Compact Multifunctional Dipole Antenna Array for MIMO Systems. A Thesis. Submitted to the Faculty. Drexel University

Transcription

1 Compact Multifunctional Dipole Antenna Array for MIMO Systems A Thesis Submitted to the Faculty of Drexel University by Mikhail Aleksandrovich Chernyavskiy in partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering / Telecommunications June 2012

2 c Copyright 2012 Mikhail Aleksandrovich Chernyavskiy. All Rights Reserved.

3 ii Dedications To my father and my grandmother, for their love and support.

4 iii Acknowledgments I would like to thank my advisor, Dr. Kapil Dandekar for his guidance, advice, and support these past three years. I am very grateful to Prathap for being a great mentor to me throughout his time in the lab, to Guillermo and David for their help with my measurements, to Magda and Kevin for never being too busy for me, and to John for his excellent advice. A big thank you to all past and present members of DWSL. It was truly a pleasure to work with each and every one of you. Last but not least, thanks to all of my friends for everything.

5 iv Table of Contents List of Tables vi List of Figures vii Abstract viii 1. Introduction Motivation Related Work Thesis Contribution Thesis Organization Background MIMO Wireless Communications Channel Capacity Fading Antennas for MIMO Systems Diversity Diversity Combining Techniques Reconfigurable & Multifunctional Antennas Antenna Array Design and Simulation Design Motivation Antenna Array Design

6 v 3.3 Antenna Array Simulation Antenna Array Construction Antenna Array Measurement Results Measured Antenna Parameters Channel Measurements Channel Measurement Results Discussion and Future Research Discussion Future Research Appendix A: Table of Symbols Appendix B: Table of Acronyms Bibliography

7 vi List of Tables 3.1 Antenna array structural parameters Area of the proposed antenna array and reference antenna array Average measured percentage capacity improvement achievable with the stacked antenna array at GHz with respect to the λ and λ/2 separations of the reference dipoles Average measured percentage capacity improvement achievable with the stacked antenna array at GHz with respect to the 2λ and λ separations of the reference dipoles

8 vii List of Figures 2.1 Block diagrams of (a) SISO and (b) MIMO communication systems Three antenna diversity techniques: (a) spatial diversity, (b) pattern diversity, and (c) polarization diversity (a) Schematic and (b) prototype of the triband antenna from [32] Schematics of the proposed antenna array design: (a) front view, (b) back view, and (c) perspective view Simulated return loss and isolation plots of the two antennas in the proposed antenna array Simulated radiation patterns in dbi of the two antennas in the proposed antenna array at GHz Simulated 3D radiation patterns of the two antennas in the proposed antenna array at GHz (a) Front and (b) back of the prototype of the proposed antenna array Measured return loss and isolation plots of the two antennas in the proposed antenna array Measured radiation patterns in dbi of the two antennas in the proposed antenna array at (a) GHz, and at (b) GHz Floor plan of test environment. Test locations of TX and RX are indicated Comparison of the proposed antenna array (left) and the reference dipoles (right) CDF of capacity of the stacked antenna array and the reference dipoles assuming 10 db SNR, measured at (a) GHz, and at (b) GHz CDF of capacity of the stacked antenna array as a function of SNR at GHz

9 viii Abstract Compact Multifunctional Dipole Antenna Array for MIMO Systems Mikhail Aleksandrovich Chernyavskiy Kapil R. Dandekar, Ph.D. A compact, stacked, multi-frequency dipole antenna array is designed and presented for use in Multiple Input Multiple Output (MIMO) wireless communication systems. The array consists of two dual band frame-printed dipoles occupying the same physical space. Each antenna can operate in the 2.4 and 5 GHz bands for wireless local area network (WLAN) applications. The array can be used in a 2 2 MIMO link at either the transmitter (TX) or receiver (RX) or both. The lack of spatial diversity that arises from having co-located antennas is counterbalanced by the pattern diversity resulting from the mutual coupling between the two antenna elements. This system takes advantage of the otherwise undesirable mutual coupling within the antenna array elements by producing pattern diversity from the shift in the radiation patterns. The proposed antenna array was simulated and manufactured and its radiation characteristics were tested. Channel measurements were also taken using the antenna array and the WARP software defined radio platform. The proposed antenna array has radiation characteristics and measured channel capacity comparable to that of specialized antennas operating in each of the frequency bands, while providing the added benefit of size reduction.

