Microwave Journal

Simulation-Driven Virtual Prototyping of Smart Products (Full Article)

This is an extended version of our July 2021 cover feature

July 14, 2021

In the past few decades, the wireless industry has been driven by constant innovations and transformations due to introduction of wireless communication standards such as 4G LTE, 5G, Bluetooth, Wi-Fi and many more.1 These advancements in the wireless industry coupled with rapid manufacturing techniques requires advanced product design that involves complex multi-physics considerations. Competition in consumer electronics market calls for measures to improve product performance with lowering the development cost and at the same time significantly reducing the time-to-market. All these challenges can be addressed by simulation-driven virtual prototyping to reduce the physical testing.2-4 Moreover, the simulation-driven virtual prototyping can be employed for designing modern smart products in order to accelerate product development speed, ensure intrinsic product qualities and improve decision-making process during the development cycle. Therefore, the simulation-driven design can be one of the important factors contributing to product completeness and timely market launching of smart products.


I. Introduction

Figure 1 shows a product development process for a smart speaker which consists of a speaker component, Printed Circuit Board (PCB), assembled electric components, and the final product with a speaker cabinet. For the smart speaker’s development, a three step simulation-driven virtual prototyping methodology is applied in this article. The first step is to design, verify and analyze the PCB layout design, the second step is design and integration for the Bluetooth antenna embedded on the PCB and positioned inside the speaker cabinet, and the third step is the smart speaker’s wireless communication performance evaluation by considering neighboring wireless products.

     Speaker                            PCB                        Assembled electronics          Final product        


Figure 1. Product development process for a smart speaker.


This article is organized as follows.  In section II, PCB design concept and main functions for a smart speaker are introduced and explained. Subsequently, rule-based verification and signal integrity (SI)/thermal analysis methods are applied to upgrade the product’s electric performances such as noise-free audio qualities and high-speed data exchange. In section III, miniature techniques are utilized to design for a Bluetooth antenna on the PCB. Besides, the designed antenna is tuned for the integration on the PCB and then the position of the PCB inside a speaker cabinet is studied for optimal performances. In section IV, the smart speaker’s wireless communication performances are simulated to evaluate the quality of the product in realistic operating environment. Particularly, the Bluetooth’s coverage of the smart speaker is estimated by calculating network parameters (data rate and throughput) as well as interference effects due to the smart speaker with other wireless devices inside home environment. Section V summarizes with conclusions.


II. PCB Design, Verification and Analysis

 PCB design for smart speaker

The current generation of smart speakers receives audio signals wirelessly using radio frequency waves. One of the most popular radio frequency standards that supports audio transmission to speakers is – Bluetooth (BT).5 The basic building blocks of a smart speaker include the wireless section (mostly Bluetooth), the charging circuitry, audio amplifier for quality audio output, a user informative output (e.g. Liquid Crystal Display, LCD) and the main controller with the memory, providing a reliable connection between these blocks. Further additions are made to enhance the feature list. A block diagram representation of a basic smart speaker is shown in Figure 2. To understand the electrical operations of the smart speaker, power supply flow is drawn in red from the USB connector toward the other functional blocks. Similarly, the blue line and green line indicate the digital signal and audio signal exchange among the blocks, respectively.



Figure 2. Block diagram of a basic smart speaker.


When designing a high-quality smart speaker, aspects such as the quality of audio signal, Bluetooth antenna performance parameters and interference with other wireless signals must be taken into consideration. The PCB with the electronic circuitry including Bluetooth IC and the antenna is shown in Figure 3. This is a 6-layer PCB with dimensions 106 x 137 mm and a total thickness of 0.7 mm. Main functionalities of PCB are audio amplifiers for high quality audio output, memory IC for storing wireless information such as pairing details and battery status, and for smart applications, USB connectivity for charging or diagnosing, charging and power supply circuitry, Bluetooth for wireless connectivity, LCD driver and display module to provide more information to the users. Lastly, the microcontroller administers the synchronous functionality of all different parts of the PCB. 

