Near-field and far-field investigation of miniaturized UWB antenna for imaging of wood

Nowadays, many types of microwave sensors exist such as special transmission sensors, guided wave transmission sensors, free-space transmission sensors, time domain reflectometry (TDR) and tomographic sensors. In some industrial applications, for instance strength grading and drying, it is critical to investigate the moisture content, density or even check if there are defects (hollow) in wood. The dielectric constant and conductivity of materials were studied in many applications, especially when microwaves (MW) were employed and used for biological tissues characterization^1–4 (characteristics of the biomedical tissues presented in order to demonstrate the methods applied for dielectric properties previously).

Microwaves applied as a non-destructive method to detect defects in wood at one frequency.⁵ It was clearly illustrated that wood’s electrical features like the dielectric constant were highly affected by parameters such as frequency, temperature, moisture content and the density at one frequency.^6,7 presented the dielectric properties of wood and oil palm trunk in different features at both THz and MW frequency range respectively. For instances, in the THz the dielectric measurement was performed in parallel and perpendicular polarization to the fibre direction of wood. A weak decrease in real part of the permittivity, and a weak increase for imaginary part of permittivity was noticed when it is parallel. Moreover, for perpendicular direction the real part reduced till 250 GHz and then had a slight increase until 500 GHz. In addition to that, dielectric properties of oil palm trunk (OPT) core were investigated using an open-ended coaxial probe method. The dielectric properties of OPT core were determined as a function of controlling factors such as microwave frequency, biomass moisture content and fibre directions (radial, tangential and cross section). The measurement performed in the frequency range of 0.5- 3.5 GHz. When frequency was increased from 1 to 3GHz, dielectric constant was decreased for all the fibre directions. The result depicted that electric field of microwave affected the interaction of OPT core biomass with electromagnetic waves⁸ (Indeed the frequency affects dielectric properties of the materials such as OPT as presented in this reference). The trend of decreasing dielectric constant with the increase of frequency was found to be similar with other biomass tested previously such as rubber wood, softwood included Black spruce, Balsam fir and Tamarak, empty fruit bunch of oil palm, oil palm shell, oil palm fibre, biochar from oil palm shell, green pea flour, lentil flour and soybean flour.⁸ Furthermore, since these waves were non-destructive, among all these applications medical imaging (MI)⁹ and MW tomography (MWT)¹⁰ were mostly used. Figure 1 shows the measured dielectric properties of the applied samples (high density wood and plywood) using the Keysight 85070D dielectric probe. After investigation, the same trend as the above papers for dielectric properties in terms of frequency obtained. Figure 2 shows the penetration depth of the electromagnetic wave at the wavelength of the incident field in meter. Its illustrated that the high-density wood shows better penetration since its conductivity was less than the ply wood. Besides, penetration depth follows the same trends for both samples as if it decreases with frequency. In addition to that, Figure 1 and Figure 2 are measured using the Keysight 85070D dielectric probe. For the measurement both samples are dried before measurement. The plywood and high density wood have the weight/moisture content of 100g and 150 g/5 % and 5% respectively. The samples were taken from the mechanic department of Universiti Teknologi Petronas (UTP).

Microwave tomography (MWT) uses tomographic sensors (like antennas in this paper to send and receive the signals) and electromagnetic principles in imaging (to investigate the electric and magnetic fields variation around the sample) and it showed more advantages in comparison with other methods like Electrical Capacitance Volume Tomography (ECVT) (MW imaging showed more promising performances and advantages in comparison with the ECVT and the other techniques presented in Table I in supplementary material orting information. According to the references presented in this table, the MW imaging technique has better outcomes in terms of the speed and none invasive effects on body as compared to others presented).¹¹ 

A microwave tomography system can have three major parts, called the sensing system (transmitting and receiving antennas), interfacing (signal conditioning) and image reconstruction algorithm (an algorithm to construct the image).¹² 

In addition, MW tomography was applied in many applications such as medical imaging in breast cancer imaging,¹³ geographical prospecting like finding mines,¹⁴ agriculture like the imaging of defects in wood¹⁵ and wood characterization.¹⁶ The geophysical imaging of root zone, trunk and moisture heterogeneity was performed in Ref. 17. Another paper worked on the MW tomography to inspect the wood.¹⁸ Imaging of wood not only used for MW region but also for higher range like THz.¹⁹ In addition to that, MW imaging of wood materials helped to inspect the damage caused by earthquake.²⁰ Besides, the antennas presented in Refs. 21 and 22 obtained narrow BW, and their antenna dimensions were larger in comparison with the antenna presented in this paper.

As aforementioned, one part of MW tomography system is the transmitter and receiver which are antennas. Since having a wide BW for MW tomography can be helpful for artifact removal and better resolution,²⁶ UWB antennas are useful choice. Based on the Federal Communication Commission (FCC), UWB antennas are licensed to use a frequency range from 3.1 GHz to 10.6 GHz, whilst in some application’s higher frequency bands (mm-waves/THz) were exploited.²³ Many types of UWB antennas exist with different shapes and characteristics such as planar UWB antennas, printed (2D) and over metal plate UWB antennas, bowl shape, leaf shape, U-shape antennas and C-shape antennas.²⁴ For instance, elliptical patches permit the operational bandwidth to span over a UWB range. In some cases, the radiating patch was optimized to improve the Radar Cross Section (RCS), while two slots were cut from the patch to obtain a good radiation performance and enhance impedance matching of the designed antenna. One of the benefits of an UWB antenna is to have broad BW, which assists in giving a high resolution for detecting small differences in the medium under test or even detecting a small movement of the chest when we breathe.²⁵ Furthermore, for MW imaging (MWI) applications, when the working BW is wide enough, high resolution in image reconstruction and clutter removal are achievable. Besides, to decrease the effect of clutter on the image and reduce the number of imaging artefacts, having more antennas to receive more scattered signals from healthy tissue and unhealthy one can be helpful; thus, the antenna dimension better to be small.²⁶ MWT is beneficial in comparison with other methods in terms of having high-resolution, not harmful for the human body and giving three-dimensional (3D) images of data compared to other methods.²⁷ applied two antennas, one circular-edge antipodal Vivaldi antenna and one corrugated balanced antipodal Vivaldi antenna. Afterward, the s-parameters, radiation characteristics of antenna like gain, transmitted and received signals of arrays antenna were investigated in the range of 3.1-10.6 GHz. The results showed improvement in feasibility of UWB imaging of the wood slice.²⁰ Presented a polarimetric radar system for sensing the concealed wood-frames damaged by earthquakes. This system employed an antenna array consisting of four linearly polarized Vivaldi antennas recording full-polarimetric radar echoes in an ultra-wideband ranging from 1 to 20 GHz. Afterwards, several surveys were done on damaged wooden wall specimens in laboratory. The experiment results indicated that the high-frequency radar waves can penetrate the wooden walls. The shape and orientation of the wooden members have shown a great sensitivity to the radar polarization. It is concluded that radar polarimetry can provide much richer information on the condition of concealed wooden structures than the conventional single-polarization subsurface penetrating radar. In Ref. 16 two log-periodic UWB antennas with dimensions of 300mm × 210 mm were used for imaging of the wood slog. They were able to directly provide images of the distributions of the dielectric properties inside the samples under test. The obtained results confirm that microwave imaging can be successfully applied for creating maps of the internal structure of wood samples and that they allow to extract information about the health and quality of the wood. Presently, the system has been proven to be able to quantitatively reconstruct targets whose maximum dimensions are about 0.1 m in green condition and about 0.25 m in dry conditions. In Ref. 22 a non-resonant ultra-wideband (UWB) antenna worked based on the quad ridged horn design. It was dual-polarization and operational in the frequency range of 0.7-10 GHz. A homogenous wooden slab used and its damping S₂₁ was investigated in desired distance and frequency to check how the S₂₁ was changing. They showed that simulations are reliable and can be used in the prediction of transmission through the wall.

