FA in similar indicators.
Since the massive demand for wireless communication and information service, the explosive growth of corresponding wireless communication systems leads to the increasing demand for compact, low cost, and low power wireless devices. Among them, many compact device designs based on wireless communication system integration modules have appeared. The Web of Science includes more than 10,000 studies on the design of compact devices for wireless communications, so compact is of great importance to integrated circuits. In [12] and [13], the authors designed the embedded filters, the broadband transmitter is designed in [14], and the multi-band antennas are devised in [15] and [16]. They are located in the RF front-end module of wireless communication, which can include integration with other circuit devices expediently. The authors of [17] achieved high performance and compact passive components through the multi-layer ceramic-based thick-film technique. The compact structure of the above devices shows that compactness is an essential characteristic of each independent device in the integration of wireless communication system modules.
In 1929, Plebanski proposed the FA circuit for the first time on the theoretical level [18]. Since then, the research on FAs has been extended to practical circuits. In 1977, Ren et al. combined a horn antenna with a filter to form the FA [19]. In 1993, Sanchez-Hernandez et al. added the filter to the edge of a patch antenna to achieve dual-frequency [20].
In addition, Abbaspourt-Tamijani et al. integrated a patch with a coplanar waveguide transmission line to form a FA. In 2002, a last-stage resonator was replaced with a resonant antenna [21]. In 2003, Queudet et al. proposed an idea of leakage waveguides as radiation structures and obtained a preliminary structure of loss-coupling between resonators [22]. The SIW FA are commonly used in the FA, the metal waveguide FA, the microstrip FA. Generally, the order of metal waveguide FA can be increased by repeating elements to form the periodic structure. There are two main methods to integrate the waveguide filter with the antenna structure, one is etching the radiation slot on the top of the waveguide and the other is opening the last stage of the cascade resonator. Comparing with other filters, it has better filtering characteristics, and it is easier to form multi-pass bands. However, as each resonant cavity cascade usually forms a metal waveguide FA, it usually has a lamination for its size and gain in the design [23], [24], [25], [26], [27], [28], [29], [30]. The microstrip FA utilizes the patch antenna as the last stage of the resonator, or etches a specially shaped slot on the top to diffuse the energy outward. Its advantage lies in its small size and simple structure. However, the filtering band of the microstrip FA is limited, the filter order is low, and the S-curve is unsatisfactory in the band. Besides, the design is too complicated, and there are too many parameters in each microstrip line [31], [32], [33], [34], [35], [36].
The authors of [37] and [38] analyzed the normative parameter of the SIW structure in leakage characteristics and wave guidance. As shown in Fig. 1, the bandgap and leakage loss may be generated because the SIW structure is flanked by the metallized via connecting the upper and lower metal plates. In order to reduce these losses, the following design rules need to be considered [37].
where s is the gap between two adjacent vias, d is the diameter of the via, λ c is the cut-off wavelength, α 1 is the total loss, and k 0 is the wave number in vacuum. Equation (1a) suggests that the gap between two adjacent (s) must be larger than the diameter (d) so that the circuit is physically realizable. Equation (1b) is required to avoid the bandgap loss in the working bandwidth. The leakage energy can be ignored by the leakage condition Equation (1c). The final Equation (1d) is not necessary but ideal, specifying that it is best to have no more than 20 vias in a wavelength. In Fig. 4, the shaded areas are functions of s/λ c and d/λ c in the SIW structure with almost no energy leakage. SIW has the similar field and dispersion characteristics to traditional waveguides when the metallized vias are closely arranged without energy leakage. However, due to the gap in the SIW side wall, a stable current cannot be formed. It only can transmit the TE mode. The authors of [38] inferred the empirical relationship between the diameter of the metallized via, SIW's dimensions, and effective waveguide width (w eff ) in Equation. (2).
where (w eff ) is the effective width in the waveguide, w is the width of SIW, d is diameter of metallized via, and s is the gap between two adjacent vias. Furthermore, after further considering the influence of d/w, a more accurate Equation.