10

11 1 Chapter 1: Introduction 1.1 Motivation Wireless communication has become an integral part of people s daily lives and a critical business tool. Wireless systems offer convenient and reliable connectivity that allows for user mobility. In addition, wireless communication allows for network access to be introduced to areas where it would be traditionally be difficult to connect to a wired network, since wireless networks are easier to deploy. Because of these attractive characteristics, wireless communication is currently at the forefront of telecommunications research. Modern communication systems require multiple antennas that support several frequency bands in a compact space [1]. Multiple input multiple output (MIMO) wireless communication [2] is a promising technology that plays an important role in new and upcoming mobile communication systems. MIMO techniques combine signals from multiple antennas to exploit the multipath in wireless channels and enable higher capacity, better coverage, and increased reliability without using extra spectrum and power resources [1]. The main advantage of MIMO systems is that they provide the ability to form parallel orthogonal transmission channels, even in rich scattering environments [3]. Many wireless communication standards require operation in multiple frequency bands. For example, an IEEE n (WiFi) device requires at least two antennas

12 Chapter 1: Introduction 2 operating in the 2.4 GHz and 5 GHz bands. At the same time, consumer devices like laptops, tablets, and smartphones continue to get smaller and thinner, leaving less and less room for antennas. These constraints provide a need for a low profile antenna array that is compatible with MIMO systems and able to operate on multiple frequencies [1]. This thesis offers one possible solution to this problem by presenting a compact multifunctional antenna array for MIMO communication systems comprised of two dual band frame-printed dipoles. The antenna array was designed, simulated, constructed, and tested. 1.2 Related Work Current research in the area of antennas for MIMO systems has been focused on electrically reconfigurable [4 13] and multimode antennas [14 16]. These antennas include spiral antennas [7, 15], dipole hybrids [4 6, 14, 15], patch antennas [8, 9, 15], and pixel antennas [13] that can reconfigure their radiation patterns, polarizations, and frequency operating bands. Patch antennas are often used as reconfigurable antennas. A reconfigurable multiport circular patch antenna was designed such that it can excite different electromagnetic modes by varying its radius with PIN diode switches [8]. Another circular patch antenna array consisting of two stacked circular disks presented in [9] is capable of changing the shape of its radiation pattern by selecting a pair of feed points connected to the two antennas, which again excite different EM modes. The authors of [10] demonstrate a compact pattern reconfigurable U-slot patch antenna.

13 Chapter 1: Introduction 3 Many somewhat more exotic antennas have also been investigated in the reconfigurable antenna literature. The authors of [7] propose a single arm Archimedean spiral antenna whose arm length can be reconfigured in length using PIN diode switches, exciting different radiation patterns in the process. A reconfigurable leaky wave antenna was designed making possible dynamic changes to the array radiation pattern [11]. A reconfigurable 2D fractal tree antenna is shown in [12]. A pixel antenna utilizing microelectromechanical switches (MEMS) capable of reconfiguring its radiation modes and operating frequency has been designed for narrowband MIMO systems [13]. Even a cubic antenna is investigated in [16]. Dipoles have been incorporated into many different reconfigurable antenna applications. A wideband reconfigurable MIMO antenna, a combination of a reconfigurable balanced dipole and a two-port chassis antenna, has been demonstrated in [4]. The authors of [5] present a compact reconfigurable antenna array consisting of two hybrid monopole/dipole elements. Each element can operate in either monopole or dipole mode. A circular polarization spiral-dipole antenna has been proposed in [15]. A dipole antenna is loaded with spirals at both of its ends to generate omnidirectional left-hand or right-hand circular polarization. More traditional dipoles are presented in [6, 14]. A linear printed dipole array is oriented in a fashion that introduces pattern diversity in [14]. An antenna array of two printed dipoles, in which each of the dipoles can be reconfigured in length using PIN diode switches can be seen in [6]. The switch configuration can be modified in accordance to changes in the environment.

14 Chapter 1: Introduction Thesis Contribution This thesis contributes to the development of an optimal antenna system for MIMO communications. The proposed antenna array is composed of two dipole antennas, each operating in multiple frequency bands, creating frequency diversity. The array is constructed in a way that introduces pattern diversity to the system. Most importantly, while typical MIMO antennas are separated to achieve decorrelation, the proposed array is compact for potential use in mobile devices. 1.4 Thesis Organization Chapter 2 provides background information regarding topics such as MIMO wireless communications, channel capacity, diversity, and antennas for MIMO systems. Chapter 3 presents the design of the proposed compact dipole antenna array and the simulation results. Chapter 4 discusses the measurement results and evaluates the antenna as part of a system. Chapter 5 concludes this thesis by providing a summary of the work, a discussion of the research in context with existing designs, as well as proposals for future work.

15 5 Chapter 2: Background 2.1 MIMO Wireless Communications In a conventional radio communication system, one transmit (TX) and one receive (RX) antenna are used to transmit information over a communication channel. This is referred to as a single input single output (SISO) system. A block diagram of a SISO system can be seen in Figure 2.1a. If a simplified channel is assumed to be time and frequency invariant, the channel is denoted by a scalar h. The scalar signal model is given as: y = hx + n (2.1) where y is the received signal, x is the transmitted signal, and n is complex additive white Gaussian noise (AWGN) with zero mean. In a noise-limited scenario, the spectral efficiency of a channel is fundamentally limited by the Shannon-Nyquist criterion [17], expressed as: ( C = log SNR h 2 ) (2.2) ( ) = log P t h 2 σn 2 (2.3) where h is the transfer function from the TX to the RX, SNR denotes the signal to noise ratio at the RX, P t is the transmitted power, and σ 2 n is the noise variance. The channel capacity can increase logarithmically with an increase in transmit power.

