Abstract
This paper presents the design of a novel fabric-based multi-band microstrip antenna in mm-wave frequencies for wearable applications. The reference patch antenna was etched on a flexible polytetrafluoroethylene (PTFE) fabric substrate with an overall dimension of 18 mm × 18 mm × 0.6 mm and optimized the patch geometry using a binary-coded genetic algorithm. The algorithm iteratively creates a new shape of the path surface, evaluates the cost function, and returns the best-fitted geometry based on the formulated fitness function. The free space and on-body simulation of the best-fitted antenna performance parameter was investigated and analyzed. In free space, the proposed antenna is resonant at five distinct frequencies: 27.8 GHz, 30.3 GHz, 40.1 GHz, 47.2 GHz, and 56.7 GHz. The antenna achieves a wide bandwidth of 0.69, 2.32, 2.22, 1.76, and 8.11 GHz and an improved broadside directivity of 10.3, 8.5, 7.8, 9.6, and 8.9 dB in free space, respectively. For on-body analysis, the antenna was simulated using a three-layer human body phantom model at three distinct distances. The gain and radiation efficiency were significantly reduced when the antenna was close to the phantom model and gradually enhanced as the gap increased. Moreover, the antenna performances were evaluated and compared by using four additional fabric substrates. Because of its excellent on-body performance with flexible textile-based substrates, the optimized antenna is a suitable candidate for multi-band body-centric communications.
1. Introduction
In recent years, there have been a number of investigations on wireless monitoring of human attributes. Several services are anticipated from wearable devices addressing the ever-growing customer requirements. Designers and researchers in the field of wearable antennas continue exploiting antenna technology to meet client needs. Globally, extensive researches have undertaken towards innovative and tiny wearable antennas capable of wirelessly communicating human body parameters to a nearby accessible network. As wearable antenna technology advances, new and diverse applications such as medical monitoring, smart diagnosis, aging care, battle-field personal care, astronaut monitoring, child protection, and location-based services have emerged [1]. Such devices are used in sports medicine to track the characteristics of athletes in training. Wearable antennas are often employed in hazardous areas to manage the work environment [2]. The wearable antenna in a well-designed system has appropriate operating bandwidth, acceptable radiation characteristics for human tissue, and high enough gain and directivity to suit the requirements. The antenna, on the other hand, must be light, flexible, compact in size, inexpensive, resistant to damage, and comfortable to wear. Microstrip antennas are an attractive option for mm-wave multi-band wearable antennas. Even though the microstrip antenna has low gain and narrow bandwidth with some other deficiencies in its performance, the antenna is advantageous over other antennas due to its low profile, easy manufacturing, and planar structure.
The dynamic development of wireless communication derives coexisted multiple standards that increase the demand for multi-functional antennas. Because of the large bandwidth available at mm-wave frequency, numerous applications such as wireless networks, mobile communication, internet of things, wireless human monitoring systems, and machine-to-machine communications will be facilitated and become more prevalent in the band for the next-generation technologies. In the case of human monitoring systems, networks in the band are critical because they allow data transmission over small ranges of cellular systems. The mm-wave band is part of the radio frequency spectrum between 30 and 300 GHz, corresponding to a free space wavelength ranging from 10 to 1 mm. As the photon energy is insufficient to take an electron from an atom, the radiation from this band is not ionized. Instead, the absorption of electromagnetic mm-wave energy by tissues results in the primary physiologic consequence of heat [3]. Thus, antennas in mm-wave for a body-centric network should be optimized according to the rules and guidelines of specific absorption rate (SAR) to protect the human tissue from injuries.
FCC allocates the following mm-wave bands for wireless applications: 24–40 GHz for 5G cellular communication, 33.4–36 GHz for vehicle speed detection, 57–59 GHz for meteorological applications, and 57–66 GHz for WPAN [4, 5]. Although mm-wave signals face several propagation obstacles, multi-band antennas for wearable devices operating at these frequencies are critical for enhancing communication capabilities with a variety of wireless protocols. The combination of automated optimization algorithms and electromagnetic analysis tools was employed to meet the complex design criteria of the wearable multi-band microstrip antennas at mm-wave frequency. Several approaches and optimization algorithms have been used in the literature to improve multi-band and performance of microstrip antennas. Employing slits and slots on the antenna structure was presented in [6]. Loading a metamaterial structure as part of the microstrip antenna is also reported [7, 9].