When s/d is smaller than 3 and d/w is smaller than 0.2, the empirical equation is very accurate. For these defects of the metal waveguide FA and the microstrip FA, the SIW FA can effectively solve these problems. Firstly, for studying the miniaturization of the SIW FA, we compare several methods of the miniaturization of the SIW filter. In the process of the SIW filter miniaturization, the SIW is firstly folded to form a multi-layer structure [39], [40], [41]. Then, a particular structure is etched on the top metal layer to form inductive and capacitive loading. The fundamental mode resonant frequency moves to the lowfrequency direction, and the physical size of the filter can also be reduced [42], [43], [44], [45], [46], [47], [48], [49]. Next, the slow-wave technology separates the electric and magnetic fields through the perturbation of the internal metallic vias to form the slow-wave effect, which causes miniaturization [50], [51], [52], [53], [54]. Finally, for the incomplete mode, by cutting a complete mode of SIW, the size can be significantly reduced with the field of unchanged fundamental mode [55], [56], [57], [58], [59], [60], [61], [62], [63], [64].
The SIW FA makes the energy diffuse outwards by adding a special structure on the basis of the original SIW filter.  antenna, realizing the integration of filtering and radiation, and further reducing the volume and cost of the military platform communication system [65]. To further simplify the design and improve the return loss in the working band, Yu et al. proposed a planar FA based on SIW in 2011. The yagi antenna is used in this design, and fifth-order FA is obtained by combining the antenna with the SIW filter [66]. To further reduce the size, Yu et al. in 2012 put forward an inductive window bandpass filter and coaxial collinear (COCO) antenna in Ku-band [67]. Also, in 2013, Xue et al. proposed a kind of SIW filter and Angle of printed dipole antenna reflector of filtering which can be applied to Q-band indoor short-range wireless communication [68]. Nevertheless, its radiation frequency and gain are still low. In order to overcome the narrow-band characteristics of the SIW FA, Hu et al. used two SIW resonant cavities to overlay vertically in 2018. They etched the resonant slot on the top to approach the patch antenna to obtain the third-order FA with high cross-polarization [69].
In the following sections, the miniaturization of the SIW FA is introduced in terms of the mode and physical structure.

II. TECHNICAL REVIEW
A. SIW FA BASED ON MULTI-LAYER SIW FA based on the multi-layer structure is a common method for miniaturization. Cavities are stacked vertically to reduce the size. The multi-layer structure is easy to meet the physical structure to achieve bypass coupling or source-toload coupling to increase the radiation null (RN), which will improve the performance of the FA. Compared with other RF devices, since the antenna has only one port, in the manufacture of the multi-layer, there is no need to consider the input and output ports on different layers, which is conducive to integration.
In [70], a third-order FA is firstly realized through three layers. The exploded view of this SIW FA is illustrated in Fig. 5(a). In [71], a novel structure is proposed to realize the highly efficient FA, in which the compensation via in the top cavity is used to adjust the resonant frequency. Using the time-domain technique, the peak value of the time-domain response corresponds to the internal coupling between resonators and the external coupling between resonators and ports. By observing the peak point of the time-domain response, the parameters in the FA are swept to get the optimal size. As shown in Fig. 5(b), the slot antenna can be formed by etching a slot on the top metal layer. In addition, the authors proposed the equivalent circuit of this FA and gave the calculation relations of coupling coefficient, S 11 , quality factor (Q), and other factors.
The above studies used simple topologies with no RNs. In order to improve the frequency selectivity, in [72], because of the bypass coupling, a double-coupling scheme was adopted to achieve a FA with four RNs, as shown in Fig. 6 As shown in Fig. 6(a), FA miniaturization is realized by double-layer substrate. The dual-slot in the middle metal layer can control the frequency of the RNs respectively. This FA has high frequency selectivity and a wider working bandwidth. However, in the middle of the structure, two RNs must be controlled together by the distance between two slots, and cannot be controlled separately.
The multi-layer technology can make the FA more compact, and hence it can be designed to transmit different information on multiple transmission channels by combining it with the patch antenna. In [73], the authors designed a FA with a four-channel multiplexing operation with two dielectric substrates and a thin foam between them. As shown in Fig. 7(a), a U-shaped slot is etched on the top metal layer, and the four frequencies are divided into high and low frequency bands. The desired four resonant frequencies are generated by adjusting the space between two substrates. Since the SIW cavity is fed directly from the microstrip line, the isolation between ports will deteriorate. Therefore, in order to increase the isolation between several channels, the SIW cavity at the bottom is coupled to the patch through the U-shaped slot to form a filter circuit. However, the radiation efficiency is lower than a conventional antenna. Besides, at high frequencies, as shown in Fig. 7(b), the poles gradually change from three to two.