16 Chapter 2: Background 6 (a) (b) Figure 2.1: Block diagrams of (a) SISO and (b) MIMO communication systems If a link in a wireless communication system is equipped with multiple antenna elements at both the transmitting and the receiving end, the system becomes a multiple input multiple output (MIMO) system. A block diagram of a MIMO system can be seen in Figure 2.1b. The channel response for a narrowband non-frequency selective MIMO system with N r receive antennas and N t transmit antennas is now

17 Chapter 2: Background 7 denoted by a channel matrix H C Nr Nt, h 11 h 12 h 1Nt h 21 h 22 h 2Nt H = h Nr1 h Nr2 h NrNt (2.4) where h ij is the transfer function, or scalar SISO channel, between the i-th RX antenna and the j-th TX antenna [17]. The vector signal model is given as: y = Hx + n (2.5) where y C Nr 1 is the received signal vector at the RX antennas, x C Nt 1 is the transmitted signal vector for the TX antennas, and n C Nr 1 is the AWGN vector at the RX antennas [6]. The channel matrix H is the mathematical representation of the transmission path of the transmitted data, which includes the multipath channel characteristics of the environment and the antenna configurations of the transmitting and receiving antenna arrays. Both the multipath channel characteristics and the antenna configurations play a large role in determining the performance of the MIMO system Channel Capacity Assuming that the flat fading channel is unknown at the transmitter and known at the receiver, the signal vector at the transmitter is composed of N t independent signals

18 Chapter 2: Background 8 with equal power. In this case, the capacity is shown to be [18]: ( C = log 2 [det I Nr + SNR )] HH N t (2.6) where I Nr is an identity matrix of size N r N r, SNR is the mean signal to noise ratio per receiver branch, det is the determinant, and the superscript denotes the complex conjugate transpose. The Kronecker model is used in this thesis to generate random channels and thus show the potential of the proposed antenna array [6]. In a spatially correlated Rayleigh-fading MIMO channel, the channel matrix H is defined by the Kronecker model as [19]: H = R 1/2 r H w R 1/2 t (2.7) where R r is the receive spatial correlation matrix, R t is the transmit spatial correlation matrix, and H w C Nr Nt is a matrix of complex Gaussian fading coefficients. In this thesis, a 2 2 MIMO system is used, i.e. N r = N t = Fading The signal radiated from a transmitting antenna is reflected, scattered, diffracted, and/or refracted by the various structures in its path. Mathematically, these factors cause the signal to experience path loss, shadowing, and fading [17]. Path loss is the reduction in field strength when an electromagnetic wave propagates though space

19 Chapter 2: Background 9 and for free space is modeled as [20]: ( ) 4πd L = 20 log 10 λ (2.8) where λ is the wavelength, d is the distance between the transmitter and the receiver, and L is the path loss in decibels. Shadowing is caused by large obstructions that obscure the main signal path between the transmitter and the receiver. These obstacles will be different for every path, causing variations with respect to the value given by the path loss model. Fading refers to rapid fluctuations in the signal and results from the interference between multiple waves reaching the receiver from the transmitter [17]. Typically, the fading caused by multipath signal propagation is considered to be a severe problem in wireless communication channels. However, MIMO systems exploit the multipath signals in order to increase the system capacity without increasing the system power or bandwidth [21]. The idea behind MIMO is that signals that travel through complex multipath environments are combined in such a way that the quality or data rate of communication for each MIMO user will be improved relative to the SISO case. Since multiple data streams can be transmitted simultaneously on the same frequency in MIMO systems, the bandwidth efficiency and capacity can be increased linearly by the number of data streams, i.e. antennas employed, with no additional overhead [17]. Multipath fading arises from the presence of multiple transmission paths between the transmitter and receiver [22]. When a signal leaves a transmitter, it can take

20 Chapter 2: Background 10 many different paths to the receiver, each with its own reflections, diffractions, etc. There are two different methods to transmit data through the utilization of these multiple streams. Spatial Multiplexing Spatial multiplexing [23] is a scheme where independent data streams are transmitted simultaneously in parallel channels from each element in an array of antennas. If N = min(n t, N r ), the bit stream in question is demultiplexed into N sub-streams, then modulated and transmitted from each antenna simultaneously. Assuming the receiver has knowledge of the channel, it can extract the signals, demodulate them, and then recombine them to yield the original bit stream [23]. This process increases spectral efficiency because the data stream can be transmitted N times as quickly as a non-multiplexed data stream. Space-Time Coding Space-time coding [23] is an alternative scheme to spatial multiplexing. It uses the multiple element antennas for diversity gain by encoding a single data stream across both time and space. In other words, multiple redundant copies of a data stream are transmitted on the N channels [23]. This process serves to increase link reliability. Any errors in one of the transmission paths can be fixed through a comparison with the other paths, therefore space-time coding can be used to improve the quality of the transmission.