An antenna designed using theoretically derived parameters does not provide sufficient performance to meet real-world communication requirements. It is a complex and time-consuming procedure to find the parameters of any antenna using simple intuition, experience, and practical measurements. Desired performance, such as multi-band operation, adequate bandwidth, and good gain, can be achieved by tuning optimum design parameters and geometries using autonomous algorithms. Optimization algorithms such as a genetic algorithm (GA) [10], ant colony optimization [11], simulated annealing algorithm, particle swarm optimization [12], differential evolution [13], bacteria foraging optimization [14], and many others can be used to optimize the design parameters of the antenna. In electromagnetic problems, a far-field simulation is required to estimate the cost function, which is time-consuming. Considering a limited computing budget, algorithms that do not require derivatives knowledge of the objective function are advantageous. Other considerations to consider while choosing a better algorithm are computational effectiveness, solution quality, convergence, and optimization accuracy. In antenna optimization, the genetic algorithm surpasses the other methods in terms of intended outputs [15].
This paper presents a wearable multi-band mm-wave microstrip antenna using a binary-coded genetic algorithm for the wireless human monitoring system. For operational reasons, antennas of this type should use fabric substrates for easy integration with clothes [16]. Simultaneously, the bandwidth of such an antenna should be as broad as possible to allow for abundant data transfer in the link in all operating bands. The radiating patch shape was divided into smaller rectangular cells and assigned a binary “0” for nonconducting cells and a binary “1” for conducting cells on the patch surface to attain optimization. The iterative “on” and “off” features of radiating surface always try a new patch geometry model and evaluate the fitness function until the fit individual appears [17]. The paper uses MATLAB and HFSS software in combination for iterative simulation. The MATLAB software was used to generate the VBS files with the modeled antenna details. HFSS calls the VBS file, simulates the antenna, and exports the resulting performance to the genetic algorithm optimizer into the primary function in MATLAB.
2. Genetic Algorithm (GA)
A genetic algorithm is an evolutionary search algorithm based on the heredity of the individual. The algorithm evaluates the cost function and selects the best-fitted individuals for reproduction. The first generation of the population produces randomly, and the rest generations generate through reproduction. The different operators of GA are responsible for the optimum design of the antenna through selection based on the formulated suitable fitness function. Crossover, mutation, and selection are the genetic operators that allow reproduction. The crossover operator reproduces a diverse population. Mutation adds a degree of diversity to the population to protect it from premature convergence [18]. In antenna optimization, GA employed to improve the directivity [19], multi-functionality [20], bandwidth [21], and size reduction [22] of antenna. It optimized the antenna’s dimensions, the feeding point’s position, the position and size of slots on antenna structure, and the position of shorting pins for appropriate propagation and resonance.
When using GA for antenna optimization to meet a specific performance goal, the optimizer and antenna structure code must be correctly programmed and interfaced. When the structure code is executed, a new antenna model iteratively appears with its recent performance in the optimization process. In each iteration, the GA optimizer begins to evaluate the fitness of each new antenna geometry, and lastly, it provides the best-fitted model with the corresponding performance parameters. The integration of GA computing code for antenna optimization using MATLAB and HFSS interface is presented in Figure 1.
3. Antenna Configuration
An appropriate wearable antenna design should be considered to minimize losses in wearable scenarios. The geometric configuration of the reference rectangular microstrip antenna suitable for GA optimization is illustrated in Figure 2. The antenna’s radiating patch and ground plane are composed of a 0.035 mm thick perfect electric conductor (PEC). A 0.6 mm thick fabric polytetrafluoroethylene (PTFE) substrate with a dielectric constant of 2.05 and loss tangent of 0.0017 is employed. It is advantageous to minimize the surface wave losses by utilizing a substrate with a low dielectric constant, which simultaneously enhances impedance bandwidth [23]. The overall dimension of the textile substrate is . The antenna’s radiating patch is in size and is fed by a 50 microstrip line that is 3.6 mm long and 0.9 mm wide. Besides the standard procedures in [24] used to calculate antenna dimensions, the parametric analysis was utilized to articulate the patch dimensions in the interest of appropriate reference antenna performance at 39.5 GHz center frequency [25]. The antenna was modeled and simulated using High-Frequency Structure Simulator (HFSS) software. The dimensions of the reference model are summarized in Table 1.
