Applications of Antenna Arrays to

Mobile Communications, Part I:

Performance Improvement, Feasibility,
and System Considerations
LAL C. GODARA, SENIOR MEMBER, IEEE

The demand for wireless mobile communications services is DBS Direct broadcasting system.
growing at an explosive rate, with the anticipation that commu- DCS Digital cellular system.
nication to a mobile device anywhere on the globe at all times will DECT Digital European cordless telecommunication.
be available in the near future.
An array of antennas mounted on vehicles, ships, aircraft, satel- DFT Discrete Fourier transform.
lites, and base stations is expected to play an important role in DOA Direction of arrival.
fulfilling the increased demand of channel requirement for these EIRP Effective isotropically radiated power.
services, as well as for the realization of the dream that a portable ESA European Space Agency.
communications device the size of a wristwatch be available at an
affordable cost for such services.
ESTEC European Space Research and Technology Center.
This paper is the first of a two-part study. It provides a com- ETACS European total access communication system.
prehensive treatment, at a level appropriate to nonspecialists, of FDD Frequency division duplex.
the use of an antenna array to enhance the efficiency of mobile FDMA Frequency division multiple access.
communications systems. It presents an overview of mobile com- FPLMTS Future Public Land-Mobile Telecommunications
munications as well as details of how an array may be used in
various mobile communications systems, including land-mobile, Systems.
indoor-radio, and satellite-based systems. It discusses advantages GEOS Geostationary satellite.
of an array of antennas in a mobile communications system, high- GMSK Gaussian minimum shift keying.
lights improvements that are possible by using multiple antennas GPS Global positioning system.
compared to a single antenna in a system, and provides details
on the feasibility of antenna arrays for mobile communications
GSM Global system for mobile communications.
applications. G/T Gain-to-noise temperature ratio.
HEOS Highly elliptical orbit satellite.
Keywords— Antenna arrays, blind signal estimation, channel
assignment, diversity combining, handoff, mobile communications, IF Intermediate frequency.
multiple access, satellite communications, wireless communica- ISI Intersymbol interference.
tions. LEOS Low earth orbit satellite.
LMS Least mean square.
NOMENCLATURE LS Least square.
AMPS Advanced mobile phone services. MEOS Medium earth orbit satellite.
BER Bit error rate. MMIC Monolithic microwave integrated circuit.
BPSK Binary phase shift keying. NMT Nordic Mobile Telephone.
CDMA Code division multiple access. PDC Personal digital cellular.
CMA Constant modulus algorithm. PHS Pocket handphone service.
C/N Carrier-to-noise ratio. PN Pseudonoise.
CT2 Second generation cordless telephone. PSK Phase shift keying.
QPSK Quadrature phase shift keying.
RF Radio frequency.
Manuscript received August 15, 1996; revised April 10, 1997. RLS Recursive least square.
The author is with University College, University of New South Wales, RMS Root mean square.
Australian Defence Force Academy, School of Electrical Engineering,
Canberra ACT 2600 Australia (e-mail: [email protected]). SCARP Smart Communication Antenna Research Pro-
Publisher Item Identifier S 0018-9219(97)05307-3. gram.

0018–9219/97$10.00  1997 IEEE

PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997 1031

 SDMA Space division multiple access. There has been a wide range of research covering de-
SER Symbol error rate. velopment of antennas suitable for mobile communications
SIR Signal-to-interference ratio. systems, and many experimental results have been reported
SMI Sample matrix inversion. to show the system requirements and feasibility. It is antic-
SNR Signal-to-noise ratio. ipated that a future mobile communications system would
TACS Total access communications system. consist of a hand-held terminal the size of a wristwatch
TDD Time division duplex. capable of steering beams toward a satellite. The system
TDMA Time division multiple access. would also consist of many radiating elements fabricated
UHF Ultrahigh frequency. by microstrip technology, each with its own phase-shifting
UMTS Universal Mobile Telecommunications Systems. network, power amplifier, and so on along with other
VHF Very high frequency. required processors manufactured by the microwave mono-
lithic integrated circuits technology. And all this is expected
to be available at an affordable price.
I. INTRODUCTION In recent years, the material applicable to array pro-
The field of wireless mobile communications is growing cessing for mobile communications has been reported in
at an explosive rate, covering many technical areas. Its many IEEE publications, as well as in many non-IEEE
sphere of influence is beyond imagination. The worldwide publications covering many specialized areas. The aim of
activities in this growth industry are perhaps an indication this two-part paper is to provide a comprehensive overview
of its importance. The demand for wireless communications of the subject at a level appropriate to readers of this
is anticipated to expand steadily, and projections indicate PROCEEDINGS by bringing all the relevant material together,
that annual activities will exceed $100 billion by the year along with a large reference list to probe further.
2000 [1]–[4]. The two papers in this study give a brief overview of
An application of antenna arrays has been suggested in the mobile communications system consisting of satellites,
recent years for mobile communications systems to over- vehicles, hand-held portables, base stations on the ground,
come the problem of limited channel bandwidth, thereby and their associated terminology. The advantages of arrays
satisfying an ever growing demand for a large number of at the base as well as at the mobile along with the system
mobiles on communications channels. It has been shown requirements are discussed. It is shown how the use of an
by many studies that when an array is appropriately used array could improve the system performance. A detailed
in a mobile communications system, it helps in improving treatment of different algorithms to estimate array weights
the system performance by increasing channel capacity and and the directions of sources is provided. Performance
spectrum efficiency, extending range coverage, tailoring evaluation of different schemes to show the improvement
beam shape, steering multiple beams to track many mobiles, in symbol and BER are considered. Relevant details of
and compensating aperture distortion electronically. It also antenna array design and fabrication are provided along
reduces multipath fading, cochannel interferences, system with the results of feasibility studies reported by many
complexity and cost, BER, and outage probability. It has authors. In short, these papers provide a complete picture
been argued that adaptive antennas and the algorithms to of the field at a level where nonexperts may be able to
control them are vital to a high-capacity communications appreciate it.
system development [4]. These papers have a general coverage. In the process of
An array of antennas may be used in a variety of ways simplifying concepts, the rigor might have been sacrificed
to improve the performance of a communications system. at times. The large and very comprehensive reference lists
Perhaps most important is its capability to cancel cochannel are by no means complete, and no effort has been made
interferences. An array works on the premise that the to include all papers on the subject. However, all possible
desired signal and unwanted cochannel interferences arrive efforts have been made to give credit where due; omissions
from different directions. The beam pattern of the array is where they exist are unintentional and are due to the
adjusted to suit the requirements by combining signals from author’s being unaware of a source’s existence.
different antennas with appropriate weighting. In Section II, a brief overview of mobile communications
The scheme needs to differentiate the desired signal is presented by discussing its working principle, various
from the cochannel interferences and normally requires multiple-access schemes, propagation characteristics, con-
either the knowledge of a reference signal, a training cepts of channel assignment, channel reuse, and handover.
signal, or the direction of the desired signal source to This section also contains discussions on satellite mobile
achieve its desired objectives. There exists a range of communications as well as on progress and trends in the
schemes to estimate the direction of sources with conflicting field. Antenna array terminology required to understand
demands of accuracy and processing power. Similarly, the paper is presented in Section III, whereas Section
there are many methods and algorithms to update the IV contains detailed treatment of how an array may be
array weights, each with its speed of convergence and used in various mobile communications systems, including
required processing time. Algorithms also exist that exploit land-mobile, indoor-radio, and satellite-based systems. Dis-
properties of signals to eliminate the need of training signals cussion on transmitting arrays and the use of arrays for blind
in some circumstances. estimation of cochannel signals is also found in this section.

1032 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

 Section V discusses the advantages of an array of anten-
nas in a mobile communications system and improvements
that are possible by using multiple rather than single
antennas in a system. It provides references to studies where
such improvements are realized and also discusses the ex-
istence of angular spreading in the mobile communications
environment and its impact on system performance. Issues
related to complexities, cost, and network implications of
the use of an array in a system are also highlighted in
this section. Section VI contains details on the design of
antenna arrays suitable for mobile communications and on
some of the experiments that have been carried out to see
the feasibility of antenna arrays for mobile communications
applications.
A detailed treatment of various beam-forming schemes,
adaptive algorithms to adjust the required weighting on
antennas, DOA estimation methods, and sensitivity of an
array system to parameter perturbations will be provided in Fig. 1. A typical setup of a base mobile system.
the second part of this paper.
connected to the telephone network, which controls many
II. OVERVIEW OF MOBILE COMMUNICATIONS base stations. It assigns a radio traffic channel to the phone
In this section, a brief overview of mobile communica- under consideration, as the control channels are used by all
tions is presented to understand its principle of working, phones in that area and cannot be used for data traffic. Once
introduce the terminology required to understand the rest the traffic channel is assigned, this information is relayed
of the paper, and provide references to issues under cur- to the phone via the base and the phone tunes itself to this
rent investigation. A very good treatment of the subject channel. The switching center then completes the rest of
discussing technologies utilized in existing and currently the call.
planned services, the performance of these services, and 2) Initiation of a Call to a Mobile: When someone calls a
trends in wireless system development can be found in [5]. mobile phone, the switching center sends a paging message
Throughout this paper, the generic term “mobile” or through several base stations. A phone tuned to a control
“mobile phone” is used to denote a communications device channel detects its number and responds by sending a
on the move, including a hand-held portable and a mobile response signal to the nearby base, which then informs
vehicle on land (also known as land mobile), a ship, or the switching center about the location of the phone.
an aircraft. To begin, let us see how a typical mobile The switching center assigns a channel, and the call is
communications system involving land mobiles and base completed.
stations works [6]–[9]. 3) Registration: A mobile is normally located by trans-
mitting a paging message from various base stations. When
a large number of base stations are involved in the paging
A. Base Station process, it becomes impractical and costly. This problem is
The area served by a mobile phone system is divided in avoided by a registration procedure where a roaming phone
small areas known as cells. Each cell contains a base station, registers with a base closer to itself. This information may
which is connected to a switching center, communicates be stored with the switching center of the area as well as
to mobile phones on the site by radio links, and connects the home switching center of the phone. The home base
these mobiles to the public switching telephone network. of the phone is the one where it is permanently registered.
A typical setup is shown in Fig. 1. Two types of radio Once a call is received for this phone, its home switching
channels are used: control channels to carry control signals center contacts the switching center where the phone is
and traffic channels to carry messages. currently roaming. Paging in the vicinity of the previous
In mobile communications literature, the transmission known location helps to locate the phone. Once it responds,
from a base station to a mobile is given many names, the call may be connected as discussed previously.
such as downstream, forward link, and downlink. The
corresponding terms for the transmission from a mobile
to a base are upstream, reverse link, and uplink. B. Multiple-Access Schemes
1) Call Initiation from a Mobile: When a mobile phone is The range of frequencies available for mobile commu-
switched on, it scans the control channels and is tuned nications is utilized in a number of ways, referred to as
to the channel with the strongest signal, usually arriving multiple-access schemes. Three basic schemes are FDMA,
from the nearest base station. The phone user identifies TDMA, and CDMA.
itself and establishes authorization to use the network. The The standard analog FDMA scheme allocates different
base station then sends this message to the switching center carrier frequencies to different users. A TDMA scheme,

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1033

useful for digital signals, allocates different time slots to
different subscribers using the same carrier frequency and
thus interleaves signals from various users in an organized
manner. Traffic in the base-to-mobile direction is separated
from that in the mobile-to-base direction by either using
different carrier frequencies or alternating in time. The two
schemes are referred to as FDD and TDD. The use of
separate frequencies for transmission in both directions does
not require as precise a synchronization of data flowing
in the two directions as does the alternative transmission
method. Details on various aspects of the TDMA scheme,
including the capacity of a system using this scheme, may
be found in [10] and [11].
A CDMA scheme, on the other hand, is a spread-
spectrum method that uses a separate code for each user.
These codes are large PN sequences that spread the spec-
trum over a larger bandwidth, simultaneously reducing
the spectral density of the signal. Various CDMA signals
occupy the same bandwidth and appear as random noise to
each other. A general discussion on various characteristics
of the CDMA scheme, its design considerations, and related
issues is provided in [8] and [12]–[15].
In theory, the capacity provided by the three multiple-
access schemes is the same and is not altered by dividing
the spectrum into frequencies, time slots, or codes, as
explained in the following example [12]. Assume that
there are six carrier frequencies available for transmission
covering the allocated spectrum. In a system using the Fig. 2. Channel usage by FDMA, TDMA, and CDMA [12].
FDMA scheme, six frequencies are assigned to six users,
and six simultaneous calls may be made. TDMA generally
requires a larger bandwidth than FDMA, so a system using The SDMA scheme is based upon the fact that a signal
this scheme creates two TDMA channels and divides each arriving from a distant source reaches different antennas in
into three time slots, serving six users. A CDMA channel an array at different times due to their spatial distribution,
requires a larger bandwidth than the other two and serves and this delay is utilized to differentiate one or more users
six calls by using six codes, as illustrated in Fig. 2. in one area from those in another area. The scheme allows
In practice, however, the performance of each system an effective transmission to take place in one cell without
differs, leading to different system capacities [11], [13]. disturbing the transmission in another cell. Without the use
Furthermore, each scheme has its advantages and disadvan- of an array, this is accomplished by having a separate base
tages, such as complexities of equipment design, robustness station for each cell and keeping cell size fixed, whereas
to system parameter variation, and so on. For example, using space diversity, the shape of a cell may be changed
TDMA processes signals from all users simultaneously, dynamically to reflect the user movement. Thus, an array
requiring complex time synchronization of the different of antennas creates an extra dimension in this arrangement
user data. This is not the case for CDMA, which pro- by providing dynamic control in space [20].
cesses individual data independently at the receiver. It
does, however, require code synchronization. It is argued C. Propagation Characteristics
in [15] that although there does not appear to be a single An understanding of propagation conditions and chan-
scheme that is the best for all situations, the CDMA scheme nel characteristics is important to the efficient use of a
possesses characteristics that give it distinct advantages transmission medium. Lately, attention has been given to
over other schemes. An interesting discussion on these understanding the propagation conditions where a mobile
and related issues encompassing the recent research on is to operate, and many experiments have been conducted
multiple access, along with a large reference list, is pro- to model the channel characteristics. Many of the results
vided in [16]. Multiple access in digital networks is also can be found in recent review papers [5], [21], [22] and
discussed in [17]. the references therein. Some other references for channel
The SDMA scheme, also referred to as space diversity, measurements and modeling are [23]–[27].
uses an array of antennas to provide control of space by 1) Fading Channels: The signal arriving at a receiver is
providing virtual channels in an angle domain [18], [19]. a combination of many components arriving from various
Using this scheme, simultaneous calls in different cells can directions as a result of multipath propagation. It depends
be established at the same carrier frequency. upon terrain conditions and local buildings and structures,

1034 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

causing the received signal power to fluctuate randomly as larger than the symbol duration and causes intersymbol
a function of distance. Fluctuations on the order of 20 dB interference (ISI) in digital communications. The ISI may
are common within the distance of one wavelength. This be reduced to a certain degree by using equalizers in TDMA
phenomenon is called fading. One may think of this signal and FDMA systems.
as a product of two variables. 4) Link Budget and Path Loss: Link budget is a name
One of the components, referred to as the short-term given to the process of estimating the power at the
fading component, changes faster than the other one and receiver site for a microwave link, taking into account the
has a Rayleigh distribution. The second component is a attenuation caused by the distance between the transmitter
long-term or slow varying quantity and has log-normal and the receiver. This reduction is referred to as the path
distribution [8], [28]. In other words, the local mean varies loss. In free space, the path loss is proportional to the
slowly with log-normal distribution, and the fast variation second power of the distance, that is, the distance power
around the local mean has Rayleigh distribution. The fluc- gradient is two. In other words, by doubling the distance
tuation in the local mean is caused by shadowing and thus between the transmitter and the receiver, the received power
is referred to as shadow fading, whereas the fast change is reduced to one-fourth of the original amount.
in signal amplitude is caused by the phase differences in For a mobile communications environment utilizing fad-
signal components and is referred to as multipath fading [5]. ing channels, the distance power gradient varies and de-
A stationary subscriber may also observe fading when the pends upon the propagation conditions. Experimental re-
differential phases of various multipath components change sults show that it ranges from a value lower than two in
fast with frequency [21]. Such fading is referred to as indoor areas with large corridors to as high as six in metal
frequency-selective fading. When the fading is independent buildings. For urban areas, the path loss between the base
of frequency, it is referred to as flat fading. and the cell site is often taken to vary as the fourth power
A movement in a mobile receiver causes it to encounter of the distance between the two [5].
fluctuations in the received power level. The rate at which Normal calculation of the link budget is made by calcu-
this happens is referred to in mobile communications lit- lating the C/N ratio where noise consists of background and
erature as the fading rate [29], and it depends upon the thermal noise and the system utility is limited by the amount
frequency of transmission and the speed of the mobile. For of this noise. However, in mobile communications systems,
example, a mobile on foot operating at 900 MHz would the interference due to other mobile units is a dominant
cause a fading rate of about 4.5 Hz, whereas a typical noise compared to the background and man-made noise.
vehicle mobile would produce a fading rate of about 70 Hz. For this reason, these systems are limited by the amount of
2) Doppler Spread: The movement in a mobile also total interference present rather than the background noise,
causes the received frequency to differ from the transmitted as in the other cases. In other words, the SIR is the limiting
frequency due to the Doppler shift resulting from its relative factor for a mobile communications system rather than the
motion. As the received signals arrive along many paths, SNR, as is the case for other communications systems.
the relative velocity of the mobile with respect to various The calculation of link budget for such interference-limited
components of the signal differs, causing the different systems involves calculating the carrier level above the
components to yield different Doppler shifts. This can be interference level contributed by all cochannel sources [27].
viewed as spreading of the transmitted frequency and is
referred to as the Doppler spread. The width of the Doppler D. Channel Assignment
spread in frequency domain is closely related to the rate of The generic term “channel” is normally used to denote a
fluctuations in the observed signal [5]. frequency in the FDMA system, a time slot in the TDMA
3) Delay Spread: Due to the multipath nature of propa- system, and a code in the CDMA system, or a combination
gation in the area where a mobile is being used, it receives of these in a mixed system. Two channels are different if
multiple and delayed copies of the same transmission, they use different combinations of these at the same place.
resulting in spreading of the signal in time. The delay spread For example, two channels in an FDMA system use two
may range from a fraction of a microsecond in urban areas different frequencies.
to something on the order of 100 s in a hilly area, which Channel assignment is a complex process where a finite
restricts the maximum signal bandwidth between 40 and number of channels are assigned to various base stations
250 kHz. This bandwidth is known as coherence bandwidth. and mobile phones. In a system with fixed channel assign-
The coherence bandwidth is defined as the inverse of the ment, channels are assigned to different cells during the
delay spread [30]. For digital modulated schemes, the signal planning stage, and the assignment rarely changes to reflect
bandwidth is the inverse of the symbol duration. the traffic needs. A channel is assigned to a mobile at the
For a signal bandwidth above the coherence bandwidth, initiation of the call and the mobile communicates with the
the different frequency components of the signal arrive base using this channel until it remains in the cell.
at a receiver at different times, and the channel becomes Dynamic channel assignment, on the other hand, is an
frequency selective. Frequency-selective channels are also efficient way of channel usage in a multiple-user environ-
known as dispersive channels, whereas the nondispersive ment. With this arrangement, a channel with the minimum
channels are referred to as flat fading channels. A channel interference is found before assignment. The interference
becomes frequency selective when the delay spread is level of all the channels used and unused is monitored