21 Chapter 2: Background Antennas for MIMO Systems In order for a MIMO communication system to have good performance, the antennas used in the system must be capable of providing a high degree of diversity [24]. It is also beneficial for the antenna array to have a compact design that is comparable in size to a single antenna. Traditionally the antennas in a MIMO system are spaced farther apart to achieve higher spatial diversity [25]. However, this is not always possible in small mobile applications. Thus it is often necessary to use antennas with different radiation patterns or polarizations, or antennas operating at different frequencies in order to achieve the required levels of diversity [2]. Section expounds on the different types of diversity and the benefits of each one. It is also possible to achieve an increased diversity level by intelligently selecting or combining the antennas at the transmitter and receiver that provide the highest levels of diversity at the two nodes. Switching circuitry can be used to select the group of antennas that provide the optimal system diversity for a given channel [2]. Section summarizes the various algorithms used for diversity combining and selection Diversity The principle of diversity is that the receiver should have more than one copy of the transmitted signal available, with each copy being received through a statistically independent channel. If the signals are uncorrelated, the fading dips in the signal will have a small probability of occurring simultaneously and therefore the multiple

22 Chapter 2: Background 12 signals can be combined at the receiver to make a signal that has a higher mean SNR than any single branch of the system has by itself. Five categories of diversity are discussed in detail in the following sections. Spatial Diversity Spatial diversity occurs when multiple physically separated antennas are used in the system. Spatial diversity takes advantage of the random nature of propagation. Many independent paths exist at any location, so the signals are uncorrelated when the locations are separated by a certain minimum distance, usually some multiple of the wavelength [22]. The independence comes from different multipath components having different amplitudes and phases when arriving at different points in space. The further apart the antennas are placed, the greater the produced phase difference, and the smaller the correlation of the signals obtained at the antennas. Figure 2.2a shows a diagram of two identical antennas with identical radiation patterns separated in order to produce spatial diversity. Since the antennas must be separated, spatial diversity is not a viable option for space constrained devices. Frequency Diversity Frequency diversity utilizes multiple frequency bands in the transmission of a signal. The same signal could be transmitted on both frequencies or the information could be split up between the frequencies. If the carrier frequencies are separated by more than the coherence bandwidth [17] of the channel, then their fading can be considered to be independent, and the probability that the signal simultaneously experiences a fade at both frequencies is low [17]. If multiband antennas are used in the system, the

23 Chapter 2: Background 13 (a) (b) (c) Figure 2.2: Three antenna diversity techniques: (a) spatial diversity, (b) pattern diversity, and (c) polarization diversity. same antenna can be used to transmit and/or receive at each frequency of interest. Limitations of frequency diversity are the availability of bandwidth and that the channel must be frequency-selective [22]. Pattern Diversity Pattern diversity (or angle diversity) [17] makes use of antennas having different radiation patterns. Each antenna picks up multipath components coming from different angular directions. The amplitudes and phases of these multipath components will be different with different antenna patterns. Therefore, their combination will be uncorrelated [22]. Various types of antennas with differing radiation patterns can be used for the sake of pattern diversity, but it is also possible to produce different radiation patterns with identical antennas by mounting them close to each other. This effect is due to strong electromagnetic interactions between the antennas, otherwise known as

24 Chapter 2: Background 14 mutual coupling [26]. Each antenna s pattern will be skewed due to the electromagnetic interactions from the other antenna [17]. Figure 2.2b shows a diagram of two co-located antennas producing two distinct radiation patterns at different angles. Polarization Diversity Because the reflection and diffraction processes depend on polarization, horizontally and vertically polarized multipath components propagate differently in a wireless channel [17]. The propagation effects of the channel depolarize the propagating beam, which leads to the fading of different polarizations being statistically independent [17]. Thus, the depolarized signal can be split into horizontal and vertical polarizations at the receiver and processed separately to produce diversity. The only limitation of polarization diversity is that, as opposed to the other diversity schemes, it is only possible to generate two diversity branches horizontally and vertically polarized (or any other two orthogonal polarizations) [22]. Figure 2.2c shows a diagram of two co-located antennas with two orthogonal polarizations. Time Diversity Since the wireless propagation channel is time-variant, signals sent and received at different times are uncorrelated [17]. If the same data stream is transmitted multiple times at intervals that exceed the coherence time [22] of the channel, the streams are subject to independent fading. An advantage of time diversity, or temporal diversity, is that multiple antennas are not required in the system. However, this diversity scheme is highly bandwidth-inefficient and requires storage to save the received data streams for processing [22].