4. Optimization Procedure
In this article, GA optimization is used to optimize the patch geometry. In the proposed solution space, the radiating patch geometry has a variety of shapes, and only a few of these configurations perform well. The GA is greedy since it continually seeks the best-fitting individual to ensure the continuation of the generation. In this optimization problem, the surface of the patch is gridded into 8 × 8 small rectangular cells having the size of as shown in Figure 3. When using a binary-coded genetic algorithm, it is evident that unique candidates are available in the solution space. Every rectangular cell on the surface has its corresponding conducting and nonconducting properties. The conducting cells are assigned by gene “1”; on the other hand, the nonconducting property of the cell is represented by gene “0.” Each candidate in the problem space has a distinctive value of fitness value. During the “on” and “off” stages of the cell, vertical overlapping was employed to protect the infinitesimal connection between cells on the patch surface [26]. The geometry of the patch surface is varied in each iteration, resulting in a new antenna with a new fitness value for each. Therefore, the performance of the antenna is updated.
The crucial role in the GA optimization process is the formulation of a good fitness function while considering the desired performance criteria. This work formulated the fitness function considering all the performance factors and their target values. The primary goal is to develop a multi-band mm-wave microstrip antenna that can operate at five unique frequencies with broad bandwidth for wearable devices. As a result, the fitness function is designed to minimize the reflection coefficient and increase the antenna’s working bandwidth. The maximum bandwidth in a single band is fixed to 8 GHz to reduce the computational time while avoiding premature convergence. The weighting coefficient will modify the predicted bandwidth in every five distinct bands: 27.5–28.5 GHz, 29.0–31.5 GHz, 38.5–40.5 GHz, 46.5–48.0 GHz, and 54.0–62.0 GHz. The formulated fitness function is presented in the following equations:where is designed asand is designed aswhere N is the number of sampling frequencies in a given band and M is the number of the operating band. The sampling frequency is in the 50 MHz used with the analysis sweep, and is the reflection coefficient value in dB at each sampling frequency. assigns the bandwidth weighting factor, and Bw (k) is bandwidth at the band in GHz.
The patch surface of the reference model is optimized using the following algorithm configurations. The total number of population clusters in a particular generation is 15 individuals. In a population, a single chromosome is represented by 64 genes. Single-point crossover with a probability of 0.8 and single-bit mutation with a mutation rate of 0.01 were used. Throughout 250 generations, the selection operator employed the tournament selection approach.
5. Results and Discussion
The optimization and simulation of the proposed pentaband antenna were carried out utilizing commercially available electromagnetic simulation software HFSS in combination with MATLAB. The geometry of the shape was optimized by providing “on” and “off” states for the rectangular cells on the patch surface until the final structure was obtained using a binary-coded genetic algorithm. The total computing performance of the multi-iterative optimization approach is influenced by the time required for electromagnetic simulation. GA returns the most robust solution out of the solution space in only 4.3 days at the 250 generations using core I7, 8 GB RAM, and 2.7 GHz processor speed, as shown in Figure 4(a). If we think for a second for each aspirant solution, the computation time would be years. Hence, GA started to converge after 103 generations, as shown in Figure 4(b).
(a)
(b)
5.1. Free Space Simulation
The optimized antenna with a fabric PTFE substrate was simulated in free space. The simulated curve for both the reference and optimized antennas is presented in Figure 5. The figure demonstrated that the reference model resonated only at a single frequency of 39.5 GHz, with a minimum peak value of −20.6 dB. However, the optimized antenna resonates at five distinct bands with center frequencies of 27.8 GHz, 30.3 GHz, 40.1 GHz, 47.2 GHz, and 56.7 GHz and peaks values of −34.5 dB, −20.1 dB, −35.2 dB, −22.4 dB, and −25.8 dB, respectively. The VSWR was determined to be less than 2 for all five operational bands, indicating that the proposed antenna was well matched to a 50 microstrip line that transmits the maximum power between the feedline and antenna.