In this paper, measurements are performed on high-density wood, soft plywood and in air. Although this might not be as realistic as the work presented in Refs. 28–31; It can still be considered as a promising method in the microwave imaging of wood in terms of the antennas’ and system dimensions, and not being destructive (recent similar such as Gamma scorpion and ECTV showed in Table I in supplementary material, applied bulky systems with large dimension antennas and were destructive like X-ray and Gamma but the proposed system is smaller with high gain and broad BW, and beneficial for detecting of defects in wood).

Since the number of antennas is directly related to the imaging accuracy, the UWB antenna is better to be low in profile. The best choice for this purpose is the planar antennas. The key challenges in MWT of wood can be listed as follows (the advantages and disadvantages of the methods are presented in Table I in supplementary material): dimensions of the applied antennas and systems, ability of obtaining 3D image with high resolution. Many methods were used in imaging of wood other than MW imaging such as Electrical Capacitance Volume Tomography (ECVT),³² High Resolution X-Ray Computed Tomography (HRXCT),²⁹ or Gamma scorpion,³³ Electric ring electrode array and Seismic tomography of trunks.¹⁷ But, each of these methods used for imaging of wood showed drawbacks that affected the imaging of wood in terms of the inadequate spatial resolution and being not portable. For instances, ECTV demonstrated the disadvantages such as Inadequate spatial resolution it can provide, the measurement resolution is dependent on ECT sensor designs and image reconstruction, highly nonlinear reconstruction problem is still considered the main obstacle to increasing resolution, nonlinearity of the problem is increased substantially (Table I in supplementary material).

Recently, many kinds of microstrip antennas were simulated and designed for imaging and tomography of the breast to detect tumours which had individual specifications. In Ref. 34, due to the small dimensions of the antenna, it could not cover the entire BW. Furthermore, the proposed antenna obtained a wider BW and more stable field pattern than that presented in Ref. 35. In Ref. 36, an antenna with packed dimensions benefited from a wide slot designed to work as a UWB antenna. Besides, the transmission response of the proposed antenna was about 5 dB in the simulation and less than 10 dB in measurement across most of the BW. A miniaturized antenna was presented in Ref. 37 but the reflection coefficient level was low within the acceptable BW. In Ref. 38, a bigger UWB antenna showed a smaller operating BW.

The proposed antenna has smaller dimensions as compared to the other recent and similar works;^61–68 thus, when the antenna is small, more antennas can be used so as more send and receive signals and better clutter removal in imaging. Furthermore, it has been shown that, with proper antenna design and signal processing, the clutters due to scatters and tag’s antenna structural backscattering can be easily removed.³⁹ In addition to that, Radar-based method used several antennas to examine the breast with low power ultrawideband (UWB) pulses. The scattered fields were received by the same or different antennas. A prerequisite for microwave radar techniques is a suitable transmitter/receiver like UWB antenna. To maximize the number of antennas to receive more data from the scattered signal, the size of antenna should be as small as possible. Planar antennas are the best option for this purpose. This will allow us to have more antennas to collect more scattered signals from the breast, and as a result it will decrease clutter in the image.^40,41 In another work, several microwave sensors (transmitters and receivers) are distributed on a circular region that may surround the object under investigation to measure scattered fields. Each sensor is alternatively activated as a transmitter, and the received signal at the rest of the sensors is captured, thus allowing to use information from all directions in the reconstruction procedure. UWB signals are proposed as illuminating signals which can provide enhanced image resolution and more clutter rejection as compared to mono-frequency reconstructions.⁴² For imaging of wood, clutter and artefact can be defined as the effect of wood’s crust (skin) or vascular tissues of heterogeneous in the wood sample.

This article is divided into four sections. The Section I presents an introduction to the present UWB antennas and MW imaging in different environments and applications. The proposed antenna’s design steps are demonstrated and then its characteristics are investigated in Section II. Then, the simulated and measurement results are illustrated in Section III. Finally, the paper concludes in Section IV.

In MW imaging, UWB antennas are used to send and receive the signals and produce an image with high acceptable resolution. The image resolution changes directly with the operating frequency and bandwidth (BW) of the antenna. Thus, higher image resolution can be obtained when the operating frequency and BW are high as well. However, increase in the frequency leads to increase of the loss in wood since the wavelength is decreased.

The proposed UWB antenna designed to radiate and receive UWB impulses with frequency content from about 3.26 GHz to about 20 GHz. To simulate the proposed UWB antenna, a droplet shape patch, stub, shorting pins and a truncated ground are used and then fed by the transmission line through a SMA port. A coaxial feed line connected to the SMA port delivers an UWB impulse to a TL at the base of the antenna. Thus, the proper dimension of the TL (width and length) helps to avoid spurious currents on the sheath of the coaxial feed line that could cause distortions in the antenna pattern and undesired variations in overall system performance and matching.

In designing procedure of the antenna, patch, transmission line (TL) and substrate dimensions, the ground length (L_g) (these feature are important to make the initial UWB antenna), and the loading of the antenna with stub (L₁₋₈) (the long stub connected to the junction and L-shape stub), shorting pins, slots cut form the patch and the ground, and strips at the back play important role in order to get an UWB antenna with wide BW and high performances (Figure 6 shows the surface current distribution of the proposed antenna around the lower and higher end of the UWB). The design steps can be explained as follows: first, a conventional elliptical patch antenna is designed at centre frequency of 12.5 GHz (the transmission line method is used to feed the antenna. The following formulas are used to obtain the actual values of the antenna’s dimensions; then the parameters’ values are optimized using CST. c=$(a2−b2)$⁠, a= $P/fμε$⁠. And μ = μ₀μ_r and ε = ε₀ε_r are the permeability and permittivity of the substrate respectively. P is an empirical constant ranging from 0.27 to 0.29. Normally p is taken as 0.275 which agrees very well with the empirical value. The eccentricity of the ellipse is defined as E_c = c/a.) and its radiation characteristics (reflection coeffect, BW, radiation efficiency, directivity and gain) are investigated (The patch and ground dimensions are optimized, as the ground length should be λ/8). The operation frequency calculation equations of antenna as a function of the patch width along with its length can be given in Refs. 43 and 44. Besides, the length of the patch (major axis of the elliptical patch (b)) gives the lower end of the ultra-wide BW and the width (minor axis of the elliptical patch (a)) helps in wider impedance BW (The wider impedance BW means wider and broader working BW which shows the reflection coefficient lower than -10 dB). The conventional UWB antenna at this center frequency (12.5 GHz) achieved the lower end at 4.2 GHz and first pole at 5 GHz.