25 Chapter 2: Background Diversity Combining Techniques It is necessary to somehow combine the signals arriving at the receiver due to the above antenna diversity techniques such that the quality of the overall signal is improved. Selection diversity selects the best signal while discarding all of the other copies of the signal. Combining diversity, on the other hand, combines all available copies of the signal using different algorithms. Generally, diversity combining leads to better performance because all present information is utilized [17]. There are three main algorithms that are implemented in diversity combining techniques. Selection Combining Selection combining [17] is mathematically the simplest diversity combining scheme. Since the fades in the individual signals do not happen simultaneously, the instantaneous SNR is monitored in all branches and the branch with the highest SNR is selected as the output signal [22]. This method is the easiest to implement but is inefficient because the useful signal power received on the non-selected branches is discarded. Maximal Ratio Combining In order to improve the output SNR even further, the signals from all branches can be combined to form the output signal. Maximal ratio combining takes all of the individual signals and performs a linear combination on them, using appropriate weighting [22]. In order to maximize the SNR at the output, a branch with higher SNR will be given a higher weighting [23]. Since the signals are not in phase, they have to first be multiplied by a complex phasor in order to bring them to zero phase

26 Chapter 2: Background 16 so they can be combined coherently. The maximal ratio combiner provides the best performance when compared to the other algorithms but at the cost of increased complexity. Equal Gain Combining Equal gain combining is similar to maximal ratio combining in that all of the branches are added together. The difference is that all the branches are weighed equally. The output SNR is better than the SNR of selection combining, but not as good as the SNR produced by maximum ratio combining. However, it is easier to implement than the maximum ratio combiner and as the number of antennas or channels in the system increases, the difference between equal gain combining and maximum ratio combining decreases [22] Reconfigurable & Multifunctional Antennas Using several antennas in an array in combination with the above-mentioned diversity combining algorithms can often provide the highest level of diversity. However, when using these algorithms, not all antennas within the array are simultaneously utilized at any given time. As a result, this becomes an impractical solution for portable devices where space is a critical constraint. Reconfigurable or multifunctional antennas provide the opportunity for a single antenna that incorporates the diversity techniques outlined in Section in its design to replace several antennas in a system. These antennas are usually more compact than the multiple antenna arrays they replace, making them much more attractive for mobile applications, as well as other applications where space is a con-

27 Chapter 2: Background 17 straint. Reconfigurable antennas can adaptively tune their radiation characteristics, polarization, or frequency of operation in response to the fluctuations in the wireless channel [2]. Various design techniques have been proposed for modifying antenna attributes [4 13, 27]. The arrangement of currents on an antenna determines the antenna s radiation distribution [2]. RF switches, material changes, and structural changes can be employed to achieve changes in an antenna s radiation pattern [2]. One very popular approach is using PIN diode switches to reconfigure the antenna structure. The antenna array in [5] consists of two elements, each of which has two possible modes by controlling states of three PIN diodes. Similarly, each of the dipoles in the reconfigurable printed dipole array in [6] can change the length of its arms using PIN diodes to produce different modes with different radiation patterns. The various antennas in [7 10, 12] all use switches to change the physical shape of the antenna and consequently the radiation pattern. Beam-steering antennas are a type of pattern reconfigurable antenna that can sweep their narrow main lobe across a wide range of angles. The reconfigurable leaky wave antenna in [11] consists of ten unit cells loaded with varactor diodes and two independent bias networks used to separately tune the varactors and steer the two beams. For polarization reconfigurability, the antenna structure, material properties, or feed configuration have to change in ways that alter current flow on the antenna [2]. There are different kinds of polarizations [28]: i) various linear polarizations, ii) right-handed and left-handed circular polarizations, and iii) elliptical polarizations.

28 Chapter 2: Background 18 Polarization is usually modified using RF switches or material changes [29]. In [27], an antenna consisting of a single octagonal microstrip patch has two ports located on perpendicular sides of the patch that excite two orthogonal polarizations of the radiated electric field. The antenna uses MEMS switches to select between the two polarization bases. Frequency reconfigurable antennas [30] can switch their frequency of operation, thus implementing frequency diversity without the need for multiple antennas. Multiband antennas can operate on multiple frequencies simultaneously, combining multiple elements in order to create antennas that operate in several independent bands. The reconfigurable dipole-chassis antennas presented in [4] are already multiband by design, but they can also sweep their operating frequencies by several hundred megahertz by varying the supplied voltage to four varactor diodes in the matching circuit. The best reconfigurable antennas allow for simultaneous changes to multiple radiation characteristics. These reconfigurations are often achieved through pixel-based approaches such as the pixel antenna in [13]. However the use of a large number of switches introduces high losses and decreases the radiation efficiency of the antenna [2]. The contribution of this thesis is a novel compact multifunctional dipole antenna array for MIMO communication systems. The stacked antenna system consists of two individual dual band frame-printed dipoles sharing the same physical space on the board. The antenna array demonstrates both pattern and frequency diversity without any losses from switches or switching networks. The frequency diversity comes from

29 Chapter 2: Background 19 both dipoles in the array being designed to resonate at multiple frequencies instead of reconfiguring their frequency of operation with switches. The pattern diversity is generated by the mutual coupling effects between the antenna array elements. The pattern and frequency diversity replace the spatial diversity typically used in MIMO systems.