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1035

periodically, and the channel assignment during the call
may be changed from the one with high interference to
the one with low interference, the so-called quiet channel
[29], [31]. As the interference environment is constantly
changing due to the movement of mobiles, this ensures that
the performance of the system is not affected adversely as
long as there are quieter channels available.
Monitoring traffic and searching for quieter channels is
a complex process with heavy computational demand. A
study [32] of a dynamic channel-assignment scheme based
upon neural networks shows the possibility of reducing
this complexity without degrading the handoff performance
compared to conventional methods and argues about the
suitability of neural networks for dynamic channel assign-
ment. Though it is not necessary that the neural networks
always provide the optimal solution to the assignment prob-
lem, some improvement in channel utilization is achievable, Fig. 3. Frequency assignment for cellular mobile communica-
as reported in [33]. Improvements in performance of mobile tions.
communications systems using neural networks for channel
assignment have also been discussed in [34]–[38].
to complete the assignment to all cells. For this case, the
There exist a number of other channel assignments be-
minimum distance between any two cells using the same
tween the two extremes of fixed and dynamic, includ-
frequency is equal to one arm of the hexagon.
ing flexible channel assignment and channel borrowing
The cells that use the same set of channels are known as
schemes [39]. Both schemes are a variation of the fixed
cochannel cells. The cells with frequency A in the figure are
channel assignment. In the former, the assignment is peri-
cochannel cells. The interference caused by the radiation
odically altered to reflect traffic needs, whereas in the latter,
from these cells is referred to as cochannel interference.
unused channels in a cell are borrowed by a congested cell.
For the system to function properly, this interference needs
The total transmitted power of the base station sometimes
to be minimized by limiting the power transmitted by the
is also considered as a criterion for channel allocation [40].
mobiles as well as by the base stations in cochannel cells.
More information on channel assignment and its effect on
The amount of transmitted power depends upon the channel
the performance of the system may be found in [39], [41],
reuse distance. For more discussion on this and related
and [42] and the references therein.
topics, see, for example, [11]–[14].
E. Channel Reuse
The total number of channels in a system is limited, F. Handoff
which in turn limits the capacity of the system to sustain A mobile phone’s movement may cause it to be far
simultaneous calls. The capacity may only be increased by away from the original base station through which the
using each traffic channel to carry many calls simultane- phone is connected to the switching center. This distance
ously. One way of accomplishing this is to use the same is detected by the switching center monitoring the signal
channel again and again. To do so, mobile phones using strength arriving from the phone at the base station. Once
the same channel have to be far away from each other in the signal becomes too weak, the switching center reassigns
order to avoid disturbance. A minimum distance between a new traffic channel via a base station closer to the phone
two cells using identical channels is required, known as the and asks the phone to tune to this new channel. This is
channel reuse distance. The capacity of the system depends known as handoff or handover, a process that is generally
on this distance. transparent to the mobile user. In a typical call, there may
The concept of channel reuse may be understood from be several handoffs.
Fig. 3, which shows how three frequencies—A, B, and The handoff needs to be performed based upon certain
C—are assigned to various hexagonal-shaped cells. As- established policy, which may involve the measured power
signment of frequencies to different cells is accomplished level and the quality of reception. The mean signal strength
by starting with a cell and assigning it a frequency, for decreases as a mobile moves away from a base, whereas
example, A. Then the three branches, which are separated it increases when a mobile approaches a base. A handoff
by 120 , are successively assigned frequencies B, C, A, policy based upon this fact may use the relative strength of
B, C, and so on. Next, a cell with frequency B is taken signals received from various base stations. The measured
as a center cell, and frequencies C, A, B, C, A, B, are as well as predicted signal strength may be involved in
assigned to the three branches, which are separated by 120 . the decision making. A policy based upon the quality of
Similarly, when a cell with frequency C is taken as a center reception may involve call blocking probability as well as
cell, frequencies A, B, C, A, B, C, are assigned to the handoff blocking probability [43]. A policy based upon
three branches. The procedure may be repeated if desired a minimum number of handoffs may lead to poor com-

1036 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

 Table 1 Comparison of Microcell and
munication quality. A good handoff scheme should aim Macrocell System Parameters
to minimize the effects of handoff on calls, including the
noticeable disturbances [39] that may be severe in some
services, such as live video transmission.
In a system where different cells use different frequen-
cies, the handoff requires switching frequencies and is
referred to as a hard handoff. In systems where all cells use
the same frequency but different codes, as in the case of
CDMA, only codes need to be switched, and the process is
known as a soft handoff. Soft handoff can be used between
sectors of the same base station. It is also being proposed for
satellite-based mobile communications systems [15]. Some monitors its received power—or the SNR, as the case may
discussion on traffic generated by handoffs may be found be—and sends the control signal to the transmitter to control
in [44]. its power as required. Sometimes a separate pilot signal is
used for this purpose.
G. Cell Splitting Power control reduces the near-far problem in CDMA
systems and helps to minimize interference near the cell
Each cell has a limited channel capacity and thus can
boundaries when used in forward link [13], [15], [39].
only sustain so many mobiles at a given time. Once the
demand in that cell exceeds this limit, the cell is subdivided
J. Mobile Radio and Cellular System
into smaller cells. Each new cell has its own base station
and frequency allocation. The power of the base station Conventional mobile radio systems use the highest possi-
transmitters is adjusted to reflect the new boundaries. The ble antennas for their base stations and radiate the maximum
power transmitted by new base stations is less than that possible power so that the mobile units can communicate
transmitted by the old one. with the base from as far away as possible. In a cellular
A consequence of cell splitting is that frequency has to system, on the other hand, the objective is to divide the
be assigned again, which affects neighboring cells. It also area to be served into small cells. Each cell is assigned its
increases the handoff rate, as the cells are now smaller own base and frequency for communication. Cells that are
and a mobile is likely to cross cell boundaries more often far apart use the same frequencies, as previously discussed.
compared to when the cells are big. Due to altered signaling 1) Microcell System: As the cells are split by reducing
conditions, this also affects the traffic in control channels. transmitted power and redefining cell boundaries, their size
becomes very small. At a radius less than about 1 km, the
H. Cell Sectorization system is referred to as a microcell system. Conversely,
a cellular system with larger cells is referred to as a
Cell sectorization refers to the case where a given cell is macrocell system. Various parameters for the two systems
subdivided into several sectors, all of which are served by are compared in Table 1 [27].
the same base station. This is normally done by employing 2) Picocell System: When cell sizes are reduced to less
directional antennas such that the energy in each sector than about 100 m, covering areas such as large rooms,
is directed by separate antennas. This has the effect of corridors, underground stations, large shopping centers, and
increased channel capacity similar to cell splitting. The so on, such cellular systems are sometimes referred to
sectors use the same base station, however, and thus do as picocell systems with antennas mounted below rooftop
not incur the cost of establishing new base stations asso- levels or in buildings. These in-building areas have different
ciated with cell splitting. This helps in reducing cochannel propagation conditions than those covered by macrocell and
interference as the energy is directed in the direction of the microcell systems and thus require different considerations
sector, which does not cause interference in the cochannel for developing channel models. Details on various models
cells, particularly in cochannel cells in the opposite direc- to predict propagation conditions may be found in [22].
tion to the sector. As in the case of cell splitting, this also Sometimes the picocell and microcell systems are also re-
affects the handoff rate. ferred to as cordless communications systems, with the term
cellular identifying a macrocell system. Mobiles within
I. Power Control these smaller cell systems are termed cordless terminals
It is important that a radio receiver receive a power level or cordless phones [1], [3], [45].
that is high enough for its proper function but not so high Providing in-building communications services using
as to disturb other receivers. To this end, two methods wireless technology and based upon cell shapes dictated
are used. One is concerned with maintaining a constant by floors and walls is a feasible alternative. It offers many
power level at the receiver by transmitter power control. advantages, including cheaper and easier installation, as it
In the other, the SNR is kept constant. In both the cases, does not require the laying of cables, etc. It is argued in
the receiver controls the power of the transmitter at the [46] that radio frequencies in the 18-GHz band are ideal
other end. For example, a base would control the power for such services, as these do not penetrate concrete and
transmitted by mobile phones, and vice versa. A receiver steel structures, which eliminates the problem of cochannel

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1037

 it may provide a number of spot beams with different
beams using different frequencies, as in cellular radio. The
communications satellites tend to complement the terrestrial
network and play a major role in areas where the latter is
not competitive or is underdeveloped.
GEOS’s have a number of drawbacks when it comes
to global voice communication. These satellites are placed
around 36 000 km above the earth, and two-way propa-
gation delay on the order of 0.6 s encountered by a signal
makes it unacceptable for voice communication. There also
exists a difficulty in covering the area of the globe far
south and far north of the equator using GEOS’s. There are
problems in these regions as the satellite appears close to the
horizon. Even using large antennas, communication beyond
75 north and south of the equator is not satisfactory.
Fig. 4. Concept of overlaying [49]. Blockage of the satellite by large buildings, even when
one is situated at about 40–45 north or south of the
interference. These frequencies offer huge bandwidth equator, is also a problem. The communications system
and require millimeter-size antennas, which are easy to requires high-power transmitters with large antennas to
manufacture and install. overcome the propagation loss suffered by the signal due
3) Mixture of Cells and Overlaying: Small cell systems to a large distance. Such conditions are not practical for
make very efficient use of the spectrum, allowing large mobile systems.
frequency reuse and resulting in an increased system To overcome these problems, a number of low-orbit
capacity. These systems are not suitable for all conditions, satellites have been proposed for mobile satellite commu-
however, due to their large handover requirement. A system nications systems. These include LEOS’s at an altitude of
of mixed cells with the concept of overlaying is discussed around 1000 km, MEOS’s at an altitude of around 10 000
in [43], [47], and [48]. In this system, a hierarchy of cells, km, and HEOS’s with varying altitudes.
with their own channel allocation and handover scheme, Though LEOS and MEOS require less power and cause
is assumed to exist. A macrocell system is assumed to be less delay, making portable mobiles viable, a large number
at the top of the hierarchy, with smaller cell systems at of satellites are required for global coverage. There is a
the bottom. A mobile with high mobility is assigned to a requirement of fast handoffs, as the satellites move rapidly
macrocell system, whereas a mobile with low mobility is and are in view only for a short time. There is also a concern
assigned to a smaller cell system. Designs incorporating of limited lifetime (five–ten years) due to orbital decay,
various combinations of different multiple-access schemes requiring regular replacement [1].
reflect the ease of handover and other traffic-management A number of satellite communications systems are
strategies. The SDMA scheme has an important role to play planned and will be in operation in the near future [50].
in this concept, with various beams placed at the bottom Among these are the ARIES, GLOBESTAR, IRIDIUM,
of the hierarchy. A typical setup [49] showing coverage of and ODYSSEY systems, which will provide full global
various systems is shown in Fig. 4. coverage with voice, data, and fax services. These are
known as big-LEO systems. The word “big” indicates that
these satellites would have enough power and bandwidth
K. Satellite Mobile Communications to provide near toll-quality voice services to hand-held
A variety of configurations involving satellites for wire- portable and vehicle mobiles, along with other services [50].
less communications exists. For example, in global posi- Little-LEO systems, such as LEOSAT, STARNET, and
tioning systems, signals received from a number of orbiting VITASET, would provide global coverage with nonvoice
satellites are used to determine the position of a receiver. services. These would provide low-bit-rate services and are
A direct broadcasting system uses satellites to transmit expected to be of small size and low mass. A number of
signals to television users. In both of these systems, the systems along with some other parameters are shown in
transmission is generally one way in nature. For a com- Table 2 [50], [51].
munications satellite system, on the other hand, a two- The IRIDIUM system is briefly discussed here as a
way communication is required. Though the requirements typical system to provide a feel for its operation, service
for different systems—such as maritime, aeronautical, and provided, and other parameters.
land-mobile systems—may be different, the basic principle IRIDIUM is proposed by Motorola to provide global
of operation remains the same. Basically, a satellite acts coverage for voice, data, fax, paging, and so on. It would
as a relay station between a mobile and the base station, have 66 satellites (LEO 780 km) in six polar orbit planes
also known as a hub station. A geostationary satellite may at 86.4 inclination, each with a mass of 700 kg, and
provide a large coverage similar to the large and fixed would be operational in 1998. Each satellite would provide
coverage of a mobile radio communications system, or 48 beams. This gives in total 3168 cells. These would

1038 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

Table 2 Comparison of Parameters for
Some Mobile Satellite Systems
In contrast to the first-generation analog systems, second-
generation systems are designed to use digital transmission,
to have a separate dedicated channel for exchange of control
information between base and mobiles, and to employ
TDMA or CDMA as a multiple-access scheme. These
systems include the Pan-European GSM and DCS 1800
systems, the North American dual-mode cellular IS-54
system, the North American IS-95 system, and the Japanese
PDC system. The GSM, DCS1800, IS-54, and PDC systems
use TDMA, whereas IS-95 uses CDMA. Both the IS-54 and
IS-95 systems are designed to operate in the frequency band
used by the AMPS system [1], [59].
The first-generation analog cordless phones are designed
to communicate with a single base, effectively replacing
a telephone cord with a wireless link to provide terminal
mobility in a small coverage area, such as within a house or
office. The second-generation digital cordless systems are
being developed with the aim of using the same terminal in
move over the surface of the earth as satellites move, and residential as well as public-access areas, such as offices,
2150 cells would be active at any time to cover the whole shopping centers, and so on, and of being able to receive
earth. The beams would be generated by three phased array and originate calls. Digital cordless systems include CT2,
panels, each generating 16 simultaneous shaped beams. The a British standard originally adopted in 1987 and later
system would complement terrestrial systems. It would use augmented in 1989, the DECT standard, and PHS of Japan.
1616–1626.5 MHz for both uplink and downlink and 23 CT2 employs an FDMA scheme, providing one channel per
GHz for intersatellite links. A mobile unit would first use carrier, whereas the DECT and PHS systems are designed
the local cellular ground-base system and then a satellite to use TDMA, providing 12 and four channels per carrier,
using one of the frequencies. A satellite would demodulate respectively [1], [3], [59].
the signal to get the address of the destination and then The third-generation mobile communications systems are
send the message to that satellite, which is in the viewing being studied worldwide under the names of UMTS and
area of the destination cell. It would use FDMA for uplink FPLMTS. The aim of these systems is to provide users
and time division multiplexing for the downlink. advanced communications services having wideband ca-
More details on various aspects of mobile communica- pabilities and using a single standard. Details on various
tions involving satellites may be found in [50]–[57]. systems under investigation can be found in [1], [3], [16],
[49], [55], and [60]. In third-generation communications
systems, satellites are going to play a major role in provid-
L. Progress and Trends in Mobile Communications ing global coverage. Adaptive antenna array processing has
Mobile communications technology has come a long the potential to provide designers with an extra dimension
way since the pioneering work at AT&T Bell Laboratories of SDMA along with FDMA, TDMA, and CDMA in
during the 1960’s and 1970’s. There, researchers coupled solving this mammoth task. Details on the use of arrays
the idea of frequency reuse with digital switching, leading for mobile communications are provided in the remainder
to the opening of the first operational cellular system in of this paper.
Chicago in October 1983 [58]. The first-generation analog
cordless phones and cellular systems became popular using
the design based upon a standard known as AMPS. Similar III. ANTENNA ARRAY TERMINOLOGY
standards based upon FDMA were developed around the This section introduces some common terms relating to
world, including TACS, NMT 450, and NMT 900 in Eu- antenna arrays.
rope; ETACS in the United Kingdom; C-450 in Germany; Antennas in general may be classified as omnidirectional,
and NTT, JTACS, and NTACS in Japan. The handoff directional, phased array, adaptive, and optimal. An omni-
decision for these systems is based upon the power received directional antenna has equal gain in all directions and is
at the base or, in the case of C-450, the round-trip delay. also known as an isotropic antenna. Directional antennas,
AMPS systems, used in Australia, the United States, on the other hand, have more gain in certain directions
Canada, and Central and South America, use an 824–849 and less in others. The direction in which the gain of
MHz band for transmission from mobiles to base and an these antennas is maximum is referred to as the boresight
869–894 MHz band for transmission from base to mobiles. direction of the antenna. The gain of directional antennas
There are 832 channels of 30-kHz width. The system in the boresight is more than that of the omnidirectional
typically makes use of 12 frequency reuse plans with antennas and is measured with respect to the gain of the
omnidirectional antennas or seven frequency reuse plans omnidirectional antennas. For example, a gain of 10 dBi
with three sectors in each cell [1]. (sometimes indicated by dBic or simply dB) means the

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1039

 noise ratio, SINR) is also referred to as optimal combining
in the mobile communications literature [29], [61], [62].