The multi-channel technology is completed in a cavity as a whole, while the antenna array requires several elements to be reasonably tiled. Through increasing the radiation cavity, we can achieve a high gain of the antenna array. In Fig. 8(a), the authors of [74] etched a slot on the top metal layer of each FA. To keep slot antennas in the same layer to form the array, a multi-layer technique is adopted. The lower SIW cavity is coupled with the upper layer to form a third-order array FA. Nevertheless, the bottom cavity wastes too much space, and the top cavity, which is a cascading form also further increases the overall size. As shown in Fig. 8(b), the authors of [75] constructed a fourth order FA with four layers to further increase the poles of the filter. Furthermore, they etched a horizontal and vertical slot to replace the traditional long slot to realize a dual-polarized antenna.
Furthermore, the authors of [76] combined the multi-layer technology with patch antennas. The FA is formed by two layers of the stacked substrate and a top layer of a right-handed circularly polarized (RHCP) patch antenna. Two cavities are coupled through the coupling slot, and upper cavity is coupled to the patch through the coupling via. Then, the single antenna elements are combined into a 2×2 antenna array, and the patches on the upper layer rotate 90 • counterclockwise, respectively. Therefore, this structure can effectively reduce the thickness of the substrate. However, there is no RN in the structure, so the frequency selectivity is mediocre.
This part mainly introduces SIW FAs by the multi-layer technology. Unlike other microwave devices, the FA formed by the integration of the filter and antenna only has one input port, which is more suitable for using multi-layer technology.

B. MULTI-MODE SIW CAVITY FA
Using multi-mode cavities is an excellent strategy to design SIW FAs with compact size. Because SIW FAs which are based on multi-mode cavities usually have smaller sizes than those based on several single-mode cavities. In addition, since the circular polarization (CP) antenna requires two orthogonal modes, it can be formed by the dual-mode cavity.
The horn FA is formed through the particular outer edge shape of the SIW structure. The authors of [77] used TE 102 and TE 201 to achieve two dual-mode cavities in the horn FA. The transmission null (TN) on either side of the working band is realized by the cross-coupling between TE 101 and TE 102 modes. However, the TNs cannot be controlled by a single variable. In Fig. 9(a), when a 1 < w 1 , the filter power divider cavity can be designed as the third resonator, and the last pole is made up of the horn. Adding an improved transition at the end of the structure (the right side of Fig. 9(a)) improve matching, radiation performance and broaden the bandwidth.
In [78], it is also a SIW horn FA, which adopts a conical transition structure. In this horn FA, four rows of inductive metallic vias are embedded into the cavity of the horn antenna thus forming three coupled resonant cavities to achieve the function of the filter. The electric field diagram of each cavity is shown in Fig. 9(b). Besides, the resonant frequency of each cavity is adjusted by changing the distance of metallic via holes. Compared with the traditional H-plane horn antenna, the bandwidth of the proposed antenna is significantly increased. However, the overall dimension is still large, and the design is too complex.
Etching a slot at the maximum electric field of a mode to disturb the resonant frequency in that mode. Therefore, by using this principle, two modes with different resonant  frequencies can coexist in a cavity simultaneously, and the dual-mode cavity technology can be realized. The authors of [79] achieved the CP antenna with TE 102 /TE 201 and TE 103 /TE 201 dual-mode cavities. Similarly, in [80], a dualmode FA is implemented on the single-layer SIW and a single cavity without apparent and additional filter structure. In Fig. 10, the two poles are a pair of split modes generated by degeneracy modes of TE 102 .