Figure 3. Assembled PCB with IC packages: (a) Top view, (b) Bottom view.

Although there are specific ICs which make sure each of the functionalities are operational, there are certain other parts on the PCB which are equally important and decide the reliability of the system. Some of them, viz., the differential nets in audio lines, the connectivity between the controller and the memory IC namely, clock, address, command and data lines, the differential data lines from USB to the controller and the custom antenna for the Bluetooth module, are crucial in determining the quality and efficiency of the smart speaker. A 3-D view of the board is shown in Figure 4, highlighting some of the essential nets such as audio lines, USB lines, and memory lines.


Figure 4. 3-D view of the PCB.


The nets marked with ‘audio lines’ are a differential pair. To ensure superior audio output, the differential lines have to match a certain specification of differential pairing. Also, the lines must be shielded with electrical ground to avoid noise seeping into them from other sources. The ‘memory lines’ and ‘USB lines’ must follow their individual high-speed protocol for a seamless experience. Specifically, the high-speed lines should satisfy their required impedances based on their speed and frequency of operation. Moreover, the memory lines must follow their DDR (Double Data Rate) specification to avoid discrepancies in data transfer to and from the memory IC. Lastly, to avoid costly design iterations and over-design, and for the general safety of the device and the user, the PCB must be checked for its thermal behavior and correct thermal anomalies, prior to manufacturing.


PCB design verification and analysis

As discussed earlier, the layouts of the audio line, USB line, and memory lines need to be carefully designed for the smart speaker’s reliable operations. Therefore, to ensure the electric qualities related to the PCB layouts, PCB verification and analysis methods are applied. The target PCB layouts for the verification and analysis are the differential audio lines and the high-speed lines (USB interface and memory bus lines) exclusively shown in Figure 5. To verify the differential audio line’s layout, rule-based electric checker, design for electric (DFE) supported in Altair PollEx,6 is utilized.  Signal integrity (SI) analysis is conducted to evaluate the PCB layout for the high-speed lines with respect to transmitting and receiving digital signals’ waveforms and voltage/time margins. In addition to SI analysis, thermal characteristics of the PCB board are analyzed. The thermal analysis checks for excessive component temperature and uneven board temperature at an early design stage.

Figure 5. Exclusive view for PCB layouts to be verified and analyzed.


The differential audio lines whose width is 0.5 mm are evaluated with respect to the paired lines’ separation and coupling rate because two positive and negative lines should be tightly coupled within a specific distance. In Table 1, the separation criterion (center-to-center distance, 0.75 mm) is determined by adding the line width (0.5 mm) and the addition space (0.25 mm) and the coupling rate criterion (better than 80 percent) is determined by considering the structures of the USB IC and connector. From the verification solver, it is obtained that the maxim separation (0.893 mm) between the two lines and the coupling rate (76.6 %) does not satisfy the criteria. Additionally, the ground shield ratio showing how much the audio lines are shielded by a ground plane meets the criterion (over 80 percent). These results can be useful for a PCB designer to decide whether modification of the layout for the audio lines should be made or not. 


Table 1. DFE results for the differential audio line.


Separation (mm)

Coupling rate (%)

Ground shield ratio (%)


= 0.75

> 80

> 80






In Figure 6, the characteristics of the USB data lines (D+/D-) are studied by the waveforms at the transmitting and receiving ports. The ideal digital signal whose pulse width is 2.08 ns (Data rate: 480 Mbps) and whose peak voltage is 0.4 V is transmitted from the USB IC and received at the USB connector. By comparison of two signals, it is observed that the received signal has a time delay which accounts for the USB lines’ length. Furthermore, the receiving signal has enough voltage margins for the USB normal operation because the high threshold (0.3 V) and low threshold (0.1 V) of the USB 2.0 specification are satisfied. For more exact compliance test, eye diagram of the received signal should be drawn and estimated with the eye mask defined by the USB 2.0 specification.7


Figure 6. Signal waveforms for the USB data line: (a) Transmitted signals at USB IC, (b) Received signals at USB connector.