After achieving the first pole and the lower end of the ultra-wide BW, the antenna is loaded by a stub connected to the junction between the patch and the transmission line (L₁₋₈) to resonate at the ISM frequency band and shift the working band to the lower band as well. The length is chosen based on the desired low resonant frequency (1.3 GHz) and the position of this stub according to the surface current distribution around the patch and the stub (Figure 7 shows the surface current distribution of the proposed antenna around the long stub connected to the junction and the L-shape stub (L_t1)). The patch shape is changed to droplet shape by cutting the elliptical shape from the edge (L_p in Figure 4). While cutting the patch, the distance between the patch and stub is considered due to the mutual coupling and surface wave may be occurred when the distance is too low as if it degrades the radiation efficiency. Besides, after adding the stub to the patch, the surface wave around the junction and the coupling are enhanced negatively; hence, the ground cut at the middle (L_sl, W_sl) to compensate this drawback, which degrades both the radiation efficiency and the reflection coefficient level. Apart from cutting the ground at the middle, another way to decrease the surface wave and coupling is to cut a slot from the patch (L_s1, g) to separate the stub from the patch and the droplet shape, which reduces the surface waves. Lack of this slot causes a reduction in the radiation efficiency and the reflection coefficient matching level. Moreover, it produces another resonance in the lower band (1.3 GHz) (A Vivaldi UWB antenna was used for inspecting a wooden frame for the range of 1-20 GHz. In Ref. 20, an antenna array consist of four arrays of linearly polarized Vivaldi antennas recorded full-Polari metric radar echoes in an ultra-wideband ranging from 1 to 20 GHz. They introduced an algorithm for 3D imaging to detect the damage in a wooden wall specimens in laboratory. They concluded that their radar system could work well to detect the conceal items in building structures). Thus, it is tried to have a resonance after 1 GHz without enhancing the antenna dimensions. Besides, the antenna can be useful for both L-band (1.3 GHz) and ISM (1.8 GHz).⁴⁵ Besides, the antenna behaves as a quarter wave monopole antenna at lower end of the ultra-wide BW at 3.26 GHz.

Next, two L-shaped stubs (L_t1) are added to the antenna as loadings to remove the stop-bands after adding the stub, to suppress the stop-bands around 8.5 GHz and to improve the reflection coefficient level. Then, the antenna is loaded by two strip lines at the back with the length of L_Load1,2 to shift the working BW to a lower band and to suppress further the stop-band that occurred at 8.5 GHz. Furthermore, another technique to reduce the surface waves, as well as shifting the bands to the lower band, is loading the antenna with shorting pins (brown dots in Figure 3 show the locations). Finally, to have a high-performance antenna, the antenna design parameters should be optimized at each step of the antenna design.

The simulated and fabricated prototypes of the proposed elliptical antenna along with the measurement setup and environment in plywood and high-density wood are presented in Figure 3 and Figure 5 respectively. The droplet antenna dimensions depict in Figure 3 are 15 mm × 15 mm × 2.4 mm. Furthermore, the antenna geometry illustrated in Figure 3 is located on the X-Y plane and the designing parameters are pointed and named accordingly (The design parameters of antenna which have affected the antenna performances are discussed in text and the rest are added in the supplementary material’s section). In addition to that, Figure 4 shows the patch geometry and how it is cut to make droplet shape to reduce the surface wave and the coupling between the patch and the stub.

In microwave (MW) imaging, two undesired parameters that affect the reconstruction of the image negatively are the clutter and the imaging artifacts. To reduce the amount of clutter and the number of artifacts in the image reconstruction process, the number of antennas should be increased. This increase in the number of antennas enhances the scattered signals in our imaging environment. Hence, having miniaturized dimensions is helpful in terms of reducing the cost of the system. According to the authors’ findings, the proposed antenna represented the lowest profile antenna applied for imaging of wood (Table II).

After designing the antenna and investigating its key parameters, the proposed antenna is simulated, and its parameters are optimized in air with dielectric constant of 1. Then the time domain characteristics of the antenna are investigated in high-density wood, and plywood (the wood sample dimensions are 45 mm × 45 mm×20 mm. Moreover, Figure 17 showed how the thickness of the sample changes the transmission coefficient result of the antenna) as working environments with dielectric constant of 3 and 2.1 respectively. Both plywood and high-density wood samples were homogenous in shape and electrical properties (after testing their dielectric properties, they presented almost same dielectric constant in each location on the sample). These types of wood are employed to show the working ability of the proposed antenna in an environment like wood. To gain an acceptable and applicable coupling between the transmitter (antenna) and wood, the antenna is simulated on PTFE substrate with a dielectric constant of 2.55, thickness of 2.4 mm and loss tangent of 0.001.⁴⁶ The substrate utilized in the proposed antenna has a lower loss tangent than other substrates like Roggers. Since the antenna has a dielectric constant of 2.55 which is close to the wood’s dielectric constant it shows better coupling (The coupling effects are analysed using both simulation and measurement in air, high density wood and plywood. The results of coupling showed in result section in Figure 20).

The key parameters in antenna designing procedure should be optimized to get the best results. These parameters can be named as patch dimensions (a, b), transmission line length and width (L_f, W_f), stub length (L_3,5), L-shape’s base length (L_t1), of length of cutting elliptical patch to make it a Droplet shape (L_p), shorting pins location. The design, simulation and optimization process of these parameters are performed in CST software. In CST, the parameters can be optimized by defining the upper and lower limitation for each parameter. The optimization method can be chosen in this software among Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and more five algorithms (GA used in our optimization due to its faster operation).

To optimize the antenna, the elliptical patch dimensions which are affecting the lower-end and higher-end of the wide bandwidth of the antenna should be considered first. Therefore, the semi-major axis of the elliptical patch has effects on the shifting of the band to lower or higher bands (directly proportional to the wavelength) and the semi-minor axis affects the BW. In addition, the width of TL is directly related to the matching of the antenna when it is equal to the characteristic impedance of the TL.

The length optimization of the stub connected to the junction is divided into two lengths, which affect the reflection coefficient result due to their distance from the edge and the resonator. Thus, only these two parts are presented here, which can be named as L₃ and L₅. The other parameters such as the gap (g) and the chamfer angle (α) do not affect the reflection coefficient result of the antenna dramatically as compared to the previously presented parameters, so they are not presented. The antenna dimensions achieved after optimization are presented in Table I.