In another approach, the multi-layer and multi-mode technology are well combined to reduce the size of the FA. In [81], the authors designed a three-order FA which is composed of two cavities. As shown in Fig. 11, two slots were etched on Ground 2 to form two modes. The electric field of the quasi-cavity mode distributes in the whole SIC and the electric field of the quasi-dual-slot mode is more concentrated near the dual slots. These two modes can change the frequency of the two RNs by controlling the distance between the two slots, forming the upper and lower nulls. Meanwhile, the double slots on the top metal layer radiate energy outwards, while the TE 110 mode inside the cavity alone cannot radiate energy. Therefore, the flexibility of this design can be improved. Although the resonant frequency of the two modes can be changed by controlling the double slot, it still needs to cooperate with multiple variables, and there is no independent control.  Additionally, to etch the slot, some special structures can be added to the maximum electric field, reducing or increasing the resonant frequency to design the dual-mode cavity. In [82], the authors designed a dual-mode SIW cavity based on TE 101 and TE 201 modes. The designs achieved controllable RN located on the upper working band, and its topology is shown in Fig. 12(b). Besides, the geometry of the FA is shown in Fig. 12(a). The mushroom structures are etched on the middle metal layer where the maximum electric field of TE 201 . Therefore, the resonant frequency of TE 201 mode can decrease faster than TE 101 mode, and two different modes can coexist in the same cavity. Miniaturization in the conventional SIW cavity resonator can be achieved through these structures. The FA based on a dual-mode cavity and two half-mode cavities is achieved by a slot on the bottom layer. The lower out-of-band RN can be controlled by the distance of the slot from the center. The authors of [83] designed a dual-band SIW cavity with TE 210 /TE 110 and TE 120 /TE 210 dual-mode, whose topologies of dual-band are shown in Fig. 13(c) and (d). Two rows of metallic vias arrays are located at the maximum electric field of TE 110 mode. Hence, the interference to TE 110 mode is more significant, resulting in the appearance of a dual-mode cavity. The slot next to the array cuts off the current of TE 110 and TE 210 modes, creating a potential difference on both sides, which can be used as an external radiation antenna. Fig. 13(a) shows that two dual-mode cavities have similar dimensions. Four special structures named metal-insulatormetal (MIM) capacitors are located at the maximum of each electric field. They can be used to disturb the frequencies of higher-order modes. The dual-mode of TE 210 /TE 102 has a similar size to the other. The MIM structure reduces the size by 48%. This structure can produce three RNs. The second and third RNs can be moved by adjusting the feeding position. Therefore, it can improve the out-of-band suppression. The size of the proposed FA is further reduced (60.39% miniaturization) by these structures. And the two working bands as shown in Fig. 13(b).
A center-fed SIW cavity-backed slot antenna can support a resonant with conical radiation patterns, but its bandwidth is too narrow. In order to increase the bandwidth, the authors of [84] investigated SIW cavity-backed slot and patch antennas, as shown in Fig. 14. And compared with patch antennas, the cavity-backed patch antenna has higher gain. The mode of this design includes the patch TM 01 mode, the cavity TM 01 mode, and the cavity TM 02 mode. Through combining these three modes, a triple-resonance FA with wide bandwidth can be obtained. When the substrate thickness is large enough, the mode of the cavity can be excited effectively to form a wide bandwidth. However, the frequency selectivity of this structure becomes worse, and two kinds of TM 01 modes are hard to be distinguished in the curve of return loss.  Through changing the space of traditional SIW vias and enlarging the size of the top metal layer by folded configuration, the authors of [85] constructed a 1/4 wavelength patch on the top metal layer with a SIW dual-mode (TE 102 /TE 201 ) cavity to achieve a third-order FA. By using the folded configuration, the antenna does not occupy additional ground dimensions. The size of the proposed FA is 69.08% smaller than other common FA. There are two parasitic modes out of the working band, one is a lower TE 101 mode which is suppressed by the shorting vias, and the other is the higher order mode at a higher frequency of the working band. When the space of sparse vias becomes larger, the Q of the SIW cavity decreases. By adjusting the position of the center asymmetrical shorting pins, the shorting pins were equivalent to a load inductance, which can improve impedance matching and axial ratio (AR) bandwidth.
The author of [86] divided the complete mode into multimode cavities by loading shorted-end vias into a single cavity. By etching slots on the metal layer, the third-order FA consists of half-TE 110 , odd TE 210 and even TE 210 mode. The half-TE 110 , odd TE 210 , even TE 210 and TE 310 mode compose the fourth-order FA. The structures of the two designs are shown in Fig. 15(a) and (b). However, this FA has no RN. Hence, although the bandwidth is high, the frequency selectivity becomes worse.