30 20 Chapter 3: Antenna Array Design and Simulation 3.1 Design Motivation Modern communication standards often necessitate the use of multiple antennas at multiple frequencies. At the same time, modern consumers demand compact wireless devices that require the antenna profile to be as small as possible. These requirements and constraints raise two challenges. First, the antennas need to be designed extremely compactly with a low profile. Secondly, the interaction between the antenna elements conventionally needs to be kept to a minimum to prevent a mutual coupling effect between the antennas. This is generally done by separating the antennas by at least a half-wavelength. However, these two constraints are at odds with one another. If it is necessary to minimize the profile of the antennas and design them to be as small and compact as possible due to the imposed space constraint, it will not be feasible to place them far apart. A possible solution to this problem would be eliminating the second constraint of no mutual coupling. Mutual coupling has been shown to be beneficial in MIMO systems [26]. The presence of other array elements in the near field of each antenna array element will distort the radiation pattern of each of the array elements in a unique fashion [26]. This distortion will produce pattern diversity in the system and improve the quality of the communications link by increasing the channel capacity [31].

31 Chapter 3: Antenna Array Design and Simulation 21 Instead of viewing mutual coupling as a drawback, as is generally done in MIMO communications, the design proposed in this thesis embraces the coupling between the two driven antenna array elements. The mutual coupling results in changes in the source currents on both of the elements, which results in a modification of the impedance presented at the element terminals [28]. More importantly, it also modifies the radiation patterns of the antennas in the array due to interactions between each of the antennas, thus providing pattern diversity to the system. The two dipoles therefore have two different radiation patterns and can pick up multipath components arriving from different directions [17]. 3.2 Antenna Array Design The design proposed in this thesis is based on the frame-printed dipole presented in [32]. The authors of [32] designed a multiband antenna consisting of a set of printed frame dipoles of different sizes, printed on a double-sided dielectric substrate. The arms of the dipoles are printed on opposite sides of the substrate and the antennas are fed from a 50 Ω coaxial cable through a microstrip-to-twinline tapered transition. The printed frame dipole is constructed by etching off a section of the arms of the traditional strip dipole. Thus, the characteristics of the frame dipole are similar to those of the strip dipole. The dipoles are nested within each other and are employed as resonators to produce a multiband response. Each of the three dipoles operates at a unique frequency band and the overall triband antenna operates at 1.8 GHz, 2.4 GHz, and 3.5 GHz. The nested dipoles can be considered parallel-connected, so the off-resonant dipoles will have a higher shunt impedance and will not debase the

32 Chapter 3: Antenna Array Design and Simulation 22 performance of the active dipole [32]. A schematic of the antenna presented in [32], as well as the milled prototype, can be seen in Figure 3.1. The antenna was not evaluated in a system. (a) (b) Figure 3.1: (a) Schematic and (b) prototype of the triband antenna from [32]. The design proposed in this thesis first modifies the antennas proposed in [32] in order for the array to operate in the 2.4 GHz and the 5 GHz bands for WLAN

33 Chapter 3: Antenna Array Design and Simulation 23 (wireless local area network) applications. This modification is done by changing the lengths of the dipole arms. Next, two of these antennas are incorporated in the space that would typically be used by only a single antenna by mirroring the front and back arms of the dipole and angling the microstrip feedlines in opposite directions in order to create two separate ports. The width of each feedline is chosen to match each port to 50 Ω. To avoid the crossing of the microstrip lines on the back of the proposed antenna array, a middle board layer containing a segment of one of the microstrip lines is added and used as a bridge. The microstrip line of one of the antennas begins on the 3rd layer, is connected to the bridge on the 2nd layer, and is connected back to the antenna feedline on the 3rd layer after passing over the microstrip line of the other antenna. Adding the middle board layer produces a three layer board and results in two virtually identical independent dual band frame dipoles that share one physical space despite each dipole having its own input port. Schematics of the proposed antenna can be seen in Figure 3.2. A summary of the main dimensions of the proposed antenna array and the material properties of the substrate is presented in Table Antenna Array Simulation The antenna array was designed and simulated in HFSS [33], a finite element method solver for electromagnetic structures. Figure 3.3 shows a plot of the simulated return loss of the two antennas (S 11 and S 22 ). As can be seen from the figure, the antennas have similar but not identical return loss curves. However, they both radiate in the

34 Chapter 3: Antenna Array Design and Simulation 24 (a) (b) (c) Figure 3.2: Schematics of the proposed antenna array design: (a) front view, (b) back view, and (c) perspective view.

35 Chapter 3: Antenna Array Design and Simulation 25 Table 3.1: Antenna array structural parameters. Antenna Outside dipole arm length mm Inside dipole arm length mm Outside dipole arm height 18 mm Inside dipole arm height 6 mm Dipole arm width 1.5 mm Left antenna feedline thickness 3 mm Right antenna feedline thickness 1.5 mm Feedline length 48 mm Board length 70 mm Board width 75 mm Board height mm Substrate Dielectric FR-4 Dielectric permittivity 4.4 Dielectric loss tangent 0.02 Dielectric thickness mm 2.4 GHz and the 5 GHz bands as desired. The two antennas have return loss values of and db at GHz, respectively, and and db at 5.32 GHz, respectively. The return loss is below the target -10 db in the bands of interest for both ports. The isolation between the two antenna ports (S 21 ) is also shown on Figure 3.3. The isolation is db at GHz and -8.0 db at 5.32 db, below the target -10 db in the 2.4 GHz band but slightly above at the higher frequency. Figure 3.4 shows the simulated azimuthal radiation patterns of each of the dipoles in dbi. The radiation patterns are similar but shifted by approximately 45, demonstrating the desired pattern diversity. The radiation patterns are likely not identical because of the bridge used in the middle layer of the antenna, as it is the only structural difference between the two antennas. Simulated 3D radiation patterns are