The proposed antenna has four pole frequencies at 4.2 GHz, 7.1 GHz, 11.2 GHz and 16.1 GHz and two resonances at lower bands at 1.3 GHz and 1.8 GHz. Hence, the physical behaviour of guided wave to radiated wave on proposed antenna is analysed, representing in the simulated surface current distribution results on the radiating patch, stubs (stub connected to the patch and the L-shape stubs) at frequencies of 1.3 GHz, 1.8 GHz, 3.26 GHz, 8.5 GHz and 20 GHz, as shown in Figure 6 and 7. Unlike narrowband antenna, the UWB antenna behaves in a traveling wave type that supports both fundamental mode of propagation at lower frequency and higher-order modes at higher frequencies. It is revealed that the patch, TL and the slot on the ground are major parts to radiate wave responding for extremely wide frequency range. On the other hand, when the current paths on patch, TL, Slot on the ground are changed or disturbed, more impact to electrical characteristics of the antenna occurs based on the quarter wave modes (Figure 6).

Moreover, the surface current distributions are very strong nearby the edges of the stub connected to the junction at 1.3 GHz. It’s obviously shown and ensured that the required length to obtain a resonance at 1.3 GHz is the same length as obtained and optimized in Figure 7 (a). The same trend goes to the surface current distribution as its very strong around the edge of the slot cut from the patch nearby the stub at 1.8 GHz (Figure 7 (b)). Moreover, it’s obviously seen that the surface current distribution of the antenna around the stubs, patch, the L-shaped stub and the TL showed in Figure 7 is more focused and has more density around the slot cut to have a resonance at 1.8 GHz. The surface current distribution of L-shape stub is presented in Figure 7 (c). This stub has been created to suppress the stop-band at 8.5 GHz and change to pass-band; the surface current distribution of the L-shape stub showed a strong current density around the stub at 8.5 GHz. The surface current distribution at 3.26 GHz and 20 GHz are presented in Figure 6. Figure 6 shows the surface current distribution of lower-end and higher-end of the working BW at 3.26 GHz and 20 GHz respectively. It’s clearly demonstrated that the density of this current is strong around the transmission line and the patch at both of these frequencies. It can be concluded from these surface current distributions and the presented results in result section, the both patch and TL dimensions affected directly on the lower-end and higher-end of the working BW. Moreover, it shows that the space between the stub and the TL is also affects the results since the current is stronger near the space on the stub.⁴⁷ 

Before discussing about the results, it better to show how antenna the UWB operates both in time and frequency domain. Typically, narrow-band antennas and propagation are described in the frequency domain. Usually the characteristic parameters are assumed to be constant over a few percent bandwidth. For ultra-wide-band (UWB) systems, the frequency-dependent characteristics of the antennas and the frequency-dependent behaviour of the channel should be considered. On the other hand, UWB systems are often realized in an impulse-based technology, and therefore the time-domain effects and properties should be known as well. For the frequency-domain description, it is assumed that the transmit antenna is excited with a continuous wave. While, for the time-domain description, it is assumed that the transmit antenna is excited with an impulse signal with the frequency f. In frequency domain the antenna transfer functions represent a two-dimensional vector with two orthogonal polarization components; but in the time domain, the antenna’s transient response becomes more adequate for the description of impulse systems. The antenna’s transient response dependents on time, but also on the angles of departures and arrivals, and polarization. Figure 8 shows the procedure of send and receive a pulse between two UWB antennas in time domain (Both transmitted pulse and output voltage are the same scale as vertical axis is the amplitude of the signal and the horizontal axis is the time in nano-seconds (ns)).

To do the measurement a Performance Network Analyzer (PNA) with model E8363C is used. First, the PNA is calibrated to obtain a perfect accuracy in measurement. Then, a frequency range from 500 MHz to 30 GHz and 1002 (While doing the measurement using the aforementioned PNA, the magnitude type put on the logarithmic to see the S-parameters in terms of frequency.) frequency steps for this range of frequency are adjusted and then the measurement in air is fulfilled (The calibration type of SOLT is used to accurately calibrate any number of ports. This type use the calibration methods of SmartCal, unguided Calibration, and ECal. The cables used for measurement are 50 ohms male cables AGILENTTECHNOLOGIES 8120-6192 with a diameter of 3.5 mm and length of 26 cm). Furthermore, to measure the scattering parameters in different medium like high-density wood and plywood, two antennas are used. One antenna connected to terminal one of the PNA as a transmitter (Tx) and the other one connected to the second one as receiver (Rx). The Tx kept fixed at the centre of the wood and the other antenna touched the other side of the wood slice according to the array’s locations showed in Figure 12 and Figure 5. In each step the Tx is fixed, and the Rx is located on array locations. Afterward, the scattering parameters (the reflection and transmission coefficient) for each array are extracted from PNA and imported to MATLAB to be evaluated for time domain considerations.

Based on both the simulation and measurement results shown in Figure 9, the proposed antenna has an acceptable reflection coefficient and working BW. Figure 9 illustrates that the proposed antenna can be assumed as a UWB antenna since it obtains more than 16.32 GHz bandwidth at centre frequency of 12.5 GHz of the entire frequency band. Besides, the measurement and simulation results are in good agreement.

It is clearly shown in Figure 9 that the resonant frequencies at both 1.3 GHz and 1.8 GHz are achieved with only a slight shift from the simulation result, but the antenna is still working, and they are within the bandwidth. Furthermore, most of the working frequency band are obtained (3.44–25 GHz) and it is just shifted slightly from 3.26 GHz in the simulation result positively. Apart from that, the stop-band after 20 GHz is removed in the measurement result and the reflection coefficient level is close to -10 dB, which is acceptable (VSWR < 2). In addition, the optimum required BW for imaging both in wood and air (up to 10.6 GHz) are obtained. The proposed antenna’s BW is enhanced to have the antenna working at higher bands, which is useful for other applications like skin cancer, which needs low penetration (previous studies applied even a low THz for skin cancer).

The first difference between the simulation and the measurement results in the reflection coefficient is due to the differences that occurred after fabrication. The next one is due to the simulated conditions and the measurement conditions are not the same, and tolerances during the fabrication process may affect the results after fabrication.

In addition to that, in CST the waveguide port is usually used for feeding the antenna from the macros section, as software does it automatically after calculating the port dimension based on microstrip line equations and substrate thickness and width of the feedline. At upper frequency range, the exposed centre pin of the SMA port causes significant radiation, so reducing the length of the centre pin limits the unwanted radiation from the connector. The simulation result in Ref. 48 showed that the shorter centre pin, the smaller VSWR is. To ease the soldering and the reliability of the connection, the centre pin is chosen to be 0.5 mm. Besides, soldering should be done carefully not to increase the resistivity of the ground using too much tin in soldering.