A. Beam Steering and Switching

The signals induced on different elements of an array
are combined to form a single output of the array. A plot
of the array response as a function of angle is normally
referred to as the array pattern or beam pattern. It is also
called a power pattern when the power response is plotted.
It shows the power received by the array at its output
from a particular direction due to a unit power source in
that direction. This process of combining the signals from
different elements is known as beam forming. The direction
in which the array has maximum response is said to be the
beam pointing direction. Thus, this is the direction where
the array has the maximum gain. For a linear array, when
signals are combined without any gain and phase change,
this is broadside to the array, that is, perpendicular to the
Fig. 5. Block diagram of a narrow-band adaptive antenna system. line joining all elements of the array.
The array pattern drops to a low value on either side of
the beam pointing direction. The place of the low value is
power radiated by this antenna is 10 dB more than that normally referred to as a null. Strictly speaking, a null is
radiated by an isotropic one. It should be noted that the a position where the array response is zero. However, the
same antenna may be used as a transmitting antenna or as a term is generally misused to indicate the low value of the
receiving antenna. The gain of an antenna remains the same pattern. The pattern between the two nulls on either side
in both cases. The gain of a receiving antenna indicates the of the beam pointing direction is known as the main lobe.
amount of power it delivers to the receiver compared to an The width of the beam (main lobe) between the two half-
omnidirectional antenna. power points is called the half-power beamwidth. A smaller
A phased array antenna uses an array of simple antennas, beamwidth results from an array with a larger extent. The
such as omnidirectional antennas, and combines the signal extent of the array is known as the aperture of the array.
induced on these antennas to form the array output. Each Thus, the aperture of the array is the distance between the
antenna forming the array is known as an element of the two farthest elements in the array. For a linear array, it is
array. The direction where the maximum gain would appear equal to the distance between the elements on either side
is controlled by adjusting the phase between different an- of the array.
tennas. The phases of signals induced on various elements For a given array, the beam may be pointed in dif-
are adjusted such that the signals due to a source in the ferent directions by mechanically moving the array. This
direction where maximum gain is required are added in- is known as mechanical steering. Beam steering can also
phase. This results in the gain of the array (or equivalently, be accomplished by appropriately delaying the signals
the gain of the combined antenna) are equal to the sum of before combining them. This process is known as electronic
the gains of all individual antennas. steering, and there are no mechanical movements. For
The term adaptive antenna is used for the phased array narrow-band signals, the phase shifters are used to change
when the gain and the phase of the signals induced on the phase of signals before combining them.
various elements are changed before combining to adjust The required delay may also be accomplished by insert-
the gain of the array in a dynamic fashion, as required ing varying lengths of coaxial cables between the antenna
elements and the combiner. Changing the combinations of
by the system. In a way, the array adapts to the situation,
different lengths of these cables leads to different point-
and the adaption process is normally under the control of
ing directions. Switching between different combinations
the system. A block diagram of a typical adaptive antenna
of beam steering networks to point beams in different
array system is shown in Fig. 5.
directions is sometimes referred to as beam switching.
An optimal antenna is one in which the gain and phase
of each antenna element is adjusted to achieve the optimal
performance of the array in some sense. For example, to B. Conventional Beam Forming
obtain maximum output SNR by canceling unwanted inter- Adjusting only the phase of signals from different el-
ferences and receiving the desired signal without distortion ements to point a beam in a desired direction is the
may be one way of adjusting gains and phases of each conventional method of beam pointing or beam forming.
element. This arrangement where the gain and phase of each The gain of each signal is kept the same. This determines
antenna element is adjusted to obtain maximum output SNR the total gain of the array in the beam pointing direction,
(sometimes also referred to as signal-to-interference-and- also known as the look direction. The shape of the antenna

1040 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

pattern in this case is fixed, that is, the positions of the these are not in the direction of the desired source. In
side lobes with respect to the main beam and their level situations where the directions of these interferences are
are unchanged. In other words, when the main beam is known, cancellation is possible by placing the nulls in the
pointed in different directions by adjusting various phases, pattern corresponding to these directions and simultane-
the relative positions of the side lobes with respect to the ously steering the main beam in the direction of the desired
main lobe do not change. This, however, may be changed signal. Beam forming in this way, where nulls are placed
by adjusting the gain and phase of each signal to shape the in the directions of interferences, is normally known as
pattern as required. The amount of change depends upon null beam forming or null steering. The cancellation of one
the number of elements in the array. interference by placing a null in the pattern uses one degree
The gain and phase applied to the signals derived from of the freedom of the array.
each element may be thought of as a single complex Null beam forming uses the directions of sources toward
quantity hereafter referred to as the weighting applied to which nulls are placed for estimating the required weighting
the signals. If there is only one element, no amount of on each element. There are other schemes that do not
weighting can change the pattern of that antenna. With two require directions of all sources. A constrained beam-
elements, however, changing the weighting of one element forming scheme uses the steering vector associated with
relative to the other may adjust the pattern to the desired the desired signal and then estimates the weights by solving
value at one place, that is, one is able to place one minima an optimization problem. Knowledge of the steering vector
or maxima anywhere in the pattern. Similarly, with three associated with the desired signal is required to protect the
elements, two positions may be specified, and so on. Thus, signal from being canceled. In situations where the steering
with an -element array, one is able to specify vector associated with the signal is not available, a reference
positions. These may be one maxima in the direction of the signal is used for this purpose.
desired signal and minimas (nulls) in the directions
of unwanted interferences. This flexibility of an -element
array to be able to fix the pattern at places is known E. Diversity Combining
as the degree of freedom of the array. For an equally spaced The discussion here concentrates on spatial diversity as
linear array, this is similar to an degree polynomial provided by an array of antennas. Unless explicitly stated,
of adjustable coefficients, with the first coefficient diversity refers to spatial diversity throughout this paper.
having the value of unity. Diversity combining [65] is used to overcome the prob-
lem of fading in radio channels and utilizes the fact that
C. Steering Vector the signals arriving at different locations fade at different
rates. A system employing a diversity combiner uses signals
The steering vector contains the responses of all elements
induced on various antennas placed a few wavelengths apart
of the array to a narrow-band source of unit power. As the
at different locations and combines these signals in one
response of the array is different in different directions, a
of many ways [66]. For example, an equal gain combiner
steering vector is associated with each directional source.
adjusts the phases of the desired signals and combines them
The uniqueness of this association depends upon the array
in-phase after equal weighting. The maximal ratio combiner
geometry [63].
applies weights in proportion to the SNR and combines the
For an array of identical elements, each component
weighted signals in-phase, whereas a selection diversity
of this vector has unit magnitude. The phase of its th
combiner selects the signal from one of the antennas for
component is equal to the phase difference between signals
processing. The selection may be based upon the power of
induced on the th element and the reference element due
the desired signal, the total power, or the SIR available at
to the source associated with the steering vector. As each
each antenna. The selection based upon the SIR is most
component of this vector denotes the phase delay caused
effective in combating cochannel interference, whereas the
by the spatial position of the corresponding element of
equal gain combiner provides the lowest outage probability
the array, this vector is also known as the space vector.
(the probability that the channel is not suitable for use).
It is also referred to as the array response vector, as
In general, increasing the number of antennas results in
it measures the response of the array due to the source
more reduction in channel fading. However, in practice,
under consideration. In multipath situations, such as exist
10–20 antennas seem to provide satisfactory results [67]. In-
in mobile communications, it also denotes the response of
formation on diversity receiver design may be found in [9].
the array to all signals arising from the source [64]. In this
Diversity combining is different from antenna array pro-
paper, the terms steering vector, space vector, and array
cessing and beam forming. It combines the signals at the
response vector are used interchangeably.
baseband or at the IF level to increase the signal level with-
out affecting the individual antenna pattern. Beam-forming
D. Null Beam Forming techniques, on the other hand, exploit the differential phase
The flexibility of array weighting to being adjusted to between different antennas normally at RF level to modify
specify the array pattern is an important property. This the antenna pattern of the whole array. In this arrangement,
may be exploited to cancel directional sources operating at once the signals are combined, the whole of the array has
the same frequency as that of the desired source, provided a single antenna pattern.

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1041

 Diversity combining is useful for combating fading and
is efficient in the absence of a common interference on all
the antennas. In the presence of cochannel interferences,
however, the array processing methods are able to provide
better performance by canceling interferences [61]. An
overview of array processing schemes will be provided in
the second part of this paper.

IV. USE OF ARRAYS IN MOBILE

COMMUNICATIONS SYSTEMS
Arrays may be used in various configurations for mo-
bile communications, some of which are discussed here.
This section will consider the improvements made possible
by using an array. The systems considered include base-
mobile, indoor-mobile, satellite-mobile, and satellite-to-
satellite communications systems. As no distinction is made
between the use of arrays for the purpose of transmis-
sion and reception in this paper, a short discussion on
transmitting arrays is provided in this section.

A. Base-Mobile System
A general description of the base-mobile system was
provided in Section II. The system consists of a base station
situated in a cell and serves a set of mobiles within the cell.
It transmits signals to each mobile and receives signals from
them. It monitors their signal strength and organizes the
handoff when mobiles cross the cell boundary. It provides
the link between the mobiles within the cell and the rest
of the network.
In this section, various scenarios are presented to show
Fig. 6. A typical setup showing different beams covering various
how an array could be used in such a system. The discussion mobiles [70].
will concentrate on the use of an array at the base station
as well as at mobiles. Some practical considerations to
implement arrays and blind estimation of the cochannel and thus may be able to use the same frequency or code,
signals are also discussed in this section. as the case may be. Fig. 6 shows a typical setup involving
1) Use of Array at a Base Station: In this section, a num- different beams covering various mobiles along with the
ber of possibilities are discussed where a base station uses directions of moving mobiles. It illustrates the situation at
an array of antennas. A base station having multiple anten- two time instants.
nas is sometimes referred to as having antenna diversity or This setup is different from the one discussed previously,
space diversity [68], [69]. where a number of beams of fixed shape cover the whole
a) Formation of multiple beams: In its simplest form, cell. Here, the beams are shaped to cover the traffic. As the
multiple antennas at the base station may be used to form mobiles move, the different beams cover different clusters
multiple beams to cover the whole cell site. For example, of mobiles, offering the benefit of transmitting the energy
three beams with a beamwidth of 120 each or six beams toward the mobiles. The arrangement is particularly useful
with a beamwidth of 60 each may be formed for the in situations where the mobiles move in clusters or along
purpose. Each beam then may be treated as a separate confined paths, such as highways. It is envisaged in [70]
cell, and the frequency assignment may be performed in that each mobile may be covered by a separate beam. Each
the usual manner. Mobiles are handed to the next beam as beam then would follow the mobile, reducing the handoff
they leave the area covered by the current beam, as is done problem to the bare minimum.
in a normal handoff process when the mobiles cross the c) Null formation: In contrast with steering beams to-
cell boundary. ward mobiles, one may adjust the antenna pattern such that
b) Formation of adaptive beams: An array of antennas it has nulls toward other mobiles. As mentioned, a null in an
with the capability to form independent beams may be used antenna pattern denotes a zero response. In practice, how-
at the base station as discussed in [20], [70], and [71]. The ever, that is seldom achievable, and one creates a pattern
array is used to find the location of each mobile, and then with a reduced response toward undesirable interferences.
beams are formed to cover different mobiles or groups of Formation of nulls in the antenna pattern toward cochan-
mobiles. Each beam may be considered as a cochannel cell, nel mobiles helps to reduce the cochannel interference in

1042 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

two ways. In the transmit mode, less energy is transmitted
from the base toward these mobiles, reducing the interfer-
ence from the base to them. In the receiving mode, this
helps to reduce the contribution from these mobiles at the
base.
d) Optimal combining: Canceling unwanted cochannel
interferences while an array is operating in receiving mode
is a very effective use of an antenna array. The process
combines signals received on various antennas in such a
way that the contribution due to unwanted interferences is
reduced while that due to a desired signal is enhanced.
Knowledge of the directions of the interferences is not
essential to the process’s functioning, while some char-
acteristics of the desired signal are required to protect it
from being canceled. Discussion on optimal combining to (a)
cancel cochannel interferences in digital mobile radio may
be found in [29] and [61].
These studies suggest the use of an array at the base
station for a TDMA system and adjustment of the array
weights during the receive cycle to enhance the signal
arriving from a particular mobile. It is argued that a system
using the same frequency for receiving and transmitting the
signals in different time slots may be able to use the com-
plex conjugate of these weights during the transmit cycle to
enhance the signal received at that mobile, simultaneously
suppressing this signal at the other mobiles. It relies on
the fact that the weights adjusted during the receive cycle
to reduce cochannel interference result in an array pattern
with nulls in the directions of cochannel mobiles. The use of
the complex conjugate of these weights during the transmit
cycle produces the same pattern, and thus no energy is (b)
transmitted toward the cochannel mobiles.
Fig. 7. Cell shape based upon traffic needs [20]. (a) Cells of fixed
By using constrained beam-forming techniques, one may shape. (b) Cells of dynamic shape.
be able to combine the formation of nulls in the directions
of unwanted mobiles while keeping the specified response
in the direction of a desired mobile. The use of constrained A study of optimal combining using a reference signal,
beam forming for mobile communications is discussed in presented in [74], indicates that better results are obtained
[19] and [71]–[73]. For proper operation, the direction of when the weights are estimated using the RLS algorithm
the desired signal is required. It is used for maintaining a than when using the LMS and SMI algorithms.
specified response in that direction. e) Dynamic cell formation: The concept of adaptive
Though constrained beam forming is very effective when beam forming may be extended to dynamically changing
the desired signal is a point source, its use in mobile cell shapes. Instead of having cells of fixed size, the use of
communications is limited, particularly in situations of array antennas allows the formation of a cell based upon
multipaths. Optimal combining using a reference signal is traffic needs, as shown in Fig. 7 [20].
more appropriate for this case. It requires a signal that An architecture to realize such a base station requires
is correlated with the desired signal. The scheme then the capability of locating and tracking the mobiles in
protects all signals that are correlated with this reference order to adapt the system parameters to meet the traffic
signal and adds them in-phase to maximize their combined requirements.
effect. It simultaneously cancels all waveforms that are Many of the DOA estimation techniques available in the
not correlated with this signal, resulting in a removal of literature assume an existence of point sources. This is
cochannel interferences. Thus, optimal combining using a not necessarily the case for land-mobile communications,
reference signal is able to make use of multipath arrivals of where scattering in the vicinity of a mobile causes a
the desired signal, whereas the constrained beam forming spreading of the source and the signal arrives at the base
treats them as interferences and cancels them. in the form of multipath, as if there were many radiating
A comparative study of a reference-signal-based algo- sources of varied power in the neighborhood of the mobile.
rithm and a method using the estimates of directions of Experiment results [75] indicate that at 870 MHz, the
sources [71] indicates the superiority of the former when estimated angular spread is around 3 when viewed from a
multipaths result in large spread of the desired signal. distance of 1 km. This corresponds to a scattering area with

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1043

a radius of about 50 m around the mobile. Considerations later updating the weights and tracking the mobile. The
for situations like this need to be given while tracking the user identification may also be used to reemphasize that
mobile and forming the beams of appropriate shapes [19], the current signal is from the desired user. In TDMA
[71]. Focusing the beam more than necessary may result in systems, a separate signal may be sent for this purpose.
a loss of gain, as found in the study reported in [76]. Similarly, for the CDMA system, a separate sequence may
Technology that allows cell shapes and sizes to be be transmitted in phase quadrature to the original signal.
changed based upon traffic conditions, channels to be The synchronization sequence transmitted in some systems
assigned dynamically as per traffic needs, and transmitter may also be used for this purpose.
power to be adapted according to receiver requirements 3) Blind Estimation of Cochannel Signals: A base station
is referred to as intelligent technology [20]. Apart from employing multiple sensors may be able to exploit the fact
variable cell size, adaptive transmitted power, and dynamic that signals arriving from different mobiles follow different
channel allocation, this technology also allows for variable paths and arrive at various elements at different times. This
speed for transmission and the use of adaptive modulation allows independent measurements of signals superimposed
and demodulation techniques. Discussion on some of these from different mobiles. This, along with the properties
issues can be found in [20] and [77]–[80]. of the modulation technique used, allows separation of
2) Practical Considerations: Some practical situations signals arriving from different mobiles. Thus, by using the
are considered here for acquiring the knowledge of the measured signals at various elements of the array at the
DOA, the array response vector, or the reference signal base, one is able simultaneously to separate all signals. The
required for the possible implementations of beam-forming process is referred to as the blind estimation of cochannel
techniques using an array. signals. It does not require knowledge of the directions
a) Using DOA information: Many systems use two or other parameters associated with mobiles, such as a
separate frequencies for transmit and receive modes. The reference signal or a training sequence, but exploits the
knowledge of the directions of mobiles obtained during temporal structure that might exist in signals inherited from
the receive mode may be used during the transmit mode the source of their generation, for example, the modulation
to transmit in the directions of mobiles in a multiplexed techniques used. Many studies on this topic have been
manner. Once the position of mobiles in a cell is determined reported in the literature [82]–[92].
during the receive mode, these mobiles are grouped It has been argued in [93] that most man-made signals
in different clusters, and then the antenna pattern is encountered in communications exhibit variation in some
adjusted such that the main beam is pointed toward one of their parameters in time, and these signals may be
cluster while nulls are formed in the directions of other modeled as cyclostationary signals. A random signal whose
clusters. The process is repeated to transmit to all the parameters vary in time with single or multiple periodicity
clusters. Transmission could take place simultaneously is referred to as a cyclostationary signal. For example, a
using parallel transmission circuits or by using a time- signal with second-order cyclostationarity is one whose
sharing arrangement of a single circuit. The nulls may autocorrelation function varies periodically with time. The
also be directed toward mobiles in other cochannel cells cyclostationarity property of communications signals has
with added implementation complexities caused by the been exploited by many authors [88], [92], [94] for blind
requirement for synchronization [81]. signal extraction.
b) Using array response vectors: A base station not re- The temporal structure inherent in BPSK signals is ex-
quiring the directions of the mobile may be able to form ploited in [82] to separate cochannel signals by solving
beams by estimating the array response vector correspond- a set of linear equations to obtain the array weights. The
ing to a mobile, as discussed in [64] for a CDMA system. method is based upon clustering the received signals as
The system uses two separate frequencies for transmission per their delays and reportedly has some implementation
and reception. In the receiving mode, the base station advantages over other similar methods [83], [95]. The
estimates the array response vector from a signal received other properties of signals that have been used for blind
from a mobile with whom the base is code locked, whereas signal extraction include constant modulus [96], [97], finite
in the transmit mode, the same is done by processing the alphabet [84], and higher order statistics [89], [98]. Use
feedback from the mobile to a special tone transmitted by of the source correlation compensation technique for blind
the base. The array response vector may also be estimated channel identification is made in [99].
from the knowledge of the user code in a CDMA system or 4) Use of Array at a Mobile: There are normally two
from the reference signal if available, as discussed in [29]. kinds of mobiles: hand-held portable mobiles and mobiles
c) Using a reference signal: The approach described in fixed on vehicles, with differing requirements of the type
[18] suggests that the base station acquires the reference of equipment they may carry. For example, it is possible
signal from the desired user and uses this to estimate the for a vehicle mobile to carry an array of antennas and the
weighting of each antenna before combining the signals to required processing equipment; however, the same would
produce the array output. It suggests that initial weights be difficult for the hand-held portable.
be estimated by using a specific signal transmitted for Application of optimal combining at a vehicular mobile
this purpose. Once the initial weights are determined, is discussed in [62] for a situation where desired signal and
the received signal then may serve as the reference for cochannel interferences arrive from many directions due to