The bandwidth of the reference model is 0.9 GHz since it operates in the range of 39.0–39.9 GHz when < −10 dB is considered. However, the optimized antenna achieved a better bandwidth in all five operating bands: 0.69 GHz (27.43–28.12 GHz), 2.32 GHz (28.94–31.26 GHz), 2.22 GHz (39.27–41.49 GHz), 1.76 GHz (46.44–48.2 GHz), and 8.92 GHz (53.83–61.94 GHz). The total bandwidth of the antenna is above 15.1 GHz in the operating bands. The bandwidth improvement was made by employing GA optimization.
The 3D gain pattern of the optimized antenna at all five resonant frequencies is presented in Figure 6. The antenna has achieved gain values of 10.2 dB, 8.4 dB, 7.7 dB, 9.5 dB, and 8.4 dB at resonating frequencies of 27.8 GHz, 30.3 GHz, 40.1 GHz, 47.2 GHz, and 56.7 GHz, respectively. As seen from the 3D pattern, they are almost in the broadside direction with slight distortions in higher frequency bands.
(a)
(b)
(c)
(d)
(e)
5.2. Simulation of the Optimized Antenna on Other Fabric Substrates
For further analysis and investigation of the improved antenna’s performance, the free space simulation was carried out by substituting the PTFE substrate with various textiles. Polyester, silk, jeans, and cotton fabrics were used as substrates with dimensions similar to PTFE. The dielectric constant and tangent loss parameters of fabric substrates are described in Table 2. Figure 7 displays the simulation result of their values. The results indicate that the optimized antenna operates in five frequency bands for all fabric substrates. When the dielectric constant decreases, the resonant frequency slightly shifts towards a higher frequency, and the total operating bandwidth of the antenna increases. The cotton-based optimized antenna has a bandwidth of 3 GHz more than the PTFE antenna. Polyester substrate performs closely to the initial PTFE optimized antenna due to the closer dielectric constants. According to simulation results, all of the evaluated substrates exhibited a gain of higher than 8 dB in each band. Table 3 summarizes the free space performance study of the improved antenna with various fabric substrates and the performance of the reference model.
Figures 8 and 9 depict the optimized antenna’s gain pattern in E and H plane with various fabric substrates, respectively. The gain patterns of the E and H planes exhibit only slight differences. The patterns guided in the broadside direction show a significant distortion at a few angles, particularly at higher frequencies.
(a)
(b)
(c)
(d)
(e)
(a)
(b)
(c)
(d)
(e)
Figures 10(a)–10(e) depict surface current distributions at resonance frequencies to highlight the electromagnetic radiation properties of the optimized antenna in the operating bands. When the proposed antenna operates at resonance frequencies, a significant surface current density was observed at the junction of the feedline and the patch, as well as a moderate current distribution over the surface of each active rectangular cell.
(a)
(b)
(c)
(d)
(e)
5.3. On-Body Simulation
In the case of wearable contrivances, the antenna should be simulated conjointly with the proximity of the human body model. The human body model should be discretized into small resolution during the joint simulation. The computational time and memory requirement are high when the model’s size is much larger than that of the operating wavelength. As mm-wave signals cannot penetrate human skin more than 0.5 mm, most research studies employ a three-layer phantom representation [26–28]. The heterogeneous and lossy nature of the human body has a direct influence on the antenna performance when it works nearby. A three-layered phantom that consists of muscle, fat, and skin is designed in HFSS, as shown in Figure 11(a). At 40 GHz, the average permittivity and conductivity of these layers are specified for muscle ( = 52.79; tan = 1.705), fat ( = 5.28; tan = 0.1), and skin ( = 31.29; tan = 5.0138) [28]. In this model, the thickness of muscle, fat, and skin layers are 5 mm, 3 mm, and 2 mm, respectively. The proposed optimized wearable antenna is fixed on a flat body phantom to study its radiation characteristics in three proximity levels to the human body. The antenna was positioned at three different distances from the phantom to analyze the antenna performance in wearable scenarios. Starting from , the antenna gradually moved away to from the phantom. Figures 11(b) show the antenna’s position at variable proximity to the model.