Figure 10 presents the antenna’s far-filed radiation pattern (E and H) at θ = 0 and 90° respectively for whole frequency band of the antenna. Besides, the horizontal and vertical component of the far-filed is depicted as well. Far-field tomography could be a good choice when time consumption is not a problem like X-ray ptychography which is a robust technique to be used in a far-field regime for a fair amount of time it can be used in near-field. Some works done on far-field algorithm for imaging purpose. (Moreover, the spatial resolution of reconstructed images can be significantly degraded. Our goal in this work is to clarify the domain of validity of the imaging model that mitigates such effects by use of a far-field approximation. Computer-simulation studies are described that demonstrate the far-field-based imaging model is highly accurate for a practical 3D PACT imaging geometry employed in an existing small animal imaging system. For use in special cases where the far-field approximation is violated, an extension of the far-field-based imaging model is proposed that divides the transducer face into a small number of rectangular patches that are each described accurately by use of the far-field approximation⁴⁹). Furthermore, Figure 10 shows the simulated radiation pattern of the antenna for electric fields (E, horizontal component of far-field) and magnetic fields (H, vertical component of the far-field). Practical measurements for the radiation pattern are rather difficult, since there is no standard antenna as receiver that operates in a medium with dielectric constant of 2-3 rather than free air. Besides, a good agreement between the simulated and the measured S11 obtained; thus, it is reasonable to rely on the simulated radiation pattern by the CST. The corresponding far-filed gain at the frequencies from 1.3 GHz to 25 GHz presented in Figure 10 are as follows: -3.25 dB, -2.6 dB, -0.9 dB, 0.42 dB, 1.88 dB, 2.36 dB, 2.76 dB, 3.26 dB, 5.6 dB, 5.4 dB, 4.83 dB, 4.57 dB.

The results presented in Figure 10 shows the radiation pattern of the proposed antenna for all working band for θ = 0° and θ= 90° planes. It is clearly illustrated that the antenna has district main lobe between 120° and 180° at 1.3 GHz. This trend remains unchanged at 1.8 GHz, 3.26 GHz, and 4.2 GHz. When frequency reaches to 6.4 GHz, the radiation pattern is altered slightly from 180° by 30° to 210°. After 6.4 GHz, the pattern does not change till 9.6 GHz which it returns back to 180°. After 9.6 GHz, the pattern is shifted to 210° at 11.2 GHz and shifted more to 240° at 16 GHz. The main lobe is unchanged at 18.6 GHz as well. Besides, this trend is not altered at 21 GHz, 24 GHz, and 25 GHz. Figure 10 is presented the radiation pattern of the proposed antenna for poles, lower-end and higher-end of the band. Furthermore, the district main lobes of the patterns at each frequency is shown with a black bold arrow towards the main lobe. Furthermore, when the simulated and the measured S₁₁ are in good agreement, the simulated radiation pattern obtained by CST can be considered as reliable result. The maximum far-field directivity of the antenna at both θ = 0° and θ = 90° is 7.3 dBi and the gain is 5.69 dB. Moreover, 3 dB beamwidth of 178° in the θ = 0° and 117° in θ = 90° plane respectively at centre frequency. In addition to that, the radiation efficiency of the antenna extracted from CST by applying FDTD analysis is 48 % at 3 GHz, 58 % at 5 GHz, 70 % at 7 GHz, 78% at 9 GHz, 75% at 11 GHz, 57% at 15 GHz, 74 % at 17 GHz, and 82 % at 20 GHz.

Since the antenna structure is complex and it consists of stubs and slots, it would be informative if the sensitivity of the antenna performance to fabrication tolerances is provided (Figure 11).

For example, the antenna performance when the layers are perfectly aligned (without tolerance) as compared to when there is a horizontal (‘hor’ in Figure 11) or vertical (‘ver’ in Figure 11) misalignment of the layers (the misalignment might cause by fabrication). Figure 11 clearly shows that when there is tolerance in fabrication of the antenna, the reflection coefficient results of the antenna change according to the amount of misalignment (gnd hor 2: 2mm, gnd ver 1: 1mm, gnd ver 2: 2mm. pat hor 1:1mm, pat hor 2:2mm, pat ver 1:1mm, pat hor 2: 2mm.). This tolerance makes some stop bands in the working BW so as more resonances in the lower band.

The measurement setup is presented in Figure 12. The wood slab dimensions are 45mm × 45mm × 20mm. The antenna positions and the angles at which they are located during the measurement are depicted in Figure 12. The same method as is done in simulation to obtain the time domain characteristics of the array antenna, is performed in measurement. In measurement, Tx kept fixed and then the array antennas A1- A9, located at the location based on Figure 12. Afterward, the time domain characteristics are investigated for each array.

To have the lowest possible distortion in the signal transmitted from the proposed antenna to the sample, the transmission response (S₂₁) needs to be flat at the desired working frequency BW.^50,51 In addition, the reflection coefficient result (S₁₁) should present an acceptable result and should not vary too much from the simulated result in air. Since S₁₁ shows how much of the wave is reflected when the wave is passing through an environment other than air, it should be less than -10 dB to be acceptable.

Figure 13 shows the simulated reflection coefficient results of the antenna in high-density wood, plywood and air. A negligible variation of the simulated reflection coefficient is observed, and the resonance shifted towards higher frequency. The reflection coefficient result in air is changed slightly when the signal passes through air to reach antenna 2 (A2 position). Besides, both the resonances at 1.3 GHz and 1.8 GHz are obtained and the lower end of the ultra-wide BW shifts slightly to the higher band. These changes are due to the distance between the two antennas and the time delay for the signal to reach the second antenna (When the distance between two antennas is increased, the time delay enhances as well. Besides, it immediately becomes clear that this factor is not independent of media size, as one might expect of a true reflection coefficient (in the control population, there was a significant inverse association between reflection coefficient and media’s length). The influence of length on reflection coefficient (V⁻/V⁺) is explained by the fact that, in taller subjects (more distance between two antennas), both transmitted voltage (V⁺) and reflected one (V⁻) travel longer distances. So, on arrival at the reflection site, the amplitude of reflected voltage (V⁻) is smaller because of damping. In addition, the reflected wave needs to travel a longer distance and is more damped as well. As a result, the taller the subject is (bigger media between two antennas), the smaller the amplitude of the reflected coefficient). The reflection coefficient result in air has the highest level because of lowest relative permittivity. Moreover, the reflection coefficient result for plywood and high-density wood are not too much different due to their close relative permittivity (2.1 and 3) but almost higher reflection coefficient for plywood due to it lower relative permittivity as compared to high-density wood. This is because of the differences in their dielectric constants and the fact that low relative primitivity shows higher reflection coefficient. In addition to that, the frequency band is shifted towards the lower band. The main reason for this shift is the dielectric loading of the wood. This variation did not affect the antenna performance when incident signal facing plywood and high-density wood because the complete operational bandwidth is still below −10 dB.