1044 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

multipaths. It is argued that due to the distributed nature facility. Depending upon the type of array system utilized,
of sources, synthesis of antenna pattern with nulls toward many scenarios are possible.
interferences is not practical. It suggests the use of optimal a) Fixed-shape beams: In a simplistic situation, beams
combining of a desired signal arriving from various paths, of fixed shape and size may be generated to cover the area
using a reference signal transmitted from the base in the of interest, allowing normal handoff as the mobiles roam
form of a pilot signal for a spread-spectrum system. An from one cell to another or as the beams move in a low
example considering the worst-case scenario indicates that orbit multiple satellite system covering different areas. Pro-
one has about 35 iterations to adjust the weights of the array posed fixed-beam antennas using multibeam phased arrays
for the vehicular mobile system and suggests the use of the for mobile communications and featuring frequency-reuse
SMI algorithm for this case. A successful implementation facility and flexibility of variable beam power allocation
of an antenna array on the roof of a mobile vehicle using have been reported in [105].
a CMA to update the weights for a TDMA system is Traditionally, phased arrays have been used in a feed
described in [100]. Developments of a dual antenna system network to control the beam coverage area. A typical
for hand-held portables are reported in [30], [101], and system may consist of a high-gain large reflector antenna
[102]. along with an array of feed elements placed in the focal
plane of the reflector in a particular geometry able to
generate a limited number of fixed-shape spot beams. A par-
B. Indoor-Mobile Radio System ticular beam is selected by choosing a combination of feed
In a base-mobile communications system, a number of elements. The steering of beams is achieved by controlling
mobiles communicate through a base station, as discussed. the phases of signals prior to the feed elements. Dual
There are, however, a number of differences between circuitry is employed for transmit and receive modes for
a land-mobile system and an in-building mobile system the signal to flow in both directions. The capability of the
owing to the type of mobiles they serve and the environment system to generate multiple spot beams with independent
in which they operate. For an in-building mobile system, the power control and frequency use makes it attractive for
mobiles are hand-held portables moving on foot, whereas mobile communications. These capabilities, along with a
for a land-mobile system, they are generally vehicles. This frequency scanning system, have been evaluated for the
results in different fading rates for the two cases. The fading European geostationary satellite system in various ESA
rate for the indoor system is much slower than the other, contracts [105], [106].
implying that optimal combining could be implemented The frequency scanning system considered in [105] and
with ease using adaptive techniques [29], [61]. [106] uses an array of active antennas to provide a high-
The delay spread problem in the two cases is also gain beam with the capability to steer it at any user location
different. For a land-mobile system, the signals generally using frequency-dependent interelement phase shift. The
arrive in delayed clusters due to reflections from large same array is used for the transmit as well as for the
objects, such as hills and buildings. For an indoor system, receive mode. In the transmit mode, the signal is supplied
there exist a large number of scattering and reflections to antennas by an array of power amplifiers, whereas
inside a building. This causes the impulse response of the in the receive mode, the antennas feed the low noise
indoor radio channel to appear like a series of pulses [103]. amplifiers of the receiver. A comparison of the two systems
The use of adaptive antennas to improve such channels is indicates that the frequency scanning system outperforms
discussed in [104]. the conventional multispot system in terms of capacity,
required power per channel, and payload mass per channel
but requires a complex beam-forming network and mobile
C. Satellite-Mobile System terminal location procedures.
In this system, the mobiles (in this case mostly vehicles) b) Dynamic beams: A system using fixed-shape beams
directly communicate with the satellite, in contrast to a does not require knowledge of the traffic conditions, as
system where a base station acts as a repeater station does a system generating spot beams of varying shapes and
by communicating with the satellite on one hand and the sizes dictated by the positions of the mobiles (Fig. 8). For
mobiles on the other. In this case, multiple antenna elements downlink transmission, this helps in transmitting the energy
may be utilized on a mobile as well as on a satellite. in the direction of mobiles by controlling the shape and
1) Array on Satellites: It is possible to use an array on coverage of the beams (as discussed in [107]), which in turn
board a satellite and provide communication in a number of reduces cochannel interferences at the mobiles. This also
ways. For example, different frequencies may be allocated helps reduce the transmitted power due to its directed nature
to beams covering different areas such that each area of transmission, relaxing the need to generate large amounts
acts as a cell. This allows frequency reuse similar to of power onboard a satellite. To generate an arbitrarily
the base-mobile system discussed previously. The major shaped beam to be pointed at a desired location, the system
difference between this and the land-mobile system is the architecture requires an advanced beam-forming network
generation of beams covering different cells rather than with independent beam-steering capability.
having different base stations for different cells. An array Recent developments in monolithic MMIC technology,
mounted onboard the satellite provides the beam-generation which allows fabrications of power amplifiers, phase

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1045

 of signals caused by reflections in the vicinity of mobiles.
The problem of multipath fading and delayed arrival is also
not serious in this case.
2) Array on Mobiles: An array of antennas may be
mounted on a mobile (vehicle) to communicate directly
with a satellite, along with its control circuitry, to steer a
beam toward the satellite. As the direction of the satellite
with respect to the mobile is changing constantly due
to vehicle movement, it requires constant tracking of
the satellite and adjusting of the direction of the beam
such that it points toward the satellite. Apart from this,
the structure of the land-mobile antenna also needs to
take into account the aesthetic aspects, which is not
the case for base-station antennas. A number of studies
have been reported in the literature covering a satellite
Fig. 8. A satellite system generating spot beams of various shapes
communications system utilizing multiple antennas for land
to cover clusters of mobiles. mobiles, which highlights these and other issues requiring
consideration. Some are briefly mentioned here.
The mobile satellite experiment studied in [111] uses
shifters, and low noise amplifiers directly coupled to a number of antennas mounted on a vehicle. A beam is
radiating elements, provides multibeam functionality using electronically steered in the direction of a satellite using
a very small space onboard the satellite. The technology phase shifters. The system consists of many satellites and
enables one to fabricate a large number of active elements a vehicle communicates with the one that is in view at the
that may be independently controlled by digital hardware. time, requiring tracking of all the satellites and switching
This, along with digital signal processing, gives one the between the satellites as required.
capability to generate a large number of independent beams Electronic steering is not utilized for the German TV-SAT
of arbitrary shapes, which may be steered at any desired 2 system [112] using multiple antennas. It employs four sets
location. of preoriented antenna elements mounted on four faces of
Such active elements with dedicated high-power and truncated pyramids. Fixed beams from each set are formed
low-noise amplifiers have been used in recent system separately and are switched depending on the orientation of
developments, including the GLOBALSTAR [108] and the mobile relative to the geostationary satellite used. The
IRIDIUM [109] systems. Panels containing a large number system is useful for larger vehicles, such as buses and trains.
of elements are mounted onboard the satellite, and these ele- Separate elements are suggested in [113] that are alter-
ments are used to generate many fixed-shape beams. These nately placed side by side in a planar configuration for
beams form cochannel cells, with each cell supporting a transmit and receive mode using digital beam-forming tech-
number of communications channels. The system allows niques for tracking two satellites in multisatellite systems.
normal handoff when mobiles move out of a cell and does The system uses low-orbit satellites.
not use the knowledge of the positions of mobiles to adapt A spherical array mount, useful for ships and aircraft
the beams. Each beam supports a group of mobiles. employing digital beam forming to control beam direction,
c) Separate beam for each mobile: Allocation of sepa- is discussed in [114]. Test results on the characteristics and
rate beams to each mobile is advocated in [110]. A satellite the suitability of an antenna array mounted on an aircraft to
acts as a relay station between mobiles and the base station, communicate directly with a satellite are reported in [115].
with communication between the base station and the satel- An experimental development of a four-element an-
lite being at a different frequency than that between mobiles tenna array to receive a 1.537-GHz pilot tone from the
and the satellite. It is envisaged that each mobile is tracked INMARSAT II F-4 geostationary satellite with a fixed
and the beam is pointed toward the desired mobile, with position receiver shows that a significant improvement in
nulls in the directions of other mobiles operating at the same link reliability can be produced using an adaptive array
frequency. Different frequencies are used to communicate at a hand-held mobile compared to an omnidirectional
with the mobiles in close vicinity. Though the system antenna [116]. It indicates the possibility of using an
as described uses different frequencies, the principle of array at hand-held mobiles communicating directly with a
operation remains the same for CDMA and TDMA systems. satellite, provided that the required signal processing could
It should be noted that the direction-finding and beam- be incorporated somewhere.
forming algorithms in this case operate in a different
environment than those operating for the base-mobile com- D. Satellite-to-Satellite Communications
munications systems. Due to the distance involved between Intersatellite communications will play an important role
the satellite and the mobiles, the signals arriving from the in mobile communications, particularly with the use of
mobiles appear more like point sources, which is not the low-orbit satellites. The position of these satellites changes
case for the signal arriving at a base station due to spreading quickly, which causes a change in their relative directions.

1046 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

Arrays of antennas can play a major role in this situation [18]–[20], [30], [70], [124]. It provides the capability
by forming beams that always point in the direction of the to reduce cochannel interferences and multipath fading,
desired satellite and have nulls in the direction of others resulting in better quality of services, such as reduced
operating in the same band. The array may be used for BER and outage probability. Its capability to form multiple
tracking the satellites as well as for canceling interferences beams could be exploited to serve many users in parallel,
arising from the transmission of other satellites. It may also resulting in an increased spectral efficiency. Its ability to
provide protection against unfriendly jamming. adapt beam shapes to suit traffic conditions is useful in
An application of adaptive arrays to cancel interfer- reducing the handoff rate, which may result in increased
ences in a satellite-to-satellite communications system using trunking efficiency.
direct-sequence spread spectrum is considered in [73]. This section discusses the advantages of an array of
A constrained beam-forming algorithm, which uses the antennas in a mobile communications system and improve-
direction of the desired signal as well as an adaptive ments that are possible by using multiple antennas in a
beam forming using the reference signal, is studied for system rather than a single one. It provides references to
this application. The study assumes that the direction of experiments and studies where such improvements have
the desired satellite in this case may be obtained from the been realized and highlights complexities associated with
knowledge of the position of the satellite as it follows a the implementation of such systems.
well-defined trajectory.
A. Reduction in Delay Spread and Multipath Fading
E. Use of Arrays in Transmit Mode Delay spread is caused by multipath propagation where
a desired signal arriving from different directions gets
Antenna arrays have an equal role to play in both the
delayed due to the different travel distances involved.
receive and transmit modes. Their use in various config-
An array with the capability to form beams in certain
urations, as discussed previously, is applicable for both
directions and nulls in others is able to cancel some of these
cases. Conceptually, the ideas of forming multiple beams
delayed arrivals in two ways. First, in the transmit mode,
to cover the cell site as well as pointing independent beams
it focuses energy in the required direction, which helps
toward a cluster of mobiles, steering nulls toward cochannel
to reduce multipath reflections causing a reduction in the
mobiles, and forming cells dynamically are equally useful
delay spread. Second, in the receive mode, an antenna array
for both the cases. However, the required signal-processing
provides compensation in multipath fading by diversity
and hardware techniques to implement the array system for
combining, by adding the signals belonging to different
the two cases differ significantly. Conventionally, most of
clusters after compensating for delays, and by canceling
the investigations involving adaptive arrays have been for
delayed signals arriving from directions other than that of
receiving antennas. Some discussion specific to transmitting
the main signal.
arrays may be found in [29], [31], [81], and [117]–[123].
1) Use of Diversity Combining: Diversity combining
The problem of transmitting cochannel signals using an
achieves a reduction in fading by increasing the signal level
antenna array to several mobiles is addressed in [118] such
based upon the level of signal strength at different antennas
that each mobile receives its desired signal with minimum
[20], [66], [125], whereas in multipath cancellation
cross talk with the other signals. It also proposes the
methods, it is achieved by adjusting the beam pattern
formation of a desired transmitting beam pattern using the
to accommodate nulls in the direction of late arrivals,
knowledge of propagation conditions acquired from the
assuming them to be as interferences. For the latter case,
signals transmitted by each mobile for this purpose.
a beam is pointed in the direction of the direct path or a
Use of the information gathered during the reception
path along which a major component of the signal arrives,
mode may be used during the transmission mode, as
causing a reduction in the energy received from other
advocated in [29], [31], and [121]. It is suggested that the
directions and thus reducing the components of multipath
complex conjugate of the array weights, estimated during
signal contributing to the receiver. Techniques to identify
the reception by optimal combining, be used in transmit
a line-of-sight path from a group of received multipath
mode using different time slots for the two modes. It should
signals have been reported in literature [126].
be noted, however, that the optimal weights may change
2) Combining Delayed Arrivals: A radiowave originating
substantially, within a dwell time of 5 ms between the
from a source arrives at a distant point in clusters after
uplink and the downlink, causing a severe degradation in
getting scattered and reflected from objects along the way
the performance of the system, as discussed in [123].
[127]. This is particularly true in scenarios with large
The formation of nulls toward cochannel interferences
buildings and hills where delayed arrivals are well sepa-
during transmit mode is described in [81] and [117]. The
rated. One could use these clustered signals constructively
discussion in [81] is for an element space method, whereas
by grouping them as per their delays compared to a
that in [117] uses a beam space technique.
signal available from the shortest path. Individual paths of
these delayed signals may be resolved by exploiting their
V. PERFORMANCE IMPROVEMENT USING AN ARRAY temporal or spatial structure.
An antenna array is able to improve the performance The resolution of paths using temporal structures depends
of a mobile communications system in a number of ways upon the bandwidth of the signal compared to the coherence

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1047

bandwidth of the channel and increases as this bandwidth spread-spectrum communications systems may be found in
increases. In a CDMA system, the paths may be resolved [138] and [139].
provided their relative delays are more than the chip period
[128]–[131]. When these paths are well separated spatially, B. Reduction in Cochannel Interference
an antenna array may be used. This could be done, for
An antenna array has the property of spatial filtering,
example, by determining their directions. Spatial diversity
which may be exploited in transmitting as well as in
combining similar to that used in the RACK receiver [132]
receiving modes to reduce cochannel interferences. In the
may also be employed to combine signals arriving in
transmitting mode, it can be used to focus radiated energy
multipaths.
by forming a directive beam in a small area where a
The signals in each cluster may be separated by user-
receiver is likely to be. This in turn means that there is
specific information present in each signal, such as the
less interference in other directions where the beam is not
frame identification number or the use of a known symbol
pointing. An analysis of a base station using multiple beams
in each frame. A simulation study reported in [133] uses the
covering various mobiles indicates that cochannel interfer-
conjugate gradient method to adjust the array weights for
ence decreases as the number of beams increases [70].
compensating multipath fading in a land-mobile communi-
Cochannel interference in transmit mode could be further
cation. The system uses TDMA, employs known symbols
reduced by forming specialized beams with nulls in the
in each frame, and provides satisfactory performance using
directions of other receivers. This scheme deliberately
six elements in the presence of up to 30 scattered waves.
reduces transmitted energy in the direction of cochannel
In some situations, it might be possible to separate
receivers and requires knowledge of their positions. An
the signals in each cluster by forming multiple beams in
implementation of a communications system using an array
the directions of each component of these clusters, with
to cancel the cochannel interference in a UHF broadcasting
nulls pointing toward the other ones. Combining signals
system is described in [140].
belonging to different users after compensating for delays
The reduction of cochannel interference in the receive
not only leads to reduction in delay spread but also reduces
mode is a major strength of antenna arrays and has been
cochannel interference.
reported widely. It does not require knowledge of the
3) Nulling Delayed Arrivals: A reduction in delay spread
cochannel interferences. If these were available, however,
using an array by nulling the delayed signals arriving
an array pattern might be synthesized with nulls in these
from different directions has been reported in [100] and
directions, as discussed, for example, in [141]. In general,
[104]. The simulation study presented in [104] considers
an adaptive array requires some information about the
indoor radio channels, uses a PN sequence as a reference
desired signal, such as the direction of its source, a reference
signal and the SMI algorithm to estimate the array weights,
signal, or a signal that is correlated with the desired signal.
and concludes that using an adaptive array, a substantial
In situations where the precise direction of the sig-
reduction in delay spread is possible. Similar conclusions
nal is known, interference cancellation may be achieved
are reached in [100] using an experimental array of four
by solving a constrained beam-forming problem, whereas
elements mounted on a vehicle. It shows that using a
when this is not the case, it may be achieved by using a
CMA, the array is able to null the delayed arrival in a
reference signal as discussed in [73] and [138], where the
time-division-multiplexed channel.
reference signal for a CDMA system is generated by code
A frequency-hopping system may also be used for cor-
synchronization. A comparison of the two schemes [73]
recting degradation due to fading. The system is useful in
for a satellite-to-satellite communications scenario when an
frequency-selective fading, where the different frequencies
approximate direction of the signal is known indicates that
fade differently. The frequency-hopping system is a spread-
the constrained scheme cancels the interference faster than
spectrum system where the carrier frequency of transmis-
the reference generation scheme.
sion is changed in a predetermined manner, in contrast to a
direct-sequence spread-spectrum system, where a pseudo-
random sequence is used to spread the spectrum. An C. Arrays in Combination with Other Techniques
introduction of enough redundancy in such a system by An antenna array may be used together with other meth-
coding the information before transmission may help the ods to enhance the performance of a system by canceling
system to correct the loss of reception caused by fading at the interference that might be present in directions other
certain frequencies. A comparison of such systems with than that of the desired signal.
a system using antenna diversity of two antennas [134] 1) Spatial Diversity and Channel Coding: A combination
shows that the performance of the two systems is almost of channel coding and spatial diversity (two antennas) for
identical. However, there is a cost associated with the a TDMA indoor wireless system [102], [142] shows that
coding method. It increases the rate of information due to the combined scheme provides a better BER than that
added redundancy, requiring higher bandwidth. This in turn without coding schemes. This reduces the effective data
reduces the capacity of the system. rate, however, and requires larger bandwidth for a given
Application of adaptive arrays for frequency-hopping data rate to be transmitted. This in turn reduces the number
communications systems is described in [135]–[137], of channels available. The problem may be alleviated by
whereas the use of adaptive arrays for direct-sequence using extra antennas to increase space diversity. This and