In summary, in most cases, multi-mode technology combined with multi-layer technology can effectively optimize the performance of the SIW FA on the original basis. As shown in [87], combining the two technologies allows a new topological structure to obtain more required nulls and increases the out-of-band suppression. A similar approach can also be applied to SIW FAs. The single-layer structure combined with multi-mode technology can produce multiple poles in a cavity, making the TN narrow. In the construction of a dual-mode cavity, in many cases, a special structure is loaded at the maximum electric field of one mode so that the structure can disturb the electric field, and the dual-mode cavity can be constructed to a large extent. This method has been applied in [88], and the result becomes ideal, even without changing the original physical size.

C. SIW CAVITY FA WITH INCOMPLETE MODE CAVITY
In general, an incomplete cavity can significantly reduce the circuit size. The circuit size only accounts for half of the complete mode by using the half mode technology. Furthermore, the quarter mode can save three-quarters of the space. The complete mode cavity can be divided into several incomplete cavities by etching slots on the metal layer at the top or bottom. The FA can also be miniaturized by using the incomplete mode. Furthermore, in many cases, etching slots introduce RNs thus further improving frequency selectivity. In addition, when the position of the etching slot of the partially constructed incomplete mode changes within a specific range, the original generated RN can be controllable, increasing the flexibility of the design requirements for the result.
In those mentioned above, the authors of [79] improved the original TE 102 /TE 201 dual-mode cavity of the SIW FA into a half-mode SIW (HMSIW) with an incomplete mode, further reducing the circuit size. The original rectangular slot is improved into an elliptical slot to allow the current to flow more smoothly along the edges, thus improving efficiency. This structure can reduce the SIW dimensions by 62.64% of the conventional SIW structures.
In [89], a third-order filter is designed by four metalized posts in the middle to divide the TE 120 mode resonators into two TE 110 modes. Besides, as shown in Fig. 16(a), the slot is etched on one side to split one of the TE 110 modes into two half TE 110 mode resonators. By adjusting y s , as shown in Fig. 16(b), the skirt selectivity can be enhanced by moving the RN close to the working band. Moreover, RN can be moved from one side of the working band to the other by changing the slot position.
The incomplete mode SIW cavity and multi-layer SIW can combine in [90]. The topology of the structure is shown in Fig. 17(c). The coupling matrix represents the filtering network, and the coupling matrix is normalized by the coupling coefficient, external Q (Q e ), and other characteristic values. The relationship between the coupling coefficients and the structural parameters of the resonators is studied. According to the theoretical and numerical analysis, the expected double-passband filter response can be obtained using an appropriate physical structure. When port 1 is excited, the  feed cavity TE 120 and the coupling post 1 are used to couple the lower half cavity of the upper substrate to form an upper working band second-order filter. When port 2 is excited, coupling post 2 and the feed cavity TE 210 coupling the upper half cavity of the top substrate form a lower working band. The electric field are shown in Fig. 17(a) and (b). Two orthogonal modes were used to realize the common feeding cavity in the design. The common radiating cavity is converted into two half-mode cavities by etching rectangular slots. At the same frequency, only half of the radiating cavities in the upper cavity are effective radiators. To some extent, this also led to the waste of space.
Based on the above FA, to further reduce the size, the authors of [91] proposed a second-order duplex FA with a single layer made by using an incomplete mode. A wider vertical slot and a thinner horizontal slot are etched in the center of the top metal layer. The wide slot divides the single band into high and low frequency bands, as shown in Fig. 18(a), and the thin slot improves HMSIW into a dual-mode cavity. Moreover, controlling the position of the loaded-vias loaded near each slot can enhance the selectivity of the working band and strengthen the port isolation. However, the poles in each band cannot be controlled individually, and the resulting two bands can vary only in small ranges. The S-Parameter of the FA is shown in Fig. 18(b). This structure can greatly reduce the circuit size by 62.08%.
In addition to split a cavity into two parts, it can be divided into several parts. This enables the filter to generate more poles, thus achieving optimal performance. The authors of [92] split the single-cavity into three parts, as shown in Fig. 19(a). At different frequencies, the electric field is concentrated in different sub-cavities, and the magnitude in other sub-cavities is negligible. These two slots combined with the capacitively coupled feeding disc excite four modes, three of which increase the impedance bandwidth. The fundamental  mode is the no radiating mode to achieve a RN and produce high stopband suppression. Therefore, the energy of three frequencies in the cavity is excited, thus forming a thirdorder filter. The RN can be easily adjusted by controlling the feeding position, thus producing a high radiation suppression in the stopband. However, this controllable RN only appears above the working band, and there is a certain limit to the frequency selectivity.