Similar to the previous USB data line analysis, the memory interface lines between the controller and the memory (Pseudo SRAM: Static Random Access Memory) are analyzed using the SI solver in PollEx and shown in Figure 7. For this analysis, one differential clock line and a group of address lines are selected from the PCB design. The controller is assumed to send the clock signal whose frequency is 133 MHz and the address signals whose data rate is 256 Mbps. The eye diagram is obtained at the memory ports. The clock signal is used as a criterion to estimate the interface with the eye mask whose high and low thresholds are 1.3 and 0.4 V, respectively. Additionally, it is assumed that the required setup time and hold time for the interface are 2 and 1.5 ns. From the eye diagram showing enough voltage margin, it can be concluded that the interface lines are well routed for the safe 256 Mbps data exchange between the controller and the memory.

 Figure 7. Eye diagram for the clock and address lines of the memory bus


Board-level thermal analysis is performed to check the main audio amplifier’s temperatures in operation by using the data sheet’s specification about the audio amp’s power rating at room temperature, 5 W and its package type, QFP (Quad Flat Package). Figure 8 shows the temperature contours in two convection conditions: one is natural convection, and the other is forced air convection whose air flow’s velocity is 5 m/s. In the condition of the natural convection, the highest temperature, 850 Celsius which is recorded below the amplifier is the maximum allowable value for the amplifier’s normal operation. As a remedy for the high temperature, the forced convection is proposed to decrease the temperature from 850 to 590 Celsius to ensure a reliable operation condition for the audio amplifier.

Figure 8. Thermal contour plots on the surface of PCB: (a) Natural convection, (b) Forced air convection (5 m/s). 

III.   Antenna Design and Integration

In this section, concept, design and placement of the Bluetooth antenna for the smart speaker is discussed. Bluetooth operates at frequencies between 2.402 and 2.480 GHz. The antenna used should have reflection coefficient of at least -10 dB in this frequency band. These parameters are crucial to validate the performance of antenna. A meander line antenna (MLA) proposed by Rashed and Tai8 to reduce the resonant length of the antenna is applied. Subsequently, the designed antenna was integrated on the PCB. The PCB is placed in its real working environment in the speaker assembly to identify the optimal location. This section demonstrates the electromagnetic characteristics of the antennas in different configurations given by changing the location and orientation of the PCB inside a speaker cabinet using commercially available electromagnetic simulation tool, Altair Feko.9 


Antenna design and integration on PCB

Meandering the antenna increases the path over which the surface current flows and that results in reducing the size of the antenna. The resonant frequency of a meander line antenna is a function of meander separation and meander spacing. If meander spacing is increased, the resonant frequency decreases. On the other hand, if the meander separation is increased, resonant frequency decreases.10 To validate the concept, we simulated a meander line antenna on a bare PCB as shown in Figure 9.




 Figure 9. Designed BT antenna in free space: (a) Meander line antenna’s shape, (b) Reflection coefficient.

FR4 is used as the PCB substrate in the simulation with relative permittivity 4.8 and dielectric loss tangent 0.017. Antenna is designed to operate at 2.4 GHz which is the desired Bluetooth frequency. As can be seen from Figure 9(b), the reflection coefficient of the antenna at 2.4 GHz is -17 dB which validates the design of the antenna. 


On a real PCB assembly, there are many components of various materials that can affect the antenna performance. To study the effect of these components, we embed the antenna module on the PCB designed and shown in Figure 3 that holds all electronics circuits. The PCB and its components are simplified to evaluate antenna performance. Figure 10 represents the antenna placed near the Bluetooth IC. Different electromagnetic material properties assigned to the various components on the PCB are shown in Table 2.