1048 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

another study [134], which compares a frequency-hopped
TDMA system with one using space diversity, show the
amount of tradeoff provided by antenna arrays.
2) Diversity Combining and Interference Canceling: An
antenna array operating in a combined mode of diversity
combining and interference canceling may be able to cancel
directional interferences using some degree of freedom and
achieve reduction in fading on the order of the remaining
degree of freedom [18], thus improving the performance of
the system as well as increasing its capacity. This increase
depends upon the correlation of fading signals at differ-
ent antennas [143]. The degradation in the performance,
however, is small for correlation up to 0.5.
3) Diversity Combining and Adaptive Equalization: A
comprehensive theory and computer simulation are pre-
Fig. 9. Base station employing SDMA [149].
sented in [68] and [69] to demonstrate the potential benefits
of a system consisting of diversity combining and adaptive
equalization using a tapped delay line filter behind each Each increase is on the order of five to ten [11], [13], [147],
antenna element. Diversity combining is useful for over- [148]. Many studies have shown that this may be further
coming the flat fading in mobile communications, whereas improved by using multiple antennas and combining the
the equalizers are normally used to reduce intersymbol signals received from them. First, the increased quality of
interferences in digital data transmission caused by delay service resulting from the reduced cochannel interferences
spread. Thus, the combined structure offers an effective and reduced multipath fading, as discussed, may be traded
means to deal with the adverse effects of dispersive as well to increase the number of users. Thus, the use of an array
as flat fading. Many examples are provided to demonstrate results in an increase in channel capacity while the quality
the reduction in the average probability of the received bits of the service provided by the system remains the same,
for a QPSK modulated system using dual diversity and that is, it is as good as that provided by a system using
different types of equalizers under various environments. a single antenna. Second, an array may be used to create
Similar results are also reported in [144]. A survey of adap- additional channels by forming multiple beams without any
tive equalization techniques along with their performance extra spectrum allocation, which results in potentially extra
characteristics, limitations, implementation complexities, users and thus increases the spectrum efficiency. Fig. 9
and use in TDMA digital mobile radio may be found in shows a typical scenario of a base-station system employing
[145]. an SDMA technique to serve many mobiles using only two
The importance of an antenna array using the tapped frequencies [149].
delay line filter to combat fading and to reduce cochannel Fundamental limits on the maximum data rate and the
interference in wideband TDMA mobile communications capacity of a multiple antenna system in a Rayleigh fading
channels is further emphasized in [102] to increase the data environment are considered in [150]. The study concludes
rate beyond about 2 Mb/s. A discussion on the performance that two users with antennas each, using optimum trans-
improvement of a QPSK modulated system employing an mitter/receiver pairs, may be able to establish up to
antenna array and a tapped delay line filter to combat channels, with each channel having about the same maxi-
multipath fading as a function of the number of taps may mum data rate as that of a single channel. This and other
be found in [146]. studies on the impact of antenna diversity on the capacity of
a wireless communications system [18] show the potential
D. Spectrum Efficiency and Capacity Improvement for a large capacity improvement in a communications
Spectrum efficiency refers to the amount of traffic a given system with the use of an antenna array. Salient features
system with certain spectrum allocation could handle. An of some of these studies are now described to show the
increase in the number of users of the mobile communi- possible improvements in various configurations.
cations system without a loss of performance causes the 1) Multiple-Beam Antenna Array at the Base Station: A
spectrum efficiency to increase. Channel capacity refers study in [81] indicates that by first using a base-station
to the maximum data rate a channel of given bandwidth array in receive mode to locate the positions of mobiles
could sustain. An improved channel capacity leads to in a cell and then transmitting in a multiplexed manner
more users of a specified data rate, implying a better toward different clusters of mobiles one at a time using
spectrum efficiency. In that sense, the two terms are used the same channel, the spectrum efficiency increases many
interchangeably in this paper following the convention used times over and depends upon the number of elements in
in most of the mobile communications literature. the array and the amount of scattering in the vicinity of
TDMA and CDMA result in an increase in channel the mobile. The study concludes that pointing beams with
capacity over the standard FDMA, allowing different time directed nulls toward other mobiles in a cell is a more
slots and different codes to be assigned to different users. efficient use of an array than reducing the cell size and the

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1049

reuse distance. The use of a multiple-beam antenna array at [61]. Using theoretical analysis and computer simulations,
the base station to improve spectral efficiency by resolving it is argued in [29] that using optimal combining at a
the angular distribution of mobiles is discussed in [70], base station with antennas may lead to up to -fold
where it is shown that spectral efficiency increases as the capacity improvement in an indoor-mobile communications
number of beams increases. Formulas, which are helpful system. Furthermore, using a given number of antennas,
in predicting interference reduction and capacity increase a substantial capacity increase is also possible with a
provided by a switched-beam antenna system employed by small increase in the SNR in a system. For example, a
a base station, also predict that the number of subscribers system using nine antennas may be able to achieve a six-
in a cell increases as the number of beams increases [151]. fold capacity increase from a 10-dB increase in SNR per
2) Antenna Array at the Base Station Employing CDMA antenna.
System: By using an array of antennas for a CDMA system, The results presented in [156] indicate that replacing the
the number of mobiles a cell may be able to sustain omnidirectional base-station antenna with an adaptive array
increases manifold (on the order of the number of elements) capable of forming multiple dynamically allocated beams
for a given outage probability and BER [64], [152], [153]. steered toward the mobiles in a building complex reduces
For example, at an outage probability of 0.01, the system the delay spread substantially. This implies that by using an
capacity increases from 31 for a single antenna system array, communications services may be extended to areas
to 115 for a five-element array. Using an array of seven within buildings where delay spread is a serious problem,
elements increases the capacity to 155 [153]. The study re- such as shopping centers, railway stations, and so on [157].
ported in [64] is for both mobile-to-base and base-to-mobile The use of multiple antennas also increases the capac-
links and derives expressions for the outage probability ity of the network by allowing many users to operate
considering the effect of cochannel interferences, whereas at the same channels, which then may be separated by
that reported in [128] is for the mobile-to-base link. The blind estimation methods. Depending upon the propagation
latter uses an RLS algorithm to estimate the weights of conditions, a large improvement in capacity is possible [87].
the array considered in simulation study and shows that
capacity improvements on the order of 60% are possible E. BER Improvement
using a four-element linear array. A consequence of a reduction in cochannel interference
3) Antenna Array at the Base Station Employing TDMA and multipath fading by using an array in a mobile commu-
System: Examples of a TDMA system employing antenna nications system to improve the communications quality is a
arrays can be found in [18], [154], and [155]. The study reduction in BER and SER for a given SNR, or a reduction
reported in [155] uses the constant modulus adaptive algo- in required SNR for a given BER. This relationship has
rithm for updating the weights of the array, whereas that in been recognized by many authors [18], [29], [61], [100],
[18] and [154] uses the SMI algorithm and considers the [133], [153], [158]–[160].
IS-54 system for its study. IS-54 is the standard for digital Experimental results reported in [100] for a high-speed
mobile radio in North America. It is for a TDMA system GMSK mobile communications system show that a four-
where each user’s slot contains a 28-b synchronization element adaptive array using the CMA is able to reduce
sequence along with 12-b for user identification and 260 the BER substantially compared to a single antenna system
data bits. The study uses the 28-b synchronization sequence in a frequency-selective fading environment. The BER in a
as the reference signal and the user identification bits to system normally reduces as the SNR is increased. These
make sure that the correct user’s signal has been acquired. results, however, show that in an irreducible error rate
The study concludes that by using optimal combining at environment, where no amount of SNR increase in a single
the base station with only two antennas, a frequency reuse antenna system would be able to reduce the BER, the
factor of four is possible except in the worst cases at the system with an adaptive array is able to achieve large
boundaries of the cell. This could be further improved to reductions.
a frequency reuse factor of three using dynamic channel A simulation study employing a multibeam system with
assignment. Furthermore, it argues that frequency reuse in cochannel interference capability [158] indicates that for a
each cell is possible using four or five antennas. This is QPSK modulated system, the SER decreases as the number
a substantial improvement compared to a frequency reuse of elements in the array increases, and in the presence of
factor of seven proposed for the current system. multipath arrivals, this decrease is better when there is a
4) Base-Station Array for Indoor-Mobile Communications: dominant path than when there are equal energy paths.
Multipath fading is a severe problem for indoor-mobile A detailed analysis and computer study of the BER in a
communications. However, due to the slow speed of mo- PSK system is presented in [61]. The study demonstrates
biles, the fading rate is much lower inside buildings than that using an array with optimal combining can achieve a
outside, typically on the order of a few hertz. This means large reduction in BER. Analysis and experimental results
that there is enough time to compute antenna weights for reported in [18] indicate that by optimally combining the
optimal combining using slow converging algorithms such signals received on multiple antennas at the base station,
as LMS, which offers the advantages of robustness and one is able to cancel the effect of other users in a flat
computational efficiency along with the possibility of its Rayleigh fading environment to such an extent that the
implementation using a single chip, as discussed in [29] and average probability of errors behaves as if the other users

1050 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

were not present. Computer simulations show that similar and [164]. The study in [66] is done for various combining
results hold for frequency-selective channels. The use of methods as a function of the number of elements in the
optimal combining in an indoor environment is studied in array as well as the number of cochannel interferences,
[29] to demonstrate its effectiveness in reducing BER for a and shows that the use of diversity combining reduces the
PSK system. An analysis of a PSK communications system outage probability. The technique to calculate the outage
presented in [159] also shows that the use of an adaptive probability of a cellular mobile system presented in [163]
array reduces BER for such a system. appears to be simpler to use than that in [66].
An increase in the number of users in a cell normally Results on the improvement in the outage probability
causes an increase in the BER. The rate of this increase using an array employing optimal combining are presented
could be made lower by using an array compared to a in [30] and [69]. The study in [69] concludes by theoretical
single antenna system, as shown in [153]. The computer analysis and computer simulation that the use of diversity
study uses the RLS algorithm to adjust the weights of an combining results in a substantial reduction in the outage
array and uses an expression for BER derived in [161]. For probability, whereas the results in [30] are derived from
an omnidirectional antenna, it is given by experiments conducted in an open-plan office environment
and compare selection diversity and optimal combining,
showing that the latter results in a reduced threshold for
a given outage probability.
A study [128] of a CDMA system using a base-station
whereas for multiple antennas it becomes antenna array derives equations for predicting the outage
probability of the system for the uplink case, that is, by
considering the signals arriving from mobiles. The study
uses user-specific codes to determine the array response
vector by correlating the array output with the code for
where is the standard -function showing the prob- the desired user. The estimated array response vector is
ability that for a zero mean, unit variance, Gaussian then used for determining the array weight employing
distributed random variable , is the processing gain of the optimal beam-forming technique. Numerical examples
the CDMA system, is the number of uses in the cell, confirm that the outage probability increases by an increase
, and is the directivity of the beam of the in the number of users in a cell as well as by the presence
multiantenna system. of multipath. Furthermore, for a given outage probability,
A comparison of the BER performance of a system using it shows that the use of an array allows an increase in the
the conjugate gradient method and the RLS algorithm to number of users in a cell, resulting in an increase in system
adjust the weights of the array indicates the superiority of capacity. The increase in capacity is on the order of the
the former [133]. The BER performance of the adaptive number of elements in an array [152].
antenna system using the LMS algorithm is compared
with the maximum entropy method in [160] and with a G. Increase in Transmission Efficiency
method based upon spatial discrete Fourier transform [162]
Electronically steerable antennas are directive compared
to show that both the latter methods provide better BER
to fixed omnidirectional antennas, that is, they have high
performance than the LMS algorithm.
gains in the direction where the beam is pointing. This fact
may be useful in extending the range of a base station,
F. Reduction in Outage Probability resulting in a bigger cell size, or may be used to reduce the
Outage probability is the probability of a channel’s being transmitted power of the mobiles. A laboratory experiment
inoperative due to increased error rate in the received data. [165] indicates that a reduction from 10 W to 250 mW
It may be caused by cochannel interference in a mobile is possible. This follows from the fact that by using a
communications system. Using an array helps to reduce highly directive antenna, the base station may be able
the outage probability by decreasing cochannel interference. to pick a weaker signal within the cell than by using
It decreases as the number of beams used by a base an omnidirectional antenna. This in turn means that the
station for land mobiles increases in a multibeam adaptive mobile has to transmit less power [18] and its battery will
antenna system [70]. The system analysis presented in [70] last longer, or it would be able to use a smaller battery,
consists of calculating the outage probability considering resulting in a smaller size and weight, which is important
one cochannel cell as well as six cochannel cells. The for hand-held mobiles.
analysis shows that the reduction in outage probability is It is also advantageous to use an antenna array at the base
slightly less in the six-cochannel-cell case compared to the station in transmit mode. In a single antenna system, all the
one-cochannel-cell case. Results on the outage probability power of the base station is transmitted by one antenna.
of the system using an omnidirectional base-station antenna However, when the base station uses an array of antennas
can be found in [28]. and transmits the same amount of power as that of the single
A detailed analysis of the outage probability of a system antenna system, the power transmitted by each antenna of
with cochannel interferences in a Nakagami fading environ- the array is much lower compared to the case where the
ment using diversity combining is presented in [66], [163], total power is transmitted by one antenna. Furthermore, for

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1051

a given SNR at the mobile site, the base station using an upon the feedback obtained from probing the mobiles could
array has to transmit less power compared to the single help eliminate this problem [118]. The mechanism works
omnidirectional antenna case due to the directive nature by transmitting a probing signal periodically. The received
of the array. This further reduces the power transmitted feedback from mobiles is used to identify the propagation
by each antenna. These reductions in transmitted power conditions, and this information is then incorporated into
level using an array allow the use of electronic components the beam-forming mechanism.
of lower power rating in the transmitting circuitry. This
results in a lower system cost, leading to a more efficient K. Cost-Effective Implementation
transmission system [124].
The use of multiple antennas on direct broadcast satel-
lites leads to a system that is cost effective due to its
H. Dynamic Channel Assignment
light weight, reduced power requirements, flexibility and
In mobile communications, channels are generally as- robustness of antenna design, ease of pattern control, and
signed in a fixed manner depending upon the position of use of more solid-state devices. It also provides a larger
a mobile and the available channels in the cell where the aperture and mechanism for error compensation by signal
mobile is positioned. As a mobile crosses the cell boundary, processing [107].
a new channel is assigned. In this arrangement, the number
of channels in a cell are normally fixed. The use of an
L. Angular Spreading and Its Impact on Performance
array provides an opportunity to change the cell boundary
and thus to allocate the number of channels in each cell as Angular spreading refers to a situation where a trans-
the demand changes due to changed traffic situations. This mitted signal gets scattered in the vicinity of the source
provides the means whereby a mobile or group of mobiles and a signal arrives at a receiver within a range of angles.
may be tracked as it moves and the cell boundary may be In a base-mobile communications system, where the base-
adjusted to suit this group. station antenna is normally high away from the ground and
Dynamic channel assignment is also possible in a fixed a mobile is close to the ground, the angular dispersion is
cell boundary system and may be able to reduce the more pronounced in the vicinity of the mobile and arrives at
frequency reuse factor up to a point where frequency reuse the base station with an angular distribution. The range of
in each cell might be possible [18]. There may be situations angles becomes smaller as the distance between a mobile
when it is not possible to reduce cochannel interferences and the base increases. Experimental results indicate that
in certain channels, and the call may have to be dropped an angular spreading of about 3 results from a distance
due to large BER caused by strong interferences. Such of 1 km [75].
a situation may arise when a desired mobile is close to Various models of the scattering situation have been
the cell boundary and the cochannel mobiles are near the reported in the literature, assuming the multipath signals
desired mobile’s base station [18]. This could be avoided by to be distributed uniformly within the spread angle [71],
dynamic channel assignment, whereby the channel of a user [143], [166] as well as with the Gaussian distribution [81],
is changed when the interference is above a certain level. [167]. Selection of the distribution function, however, does
not appear to be critical as long as the spread is small
I. Reduction in Handoff Rate around the nominal direction [75].
A dispersion of the radio environment results in the
When the number of mobiles in a cell exceeds its distortion of the perceived antenna side-lobe levels at the
capacity, cell splitting is used to create new cells, each base station [168] as well as an increase in correlation of
with its own base station and new frequency assignment. A fading at different elements of the array [143], both of
consequence of this is an increased handoff due to reduced which affect the performance of the system. The problem
cell size. This may be reduced using an array of antennas. of fading correlations is studied in [169]. It is shown by
Instead of cell splitting, the capacity is increased by creating deriving the relationships between the angle of arrival,
independent beams using more antennas. Each beam is beamwidth, and correlation of fading that a larger antenna
adapted or adjusted as the mobile locations change. The spacing is required to reduce the correlation when the angle
beam follows a cluster of mobiles or a single mobile, as of arrival approaches parallel to the array, resulting in a
the case may be, and no handoff is necessary as long reduction of the beamwidth.
as the mobiles served by different beams using the same An increased correlation of fading between various ele-
frequency do not cross each other. A comparison of the ments greater than 0.8 causes signals at all of the antennas
SDMA method with other schemes indicates that the former to fade away simultaneously, rendering an array in the
offers many practical advantages over the other methods of maximal ratio combining mode ineffective against fading
capacity increase [149]. correction [31]. The array, however, is able to suppress
interferences, as independent fading is not required for
J. Reduction in Cross Talks interference suppression [143]. A detailed investigation of
Cross talks may be caused by unknown propagation the effect of fading correlation on the performance of
conditions when an array is transmitting multiple cochannel adaptive arrays to combat fading is carried out in [143].
signals to several receivers. Adaptive transmitters based The investigation concludes that correlation of up to 0.5