Alternatively, two long parallel slots can be etched on the metal layer. In [93], two slots are etched on the top and a longer slot on the bottom metal layer, the resonant frequencies of the two original modes are significantly reduced. The electric fields at the frequencies of the three poles are shown in Fig. 20. This structure produces an upper RN and a lower RN controlled by the position of the long slot, which results in a quasi-elliptic filtering response. In addition, to obtain good radiation patterns, there are slight overlaps between the two top slots and the bottom slot. However, both RNs are controlled by the same variable and cannot be changed independently.
In [94], etching a long slot on the middle metal layer will divide a cavity into two half-modes and make the middle cavity achieve a dual-mode resonator, as shown in Fig. 21. A short slot on the side produces two controllable TNs, and the l 7 and W 4 control lower and upper RNs, respectively. A patch antenna is added to the top layer to form TM 010 as the third resonant mode. And shorting pins is added to the upper cavity to improve the gain and reduce the H-plane cross-pol. The device size is reduced by about 65.53% compared to conventional devices through the dual half-mode cavity.
The authors of [95] firstly designed a cascade of secondorder filter and etched a slot on the center of the metal layer in the second cavity. When the third-order FA is formed, the  length of the slot is close to the width of the cavity, the second cavity becomes a dual-mode cavity. A capacitor ring is placed in the first cavity to reduce the electrical size of the cavity but slightly reduce the unloaded Q (Q u ) of the cavity. The RN is generated by the cross coupling of the two half-modes through the topology shown in Fig. 22. The frequency of the RN can be controlled by changing the position of the slot.
Two RNs can be generated when the patch is placed side by side with the SIW cavity. In order to reduce the overall size, the authors of [96] tried to change the original SIW into HMSIW, which is only half of the original size and has the same resonance frequency. The top view of the antenna is shown in Fig. 23. The parasitic patch was then added to the side of the half-mode cavity to create a coupling between the patch and the cavity, resulting in a second-order FA. The gap between the half-mode cavity and the parasitic patch and the distance of shorting pins on the patch to the center control the upper and lower RNs, respectively. In the working band, the TM 01 mode of the patch is excited so sufficient that the patch can radiate energy outward.
The aforementioned incomplete mode technique shows that the incomplete mode miniaturizes the SIW FA to a large extent. Since a common method for the SIW FA to radiate outward is to etch the radiation slot on the outer metal layer, a half-mode cavity whose size is almost half of the complete cavity can be realized using the etching slot by optimizing and modifying the size and position of the radiation slot. Combined with other special structures, RNs and poles are added to optimize the SIW FA. However, half-mode cavities can decrease the Q u , resulting in lower frequency selectivity and an increase in overall circuit losses.

D. HYBRID SIW FA
By etching planar resonant structure such as Grounded Coplanar waveguide, Complementary Split-Ring Resonator on the metal layer at the top or bottom of the FAs, extra poles can be formed by those planar resonant structures to increase frequency selectivity without changing the size of the original circuit. As a special structure, U-slot takes up very little space but provides additional RNs or poles. By generating an electric field in the U-shaped slot, the current on both sides of the slot is equal magnitude and in opposite so that the RN can be generated. In addition, a new mode is applied to the FA by changing the shape of the SIW resonator.
The authors of [97] combined one circular patch resonator, one square SIW cavity, and nine metallic pins to form a FA with three poles, as shown in Fig. 24(b). Since the electric field distribution of TM 11 mode is similar to the central field of TE 21 mode, the bandwidth can be expanded. As the TE 11 mode of the rectangular cavity is centrosymmetric, a lower RN can be generated. As shown in Fig. 24(a), the RN in the upper stopband is due to the topology of the antenna. By adding metal probes to the centerline, the lower RN generated by the TE 11 mode can be moved upwards and minimize the effect on TE 21 . Due to the asymmetry of the antenna, the lower stopband becomes worse, and hence two pairs of metal probes have to be shifted.
