Diabetes is one of the most common chronic diseases in the world. The daily count of patients suffering from this disease is increasing at an alarming rate. Diabetes demands regular medical attention and risk-reduction tactics. More than 9% of the global population is diagnosed with diabetes, with approximately 23% going undiagnosed (Narayan et al., 2000). Diabetes is the seventh biggest contributing cause of death, according to recent reports from the American Diabetes Association (Kaul et al., 2013) and the World Health Organization (WHO). Timely diagnosis and treatment are vital for patients with diabetes. So, research for early detection, novel preventative measures, and improved treatments is in high demand. Early detection of prediabetes and monitoring of blood glucose levels (BGLs) of the body could help in treatment or prevent diabetes altogether if patients adopt healthier practices. The idea of BGL monitoring is both affordable and practical in developing nations. However, the conventional method of finger pricking to measure BGL is painful and prone to risk of infection. A non-invasive, safe, and painless method for BGL monitoring eases the process. Various non-invasive methods, such as thermal (Oliver et al., 2009; Cho et al., 2004; Tang et al., 2008), optical (Rawer et al., 2004; Malik and Coté, 2010; Nicolai et al., 2007), and electromagnetic methods (Gourzi et al., 2005; Tura et al., 2010; So et al., 2012), are currently being studied for BGL monitoring. However, these methods are expensive and require large instruments to carry out the measurements. BGL monitoring through RF/microwaves is a new method that is better than the existing methods for BGL monitoring (Satish et al., 2021; Kandwal et al., 2021; Rahayu et al., 2023). The RF sensors required for the measurements are small and, hence, easy to handle. Moreover, the process of RF measurement of BGL is cheap and simple. These factors make RF/microwave-based sensing devices for monitoring BGL popular and in high demand.

The setup for the RF/microwave sensing of BGL requires an antenna. The antenna radiates, and when a human body is placed near the antenna, the body reflects the radiation at a specific frequency known as the resonant frequency. The value of the return loss and the resonant frequency when an antenna is in proximity to a human body is different from when an antenna radiates in a free space. This difference varies with the blood glucose concentration of the body. This is explained in more detail in the later sections. The difference in return loss and resonant frequency is correlated with the actual BGL, and the readings are measured. This process is an easy, inexpensive, and quick method to measure BGL in humans.

The prime concern for this RF/microwave BGL sensing method is antenna design. The antenna must be miniaturized so that it is portable. The specific absorption rating (SAR) must be well within the prescribed limits so that the radiations are not harmful to the human body. The radiation penetration must be deep enough to reach the blood layer of the human body. For this, the gain of the antenna must be enough so that the major lobe of the antenna has a high magnitude to reach the appropriate distance in the human body for BGL measurement.

The above-mentioned points were considered while designing the sensor for BGL monitoring in this work. An electromagnetic (EM) sensor designed for this purpose consisted of an antenna and an artificial magnetic conductor (AMC). The sensor had a low profile and showed a gain of 3.6 dBi and an SAR of 0.036 W/kg.

The overall article is organized as follows. The detailed design with a numerical analysis of the proposed EM sensor is given in Section 2. Section 3 discusses the simulated results of the sensor. Section 4 is dedicated to the SAR results and penetration depth of the proposed sensor. The simulated results of the proposed sensor on a human phantom with different BGL values are presented in Section 5, whereas Section 6 provides the measured results of the proposed sensor on different aqueous glucose solutions and a comparison of simulated and measured results. Section 7 concludes the article.

Figure 4 shows the final CDRA geometry with the dimensions labeled. Table 1 shows the physical dimensions associated with the CDRA design. Figure 5A shows the S₁₁ parameter graph of the final antenna design (Step VI). An impedance bandwidth from 2.35 GHz to 2.6 GHz is exhibited by the final optimized antenna design. Table 1 discusses various parameters and their respective dimensions associated with the designed CDRA.

Figure 5B shows the simulated directivity pattern of the designed CDRA backed by the designed DGS at the resonant frequency of 2.4 GHz. From the pattern, we can infer that the CDRA shows a bidirectional pattern of radiation with a gain calculated as 3.58 dBi at 2.4 GHz. Figure 6A shows the E-field pattern of the CDRA at its resonant frequency, that is, 2.4 GHz. Figure 6A shows the E-field at the top of the DR (z = H = 6 mm) and at 2.4 GHz. With reference to Figure 6A, the typical HEM11δ -mode field distributions are observed. Figure 6B shows the surface current distribution of the designed CDRA at the resonant frequency of 2.4 GHz. The surface current plot explains the radiation pattern of the CDRA. The directions of the surface current at the opposite ends of the feed are opposite to each other. This leads to the bidirectional radiation pattern of the CDRA. An artificial magnetic conductor (AMC) was designed to improve the radiation properties of the CDRA. The following section describes the design procedure of the AMC.