1052 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

causes little degradation, but a higher correlation decreases quired for antenna arrays, and experimental studies to
performance significantly. demonstrate the suitability of the arrays for mobile com-
Transmission strategies in a multihop packet radio net- munications are briefly discussed.
work normally are concerned with designing routing al-
gorithms for sending packets from the source to the des- A. Developments of Antenna Elements and Arrays
tination. A study presented in [170] indicates that fading Antenna design is one of the biggest challenges in mobile
affects the performance of a multihop system using routing communications system development. An antenna mounted
algorithms. This, however, does not seem to be the case on a vehicle has stringent requirements. Mechanically,
for a single-hop system. it must conform with the vehicle as much possible for
aesthetic and aerodynamic reasons. It must be cheap and
M. Cost, Complexity, and Network Implication suitable for mass production, easy to install, strong enough
to withstand forces during vehicle movement, light weight,
It follows from the discussion in previous sections that
and compact. Electrically, it should have the required gain,
a system using an adaptive array to improve the perfor-
bandwidth, wide-angle scanning capability, and low losses.
mance of mobile communications requires estimation and
The gain of an antenna is at its maximum in the boresight
optimization of numerous time-varying system parameters
direction of the antenna and drops off in directions away
in a dynamically changing environment.
from the boresight direction. The amount of gain required
Though the specific parameters would depend upon the
depends upon the C/N ratio available at the antenna sight for
type of system under consideration and the mode of array
receiving antennas, the C/N ratio required by the receiver,
processing incorporated in the system, it is clear that
and the losses caused by cables, mixers, and the receiving
implementation of such a system would require a complex
circuitry in general as well as the noise generated therein.
network and system architecture.
All of these losses and noise in the receiving systems are
Implementation of schemes such as dynamic handoff,
normally specified by the system noise temperature. It is not
dynamic channel assignment, dynamic beam shaping to
the physical temperature of the system but indicates that a
incorporate clusters of mobiles, dynamic nulling of cochan-
noise source at this temperature would generate the same
nel interferences, and procurement of the knowledge of
amount of noise as that generated by the system referred
the desired mobile—in the form of a reference signal, its
to at the antenna terminal.
direction, or the array response vector associated with the
The concept of noise temperature follows from the fact
mobile—for the purpose of beam forming would require a
that a noise source at temperature (absolute temperature)
complex control structure. Then there is a question of the
would generate the noise power , with denoting
time required to update the system parameters.
Boltzmann’s constant and the bandwidth of interest.
The system requires quick updates on the positions of fast
Thus, a device of bandwidth generating this much noise
moving mobiles, whereas its response time is limited by the
is said to have the noise temperature . In view of this, the
time required by DOA estimation and tracking schemes to
receiving antennas are also specified by their gain-to-noise
update the positions of mobiles and by the beam-forming
temperature ratio, in short G/T dB/K. An antenna used for
algorithms to converge to a satisfactory level. Though some
voice and high-speed data around 20 Kb/s typically is of a
of these algorithms may be implemented in parallel to
medium gain around 11 dBi, which amounts to G/T around
increase the signal processing power, this adds to the cost
13 dB/K.
of the system.
For transmitting antennas, the antenna gain and amount
The system cost includes not only the cost of building
of power supplied to the antenna are combined, and EIRP
hardware to implement the control structure but the cost
is normally used to specify it. It is a product of the supplied
of building active antennas along with the required phase
power and the antenna gain and indicates that for the
shifters, high-power amplifiers, and circuitry associated to
calculation of the flux density at distance away from
drive these components. Most of these components are
this source, the EIRP is the power radiated by an isotropic
excessively expensive at present and are not required for
source, that is, a source radiating uniformly in all directions.
a system without SDMA. This is currently a disadvantage
The flux density at a point distance away from an
of using an adaptive array in mobile communications. Apart
isotropic source is given by the transmitted power divided
from the cost of implementation, many reliability questions
by the surface area of a sphere of radius centered at the
also need careful consideration.
source, .
Some network architecture and traffic-implication issues
In recent years, there has been a wide range of interest
of mobile communications in general are considered in [48]
in developing antenna elements, arrays of antennas, feed
and [171]–[173]. However, there is no consideration of the
systems suitable for such arrays, and other related com-
complications arising from the use of an array in such
ponents covering a wide range of frequency bands. The
systems.
frequency bands in satellite communications literature are
normally referred to by letters. The frequencies for L-band
VI. FEASIBILITY OF ANTENNA ARRAY SYSTEMS typically are around 1.5 GHz for downlink and 1.6 GHz for
In this section, developments in the design of arrays, uplink. The band is denoted as a 1.6/1.5 GHz band. Higher
antenna elements suitable for arrays, tracking systems re- frequencies are used for uplink and lower ones for downlink

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1053

Table 3 Letter Designation for the Frequency Spectrum
conductors but have gaps at regular intervals. The inter-
ring spacing, gap size, and slot width are the design
parameters. A center-fed slotted disk antenna suitable for
vehicle mounting is reported in [183]. Discussions on a
linear tapered slot antenna may be found in [188]. A
variation of the linear tapered slot antenna reported in [182]
has the capability of easy packaging and integration with
MMIC devices with a gain of 9 dBi at the design frequency
of 10 GHz.
A circular patch antenna featuring a gain of 8.5 dBi, com-
pensated by an inductance–capacitance circuit to increase
its band, and placed in a cavity to isolate it from other
antennas has been reported in [177]. The patch antennas
may be stacked to increase the gain, as demonstrated in
[181], by using two sets of ring antennas to obtain a gain
of around 9 dBi. The development and fabrication of arrays
using circular patch antennas has been discussed in the
literature, including conformal arrays of semispherical and
partially spherical shapes using 16 elements [180] and a
planar array using alternative elements for transmitting and
receiving a signal [178].
Array design suitable for mobile communications is an
because transmission loss increases as the frequency of active area of research, and many prototypes have been
transmission increases, requiring more transmitted power, built, analyzed, and tested. These designs include:
and the fact that it is easy to generate power at the ground
• an eight-element array of spiral antennas with a me-
compared to onboard a satellite.
chanical steering capability with a gain of 15 dBi
The other bands include the S-band covering 2.5/2.4 [186];
GHz, C-band covering 6/5 GHz, Ku-band covering 14/12
GHz, and Ka-band covering 30/20 GHz. Letter designation • a 12-element array mounted on a truncated pyramid
for the entire frequency spectrum as defined by IEEE with electronic steering capability with a gain of 10
Standard 521 [174] is provided in Table 3. dBi [185];
An excellent treatment of vehicle antennas for mobile • a single-layer slotted leaky waveguide array with 16
communications is provided in [175] and concludes that the waveguides suitable for direct broadcast reception
MMIC phased array antenna technology will play a key role from a satellite in Japan [189];
in future mobile communications. Using this technology, a • a spherical array with beam-forming capabilities [180];
hand-held portable monitor consisting of an antenna array • a set of four planar arrays, each with 32 patch antennas
along with its feed circuitry, low noise amplifiers, power suitable to operate at 12 GHz for mobile reception of
amplifiers, phase shifters, and so on will be able to access direct broadcasting satellites in Germany [112].
a satellite in the Ka-band [176]. An analysis of the slotted leaky waveguide array system
Current developments in basic antenna elements suitable can be found in [190]–[192]. These and other array design
for building an array for vehicle antennas in a satellite- activities [109], [111], [193], [194] demonstrate the suitabil-
to-mobile communications system include microstrip patch ity of antenna arrays for mobile communications systems
antennas [177]–[180], a ring patch [181], slot antennas from an antenna-design point of view.
[182], [183], crossed dipole antennas [184], traveling-wave-
type antennas [185], and spiral-shaped antennas [186].
Many of these variously shaped building-block antennas B. Tracking System
are fabricated using microstrip technology. A microstrip Land-mobile satellite communications using a directional
consists of a nonconducting surface referred to as a sub- antenna requires knowledge of the direction of the satellite
strate covered with a conductor. The shape of an antenna to steer the beam toward it. As the vehicle moves, tracking
is formed by an etching process to remove the unwanted the satellite becomes important. The accuracy of the track-
conductor. The thickness of the substrate, number of layers, ing depends upon the type of antenna system. For example,
and type of material used for the substrate are among the in a system where the shape of the antenna pattern remains
parameters to control the radiation properties of the antenna. fixed during the steering process—such as a mechanically
Vehicle antennas for a base-station communications system steered array, where the boresight direction of the array
providing space diversity include a trunk lid antenna and a is pointed toward the satellite, as well as an electronically
disk-loaded monopole antenna [187] operating at 900 MHz. steered array, where only phase is adjusted to steer the
A slotted disk antenna is made up of circular rings beam—the accuracy required in the direction of the satellite
of slots. The circular rings are not made of continuous is on the order of a half beamwidth, as any higher error

1054 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

would put the satellite out of the main lobe. It is argued in that the optimal combiner is able to track the time-varying
[111] that the maximum acquisition time permitted for an environment without degrading the signal quality. Further-
electronically steered array is 10 s and the time required more, a comparison of diversity combining and optimal
from acquisition to optimum pointing is less than 2 s. A combining shows that an eight-element array provides an
tracking system needs to take into account the maneuvering improvement in SNR of around 5 dB compared to a single
of vehicles resulting in velocities on the order of 45 /s and antenna using diversity combining and a further 5 dB using
accelerations up to 45 /s [111], as well the convergence optimal combining. The system is designed to operate at
speed of the algorithm used when an adaptive scheme is 1.89 GHz and concludes that the adaptive antenna array
employed to adjust the array weights [154]. techniques have now reached a stage where they may be
Two types of systems are normally used for satellite considered for commercial mobile radio communications.
tracking purposes: a closed-loop system where the received An experiment conducted in central Tokyo using a four-
signal from a satellite is used and an open-loop system, element antenna array mounted on the roof of a van shows
which does not require this signal. that when weights are updated using the CMA, the antenna
A classical closed-loop method known as a monopulse pattern may be adjusted by digital beam-forming concepts
scheme generates two signals, sum and difference, from the to cancel delayed signal to overcome multipath fading
signal received from a satellite. A sum signal is generated [100], [125]. The system employs a TDMA scheme and
by adding the two signals in phase and is used to recover uses Gaussian-filtered minimum shift keying transmission
information. The difference signal is generated by adding at a rate of 256 Kb/s. The system operates at 1.5 GHz,
the two signals from symmetrical elements in the array out and the experimental results demonstrate that the array is
of phase, such that when the beam is pointing to the satel- successful in directing nulls toward the delayed signals due
lite, a null is formed. A number of null steering algorithms to multipath. The study concludes by suggesting that the
exist in the literature [195], [196]. In the course of vehicle adaptive arrays are expected to function well as a counter-
movement when the beam is not pointing correctly, the measure to selective fading in a high-speed communications
array is steered to achieve the perfect null in the difference environment.
beam. A variation on this scheme is implemented in [185] An experiment conducted in south France using arrays
using an array operating at L-band. The results of the field of elements operating at L-band used a circular as well as
trial indicate the feasibility of a low-cost antenna system a pyramid-shaped array of five patched elements [199]. It
[185]. used a reference signal generated from the demodulator and
An open-loop system does not require the signal from the used the matrix inversion algorithm to update the weights.
satellite in the form of feedback. It uses the output from The results at the transmission rate of 200 Kb/s indicate
sensors such as a magnetic compass and optical fiber gyro that the digitally implemented algorithm is able to adjust
and is useful in situations where the satellite signal is not the weights in multipath conditions. The experimental and
stable. This is the case for a land-mobile communications simulation results show the tracking, jammer canceling, and
system compared to a maritime communications system multipath correcting properties of the algorithm. The study
[197]. A phased array tracking system using a hybrid of concludes that for a further improvement in the receiver
geomagnetic sensors and optical fiber gyro shows a good performance, the array needs to be enlarged using more
performance, as discussed in [197] and [198]. Geometric elements.
sensors provide the absolute direction, whereas the gyros The other system developments and experiments to carry
provide the relative variations in direction and are useful to out the feasibility studies of antenna arrays for mobile
compliment the sensors when their capabilities are hindered communications include the study of an array mounted
by local environment conditions such as buildings, bridges, on an aircraft using the ETS-V satellite in Japan [115],
and power poles [197]. the study of an array mounted on the roof of a car
to receive the German SAT-2 [112], a feasibility study
C. Experimental Systems of a frequency scanning satellite system [105], [106],
A number of projects have been undertaken to conduct hardware implementation of a four-element adaptive array
experiments involving antenna arrays by developing sys- for high-speed digital mobile communications [125], and
tems and designing hardware. The results of the field trials a laboratory prototype of a low-power system for mobile
have been reported in the literature. communications [165]. The setting up of a consortium to
A project [30] known as the Smart Communication investigate the feasibility of adaptive array technology in
Antenna Research Program (SCARP) has been undertaken third-generation mobile communications systems in Europe
by ERA Technology, U.K., and supported by many in- is reported in [157].
dustrial organizations in the United Kingdom. It involves Some other developments reported in the literature re-
eight dipole antennas mounted on an eight-faced pyramid lated to systems employing adaptive antennas include a
to provide hemispherical coverage and has studied the multiway power divider for a planar slotted waveguide
performance of a system in the fading channels using a array suitable for vehicles to communicate with direct
reference signal to update the weights of optimal combin- broadcasting satellites [190], a digital demultiplexer for an
ing. The weights are updated using normalized least-squares adaptive satellite array [110], a Ku-band solid-state power
algorithms from a set of 100 snapshots. The results indicate amplifier for European satellite systems, and a low-cost feed

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1055

system suitable for vehicle antennas, which does not require munications,” IEEE Commun. Mag., vol. 33, pp. 50–57, Jan.
the use of phase shifters at each element [111]. 1995.
[11] K. Raith and J. Uddenfeldt, “Capacity of digital cellular TDMA
An interesting overview of key payload technologies systems,” IEEE Trans. Veh. Technol., vol. 40, pp. 323–332,
for future satellite personal communications, dealing with 1991.
payload architectures involving onboard digital signal pro- [12] W. C. Y. Lee, “Overview of cellular CDMA,” IEEE Trans. Veh.
Technol., vol. 40, pp. 291–302, 1991.
cessing and narrow-band beam forming suitable for MEO [13] K. S. Gilhousen, I. M. Jacobs, R. Padovani, A. J. Viterbi, L.
and GEO satellites, is provided in [200]. It discusses A. Weaver, Jr., and C. E. Wheatley III, “On the capacity of
many implementation issues relevant to present and future cellular CDMA system,” IEEE Trans. Veh. Technol., vol. 40,
pp. 303–312, 1991.
activities of the ESA. [14] R. L. Pickholtz, L. W. Milstein, and D. L. Schilling, “Spread
spectrum for mobile communications,” IEEE Trans. Veh. Tech-
nol., vol. 40, pp. 313–322, 1991.
VII. CONCLUDING REMARKS [15] R. Kohno, R. Meidan, and L. B. Milstein, “Spread spectrum
This paper has attempted to provide a snapshot of a dy- access methods for wireless communications,” IEEE Commun.
Mag., vol. 33, pp. 58–67, Jan. 1995.
namic and fast-growing industry at a level where nonexperts [16] P. W. Baier, P. Jung, and A. Klein, “Taking the challenge
in the field are able to appreciate it. After presenting an of multiple access for third generation cellular mobile radio
overview of mobile communications, this paper showed the systems—A European view,” IEEE Commun. Mag., vol. 34,
pp. 82–89, Feb. 1996.
current state of research in the use of antenna arrays to [17] N. Abramson, “Multiple access in wireless digital networks,”
meet the projected demand of increased channel capacity Proc. IEEE,, vol. 82, pp. 1360–1370, 1994.
[18] J. H. Winters, J. Salz, and R. D Gitlin, “The impact of antenna
by describing how an array may be used in various mobile diversity on the capacity of wireless communication systems,”
communications systems and what improvements may be IEEE Trans. Commun., vol. 42, pp. 1740–1751, 1994.
expected using such systems. Many theoretical, experimen- [19] V. A. N. Barroso, M. J. Rendas, and J. P Gomes, “Im-
pact of array processing techniques on the design of mobile
tal, and computer-simulated studies were cited to show the communication systems,” in Proc. IEEE 7th Mediterranean
feasibility of such systems. Electrotechnical Conf., Antalya, Turkey, 1994, pp. 1291–1294.
This paper has painted a very bright picture of the field [20] M. Mizuno and T Ohgane, “Application of adaptive array
antennas to radio communications,” Electron. Commun. Japan,
and at times appears ambitious. Even though many studies vol. 77, pp. 48–59, 1994.
point toward the feasibility of such systems, the systems are [21] H. L. Bertoni, W. Honcharenko, L. R. Maceil, and H. H. Xia,
in their infancy, particularly when it comes to implementa- “UHF propagation prediction for wireless personal communi-
cations,” Proc. IEEE,, vol. 82, pp. 1333–1359, 1994.
tion issues. Many questions related to network architecture, [22] B. H. Fleury and P. E. Leuthold, “Radiowave propagation in
system architecture, and required control strategies need mobile communications: An overview of European research,”
answering before cost-effective mobile communications IEEE Commun. Mag., vol. 34, pp. 70–81, Feb. 1996.
[23] L. Dossi, G. Tartara, and F. Tallone, “Statistical analysis of
systems using adaptive antenna arrays can be developed. measured impulse response functions at 2.0 GHz indoor ra-
dio channels,” IEEE J. Select. Areas Commun., vol. 14, pp.
405–410, 1996.
ACKNOWLEDGMENT [24] G. E. Corazza, V. D. Esposti, M. Frullone, and G. Riva, “A
The author wishes to thank Lt. Col. M. Ryan for pro- characterization of indoor space and frequency diversity by ray
tracing modeling,” IEEE J. Select. Areas Commun., vol. 14, pp.
viding the software to manage the reference list contained 411–419, 1996.
in this paper, J. Calder for his constant encouragement [25] M. A. Panjwani, A. L. Abott, and T. S. Rappaport, “Interactive
during the course of writing this paper, and the anonymous computation of coverage regions for wireless communication
in multifloored indoor environments,” IEEE J. Select. Areas
reviewers for their helpful comments. Commun., vol. 14, pp. 420–430, 1996.
[26] L. Talbi and G. Y. Delisle, “Experimental characterization of
EHF multipath indoor radio channels,” IEEE J. Select. Areas
REFERENCES Commun., vol. 14, pp. 431–440, 1996.
[1] J. E. Padgett, C. G. Gunther, and T. Hattori, “Overview of [27] J. B. Andersen, T. S. Rappaport, and S. Yoshida, “Propaga-
wireless personal communications,” IEEE Commun. Mag., vol. tion measurements and models for wireless communications
33, pp. 28–41, Jan. 1995. channels,” IEEE Commun. Mag., vol. 33, pp. 42–49, Jan. 1995.
[2] B. Jabbari, “Scanning the special issue on wireless networks [28] R. C. French, “The effect of fading and shadowing on channel
for mobile and personal communications,” Proc. IEEE, vol. 82, reuse in mobile radio,” IEEE Trans. Veh. Technol., vol. 28, pp.
pp. 1331–1332, 1994. 171–181, 1979.
[3] R. Pandya, “Emerging mobile and personal communication [29] J. H. Winters, “Optimum combining for indoor radio systems
systems,” IEEE Commun. Mag., vol. 33, pp. 44–52, June 1995. with multiple users,” IEEE Trans. Commun., vol. COM-35, pp.
[4] T. S. Rappaport, “Wireless personal communications: Trends 1222–1230, 1987.
and challenges,” IEEE Antennas Propagat. Mag., vol. 33, pp. [30] M. Barrett and R. Arnott, “Adaptive antennas for mobile com-
19–29, Oct. 1991. munications,” Electron. Commun. Eng. J., vol. 6, pp. 203–214,
[5] K. Pahlavan and A. H. Levesque, “Wireless data communica- 1994.
tions,” Proc. IEEE, vol. 82, pp. 1398–1430, 1994. [31] W. C. Jakes, Jr., Ed., Microwave Mobile Communications.
[6] D. S. Goodman, “Personal communications,” in Mobile Com- New York: Wiley, 1974.
munication, C. G. Gunther, Ed. Berlin: Springer-Verlag, Int. [32] P. T. H. Chan, M. Palaniswami, and D. Everitt, “Neural
Zurich Seminar Digital Commun., 1994. network based dynamic channel assignment for cellular mobile
[7] J. Oetting, “Cellular mobile radio—An emerging technology,” communication systems,” IEEE Trans. Veh. Technol., vol. 43,
IEEE Commun. Mag., vol. 31, pp. 10–15, 1993. pp. 279–288, 1994.
[8] W. C. Y. Lee, Mobile Communication Design Fundamentals. [33] M. Sengoku, K. Nakano, Y. Yamaguchi, and T. Abe, “Channel
New York: Wiley, 1993. assignment in a cellular mobile communication system and
[9] , Mobile Cellular Telecommunications. New York: application of neural network,” Electron. Commun. Japan, vol.
McGraw-Hill, 1995. 75, pp. 24–36, 1992.
[10] D. D. Falconer, F. Adachi, and B. Gudmundson, “Time di- [34] E. D. Re, R. Fantacci, and L. Ronga, “A dynamic channel
vision multiple access methods for wireless personal com- allocation technique based on Hopfield neural networks,” IEEE