A three-pole FA is proposed, and the three poles are respectively provided by the half-mode resonator and the U-shaped slot on the bottom surface of the cavity. The authors of [98] VOLUME 10, 2022  implemented a third-order FA with two U-shaped slots and one long slot. As shown in Fig. 25(a) and (b), the long slot caused a single cavity to produce a dual-mode cavity with an out-of-band RN. U-shaped slot 1 caused the FA to generate a below RN to increase frequency selectivity. U-shaped slot 2 made the right half cavity excite the half-TE 110 mode. The electric field of this mode is compressed in the cavity to generate resonant frequency points at higher frequencies, thus causing three poles to increase the bandwidth. The upper and lower RNs are controlled by g and S 2 , respectively. However, as shown in Fig. 25(c), these two RNs are not controllable. These nulls change as the overall bandwidth changes. Overall, U-slot can reduce the FA by 81.4% compared with the control group in this paper.
In addition to combining two U-slots, one U-slot can also be modified to produce a special structure. In [99], The U-slot makes the SIW cavity form the electric and magnetic mixed coupling. Loading electric and magnetic mixed coupling structures weakens the strength of TE 120 , which leads to unequal magnitude of TE 120 on the two sides of the slot. The fundamental and TE 120 modes interact, and their magnetic  fields cancel each other. The magnetic field in the left slot weakens, producing an electric and magnetic mixed mode, as shown in Fig. 26. The coupling structure is not in contact with the cavity and can form electrically coupling. The electric and magnetic mixed coupling structure is shared with the cavity and can form magnetic coupling, generating the upper RN. The current on the both sides of the slot is equal magnitude and in opposite. Therefore, there is no transverse radiation. The extended length below the U-shaped slot controls the lower RN. However, the lower RN change is based on the bandwidth and not controllable in the practical sense.
In addition to the overall rectangular SIW cavity, it can also be directly designed to be circular. The authors of [100] composed a FA of a monopole and two stacked annular waveguide cavities can reduce the dimensions by 32.38%, as shown in Fig. 27, to achieve a third-order FA. The upper RN results from cross-coupling the source and the non-adjacent resonator. By changing r 4 and t 1 , the frequency of the RN can be controlled.
In summary, some SIW FAs with the hybrid structure are introduced. On the one hand, the order of the FA is increased, or the bandwidth is flexible by adding a particular structure. On the other hand, by changing the external shape of the whole structure, a new resonance mode is introduced to construct a SIW FA which is different from the traditional rectangular shape.

E. SLOW-WAVE FA
In addition to the slow-wave technology of loading internal metallic vias mentioned above, surface plasmon polaritons (SPPs) structure can also be loaded to achieve the slow-wave technology [101], [102], [103], [104], [105], [106], [107]. In [108], the authors proposed a new type of SIW-SPP composite guided wave transmission line and used this transmission line to construct a bandpass filter. Combining the SIW with effective surface plasmon polaritons (ESPPs) improves bandwidth flexibility and provides a stable stopband. ESPPs can propagate along the interface of the effective epsilon-negative and the epsilon-positive structures. This structure has a certain foundation in the SIW filter. In [109], the authors tried to propagate in SIW of different dielectrics. In [110], a dual-band filter based on two different ESPPs modes was constructed, but the bandwidth was not controlled by independent parameters.
As shown in Fig. 28, in [111], the structural length L SPP of ESPPs mainly affects the out-of-band suppression of the filter, and the bandgap between the ESPPs causes the stopband performance of the filter. The width of SIW and the filling dielectric can control the start frequency of the working band. These can achieve the purpose of controllable bandwidth. Therefore, the authors design a filter based on ESPPs induced by structural dispersion and combine it with a yagi antenna to construct a FA with controllable bandwidth and a stable stopband. Therefore, the design of this structure can be used to fabricate UWB FAs.
To sum up, comparison of the different miniaturization techniques in SIW FAs is presented in Table. 1.

III. APPLICATION
With the technology matures, SIW FAs have been gradually transferred from academia and research institutions to industrial R&D departments and commercial suppliers. From the beginning of the SIW technology development to the present, SIW FA is mainly used in 5G (5th generation mobile communication technology) and wireless local area network (WLAN) communication, as well as the satellite radar. Because of the advantages of high gain and wide bandwidth, the array antenna is an ideal choice for the above practical applications.