Figure 14 illustrates the transmission response simulation results of the antenna in air, high-density wood and plywood. The plywood transmission response is better than that high-density wood. For both plywood and high-density wood, the transmission response presents satisfactory results until 10 GHz, and even in air the result only varies around 5 dB except at two resonant frequency points. After 10 GHz the results is degraded by almost 10 dB until reaching 15 GHz, and for higher frequencies this degrading goes still further. Overall (at most of the BW), the transmission response result for air is better than for the other two environments except for the frequency band of 20-23 GHz which the transmission coefficient increased by 10 dB. That is due to the higher dielectric constant of high-density wood and plywood compared to air. Overall, the transmission response result for air is better than for the other two environments. That is due to the higher dielectric constant of high-density wood and plywood compared to air. This higher dielectric constant is reduced the signal and wave amplitude (transmitted form the antennas) when facing environments with different dielectric constants (Dielectric Constant or Relative Permittivity: Considering the dielectric constant of a material is important for signal integrity and impedance considerations, which are critical factors for high-frequency electrical performance. The dielectric constant varies with frequency and generally decreases as frequency increases; some materials have less of a change in relative permittivity than others. Materials suitable for high frequency applications are those whose dielectric constant remains relatively the same over a wide frequency range–from a few 100MHz to several GHz. Dielectric Loss Tangent or Dissipation Factor: A material’s loss tangent gives a measure of power lost due to the material. The lower a material’s loss tangent, the less power lost. The Tan δ of most PCB materials range from 0.02 for most commonly used materials to 0.001 for very low-loss high-end materials. It also varies with frequency, increasing as frequency increases. Loss tangent isn’t usually a critical consideration for digital circuitry, except at very high frequencies above 1 GHz. However, it is a very important parameter for analogue signals, as it determines the degree of signal attenuation and thus affects the signal to noise ratio at various points along signal traces).

To minimize the distortion in the transmitted signal through the wood, it is required that the S₂₁ be as flat as possible in the required frequency bandwidth.²⁶ Figure 15 depicts the measured transmission response of the proposed antenna based on the measurement setup in Figure 12. To perform the measurement of transmission coefficient, two antennas are located at both side of the wood sample (plywood and high-density wood) with thickness of 20 mm and in air with the distance of 20 mm (to perform the measurement of transmission coefficient, two antennas are located at both side of the wood sample (plywood and high-density wood) (position A5 from Fig. 12) with thickness of 20 mm and in air with the distance of 20 mm.). The simulated S₂₁ in Figure 14 shows almost only 5 dB variation in the frequency range of 3- 10 GHz (except at 7.5 GHz which it was altered more to 10 dB) for air while this level is reduced at higher frequencies for air, high-density wood and plywood (the variation is illustrated with dashed red line in shape) while this level is reduced at higher frequencies. The S₂₁ variation is decreased by almost 10 dB for measurement in air. The same trend goes with the measurement of plywood and high-density wood at frequency range of 3-10 GHz. After 10 GHz the transmitted power is dissipated within the wood because of the lossy material (higher relative permittivity). Moreover, when the frequency is higher than 12.5 GHz to 15 GHz, the transmission response is degraded by more than 15 dB. This reduction in S₂₁ and distortion are due to the decreasing of the wavelength at higher frequency, and due to the higher dielectric constant of high-density wood and plywood compared to air. Since the substrate chosen for this antenna had a dielectric constant (2.55) near to both high-density wood and plywood, their results did not differ greatly.

As aforementioned, the sample’s thickness is chosen to be only 20mm initially to accelerate the simulation time. Then its extended to 40 mm, and 80 mm to investigated if the system is flexible and can work in different dimensions (The X, Y dimensions are the same for all cases as 45 mm × 45 mm).

Figure 16 shows the reflection coefficient result in different thickness of the samples for both plywood and high-density wood. Not too much changes were occurred when the thickness increased; only slight shift at 4.2 GHz and 10 GHz in wood occurred. The first resonance was removed when the sample changed to high density wood and more ripples in band obtained caused by the higher conductivity of the plywood (The first resonance was removed when the sample changed to high density wood and more ripples in band obtained caused by the higher conductivity of the plywood. The variation of S-parameters are higher at lower conductivity.⁵² When the signal faces a medium with higher conductivity, it dissipated more and more attenuation in reflection coefficient level occurs). In addition to that, the transmission coefficient of the proposed antenna is investigated in different thickness of the sample for both high-density wood and plywood (Figure 17). It is demonstrated that isolation increases with thickness for plywood. Moreover, the transmission coefficient of plywood shows higher level for frequencies lower than 10 GHz and it is degraded after 5 GHz to 30 GHz gradually and lower below the 5 GHz. Furthermore, same trend is followed by transmission coefficient of high-density wood, but better level and same trend obtained for the high-density wood. In addition to that, this slight alteration might be due to the higher conductivity of the plywood which causes less penetration (1× 10⁻⁷ for plywood and almost 1 × 10⁻¹³ for high-density wood).

After investigation of reflection coefficient and transmission coefficient in different thickness of the samples, the effects of each array on its adjacent array’s reflection and transmission coefficient results are investigated to show the level of the mutual coupling and the isolation among the arrays. Because, there might be a range of frequencies that indicate good isolation between elements and its equally important to identify the range of frequencies for which there is poor isolation. In addition, poor tomographic reconstruction results at these frequencies is expected.

The simulated amplitude reflection coefficient results of each antenna array in the presence of plywood and at the presence of the other array antennas are presented in Figure 18. As indicated in Figure 18, the reflection coefficient has higher level at some frequencies between 5 GHz–7.5 GHz and 17 GHz-21.5 GHz. This change in the reflection coefficient indicates the loading effects due to the presence other antennas. It can also give an indication as to which working frequencies should be selected for the microwave tomography (MWT) operation. To do the measurement only two antennas are used at each stage of the measurement. For example, antenna A1 and antenna A5 are used and the reflection/transmission coefficient are measured. The A1 fixed and the other antennas are replaced and their results are recorded (this method is used due to having a PNA with only two port to perform the measurement).

Besides, the image reconstruction error is lower at frequencies where the reflection coefficient is invariant. Figure 19 shows the reflection coefficient phase of the proposed antenna when its alone and when nine arrays are located around it in both X and Y-directions. It is clearly illustrated when antenna is located among the arrays, the reflection coefficient phase of the antenna slightly changed due to the loading effects of the other arrays on each other.

Furthermore, it shows linear phase variations except for A4 which is farther than Tx and its reflection coefficient is so affected by loading effects of A2, A3 and A6.

Figure 20 shows the transmission coefficient results between antenna (Tx) and other antennas, which highlights the level of mutual coupling. As shown in this Figure, the mutual coupling is less than –20 dB at all the band except coupling with antenna 1 which is less than -13 dB at frequencies around 5 GHz and there is a good isolation between elements at these frequencies. Furthermore, it is obviously demonstrated that the reflection coefficient result of the antenna is not altered a lot except at 8.6 GHz which a stop-band is occurred. In addition, some other changes at the other frequencies such as 11 GHz, 9.6 GHz, and from 3.3 GHz to 8.6 GHz turn up. These changes in the reflection coefficient results indicate the loading effects due to the presence of the other antennas.