Electromagnetic bandgap (EBG) structures have gained increasing attention in many antenna designs in recent years (Ashyap et al., 2020; Dewan et al., 2017a). The remarkable characteristics, which include high-impedance surfaces within a restricted frequency range and restricting the propagation of surface waves within the bandgap frequency range, are the cause of their popularity. Photonic bandgap structures, which are realized in periodic configurations, serve as the foundation for the performance of EBGs. Applications of EBG structures reveal a variety of intriguing features (Dewan et al., 2017a). It is commonly known that the original current and the image current produced by a perfect magnetic conductor (PMC) are in parallel. Over the intended frequency range, AMC structures exhibit behavior like PMCs with an in-phase reflection coefficient (Dewan et al., 2017b). These reflected waves, which are generated by the AMC, constructively interfere with the source wave. The combined effect from both the reflected wave and source wave improves the radiation efficiency and gain of the antenna (Kumar et al., 2020). Periodic patches printed with or without vias on a grounded substrate make up AMCs.

In this work, a beehive-shaped metasurface is designed to enhance the radiation performance of the designed CDRA. This metasurface, shown in Figure 7A, was designed on a FR-4 substrate of thickness = 3.2 mm. The back of the AMC is a full ground plane. The ground plane and the beehive structure are made of copper. The designed procedures start with selecting the unit cell geometry on the requisite material.

A hexagon beehive-shaped structure is chosen as the design for AMC, as these are compact. Because they have less space between the sides of the hexagon (unit cell of metasurface), the resultant magnetic coupling is higher than the rest of the split-ring resonator (SRR) structures. Figure 7B shows the unit cell and the 3 × 3 AMC structure designed in this work. Table 2 describes the physical dimensions associated with the designed AMC.

The reflection phase vs. frequency plot (Figure 7C) shows that the designed AMC generates reflected waves that are similar in the direction of the original current in the target operating band of 2.4–2.5 GHz. The bandwidth of AMC is from −90 ° (2.3 GHz) to +90 ° (2.6 GHz), with 0 ° as the designated resonant frequency (2.48 GHz).

Figure 8A shows the perspective view of the designed CDRA backed by the designed beehive-shaped AMC. Figure 8B shows the simulated return loss vs. frequency plot. As is evident from Figure 8B, the CDRA + AMC operated in the ISM band, resonating at 2.584 GHz.

Figure 8C shows the simulated directivity pattern of the designed CDRA backed by the designed AMC. From the pattern, we can infer that after adding the AMC to the CDRA, the whole structure showed improved radiation characteristics. The AMC reduces the back scattering of the radiation and, hence, improves the front-to-back ratio. The directivity was improved to 7.69 dBi.

The proposed antenna is fabricated and experimentally tested for validation of its radiation properties, such as impedance bandwidth and gain, so that it can be successfully used in blood glucose monitoring applications.

The Debye model provides a precise and efficient depiction of biological tissues across a broad frequency range. Recently, this model has been employed to simplify the analysis of experimental data acquired from diverse human tissues such as the brain, fat, breast, skin, bone, liver, and others The Debye Equation 5 is a mathematical model that describes the frequency-dependent behavior of complex impedance in biological tissues (Karacolak et al., 2013). It is commonly used to represent the electrical properties of tissues, considering their capacitive and resistive characteristics. Z ω = R   ·   1 + j Q   ·   ω τ . (5)

In this equation, Z(ω) represents the complex impedance at a given angular frequency ω, R is the DC resistance, Q is the Debye parameter, which depends on the relaxation time, τ is the characteristic relaxation time, and α is the shape parameter that determines how the relaxation time must be distributed. This equation allows for a more accurate representation of the impedance response of biological tissues over a wide range of frequencies, providing insights into their electrical behavior. The equation is useful in various biomedical applications, such as bioimpedance measurements, tissue characterization, and modeling of electrical stimulation.

Karacolak et al. (2013) use the Cole–Cole model for dielectric properties of blood plasma, which are dependent on glucose levels for CBGM. They show a correlation between electrical properties (relative permittivity, ε r , and conductivity, σ) of blood plasma and plasma glucose concentration. This model provides a mathematical framework that describes the frequency-dependent behavior of the complex impedance of the blood plasma. This model is applied to characterize the relationship between glucose concentration and the dielectric properties of blood plasma. It allows for the estimation of glucose levels based on the measured impedance at various frequencies.