Figure 10. Antenna integrated on the PCB.


Table 2. Material properties of components on the PCB.

Material Name

Relative permittivity

Dielectric Loss tangent














After the antenna is placed on the PCB, it is observed that the resonant characteristics have changed. The reflection coefficient plot in Figure 11 shows that antenna would suffer significant losses due to mismatch as reflection coefficient around 2.44 GHz has reduced from -17 to -3.95 db. To restore the performance, a matching circuit to the antenna is designed and added. The components of the circuit are shown in Figure 12.

Figure 11. Reflection coefficient after antenna integration on the PCB.




Figure 12. Matching circuit.


Once the matching circuit is added, antenna performance is observed by plotting the reflection coefficient and radiation pattern that displays the values of realized gain as seen in Figure 13.  It is observed that the resonance occurs at 2.4 GHz and the value of reflection coefficient at the frequency is much improved. The reflection coefficient at 2.44 GHz is -25 dB. The antenna operates for a bandwidth of 250 MHz around the desired frequency. This means that the received signal will not suffer mismatch losses. The doughnut shaped radiation patterns similar to a dipole antenna’s validates our BT antenna design on the PCB.

Figure 13. Antenna characteristics after adding matching circuit: (a) Reflection coefficient, (b) 3D radiation pattern, (c) 2D radiation pattern.


Antenna Placement inside a Speaker cabinet

Now that we have integrated the antenna on the PCB and checked its performance, it is needed to perform an in-situ analysis where the antenna with the PCB to be placed  in its real environment inside a smart speaker cabinet shown in Figure 14. Antenna placement analysis helps in optimizing the location of the antenna inside its actual environment. It also aids in analyzing the overall performance of the product. Simulating such an environment can improve or ease the decision-making process and reduce costs significantly. The other components inside the speaker include metallic heat exchanger, fan for cooling, acoustic port and speaker module. The speaker cabinet chosen is made of balsa wood that has dielectric constant of 1.3. The fan for cooling is made of Teflon. The length and the breadth of the speaker cabinet is 355 and 305 mm while the height is 450 mm.


When selecting the location and orientation to mount the PCB with the antenna, several factors like spacing inside the cabinet, mount support options and thermal efficiency can limit the number of iterations to be simulated. We simulated the performance of the antenna by changing the location and orientation to make a best possible location that considers some practical constraints. The parametric sweep option in Feko is used to automate the simulation process for all possible locations.


Location 1 includes the PCB antenna placed near the back wall of the Bluetooth speaker in such a way that the antenna is facing toward the back side of the speaker as shown in Figure 14 (a). Location 2 and 3 include the PCB mounted on the same back wall, facing inward, but the antenna is located on the top and bottom side respectively as shown in Figure 14 (b) and (c). Location 4 and 5 include the PCB mounted in the free space between the components, facing inward and close to the back wall and the front wall respectively as shown in Figure 14 (d) and (e).

Fig. 14. Antenna placement: (a) Location 1, (b) Location 2, (c) Location 3, (d) Location 4, (e) Location 5.


The reflection coefficients from all other iterations are plotted on the same cartesian graph as shown in Figure 15 to make a decision on the best PCB location. Except for location 1, reflection coefficient is better than -10 dB for all other locations.

Figure 15.  Comparison of reflection coefficients: Location 1, 2, 3, 4 and 5.


The 3D radiation patterns of locations 2, 3, 4 and 5 are shown in Figure 16 and respective 2D polar plots for both principal planes are shown in Figure 17.

Figure 16. 3D Radiation patterns: (a) Location 2, (b) Location 3, (c) Location 4, (d) Location 5.


Figure 17.  Comparison plot of total realized gain for locations 2, 3, 4 and 5: (a) theta = 90,  (b)  phi = 0, (c) phi = 90.