Transmission coefficient amplitude and phase evaluation are used to identify the level of mutual coupling. Figures 20 and 21 present good isolation at most of the frequency band since the transmission coefficient amplitude is between - 30 dB and -40 dB at most of the BW. Although, at some frequencies its near to - 20 dB. When one antenna is located at the front of the proposed antenna at Phi= 0, frequencies from 3.5 GHz to 5 GHz show transmission coefficient of near – 20 dB thus at these frequencies higher mutual coupling and lower isolation occurred in comparison with the other frequencies. Furthermore, not too much variation in phase is noticed in Figure 21 as shape of the plot for phase is changed a lot but shifted in frequencies at each degrees (a signal has linear phase when phase response is a linear function of frequency). In addition to that, the phase variation is linear at most of the frequency band except at few bands of the frequency for each arrays at presented φ and θ. These phase variations at different frequency bands and degrees are occurred because of the antenna loading effects of the other arrays around it. These places of the non-linarites are illustrated by red circles on the Figure. To have better understanding of the phase, only two arrays of each direction are presented.

Figure 22 and Figure 25 indicate the transmitted and received signals respectively.

In Figure 22 h(t) is the time domain impulse response (The initial impulse which use as input for the antenna is a Gaussian signal with a maximum amplitude of 1) and |h⁺(t)| is the envelope of the impulse response which localize the distribution of energy versus time and it can be a direct measure for the dispersion of an antenna. The peak value p (θ, ψ) of the envelope shows the strongest peak of the antennas time domain transient response envelope. The envelope width (τ) demonstrates the broadening of the radiated impulse and is determined as the magnitude of analytic envelope at half maximum. It should not exceed a few hundred picoseconds to obtain the high data rate and high resolution in communications (High resolution is important in tomography to have better image of the sample and the defects in it). Furthermore, the ringing of a UWB antenna is an undesired parameter and normally caused by resonances due to energy storage or multiple reflections. Besides, it is defined the time until the envelope has fallen from the peak value to a certain lower bound and it should be negligibly small less than a few envelop width. The energy or the ringing is not used at all and it can be eliminated by e.g. absorbing materials (The absorbing materials can be added around the antenna or the system. Apart from the absorbing material, eliminating the ringing can be done in imaging section using the proper algorithm and time window (the algorithm and time widow consideration are not in scope of this article)).

Based on the measurement setup presented in Figure 12, antenna Tx keeps fixed and antenna A1-A9 are moved around the antenna Tx at different angles (Rx as A5). Antenna Tx transmits the signal depicted in Figure 22 and then the other antennas receive the signal presented in Figure 25 for different arrays (A5-A9). The simulated received signals in air show the highest amplitude. When the environment changes to high-density wood and plywood, due to the higher dielectric constant, the amplitude of the received signals is changed and decreased. The plywood received signals show better results than high-density wood due to the lower dielectric constant compared to the high-density wood. Besides, the total shape of the signal when transmitted from antenna Tx and after passing through the environment with different dielectric constants is not changed (Figure 25 shows the received signals in different angles (φ) from 0 to 180 degrees. It’s obvious that the signals’ shape did not change dramatically except the signal’s amplitude (The time delay is presented in section C.). The signal’s similarities and low distortion in signals were proven with fidelity percentage later in Figure 26). Hence, the proposed antenna can be an acceptable device to act as a transceiver in a medium such as wood (The pulse repetition frequency (PRF) is the number of pulses of a repeating signal in a specific time unit, normally measured in pulses per second. The term is used for example in radar technology). In simulated and measured results of the time domain considerations, for each MW frequency, a resonant effect might occur that register in receiving antenna as an amplification of the signal in frequency domain. This explains why some materials like plywood can express higher transmission. In addition, plywood has less dielectric constant compared to the high-density wood. Thus, the transmitted signal can be penetrated more into the material. Figures 23 and 24 illustrate frequency domain spectrum of the transmitted signal and time domain variant of reflection coefficient and transmission coefficient. This can be done by Fourier transform (FT). The FT will be calculated in the frequency band between F_min and F_max.

Mathematically, the FT returns positive and negative frequencies. For real-valued functions, the positive and negative side of the FT are the complex conjugate of each other and it is common to plot only the positive frequencies. If this is desired, F_min should be larger than or equal to zero. The number of samples of the input function and the resulting FT are critical for accurate and physically meaningful results. Due to the Nyquist-Shannon sampling theorem, the input function should be sampled with at least twice the specified F_max.

As reported in Ref. 53, the S-Parameters of the proposed antenna can be attained from time domain data signals based on Fourier theory and techniques. In this paper, the S-parameters and time domain signals of the designed antenna are provided to validate the proposed design method. The reported method in Ref. 53 could not performed in our Lab due to the lack of equipment and limitations. However, one way of analyzing the radiation and detection of these pulses would involve the traditional frequency-domain (FD) antenna parameters on a frequency-by-frequency basis. FD parametrization of the antennas lends itself conveniently to producing a comprehensive transmit-receive system description.

However, because of the broad frequency band of the short-pulsed fields, direct treatment of the antennas in the time domain (TD) may lead to more efficient and physically transparent representations. The dividends of a self-sustained TD description call for a development of a complete characterization of a transmit-receive antenna system.⁵⁴ This system characterization should be equivalent to the FD parameters and possess their important features, in particular, the ability to link all parts of the transmit-receive system together, in a manner equivalent to the FD Friis equation. The TD system operators are defined with reference to the current-wave incident upon the antenna terminals.

It is required to calculate the signal fidelity, since signal distortion is a critical issue. Fidelity can be considered as the magnitude of the cross-correlation when it reaches its maximum between the transmitted and received pulse. After achieving the received signals in the time domain from the last section, the fidelity, F, can be obtained as follows:⁵⁵ 

x(t) is the transmitted signal and y(t) is the received signal which has been shifted by τ. The antenna’s fidelity with 20 mm thick wood is depicted in Figure 26 for different angles. To achieve the fidelity factor one array is kept at the centre (transmitter) and the other arrays are located around the antenna in different angles. Then a pulse covers the frequency range of 3.26 to 20 GHz is sent from the transmitter. Based on the high fidelity presented in Figure 26, there is low distortion in the transmitted signal, with a percentage of over 65 %, and the proposed antenna can be highly recommended for use in the microwave imaging of wood.⁵⁶ In addition, Figure 26 illustrates the high percentage of fidelity in all three environments. It is obvious that the fidelity percentage is the highest for both the measured and simulated results in air, followed by the results for plywood and high-density wood.

The lowest percentage is for the high-density wood, which is because it’s highest dielectric constant compared to air and plywood. Furthermore, it demonstrates that how uniform the illumination is and how good isolation among the arrays can be. Thus, better isolation and more uniformity are illustrated in air and then followed by plywood and the high-density wood.

The fidelity results of the proposed antenna located at the center based on the arrays arrangement in Figure 12, are illustrated in Figure 27. The X-axis shows the antennas (A₁-A₉) and the Y-axis demonstrates the fidelity percentage for each antenna.