In this study, it is assumed that other minerals present in the blood plasma, such as calcium, chloride, potassium, and magnesium, have negligible impact on the electrical properties (Karacolak et al., 2013).

It is evident from the study that the dielectric constant and the conductivity are inversely related to the glucose concentration of the sample. This difference is more apparent in the higher frequency range.

In the proposed research work, the dielectric parameters, namely, the dielectric constant ( ε ′ and dielectric loss ( ε ″ or tan ⁡ δ , are derived from the complex dielectric Debye equation using Equation 6 (Karacolak et al., 2013): ε * = ε ′ − ε ″ = ε ∞ + ε s − ε ∞ 1 + j ω τ + σ s j ω ε 0 , (6)

Where ε ∞ is permittivity at infinite frequency, ε ∞ is static permittivity, ω is angular frequency, τ is relaxation time, σ s is static conductance, and ε 0 is permittivity of free space.

Static conductance is omitted for materials with low conductivity. Equations Equations 7–9 are the equations for calculating ε ∞ , ε s , and τ . ε ∞ = a n X 2 + b n X + c n , (7) ε s = a n X 2 + b n X + c n , (8) τ = a n X 2 + b n X + c n , (9)

Where X is glucose concentration in mg/dL. Table 3 lists the different coefficients for Equations 7–9.

Table 4 shows the dielectric parameters calculated for glucose levels of 80 mg/dL (normal), 110 mg/dL, 130 mg/dL, 150 mg/dL, and 170 mg/dL (hyper-glycemic). Table 4 shows ε ′ and ε ″ for the above-mentioned glucose levels at the resonant frequency of 2.4 GHz .

A human phantom with skin, fat, and blood tissue layers was then designed using CST software for in-silico analysis.

The proposed BGL monitoring purpose uses the CDRA near the bio mimic of human tissues. For an RF device to be used in such proximity to the human body, a specific absorption rate of less than or equal to 1.6 W/kg per 1 g of human tissue must be reported to ensure its safety for human exposure (Kaur K. and Kaur A., 2022).

Equation 10 below gives the expression for SAR as SAR = E 2 × σ ρ , (10)

Where ρ is the volume density of the tissue in Kg, σ is the conductivity of the bio mimic of human forearm tissues in S/m, and E is the measure of the electric field incident on the phantom.

Table 5 discusses the dielectric parameters of various human tissues at 2.4 GHz. A human phantom consisting of these human tissues (skin, fat, and blood) was designed as shown in Figure 11A. Figure 11B shows the placement of the designed CDRA near the bio phantom. Figure 11C shows the SAR plot of the human phantom at the resonant frequency of 2.4 GHz when the designed CDRA is fed 1 mW of power.

The CDRA sensor was then also simulated for better validation using the forearm of the bio model, named “Hugo,” provided by the CST MWS. “Hugo” is a 38-year-old male model consisting of 31 human-mimicking tissues. Figure 12 shows the placement of the antenna with respect to the bio model and its corresponding simulated SAR results.

After the SAR analysis, the CDRA was then simulated with bio phantoms with different blood glucose values. Figure 13 shows the simulated S₁₁ results when the dielectric parameters of the blood layer were changed according to the glucose level as calculated in Section 7 using the Cole–Cole model. Table 6 shows the simulated S₁₁ values for different BGLs. Based on these values, the sensitivity of the CDRA as a BGL monitoring device was calculated as 4.5 × 10 − 3 dB/mg/dL.

A novel RF structure consisting of a cylindrical dielectric resonator antenna (CDRA) and an artificial magnetic conductor (AMC) operating in the ISM band (2.4–2.5 GHz) was designed in this literature. This structure can be used in various biomedical applications, such as monitoring blood glucose levels. The overall structure showed a gain of 7.7 dBi with an impedance bandwidth of 330 MHz. The antenna was tested with the “Hugo” bio model (38-year-old male) provided by CST MWS, and SAR was simulated. The antenna showed a permissible value of SAR of 0.036 W/kg. For BGL monitoring, the antenna is simulated and analyzed for different BGL values. For this purpose, the Debye model is referred to for extracting the dielectric properties of human blood for different glucose levels. The proposed BGL monitoring structure showed a sensitivity of 4.5 × 10 − 3 dB/mg/dL. The proposed sensor has a simple design and is quite economical, which makes it a good choice for a preliminary monitoring device for hyperglycemia.