1056 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

 Trans. Veh. Technol., vol. 45, pp. 26–32, 1996. works,” IEEE Trans. Veh. Technol., vol. 40, pp. 366–374, 1991.
[35] N. Funabiki and S. Nishikawa, “An improved neural net- [60] J. S. Dasilva, B. Barni, and B. A. Fernandez, “European mobile
work for channel assignment problems in cellular mobile com- communications on the move,” IEEE Commun. Mag., vol. 34,
munication systems,” IEICE Trans. Commun., vol. E78, pp. pp. 60–69, Feb. 1996.
1187–1196, 1995. [61] J. H. Winters, “Optimum combining in digital mobile radio with
[36] K. Shimada, M. Sengoku, and T. Abe, “A flexible channel as- co-channel interference,” IEEE J. Select. Areas Commun., vol.
signment strategy using an artificial neural network in a cellular SAC-2, pp. 528–539, 1984.
mobile communication system,” IEICE Trans. Fundamentals [62] R. G. Vaughan, “On optimum combining at the mobile,” IEEE
Electron. Commun. Comput. Sci., vol. E78-A, pp. 693–700, Trans. Veh. Technol., vol. 37, pp. 181–188, 1988.
1995. [63] L. C. Godara and A. Cantoni, “Uniqueness and linear indepen-
[37] D. Kunz, “Channel assignment for cellular radio using neural dence of steering vectors in array space,” J. Acoust. Soc. Amer.,
networks,” IEEE Trans. Veh. Technol., vol. 40, pp. 188–193, vol. 70, pp. 467–475, 1981.
1991. [64] A. F. Naguib, A. Paulraj, and T. Kailath, “Capacity improve-
[38] P. T. H. Chan, M. Palaniswami, and D. Everitt, “Dynamic ment with base-station antenna arrays in cellular CDMA,” IEEE
channel assignment for cellular mobile communication system Trans. Veh. Technol., vol. 43, pp. 691–698, 1994.
using feedforward neural networks,” in Proc. Int. Joint Conf. [65] D. G. Brennan, “Linear diversity combining techniques,” Proc.
Neural Networks, vol. 2, Singapore, Nov. 1991, pp. 1242–1247. IRE, vol. 47, pp. 1075–1102, 1959.
[39] D. E. Everitt, “Traffic engineering of the radio interface for [66] A. A. Abu-Dayya and N. C. Beaulieu, “Outage probability
cellular mobile networks,” Proc. IEEE, vol. 82, pp. 1371–1382, of diverse cellular systems with co-channel interference in
1994. Nakagami fading,” IEEE Trans. Veh. Technol., vol. 41, pp.
[40] R. D. Yates and C. Y. Huang, “Integrated power control and 343–355, 1992.
base station assignment,” IEEE Trans. Veh. Technol., vol. 44, [67] R. G. Vaughan and N. L. Scott, “Closely spaced monopoles for
pp. 638–644, 1995. mobile communications,” Radio Sci., vol. 28, pp. 1259–1266,
[41] T. S. P. Yum and K. L. Yeung, “Blocking and handoff perfor- 1993.
mance analysis of directed retry in cellular mobile systems,” [68] P. Balaban and J. Salz, “Optimum diversity combining and
IEEE Trans. Veh. Technol., vol. 44, pp. 645–650, 1995. equalization in digital data transmission with application to
[42] S. Tekinay and B. Jabbary, “Handover and channel assignment cellular mobile radio—Part I: Theoretical considerations,” IEEE
in mobile cellular networks,” IEEE Commun. Mag., vol. 29, Trans. Commun., vol. 40, pp. 885–894, 1992.
pp. 42–46, Nov. 1991. [69] , “Optimum diversity combining and equalization in dig-
[43] G. P. Pollini, “Trends in handover design,” IEEE Commun. ital data transmission with application to cellular mobile ra-
Mag., vol. 34, pp. 82–90, Mar. 1996. dio—Part II: Numerical results,” IEEE Trans. Commun., vol.
[44] G. P. Pollini, K. S. M. Hellstern, and D. J. Goodman, “Signaling 40, pp. 895–907, 1992.
traffic volume generated by mobile and personal communica- [70] S. C. Swales, M. A. Beach, D. J. Edwards, and J. P. McGeehan,
tions,” IEEE Commun. Mag., vol. 33, pp. 60–65, June 1995. “The performance enhancement of multibeam adaptive base-
[45] W. Tuttlebee, “Cordless personal telecommunications,” IEEE station antennas for cellular land mobile radio systems,” IEEE
Commun. Mag., vol. 30, pp. 42–53, Dec. 1992. Trans. Veh. Technol., vol. 39, pp. 56–67, 1990.
[46] T. A. Freeburg, “Enabling technologies for wireless in-building [71] S. Anderson, M. Millnert, M. Viberg, and B. Wahlberg, “An
network communications—Four technical challenges, four so- adaptive array for mobile communication systems,” IEEE Trans.
lutions,” IEEE Commun. Mag., vol. 29, pp. 58–64, Apr. 1991. Veh. Technol., vol. 40, pp. 230–236, 1991.
[47] S. S. Rappaport and L. R. Hu, “Microcellular communi- [72] S. C. Swales, M. A. Beach, D. J. Edwards, and J. P. McGeehan,
cation systems with hierarchical macrocell overlays: Traffic “The realization of a multi-beam adaptive base-station antenna
performance models and analysis,” Proc. IEEE, vol. 82, pp. for cellular land mobile radio systems,” in Proc. IEEE Veh.
1383–1397, 1994. Technol. Conf., San Francisco, CA, 1989, pp. 341–348.
[48] R. Steel, J. Whitehead, and W. C. Wong, “System aspects of [73] M. A. Jones and M. A. Wickert, “Direct sequence spread
cellular radio,” IEEE Commun. Mag., vol. 33, pp. 80–86, Jan. spectrum using directionally constrained adaptive beamforming
1995. to null interference,” IEEE J. Select. Areas Commun., vol. 13,
[49] J. S. Dasilva, B. Arroyo, B. Barni, and D. Ikonomou, “European pp. 71–79, 1995.
third generation mobile systems,” IEEE Commun. Mag., vol. [74] J. Fernandez, I. R. Corden, and M. Barrett, “Adaptive array
34, pp. 68–83, Oct. 1996. algorithms for optimal combining in digital mobile communi-
[50] W. W. Wu, E. F. Miller, W. L. Pritchard, and R. L. Pickholtz, cation systems,” in Proc. Inst. Elect. Eng. Int. Conf. Antennas
“Mobile satellite communications,” Proc. IEEE, vol. 82, pp. Propagation, Edinburgh, Scotland, 1993, pp. 983–986.
1431–1448, 1994. [75] T. Trump and B. Ottersten, “Estimation of nominal direction
[51] F. Abrishamkar and Z. Siveski, “PCS global mobile satellites,” of arrival and angular spread using an array of sensors,” Signal
IEEE Commun. Mag., vol. 34, pp. 132–136, Sept. 1996. Process.,vol. 50, pp. 57–69, 1996.
[52] J. R. Norbury, “Satellite land mobile communication systems,” [76] N. Nakajima, T. Hori, and T. Nara, “Configuration of multibeam
Inst. Elect. Eng. Electron. Commun. Eng. J., pp. 245–253, base-station antenna and propagation characteristics,” in Proc.
Nov./Dec. 1989. IEICE Nat. Conv., vol. 3, 1985, p. 269.
[53] F. Ananasso and F. D. Priscoli, “The role of satellite in personal [77] K. Okada and F. Kubota, “On the dynamic channel assignment
communication services,” IEEE J. Select. Areas Commun., vol. in small zone systems for mobile communications,” Rev. CRL,
13, pp. 180–196, 1995. vol. 36, no. 179, pp. 113–123, June 1990.
[54] R. D. Gaudenzi, F. Giannetti, and M. Luise, “Advances in [78] S. Tekinary and B. Jabbary, “Handover and channel assignment
satellite CDMA transmission for mobile and personal commu- in mobile cellular networks,” IEEE Commun. Mag., vol. 29,
nications,” Proc. IEEE, vol. 84, pp. 18–39, 1996. pp. 42–46, Nov. 1991.
[55] E. D. Re, “A coordinated European effort for the definition [79] M. Yokoyama, “Decentralization, and distribution in network
of a satellite integrated environment for future mobile com- control of mobile radio communications,” IEICE Trans., vol.
munications,” IEEE Commun. Mag., vol. 34, pp. 98–104, Feb. E73, no. 10, pp. 1579–1586, Oct. 1990.
1996. [80] S. Yasuda, “Effect of adaptive frequency re-use in mobile
[56] E. D. Re, C. L. Devieux, Jr., S. Kato, S. Raghavan, D. communication dynamic channel assignment,” in Proc. IEICE
Taylor, and R. Ziemer, Eds., “Special issue on mobile satellite Spring Nat. Conv., vol. B, 860, 1989, p. 860.
communications for seamless PCS,” IEEE J. Select. Areas [81] P. Zetterberg and B. Ottersten, “The spectrum efficiency of
Commun., vol. 13, Feb. 1995. a base station antenna array system for spatially selective
[57] R. Laane, Ed., “Special issue on satellite and terrestrial systems transmission,” IEEE Trans. Veh. Technol., vol. 44, pp. 651–660,
and services for travelers,” IEEE Commun. Mag., vol. 29, Nov. 1995.
1991. [82] K. Anand, G. Mathew, and V. U. Reddy, “Blind separation
[58] W. W. Erdman, “Wireless communications: A decade of of multipath co-channel BPSK signals arriving in at an antenna
progress,” IEEE Commun. Mag., vol. 31, pp. 48–51, Dec. array,” IEEE Signal Processing Lett., vol. 2, pp. 176–178, 1995.
1993. [83] S. Talwar, M. Viberg, and A. Paulraj, “Blind estimation of
[59] D. J. Goodman, “Second generation wireless information net- multiple co-channel digital signals using an antenna array,”

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1057

 IEEE Signal Processing Lett., vol. 1, pp. 29–31, 1994. Vernucci, M. Ruggieri, and T. Jones, “A comparative analysis
[84] S. Talwar and A. Paulraj, “Performance analysis of blind of frequency scanning and multispot beam satellite systems for
digital signal copy algorithms,” in Proc. MILCOM, vol. I, Fort mobile communication,” Int. J. Satell. Commun., vol. 13, pp.
Monmouth, NJ, 1994, pp. 123–128. 85–104, 1995.
[85] S. Talwar, M. Viberg, and A. Paulraj, “Blind estimation of [107] C. R. Horton and K. Abend, “Adaptive array antenna for
multiple co-channel digital signals using an antenna array—Part satellite cellular and direct broadcast communications,” in Proc.
I: Algorithms,” IEEE Trans. Signal Processing, vol. 44, pp. 3rd Int. Mobile Satellite Conf. (IMSC), Pasadena, CA, 1993, pp.
1184–1197, 1996. 47–52.
[86] S. V. Schell and W. A. Gardner, “Programmable canonical [108] E. Hirshfield, “The Globalstar system,” Appl. Microwave Wire-
correlation analysis: A flexible framework for blind adaptive less, pp. 26–41, Summer 1995.
spatial filtering,” IEEE Trans. Signal Processing, vol. 43, pp. [109] J. J. Schuss, J. Upton, B. Myers, T. Sikina, A. Rohwer, P.
2898–2908, 1995. Makridakas, R. Francois, and L. Wardle, “The IRIDIUM main
[87] , “Blind adaptive antenna arrays in cellular communica- mission antenna concepts,” in Proc. IEEE Int. Symp. Phased
tions for increased capacity,” in Proc. Virginia Tech 3rd Symp. Array Systems Tech., Boston, MA, 1996, pp. 411–415.
Wireless Personal Communications, Blacksburg, VA, 1993, pp. [110] T. Gebauer and H. G. Gockler, “Channel-individual adaptive
6/1–6/2. beamforming for mobile satellite communications,” IEEE J.
[88] Q. Wu, K. M. Wong, and R. Ho, “Fast algorithm for adaptive Select. Areas Commun., vol. 13, pp. 439–448, 1995.
beamforming of cyclic signals,” Inst. Elect. Eng. Proc.—Radar, [111] D. G. Bodnar, B. K. Rainer, and Y. Rahmatsamii, “A novel ar-
Sonar Navigation, vol. 141, pp. 312–318, 1994. ray antenna for MSAT application,” IEEE Trans. Veh. Technol.,
[89] L. Tong, Y. Inouye, and R. W. Liu, “Waveform preserving vol. 38, pp. 86–94, 1989.
blind estimation of multiple independent sources,” IEEE Trans. [112] H. J. Schrewe, “An adaptive antenna array for mobile reception
Signal Processing, vol. 41, pp. 2461–2470, 1993. of DBS-satellites,” in Proc. IEEE 44th Veh. Technol. Conf.,
[90] J. F. Cardoso and A. Souloumiac, “Blind beamforming for non- Stockholm, Sweden, 1994, pp. 1494–1497.
Gaussian signals,” Inst. Elect. Eng. Proc.—F, 1993, vol. 140, [113] R. Suzuki, Y. Matsumoto, R. Miura, and N. Hamamoto, “Mo-
pp. 362–370. bile TDM/TDMA system with active array antenna,” in Proc.
[91] A. J. V. D. Veen, S. Talwar, and A. Paulraj, “Blind estimation IEEE Global Telecommun. Conf. (GLOBECOM), Houston, TX,
of multiple digital signals transmitted over FIR channels,” IEEE 1991, pp. 1569–1573.
Signal Processing Lett., vol. 2, pp. 99–102, 1995. [114] W. Chujo and K. Kashiki, “Spherical array antenna using digital
[92] Q. Wu and K. M. Wong, “Blind adaptive beamforming for beamforming techniques for mobile satellite communications,”
cyclostationary signals,” IEEE Trans. Signal Processing, vol. Electron. Commun. Japan, (English trans. of Denshi Tsushin
44, pp. 2757–2767, 1996. Gakkai Ronbunshi), vol. 75, pp. 76–86, 1992.
[93] W. A. Gardner, “Exploitation of spectral redundancy in cyclo- [115] S. Ohmori, Y. Hase, H. Wakana, and S. Taira, “Experiments on
stationary signals,” IEEE Signal Processing Mag., vol. 8, pp. aeronautical satellite communications using ETS-V satellite,”
14–37, 1991. IEEE Trans. Aerosp. Electron. Syst., vol. 28, pp. 788–796, 1992.
[94] B. G. Agee, S. V. Schell, and W. A. Gardner, “Spectral self [116] R. M. Allnutt, T. Pratt, and A. Dissanayake, “A study in
coherence restoral: A new approach to blind adaptive signal small scale antenna diversity as a means of reducing effects of
extraction using antenna arrays,” Proc. IEEE, vol. 78, pp. satellite motion induced multipath fading for handheld satellite
753–767, 1990. communication systems,” in Proc. Inst. Elect. Eng. 9th Int. Conf.
[95] A. Swindlehurst and J. Yang, “Using least squares to improve Antennas Propagation, Eindhoven, The Netherlands, 1995, pp.
blind signal copy performance,” IEEE Signal Processing Lett., 135–139.
vol. 1, pp. 80–82, 1994. [117] I. Chiba, T. Takahashi, and Y. Karasawa, “Transmitting null
[96] B. Agee, “Blind separation and capture of communication beamforming with beam space adaptive array antennas,” in
signals using a multitarget constant modulus beamformer,” in Proc. IEEE 44th Vehicular Technology Conf., Stockholm, Swe-
Proc. IEEE MILCOM, Bedford, MA, 1989, pp. 340–346. den, 1994, pp. 1498–1502.
[97] I. Parra, G. Xu, and H. Liu, “Least squares projective constant [118] D. Gerlach and A. Paulraj, “Adaptive transmitting antenna
modulus approach,” in Proc. IEEE Int. Symp. Personal, Indoor, arrays with feedback,” IEEE Signal Processing Lett., vol. 1,
Mobile Radio Communications, Toronto, Canada, 1995, pp. pp. 150–152, 1994.
673–676. [119] G. Raleigh, S. Diggavi, V. Jones, and A. Paulraj, “A blind adap-
[98] J. F. Cardoso, “Source separation using higher order mo- tive transmit antenna algorithm for wireless communications,”
ments,” in Proc. IEEE ICASSP, Albuquerque, NM, 1990, pp. in Proc. IEEE ICC, Seattle, WA, 1995, pp. 1494–1499.
2109–2112. [120] G. Xu and H. Liu, “An effective transmission beamforming
[99] Y. Hua, H. Yang, and W. Qiu, “Source correlation compen- scheme for frequency division duplex digital wireless commu-
sation for blind channel identification based on second-order nication systems,” in Proc. IEEE ICASSP, Detroit, MI, 1995,
statistics,” IEEE Signal Processing Lett., vol. 1, pp. 119–120, pp. 1729–1732.
1994. [121] P. S. Henry and B. S. Glance, “A new approach to high capacity
[100] T. Ohgane, N . Matsuzawa, T. Shimura, M. Mizuno, and digital mobile radio,” Bell Syst. Tech. J., vol. 60, pp. 1891–1904,
H. Sasaoka, “BER performance of CMA adaptive array for 1981.
high-speed GMSK mobile communication—A description of [122] C. Farsakh and J. A. Nossek, “Channel allocation and down link
measurements in central Tokyo,” IEEE Trans. Veh. Technol., beamforming in an SDMA mobile radio system,” in Proc. IEEE
vol. 42, pp. 484–490, 1993. Int. Symp. Personal, Indoor Mobile Radio Communications,
[101] K. Tsunekawa, “Diversity antenna for portable telephones,” in Toronto, Canada, 1995, pp. 687–691.
Proc. IEEE Vehicular Technology Conf., San Francisco, CA, [123] J. Wigard, P. E. Mogensen, F. Frederiksen, and O. Norklit,
1989, pp. 50–56. “Evaluation of optimum diversity combining in DECT,” in
[102] C. L. B. Despins, D. D. Falconer, and S. A. Mahmoud, “Com- Proc. IEEE Int. Symp. Personal, Indoor Mobile Radio Commu-
pound strategies of coding, equalization and space diversity for nications, Toronto, Canada, 1995, pp. 507–511.
wideband TDMA indoor wireless channels,” IEEE Trans. Veh. [124] M. Goldburg and R. Roy, “The Impacts of SDMA on PCS
Technol., vol. 41, pp. 369–379, 1992. system design,” in Proc. IEEE 1994 3rd Annu. Int. Conf.
[103] H. Hashemi, “The indoor radio propagation channels,” Proc. Universal Personal Communications, San Diego, CA, 1994, pp.
IEEE, vol. 81, pp. 943–968, 1993. 242–246.
[104] C. Passerini, M. Missiroli, G. Riva, and M. Frullone, “Adaptive [125] T. Ohgane, T. Shimura, N. Matsuzawa, and H. Sasaoka, “An
antenna arrays for reducing the delay spread in indoor radio implementation of a CMA adaptive array for high speed GMSK
channels,” Inst. Elect. Eng. Electron. Lett., vol. 32, pp. 280–281, transmission in mobile communications,” IEEE Trans. Veh.
1996. Technol., vol. 42, pp. 282–288, 1993.
[105] P. Russo, A. D’ippolito, M. Ruggieri, N. Volpicella, G. Scialino, [126] R. W. Klukas and M. Fattouche, “Radio signal direction finding
S. Vatta, and T. Jones, “A frequency scanning satellite system in the urban radio environment,” in Proc. Nat. Tech. Meeting
for land mobile communications,” Int. J. Satell. Commun., vol. Institute Navigation, San Francisco, CA, pp. 151–160, 1993.
11, pp. 87–103, 1993. [127] G. Morrison, M. Fattouche, and D. Tholl, “Parametric modeling
[106] C. Cornacchini, R. Crescimbeni, A. D’ippolito, P. Russo, A. and spectral estimation of indoor radio propagation data,” in