As can be seen from the simulations, location 1 can be discarded due to poor reflection coefficient performance. Similarly, for location 4, radiation pattern appears to be focused on one side of the speaker with very low gain values in other directions. For optimal performance, the antenna should have uniform gain in all directions. Locations 2 and 3 are suitable for the placement of the antenna for optimal performance. For further analysis such as coverage and interference we select Location 2 as the best position.


IV.       Wireless Coverage and Interference for Smart Products

With the advent of 5G and Internet of Things (IoT), the trend is towards smart household electronics, including speakers. These smart devices operate on several technologies, like Wi-Fi, Bluetooth, LTE, ZigBee, etc. for various applications. Because of the congested spectrum, some of these technologies operate within closely separated frequency bands, leading to interference.11 Especially, the Bluetooth and Wi-Fi co-exist and operate around 2.4 GHz and the interference from Bluetooth affects the Wi-Fi throughput and vice versa. Therefore, it is important to analyze the interference levels through virtual prototyping at the early design stages to avoid costly rework after the final design. The effect of interference from Wi-Fi on the Bluetooth speaker is evaluated inside a multi-story residential building using Altair’s wireless propagation and radio network planning software, WinProp,9 as shown in Figure 18.


Figure 18. Multi-story residential building: (a) Complete house, (b) Cross-sectional view.


For an accurate analysis, the residential building is modelled as a detailed multi-story design with thick walls, flooring, staircase, fireplace, cabinets, doors, windows and roof. The Bluetooth speaker is placed in one corner of the living room while the Wi-Fi router is in the corner of an adjacent room, as shown in Figure 18 (b). The speaker is operating on the latest Bluetooth 5 technology12 while three different Wi-Fi technologies, namely the 802.11b, 802.11n and 802.11ax [13, 14, 15] are considered for the router.


Bluetooth Coverage

The Bluetooth speaker has a near omni coverage along the horizon, as depicted in the previous section. This section describes the RF channel characteristics of the speaker inside the building when operating on the Bluetooth 5 technology. This technology supports a maximum data rate of 3 Mbps for Up Link (UP) and Down Link (DL) through the enhanced data rate transmission mode, while the high speed mode supports 2 Mbps and the long-range modes support 500 Kbps and 125 Kbps in UL and DL. The results from the network analysis, as shown in Figure 19, illustrates that the enhanced data rate transmission mode is serving the entire house except for a small area in the corner on the opposite side of the speaker location. The Bluetooth technology is a packet-based protocol with a master/slave architecture, with one master communicating with up to seven slaves in a piconet. Therefore, for this CDMA based technology, the maximum number of codes available for user on one carrier are seven. Ideally, these codes will be perfectly orthogonal to each other, however, in a real-life environment like a residential building, a reasonable 0.8 orthogonality factor is considered. This results in a maximum achievable DL throughput of 19 MBps, as shown in Figure 19 (b).





Figure 19. Bluetooth network parameters: (a) Maximum DL data rate, (b) Maximum achievable throughput in DL.


Interference between Bluetooth 5 and 802.11n

As 802.11n supports Multiple Input Multiple Output (MIMO) systems, a router with two antennas is used for the analysis where each antenna carries one data stream in a 2x2 MIMO scenario. The radiation patterns of the router antennas at the 2412 MHz carrier frequency are showed in Figure 20.

Figure 20. Radiation patterns of the two antennas for the 2x2 MIMO router at 2412 MHz.


The router antennas are well matched for the 2.4 GHz Wi-Fi frequency bands, as shown in Figure 21 (a). For MIMO, it is not only required for the antennas to be well matched at the carrier frequency but also well isolated to avoid interstream interference between the data streams. It is clearly illustrated in Fig. 21 (a) that the two antennas have good isolation (~25 dB isolation), but a true indication of the independent behavior of the antennas is the Envelope Correlation Coefficient (ECC), as shown in Figure 21 (b). An ECC value of 0.5 is ok, higher than 0.5 is considered bad, and 0.3 or less is considered pretty good for MIMO applications.