(a)
(b)
Figure 12 shows the performance curves of the optimized PTFE substrate antenna corresponding to various proximity placements from the human body model. When the antenna is placed 2 mm, 5 mm, and 10 mm away from the phantom, the center frequency shifts towards lower, as shown in the graph. At the same time, the antenna resonates at five distinct frequencies in each level of proximity. The shifting was not noticeable when the antenna was close to free space at 10 mm proximity. The bandwidth of the optimized antenna was highly affected at placement, resulting in about 10.9 GHz over all pentaband operations, which is 28% lower than the free space bandwidth coverage.
The gain and radiation efficiency were adversely affected when the antenna operated close to the human body model. When the gap between the antenna and the phantom model gradually increases, radiation efficiency and gain also increase. At a distance of 2 mm, the minimum gain value was observed, which is 4.7 dB. The maximum gain in the on-body simulation was 8.4 dB at 10 mm proximity. However, this value was substantially lower when compared to the free space findings due to the lossy nature of human tissues. Table 4 summarizes the results of the on-body simulation of the optimized antenna with a PTFE substrate. The gain patterns of the antenna in the E-plane and H-plane for various distances are shown in Figures 13 and 14, respectively. The radiation patterns do not change much with multiple antenna placements except for power level variations.
(a)
(b)
(c)
(d)
(e)
(a)
(b)
(c)
(d)
(e)
5.4. Comparison with Related Works
Table 5 summarizes the results of genetically improved antennas and some other similar research studies. The originality of this article has been recognized in terms of its pentaband operation and other performance metrics. For example, in reference [16], a wearable UWB antenna for mm-wave devices was proposed. The antenna was only capable of operating in a single band with a bandwidth of 12.1 GHz. However, this genetically modified wearable antenna can operate in five frequency bands with a total bandwidth of 15.1 GHz. As seen from the comparison table the highest gain was reported by [20], which is 8.7 dB. Nevertheless, the proposed antenna has a maximum gain of 10.2, which indicates its superiority. Aside from the fact that the improved antenna utilizes a flexible textile-based substrate, it also outperforms related works in both free space and on-body simulations, making the optimized antenna a strong choice for wearable devices.
6. Conclusion
This work aimed to optimize and analyze a novel multi-band mm-wave textile-based antenna for wearable applications. A binary-coded genetic algorithm was utilized to optimize the reference patch geometry targeted to enhance the multi-band operation and widen the bandwidth. The improved antenna was simulated in free space and at various distances from a phantom human body model. In free space simulation, the antenna operates in five bands: 27.43–28.12 GHz, 28.94–31.26 GHz, 39.27–41.49 GHz, 46.44–48.2 GHz, and 53.83–61.94 GHz a total bandwidth of 15.11 GHz. The broadside directivity was 10.3, 8.5, 7.8, 9.6, and 8.9 dB in the respective bands. The gain was highly affected when the antenna was close to the human body model compared to free space findings. The antenna was also tested on various flexible textile substrates such as cotton, jeans, silk, and polyester. The finding shows that the antenna has adequate bandwidth coverage in all five distinct bands with sufficient gain. In the future, the performance of the antenna will be tested with moist substrates and different bending angles of the antenna. Due to a lack of adequate laboratory facilities, the paper has only provided simulations at this stage. Therefore, prototype fabrication and measurements will be central to our future work. In summary, the proposed antenna is suitable for multi-band mm-wave wearable devices, and a genetic algorithm based on binary-coded genetic programming could be used to improve the multi-functionality and impedance bandwidth of the antenna.
Data Availability
The data used to support the findings of this study are included within the article.
Conflicts of Interest
The authors declare that they have no conflicts of interest.