36 Chapter 3: Antenna Array Design and Simulation 26 Figure 3.3: Simulated return loss and isolation plots of the two antennas in the proposed antenna array. displayed in Figure 3.5. The gain values in these plots are displayed in absolute units, not dbi, in order to emphasize the differences between the two radiation patterns. One of the antennas has a slightly higher maximum gain value and the shifted radiation patterns are even more pronounced in Figure 3.5. The simulated antenna radiation efficiency was found to be at GHz and at 5.32 GHz. The relatively low efficiency is expected when using a low ɛ r material like FR-4. When the dielectric permittivity value increases, the radiation efficiency decreases [2].

37 Chapter 3: Antenna Array Design and Simulation ± Figure 3.4: Simulated radiation patterns in dbi of the two antennas in the proposed antenna array at GHz. 3.4 Antenna Array Construction The proposed antenna array allows for a low-cost manufacturing solution. The antennas were milled using a T-Tech Quick Circuit 5000S-FA milling machine [34] on FR-4 dielectric substrate of 31 mil thickness. After milling, the three layers were connected using copper interconnects, and the layers were adhered together using a LPKF MultiPress S multiboard press [35]. The final prototype can be seen in Figure 3.6. The bridge in the middle layer can be seen clearly through the back of the antenna array, in Figure 3.6b.

38 Chapter 3: Antenna Array Design and Simulation 28 (a) (b) Figure 3.5: Simulated 3D radiation patterns of the two antennas in the proposed antenna array at GHz.

39 29 Chapter 3: Antenna Array Design and Simulation (a) (b) Figure 3.6: (a) Front and (b) back of the prototype of the proposed antenna array.

40 30 Chapter 4: Antenna Array Measurement Results 4.1 Measured Antenna Parameters The scattering parameters of the dipoles were measured using an Agilent N5230A Vector Network Analyzer (VNA). The decision was made to perform 5 GHz measurements at GHz instead of 5.32 GHz because i) the S-parameters at GHz on the manufactured antenna appear superior to those at 5.32 GHz and ii) the GHz frequency, corresponding to Channel 161, is also free from outside interference. Since radiation at either frequency is sufficient for 5 GHz channel measurements, the antenna did not have to be redesigned, but rather the target frequency in the 5 GHz band could simply be changed since the intended design did not have a specific target frequency in mind. Figure 4.1 shows the measured return loss curves of the two antennas. The two antennas have return loss values of and db at GHz, respectively, and and db at GHz, respectively, both well below the target -10 db value. The return loss values are comparable to the results of the simulations. The other peaks seen in Figure 4.1 are caused by coupling between the antennas. The isolation between the two antennas is db at GHz and db at db, again both well below the target -10 db and exceeding the simulated results. Radiation patterns were measured in the Drexel anechoic chamber. Figure 4.2 shows the measured azimuthal radiation patterns in dbi taken at and GHz.

41 Chapter 4: Antenna Array Measurement Results 31 Figure 4.1: Measured return loss and isolation plots of the two antennas in the proposed antenna array. The measured radiation patterns are similar to the simulated ones at GHz, with a comparable shift between the two patterns approximately 45. The GHz radiation patterns do not look as similar to simulated results, but still exhibit some pattern diversity. Unfortunately, accurate gain values were not able to be measured in the 5 GHz band due to limitations of the anechoic chamber. The chamber is not able to be calibrated for the higher frequency band. 4.2 Channel Measurements The antenna array was tested as part of a system by taking channel measurements in an indoor environment using the Wireless Open-Access Research Platform (WARP),

42 Chapter 4: Antenna Array Measurement Results ± (a) ± (b) Figure 4.2: Measured radiation patterns in dbi of the two antennas in the proposed antenna array at (a) GHz, and at (b) GHz.

43 Chapter 4: Antenna Array Measurement Results 33 a software defined radio testbed developed by Rice University [36]. A 2 2 MIMO orthogonal frequency division multiplexing (OFDM) implementation of WARP was used. Measurements were performed on channel 14 of the band (centered at GHz) and at channel 161 (centered at GHz), each with a 20 MHz bandwidth. The measurements were taken in Drexel s Wireless Systems Laboratory on the 3rd floor of the Bossone Research building in Drexel University. A floor plan of the environment, along with the locations of the TX and RX can be seen in Figure 4.3. The receiver remained stationary while a TDK PP-02 field probe positioner [37] was used to sweep the transmitter along a 1.5 m 1.5 m grid in the horizontal and vertical directions in 50 cm steps. The channel matrix was measured for each location of the transmitter for both frequencies. Due to the vertical displacement of the transmitter, the measurements consisted of a combination of both line-of-sight (LOS) and nonline-of-sight (NLOS) links. The printed dipoles described in [38] were used as reference antennas for comparison of the proposed design. Since the antennas in [38] are only designed for GHz, the design was scaled and analogous dipoles were manufactured for the 5 GHz band. One of the 2.4 GHz dipoles and one of the 5 GHz dipoles joined by a splitter were used at each port of the 2 2 MIMO system and the measurements were repeated as described above for the proposed antenna array. Because of uniform power allocation, each of the reference dipoles received half of the total input power [39]. This was done because each of the two antennas in the proposed antenna array would be replacing two of the reference dipoles in a communication system. WARP was used to measure the channel matrix at each location of the transmitter