Figure 21. Matching and isolation plots of Wi-Fi router: (a) S-parameter data and (b) Envelope correlation coefficient plot.


The 802.11n being an OFDM technology using Time Division Duplex (TDD) separation, the maximum achievable throughput will be the same as the maximum achievable data rate. From the Figure 22, it is clear that having the access point in the corner of the room next to the living room will provide good coverage for most part of the house.

Figure 22. Maximum achievable throughput with the 802.11n Wi-Fi technology inside the residential building.


The Wi-Fi is operating on the 2412 MHz carrier with the Bluetooth on 2442 MHz. Apart from the two carriers being close to each other, they are also physically located closely inside the house. This leads to a drop in the Bluetooth throughput because of the leakage from Wi-Fi into the Bluetooth frequency band, especially in the areas close to the Wi-Fi router, as shown in Figure 23.


Figure 23. Effect of Wi-Fi on the Bluetooth throughput. White space near the Wi-Fi router indicates the drop in Bluetooth coverage due to the interference.


The interference from Wi-Fi on the Bluetooth can be mitigated by considering additional filtering in the Bluetooth module. When the interference due to leakage from Wi-Fi is attenuated by additional 20 dB with better filtering, the Bluetooth throughput is improved reasonably, as shown in Figure 24.


Figure 24. Bluetooth throughput with additional attenuation of leakage from Wi-Fi.


Interference between Bluetooth 5 and 802.11ax

The 802.11ax technology is introduced to provide higher data rates than its predecessor 802.11ac and, as such, supports several transmission modes like BPSK, QPSK. 16 QAM, 64 QAM and 256 QAM. The 802.11ax operates in the 5 GHz frequency band and supports up to 8x8 MIMO. However, there are not many user equipment supporting eight data streams and as such a four-antenna router is chosen for this study, as shown in Figure 25.



Figure 25. Four antenna Wi-Fi router for 802.11az protocol.


With the S11 being less than -10 dB, all the four antennas of the router are well matched over the entire 5 GHz Wi-Fi frequency bands. Not only matching, these antennas exhibit good isolation between them for MIMO operation, as shown in Figure 26.


Figure 26. S-parameters of the four-antenna router.


The 5 GHz band supports higher bandwidths that help in achieving the higher data rates. This is evident from the increased maximum achievable throughput from 18 MBps for 802.11n (Figuer 23) to 700 MBps as shown in Figure 27(a). However, these high frequency signals are attenuated much more than the 2.4 GHz signals used for 802.11n because of the shorter wavelength. As a result, the Wi-Fi router with 802.11ax technology can provide the coverage only for a section of the house close to the access point as seen in Figuer 27 (a). The router in this case is operating on the 5530 MHz carrier with the four antennas supporting four individual data streams in 4x4 MIMO scenario. From this analysis, it is evident that while higher data rates are achieved using the 5 GHz frequency bands, this comes at the expense of limited coverage. As such, one could use the 2.4 GHz frequency bands for broader coverage while using the 5 GHz bands for better speed, which is common in the present-day Wi-Fi access points. One could also use repeaters when operating on the 5 GHz bands to provide a broader coverage, as illustrated in Figuer 27(b).





Figure 27. Maximum achievable DL throughput of 802.11 ax inside the house: (a) Without repeater, (b) With repeater.


Another advantage of using the 5 GHz frequency bands for the Wi-Fi is its non-interference with the 2.4 GHz Bluetooth, as shown in Figure 28. As illustrated, the Wi-Fi access point has minimal effect on the Bluetooth throughput.



Figure 28. Bluetooth throughput when considering the leakage from the 5 GHz Wi-Fi access point.