To show how many receivers can uniformly be illuminated by the transmitter, antenna Tx transmitted a pulse in the working frequency band and nine arrays are located around the antenna (Figure 12) to receive the transmitted pulse from the Tx. The errors and changes in transmitted and received pulses are investigated by the fidelity percentage between arrays located in X and Y directions. Besides, the isolation between the arrays can be shown by changes in fidelity percentage. The fidelity percentage (similarity between the send and receive pulses) is quite high from 97 % to 99 %. The antenna A1 obtained the lowest fidelity among the other arrays. In addition, A₆, A₃ and A₄ show almost the same percentage and A₅ got the highest percentage. Therefore, it can be concluded that a good isolation exists among the arrays and they are uniformly illuminated by transmitter.

When the distortion in the signal phase becomes critical, a parameter like group delay (GD) should be considered. When the received signals show too much fluctuation, a low percentage of fidelity will be obtained. Figure 28 illustrates the proposed antenna’s GD for various distances (5–75 mm) between the transmitter (antenna Tx) and receiver (A-A9). The increment is because of the heterogeneity in various environments and frequencies. Since group delay has a direct effect on widening the resolution cell, the antenna should be designed in a way to keep the instantaneous error less than one resolution cell and the FFT should be 1/T. The maximum group delay can be calculated as follows:

In Equation (2)f_s is the low resonance frequency.⁵⁷ In a medium with high thickness, the lowest resonance frequency should be reduced. For this paper’s purpose the lowest resonance is at 1.3 GHz and the maximum GD becomes 3.8 ns (The maximum GD is 3.8 ns at 1.3 GHz which is the lowest resonance. The maximum GD in Fig. 28 is based on the distance at 75 mm which is 2.5 ns).

The simulated near-field radiation intensity of an antenna is an important parameter in MW imaging. In the radiating near-field region, the average energy density remains constant at different distances from the antenna; although there are localized energy fluctuations. Normally, the near-field test system measures the energy in the radiating near-field region and converts those measurements by a Fourier transform into the far-field result. The proposed antenna’s radiation pattern is drawn at 5 cm distance from the antenna at the XZ-plane as shown in Figure 29. It is clearly presented in Figure 29 that the antenna radiates most of its power to the wood. Hence, it can be a good medium for MW imaging in wood. For acquiring the near-field pattern in CST, for both the E-field and H-field, 360 probes (in ideal case to get a smooth pattern) are placed at 5 cm distance around the antenna (in this paper only 180 probes are put around the antenna). Next, these fields in each direction (X, Y, Z) are extracted from CST and then imported to MATLAB to draw the radiation pattern of the antenna (by increasing the number of probes a smoother shape can be obtained). The near-field data (amplitude and phase distribution) is acquired by using a probe to scan the field over a preselected geometrical surface, which is a sphere. The calculated data can be transformed to the far-field using analytical methods to prove that the achieved near-field pattern is correct.⁵⁸ 

The half energy beamwidth (HEBW) and half energy beam (HEB) are based on the energy radiated by the antenna. The energy in and around the antenna structure is calculated by summing time samples of the instantaneous Poynting vector over the duration of the simulation time. This term was called the energy flux density (EFD).⁵⁹ In the near-field, the HEBW is defined on a plane orthogonal to the main radiation beam and situated at a given distance from the antenna aperture. The HEBW describes the region over which the energy is greater than half of the maximum value on the selected plane. As shown in Figure 30, the antenna is aligned with the y-axis, so the HEBW is evaluated in the x- and z-directions. The HEB attempts to provide a more general representation of the radiation beam in the near-field. The HEB representation in the near-field is powerful since, by including the beam origin information, it models the radiation behaviour as function of the actual antenna structure and size. This makes radiation pattern comparison between antennas more straightforward and makes prediction of the radiation coverage of an object placed close to the antenna possible.⁶⁰ 

To define the beamwidth, the EFD is computed on X and Z planes ranging from the antenna feed point to 50 mm away from the aperture. These data are shown for both antennas in Figure 31 for the X plane. The broadside radiation behaviour is evident. A slight increase in the back radiation is noticed when frequency is increased. When the radiation behaviour is analysed in the frequency domain, it is observed that grating lobes around the main lobe of the radiation pattern increase with frequency.

Electromagnetic waves are used to transport information through a wireless medium or a guiding structure, from one point to the other. The quantity used to describe the power density associated with an electromagnetic wave is the time average Poynting vector (average power density) which is a cross product of E and H^*.⁶⁰ 

The real part of excitation pulse presented as equation (1) presented in Ref. 60 is averaged over the propagated power density, and the imaginary part represents the reactive (stored) power density associated with the electromagnetic fields. The real part is responsible for delivering power from one antenna to another antenna. Figure 32 shows the real part of the Poynting vector when one antenna (Tx) is propagating in the ZX plane for the other 9 arrays. In addition to that, good penetration of electromagnetic waves inside the wood sample is observed from Figure 32 improves the power efficiency of the link and the ability to detect the presence of defects in wood.

The proposed UWB antenna is compared with recent similar existing antennas for imaging purposes, as shown in Table II. In addition, the antenna’s performance is checked in terms of some parameters such as applications, 10-dB BW, dimensions, directivity and gain. However, while the proposed antenna might not have higher gain than some works presented in Ref. 13, a good fractional bandwidth (FBW >130.4%) is still achieved (f1=3.26 GHz, f2= 20 GHz, fc=12.5 GHz). Since the proposed UWB antenna is low in profile, more antennas can be used for MW imaging of wood. In addition, a high performance can be achieved when the antenna dimensions are remained small compared to the recent works presented in Table II.

A UWB antenna with miniaturized size is presented for microwave imaging in wood. The proposed UWB elliptical microstrip antenna is designed to radiate in the wood environment. The proposed antenna comprises an elliptical patch, fed by a transmission line. A stub is connected to the junction between the patch and transmission line, and a slot cut from the patch are used to have two more resonances at 1.3 GHz and 1.8 GHz. It has a good feed match at the 3.26 to 20 GHz frequency band (almost lower than -13 dB at most of the working BW) and 3 dB beamwidth of 178° in the θ = 0° and 117° in θ = 90° plane respectively. In addition, a maximum gain of 5.69 dB (3.87 dB near-field) and directivity of 7.3 dBi are achieved. Both the simulated and measured results for the transmitted and received signals show low distortion through two types of wood and air, since the antenna’s transmission response (S₂₁) is almost 5 dB at most of the frequency band (3.26–20 GHz) and (21.7–25); hence, the UWB antenna is suitable and adequate for imaging of wood, especially when both the simulated and measured received signals in different environments show a good agreement. Moreover, the measured signals’ shape is not changed in the different environments. In addition, the proposed antenna shows high fidelity in both received and transmitted signals in various media. In accordance with the achieved results, the manuscript supports the possibility that the antenna is a strong candidate for wood imaging.

This section presents the supplementary material to support the explanation in the main text. The explanation in this section are linked to the main text. First, a table is presented in order to compare the techniques for MW imaging. Then, some steps in antenna optimization with less importance from those presented in the main text are illustrated just to explain more.

This work was supported by YUTP grants of graduate study office of Universiti Teknologi PETRONAS (UTP).