44 Chapter 4: Antenna Array Measurement Results 34 Figure 4.3: Floor plan of test environment. Test locations of TX and RX are indicated. for both frequencies and for different separation distances of the reference antennas. The separation distances used were λ and λ/2 for the 2.4 GHz band and 2λ and λ for the 5 GHz band. The λ/2 separation was impossible to achieve with the shortened wavelength corresponding to the 5 GHz band because of the way the antennas were mounted. Figure 4.4 shows the proposed antenna array next to the reference dipoles. Table 4.1 displays the area of each antenna array. The area of the proposed system is reduced by approximately 58% with respect to a reference system composed of two 2.4 GHz dipoles and two 5 GHz dipoles separated by λ/2 in order to generate spatial diversity in the system.

45 Chapter 4: Antenna Array Measurement Results 35 Figure 4.4: Comparison of the proposed antenna array (left) and the reference dipoles (right). Table 4.1: Area of the proposed antenna array and reference antenna array. Antenna Area (cm 2 ) Reference dipole array Proposed antenna array 52.5 Channel capacity is selected as the performance metric because it allows the study of the antenna array performance independently from the system modulation and the adopted coding technique [2]. To determine the capacity of the MIMO OFDM link, a Frobenius normalization of the channel matrix for each subcarrier was computed in order to remove the differences in path loss among the different channel matrices while preserving the relative antenna gain effects [6]. The normalization factor is

46 Chapter 4: Antenna Array Measurement Results 36 defined as [6]: N F = H ref 2 F N t N r (4.1) Since the channel was characterized over a broad frequency band, the capacity of the wideband channel was defined as an average value of the capacities over all the m subcarriers of the MIMO OFDM system [40]. The capacity was additionally averaged over the k samples in order to minimize the impact arising from the minor differences in spatial orientation between the stacked dipole array and the reference dipoles. Therefore, an estimator of the Shannon capacity is the average of the capacities over the m subcarriers and the k samples, and was determined using [10]: C = 1 km k j ( m log 2 [det i I Nr + SNR N t H ij H ij N 2 F ij )] (4.2) where m is the total number of subcarriers, k is the total number of samples, and N Fij is the Frobenius norm for each subcarrier and sample. In this experiment, m = 52 subcarriers and k = 3200 samples (200 samples at each of the 16 locations) were used. There are 64 total subcarriers, but only 52 are used for data transmission. A separate Frobenius norm was calculated for each frequency band. The optimal solution for the reconfigurable antenna was the one that guaranteed the highest average capacity Channel Measurement Results Figure 4.5 shows the CDF plots of the capacity of the stacked antenna array and the reference dipoles in the indoor environment shown in Figure 4.3 for both measured

47 Chapter 4: Antenna Array Measurement Results 37 frequencies assuming an SNR of 10 db. The proposed antenna array significantly increases MIMO link capacity with respect to a conventional dipole system. As can be seen in Figure 4.5a, the proposed antenna array convincingly outperforms the reference dipoles at both spacings in the 2.4 GHz band. It does not perform as well in the 5 GHz band, but it is still comparable to the reference dipoles at the higher frequency while providing a form factor improvement. Table 4.2 shows the average percentage capacity improvement in the 2.4 GHz band by using the stacked antenna array with respect to the reference dipoles for different values of SNR in a MIMO OFDM system. Table 4.3 shows the same in the 5 GHz band. At an SNR of 10 db, the proposed antenna array outperforms the optimal configuration of the reference dipoles by 22% in the 2.4 GHz band, and the capacity improvement ranges from 33% at 5 db SNR to 8% at 30 db SNR. As can be seen, the proposed antenna array performs better with respect to the reference antennas at low SNR values. Figure 4.6 shows the capacity of the stacked antenna array measured at GHz as a function of SNR. The performance of the antenna in the 5 GHz band was roughly comparable to the performance of the reference dipoles, at both separations and at all SNR values. The median capacity improvement ranges from 3% at 5 db SNR to -3% at 30 db SNR. The performance of the proposed antenna array is much worse in the 5 GHz frequency band than in the 2.4 GHz band, but even at the higher frequency the performance is still similar to the reference dipoles. Again, the proposed antenna array performs better at low SNR values. It is likely that the pattern diversity of the proposed antenna array was not as