In the early stage, a X-band SIW array FA for radar was firstly studied in [112], and then the application of SIW FA array in automotive radar was studied. As shown in Fig. 29, there is the application scope of the vehicle medium-range radar (MRR) and long-range radar (LRR). The array in [113] used a 1 × 16 SIW slot as a unit. Due to the bandwidth requirements of LRR, the array is composed of six identical SIW arrays, and the antenna can support MRR and LRR at the same time. The electromagnetic energy is fed to the 2 × 4 array using the metallic blind holes and metallic vias from the power divider in [114]. Meanwhile, changing the shape and position of the metal layer slot at the top and combing the eight-channel SIW. Then, the wideband millimeterwave SIW cavity-backed slot array antenna can be obtained which is suitable for 79 GHz automotive radar.
Due to its relatively small size, SIW array FAs has been widely used in ground communications such as 5G and WLAN. As 5G is applied in the 28 GHz band, the path loss is severe. Hence, directional antennas or phased arrays need to be designed for achieving the desired SNR. Generally, the array configuration for 5G applications mainly includes a linear array and a rectangular array. However, this array has a fixed asymmetry and the main lobe of the rectangular array radiogram has again fluctuation of about 35 dB under different azimuth cutting. Therefore, the authors of [115] used the circular structure, and the circular array has a symmetric structure, to achieve stable and high gain. In wireless communication systems, radial line slot array (RLSA) antenna has become a popular antenna suitable for wireless communication systems because of its flat and low-profile characteristics. In [116], the disk structure is also employed to design an RLSA antenna for wireless communication through the SIW structure.
Furthermore, for the multiple input multiple output (MIMO) system, a rectangular SIW array antenna is proposed to avoid mutual coupling between antennas [117]. Similarly, using the advantages of the SIW antenna, [118] designed a slotted SIW array antenna which is suitable for WLAN. The antenna surface adopts several longitudinal slot array structures. By optimizing the number of branches in an array and the number of slots in one SIW branch, the gain can meet the desired requirements.

IV. CONCLUSION
In this paper, the latest development and miniaturization techniques of SIW FA are reviewed in detail. The advantages of integrated FA compared with discrete filter and antenna are introduced. By combining with many different SIW FA examples, different methods of reducing circuit size are explained. The results show that the miniaturization technology can be realized by using the multi-layer, multi-mode, incomplete mode and slow-wave technology, as well as the summarized hybrid technology (sub-mode and other miniaturization techniques).
So far, array antennas have been used to miniaturize SIW FA in practical applications, while other SIW FA miniaturization methods are still in the theoretical research stage. In order to achieve the miniaturization of the SIW FA, dielectric substrates with high permittivity can be selected. In addition, the FA can be further miniaturized by creating more new topologies and combining known topologies into multimode structures. Similar to [55], the multi-mode technology and slow-wave technology have been relatively mature in SIW filters, and the SIW FA can also refer to the miniaturization methods which ate commonly used in filters. Nevertheless, miniaturization inevitably reduces the Q and IL. In most cases, materials, structures, and patterns should be balanced so that the final structure can be as small as possible.
Processing costs and other factors have no significant effect on the advantages of the SIW structure.
The FA and its miniaturization technology can meet the requirements of current and future wireless communication systems. Compared with the traditional combination of filter and antenna, the FA in the communication system is more efficient in terms of the size, cost, and loss of the whole subsystem. We believe that tunability and reconfigurability will play an essential role in achieving the flexibility of SIW FA structures. They are typical of particular interest to the multi-band and multi-mode FA. The SIW can easily accommodate tuning devices or materials through fieldcircuit interactions in the individual electric and magnetic tuning/switching. Combining the multi-layer technology provides more space for tuning devices or materials to accommodate and drive. Therefore, the constructed FA tends to be idealized in a certain range by controlling the tuning devices or materials.
In addition, it can be known from [119] and [120] that the conductor and medium losses can be enhanced by using the complementary split-ring resonator (CSRR), and the energy can be fully absorbed by using the inherent loss of materials, so as to realize the common-mode absorption of broadband. In this way, the reflection of energy is suppressed. Therefore, CSRR combined with SIW structure and artificial microstructure design is expected to realize the absorption filter. According to the method mentioned above, the SIW FA can be further designed to meet the design conditions.