V.  Conclusions

In this paper, the simulation-driven virtual prototyping techniques are introduced into the smart product’s development cycle to reduce the development time and ensure intrinsic qualities of the smart product. Specifically, the techniques are coupled with the three different development stages of the wireless speaker: PCB layout, antenna design and integration, and wireless coverage and interference evaluation.


In the PCB layout stage, the PCB design considerations for the smart product’s electric performance are discussed and the corresponding PCB layouts are verified by the design rule checker, DFE and closely studied by the SI and thermal analysis. Thanks to the verification and analysis process, it is illustrated to maximize audio quality and high-speed data rate of the smart product at the PCB design step.


In the antenna design and integration stage, a meander-line antenna is designed and integrated on the PCB by optimizing antenna shape and using a matching circuit. It is also discussed in detail that virtual simulations of PCB with antenna for placement in optimal location inside a speaker cabinet considering the neighboring electrical and mechanical components for the best antenna performance possible.


In the wireless coverage and interference stage, the Bluetooth performance of the smart speaker inside a house are evaluated in terms of wave propagation and network parameters. In addition, co-existence situations of the smart speaker with nearby wireless devices in a real life are analyzed by considering the various Bluetooth and Wi-Fi air interface standards.


In conclusion, the simulation-driven virtual prototyping is an essential solution to make the smart products more cost-effective, high quality, and reliable from the early development stage. According to the market trends, cost reduction, compact size, and seamless connectivity of smart wireless products become increasingly required. To satisfy the market requirements, the simulation-driven design and virtual prototyping should be readily and widely coupled with the development process of the modern electrical and mechanical products.



  1.  White paper: Innovations for new business opportunities – 5G-PPP: https://5g-ppp.eu/wp-content/uploads/2017/01/5GPPP-brochure-MWC17.pdf
  2. Ulf Sellgren, “Simulation-Driven Design – Motives, Means, and Opportunities”, Doctoral Thesis, The Royal Institute of Technology, 1999.
  3. Slawomir Koziel, Xin-She Yang, and Qi-Jun Zhang, “Simulation-Driven Design Optimization and Modeling for Microwave Engineering”, Singapore, 2013. 
  4. G. Gary Wang, “Definition and Review of Virtual Prototyping”, ASME Journal of Computing and Information Science in Engineering, Vol. 2, No. 3, pp. 232-236, 2002.
  5. M. A. Erstina, "Design and Construction of Wireless Speaker Using Bluetooth." African Journal of Applied Research, Vol. 1, No. 1, pp. 413-426, 2016.
  6. Altair PollEx: https://www.altair.com/pollex/
  7. Universal Serial Bus Specification, April 2000.
  8. J. Rashed and C. T. Tai, “A new class of resonant antennas”, IEEE Trans. Antennas Propagat., Vol. 39, pp. 1428–1430, Sept. 1991.
  9. Altair Feko: https://www.altair.com/feko/
  10. S. R Best and J. D. Morrow, “Limitations of Inductive Circuit Model Representations of meander line antennas”, IEEE Trans. Antennas and Propagation Society, International Symposium, vol. 1, pp 852-855, June 2003.
  11. Ulf WetzkerIngmar SplittMarco ZimmerlingCarlo Alberto Boano and Kay Römer, “Troubleshooting Wireless Coexistence Problems in the Industrial Internet of Things,” IEEE Int. Conf. on Computational Science and Eng, Paris, France, 2016.
  12. Bluetooth core specification version 5.0: http://www.bluetooth.com.
  13. 802.11b standard core specifications: https://standards.ieee.org/standard/802_11b-Cor1-2001.html.
  14. 802.11n standard core specifications: https://standards.ieee.org/standard/802_11n-2009.html.
  15. Evgeny Khorov, Anton Kiryanov, Andrey Lyakhov, and Giuseppe Bianchi, “A Tutorial on IEEE 802.11ax High Efficiency WLANs,” IEEE Comm. Surveys & Tutorials, Vol. 21, No. 1, First Quarter 2019.