1058 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997

 Proc. Wireless 1992, T. R. Labs, Calgary, Alberta, Canada. [152] B. Suard, A. F. Naguib, G. Xu, and A. Paulraj, “Performance of
[128] A. F. Naguib and A. Paulraj, “Performance of CDMA cellular CDMA mobile communication systems using antenna arrays,”
networks with base-station antenna arrays,” in Proc. IEEE Int. in Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing
Zurich Seminar Communications, Zurich, Switzerland, 1994, (ICASSP), Minneapolis, MN, 1993, pp. 153–156.
pp. 87–100. [153] Y. Wang and J. R. Cruz, “Adaptive antenna arrays for the
[129] G. L. Turin, “Introduction to spread-spectrum antimultipath reverse link of CDMA cellular communication systems,” Inst.
techniques and their application to urban digital radio,” Proc. Elect. Eng. Electron. Lett., vol. 30, pp. 1017–1018, 1994.
IEEE, vol. 68, pp. 328–353, 1980. [154] J. H. Winters, “Signal acquisition and tracking with adaptive
[130] J. S. Lehnert and M. B. Pursely, “Multipath diversity reception arrays in the digital mobile radio system IS-54 with flat fading,”
of spread spectrum multiple access communications,” IEEE IEEE Trans. Veh. Technol., vol. 42, pp. 377–384, 1993.
Trans. Commun. vol. COM-35, pp. 1189–1198, 1987. [155] T. Ohgane, “Characteristics of CMA adaptive array for selective
[131] G. Turin, “The effect of multipath and fading on the perfor- fading compensation in digital land mobile radio communica-
mance of direct sequence CDMA systems,” IEEE J. Select. tions,” Electron. Commun. Japan, vol. 74, pp. 43–53, 1991.
Areas Commun., vol. 2, pp. 597–603, 1984. [156] M. A. Beach, P. Guemas, and A. R. Nix, “Capacity and ser-
[132] R. Price and P. E. Green, “A communication technique for vice extension for wireless networks using adaptive antennas,”
multipath channels,” Proc. IRE, vol. 46, pp. 555–570, 1958. Electron. Lett., vol. 30, pp. 1813–1814, 1994.
[133] S. Choi and T. K. Sarkar, “Adaptive antenna array utilizing the [157] M. A. Beach, H. Xue, R. Davies, J. P. McGeehan, and M.
conjugate gradient method for multipath mobile communica- Barrett, “Adaptive antennas for third generation systems,” in
tion,” Signal Process., vol. 29, pp. 319–333, 1992. Proc. Inst. Elect. Eng. Colloquium Mobile Communications
[134] L. F. Chang and P. J. Porter, “Performance comparison of Toward Year 2000, London, U. K., 1994, pp. 10/1–10/6.
antenna diversity and slow frequency hopping for the TDMA [158] P. Kawala and U. H. Sheikh, “Adaptive multiple-beam array for
portable radio channel,” IEEE Trans. Veh. Technol., vol. 38, pp. wireless communications,” Inst. Elect. Eng. 8th Int. Conf. An-
222–229, 1989. tennas Propagation, Edinburgh, Scotland, 1993, pp. 970–974.
[135] D. J. Torrieri and K. Bakhru, “An anticipative adaptive array [159] M. W. Ganz and R. T. Compton, Jr., “Protection of PSK
for frequency-hopping communications,” IEEE Trans. Aerosp. communication systems with adaptive arrays,” IEEE Trans.
Electron. Syst., vol. 24, pp. 449–456, 1988. Aerosp. Electron. Syst., vol. AES-23, pp. 528–536, 1987.
[136] K. Bakhru and D. J. Torrieri, “The maximin algorithm for [160] M. Nagatsuka, N. Ishii, R. Kohno, and H. Imai, “Adaptive array
adaptive array and frequency-hopping communications,” IEEE antenna based on spatial spectral estimation using maximum
Trans. Antennas Propagat., vol. AP-32, pp. 919–927, 1984. entropy method,” IEICE Trans. Commun., vol. E77-B, pp.
[137] D. J. Torrieri and K. Bakhru, “Frequency compensation in an 624–633, 1994.
adaptive antenna system for frequency-hopping communica- [161] J. C. Liberti and T. S. Rappaport, “Analytical results for
tion,” IEEE Trans. Aerosp. Electron. Syst., vol. AES-23, pp. reverse channel performance improvements in CDMA cellular
448–467, 1987. communication systems employing adaptive antennas,” in Proc.
[138] R. T. Compton Jr., “An adaptive array in a spread-spectrum IEEE Globecom’93, Houston, TX, pp. 42–47.
communication system,” Proc. IEEE, vol. 66, pp. 289–298, [162] C. Yim, R. Kohno, and H. Imai, “Adaptive array antenna based
1978. on estimation of arrival angles using DFT on spatial domain,”
[139] D. M. Dlugos and R. A. Scholtz, “Acquisition of spread Electron. Commun. Japan, vol. 76, pp. 96–108, 1993.
spectrum signals by an adaptive array,” IEEE Trans. Acoust., [163] Q. T. Zhang, “Outage probability of cellular mobile radio in the
Speech, Signal Processing, vol. 37, pp. 1253–1270, 1989. presence of multiple Nakagami interferers with arbitrary fading
[140] M. D. Windram, L. Brunt, and E. J. Wilson, “Adaptive antennas parameters,” IEEE Trans. Veh. Technol., vol. 44, pp. 661–667,
for u. h. f. broadcast reception,” Proc. Inst. Elect. Eng., vol. 1995.
127, Pt. F, pp. 249–256, 1980. [164] , “Outage probability in cellular mobile radio due to
[141] S. Choi, T. K. Sarkar, and S. S. Lee, “Design of two- Nakagami signal and interferers with arbitrary parameters,”
dimensional Tseng window and its application to antenna IEEE Trans. Veh. Technol., vol. 45, pp. 394–372, 1996.
array for the detection of AM signal in the presence of strong [165] S. A. Harbin and B. K. Rainer, “Low-power wireless mobile
jammers in mobile communication,” Signal Process., vol. 34, communications system,” in Proc. IEEE 44th Vehicular Tech-
pp. 297–310, 1993. nology Conf., vol. 1, Stockholm, Sweden, 1994, pp. 673–676.
[142] C. Despins, D. Falconer, and S. Mahmoud, “Coding and opti- [166] Y. Yamada, K. Kagoshima, and K. Tsunekawa, “Diversity
mum baseband combining for wideband TDMA indoor wireless antennas for base and mobile situation in land mobile com-
channels,” Canadian J. Elect. Comput. Eng., vol. 16, pp. 53–62, munication systems,” IEICE Trans., vol. E74, pp. 3202–3209,
Apr. 1991. 1991.
[143] J. Salz and J. H. Winters, “Effect of fading correlation on [167] F. Adachi, M. T. Feeney, A. G. Williamson, and J. D. Parsons,
adaptive arrays in digital mobile radio,” IEEE Trans. Veh. “Cross correlation between the envelopes of 900 MHz signals
Technol., vol. 43, pp. 1049–1057, 1994. received at a mobile radio base station site,” Proc. Inst. Elect.
[144] P. Balaban and J. Salz, “Dual diversity combining and equaliza- Eng., vol. 133, Pt. F, 1986, pp. 506–512.
tion in digital cellular mobile radio,” IEEE Trans. Veh. Technol., [168] P. C. F. Eggers, “Angular dispersive mobile radio environments
vol. 40, pp. 342–354, 1991. sensed by highly directive base station antennas,” in Proc. IEEE
[145] J. G. Proakis, “Adaptive equalization for TDMA digital mobile Int. Symp. Personal, Indoor Mobile Radio Communications,
radio,” IEEE Trans. Veh. Technol., vol. 40, pp. 333–341, 1991. Toronto, Canada, 1995, pp. 522–526.
[146] N. Ishii and R. Kohno, “Spatial and temporal equalization based [169] W. C. Y. Lee, “Effects of correlation between two mobile radio
on an adaptive tapped-delay-line array antenna,” IEICE Trans. base station antennas,” IEEE Trans. Commun., vol. COM-21,
Communications, vol. E78-B, pp. 1162–1169, Aug. 1995. pp. 1214–1224, 1973.
[147] S. Sivanand, “On adaptive arrays in mobile communication,” [170] V. Wong and C. Leung, “Effect of Rayleigh fading in a multihop
in Proc. IEEE Nat. Telesystems Conf., Atlanta, GA, 1993, pp. mobile packet radio network with capture,” IEEE Trans. Veh.
55–58. Technol., vol. 44, pp. 630–637, 1995.
[148] K. S. Gilhousen, I. M. Jacobs, R. Padovani, A. Viterbi, L. A. [171] P. E. Wirth, “Teletraffic implications of database architectures
Weaver, and C. Wheatly, “On the capacity of a cellular CDMA in mobile and personal communications,” IEEE Commun. Mag.,
system,” IEEE Trans. Veh. Technol., vol. 40, pp. 303–312, 1991. vol. 33, pp. 54–59, June 1995.
[149] M. Tangemann and R. Rheinschmitt, “Comparison of upgrade [172] B. Jabbari, G. Colombo, A. Nakajima, and J. Kulkarni, “Net-
techniques for mobile communication systems,” in Proc. IEEE work issues for wireless communications,” IEEE Commun.
Int. Conf. Commununications, New Orleans, LA, 1994, pp. Mag., vol. 33, pp. 88–98, Jan. 1995.
201–205. [173] W. D. Ivancic, M. J. Shalkhauser, and J. A. Quintana, “A net-
[150] J. H. Winters, “On the capacity of radio communication systems work architecture for a geostationary communication satellite,”
with diversity in Rayleigh fading environment,” IEEE J. Select. IEEE Commun. Mag., vol. 32, pp. 72–84, July 1994.
Areas Commun., vol. SAC-5, pp. 871–878, 1987. [174] M. I. Skolnik, Ed., Radar Handbook, 2nd ed. New York:
[151] A. R. Lopez, “Performance predictions for cellular switched McGraw-Hill, 1991.
beam intelligent antenna system,” IEEE Commun. Mag., vol. [175] S. Ohmori, “Vehicle antennas for mobile satellite communica-
34, pp. 152–154, Oct. 1996. tions,” IEICE Trans., vol. E74, pp. 3210–3221, 1991.

GODARA: ANTENNA ARRAYS AND MOBILE COMMUNICATIONS—PART I 1059

[176] P. Stabrook, J. Huang, W. Rafferty, and M. K. Sue, “A 20/30 [192] J. Hirokawa, M. Ando, and N. Goto, “Waveguide junction with
GHz personal access satellite system design,” in Proc. Int. Conf. an inductive post,” IEICE Trans. Electron., vol. E75-C, pp.
Commun., vol. 7.4, Boston, MA, 1989, pp. 1–7. 348–351, 1992.
[177] N. C. Karmakar and M. E. Bialkowski, “On the modeling [193] L. Rexberg, U. Jostell, M. Andersson, and T. Lorentzon, “Feed
of a cavity enclosed broadband circular patch antenna for L- array element for mobile communication service systems,” in
band land mobile satellite communications,” Microwave Opt. Proc. IEEE Antennas Propagation Int. Symp., Seattle, WA,
Technol. Lett., vol. 7, pp. 784–785, 1994. 1994, pp. 902–905.
[178] R. Miura, R. Suzuki, N. Hamamoto, and Y. Matsumoto, “Active [194] N. Kuga and H. Arai, “A flat four-beam switched array
array antenna with ceramic substrate for TDM/TDMA mobile antenna,” IEEE Trans. Antennas Propagat., vol. 44, pp.
satellite communications,” in Proc. IEEE Global Telecommun. 1227–1230, 1996.
Conf. (GLOBECOM), Houston, TX, 1993, pp. 564–568. [195] K. W. Lo, “Null steering in monopulse planar array using
[179] R. N. Simons, E. S. Kelly, R. Q. Lee, and S. R. Taub, amplitude only control,” Electron. Lett., vol. 26, pp. 1700–1701,
“Radial microstrip slot line feed network for circular mobile 1990.
communications array,” in Proc. IEEE Int. Symp. Antennas [196] T. B. Vu, “Pattern nulling in difference pattern by amplitude
Propagation, Seattle, WA, 1994, pp. 1024–1027. control,” IEEE Trans. Antennas Propagat., vol. AP-33, pp.
[180] W. Chujo, Y. Konishi, Y. Ohtaki, and M Fujise, “Conformal 669–671, 1985.
array antenna for mobile satellite communications,” Electron. [197] S. Ohmori, S. Yamamoto, K. Tanaka, M. Matsunaga, and M.
Commun. Japan (English trans. of Denshi Tsushin Gakkai Tsuchiy, “Phased array tracking antenna system for vehicles,”
Ronbunshi), vol. 75, pp. 97–108, 1992. in Proc. Australia’s Electron. Conv. (IREECON), Sydney, Aus-
[181] J. C. Batchelor, K. Voudouris, and R. J. Langley, “Dual mode tralia, 1991, pp. 17–20.
and stacked concentric ring patch antenna array,” Inst. Elect. [198] S. Ohmori, K. Mano, and K. Tanaka, “A phased array tracking
Eng. Electron. Lett., vol. 29, pp. 1319–1320, 1993. antenna for vehicles,” in Proc. 2nd Int. Mobile Satellite Conf.
[182] R. N. Simons, N. I. Dib, R. Q. Lee, and L. P. B. Katehi, (IMSC) Ottawa, 1990, pp. 519–522.
“Integrated unipolar transition for linearly tapered slot antenna,” [199] J. F. Diouris, B. Feuvrie, and J. Saillard, “Adaptive multisensor
IEEE Trans. Antennas Propagat., vol. 43, pp. 998–1002, 1995. receiver for mobile communications,” Ann. Telecommun., vol.
[183] A. Petosa, P. C. Strickland, and J. S. Wight, “Analysis of a 48, pp. 35–46, 1993.
low profile slotted microstrip antenna for cellular mobile radio,” [200] J. Ventura-Traveset, I. Stojkovic, F. Coromina, J. Benedicto,
IEEE Trans. Veh. Technol., vol. 41, pp. 363–368, 1992. and F. Petz, “Key payload technologies for future satellite
[184] Y. Kazama, “One-point feed printed crossed dipole with a personal communications: A European perspective,” Int. J.
reflector antenna combined with a reactively loaded parasitic Satell. Commun., vol. 13, pp. 117–135, 1995.
loop for land mobile satellite communication antennas,” Proc.
Inst. Elect. Eng., vol. 140, Pt. H, 1993, pp. 417–420.
[185] J. Aurinsalo, I. Huhtinen, A. Hujanen, A. Jongejans, I. Marttila,
R. Mutanen, I. Pitkanen, P. Rinous, T. Salminen, S. Tallqvist,
and M. N. Travers, “Steerable L-band antennas for land mobile
terminals,” in Proc. IEEE 44th Vehicular Technology Conf., Lal C. Godara (Senior Member, IEEE) received
Stockholm, Sweden, 1994, pp. 600–604. the B.E.(Hons.) degree from the Birla Institute
[186] A. Kuramoto, T. Yamane, R. Simizu, and N. Endo, “Compact of Technology and Science, Pilani, India, the
mobile antenna system with a new developed simple auto M.Tech. degree from the Indian Institute of
tracking method for satellite communications,” Trans. IEICE, Science, Bangalore, India, and the Ph.D. degree
vol. J71-B, pp. 1389–1392, 1988. from the University of Newcastle, Shortland,
[187] S. Hosono, N. Goto, and H. Arai, “A flat diversity antenna Australia.
by disk loaded monopole and three notches,” IEEE Trans. Veh. He has served at many universities and re-
Technol., vol. 43, pp. 353–358, 1994. search establishments, including Birla Institute
[188] K. S. Yngvesson, T. L. Korzeniowski, and Y. S. Kim, “The of Technology and Science; Australian National
tapered slot antenna—A new integrated element for millimeter University, Canberra; University of Canberra
wave application,” IEEE Trans. Microwave Theory Tech., vol. (previously Canberra College of Advanced Education); and the Defence
37, pp. 365–374, 1989. Research and Development Laboratory, Hyderabad. He has visited Stan-
[189] J. Hirokawa, M. Ando, N. Goto, N. Takahashi, T. Ojima, and M. ford University, Palo Alto, CA, Yale University, New Haven, CT, and
Vematsu, “A single layer slotted leaky waveguide array antenna Syracuse University, NY, as a Research Scholar. He currently is a Senior
for mobile reception of direct broadcast from satellite,” IEEE Lecturer at University College, University of New South Wales, Australian
Trans. Veh. Technol., vol. 44, pp. 749–755, 1995. Defence Force Academy, Canberra. His research interests include adaptive
[190] J. Hirokawa, M. Ando, and N. Goto, “A single layer multiple antenna array processing, mobile communications, spectral analysis, and
way power divider for a planner slotting waveguide array,” echo cancelling.
IEICE Trans. Commun. vol. E75-B, pp. 781–787, 1992. Dr. Godara is a Fellow of the Acoustical Society of America. He helped
[191] J. Hirokawa, K. Sakurai, M. Ando, and N. Goto, “An analysis found the Australian Capital Territory Section of the IEEE in 1988 and
of a waveguide junction with an inductive post,” IEEE Trans. was its Founding Chairman for three years. He was Chairman of the IEEE
Microwave Theory Tech., vol. 39, pp. 563–566, 1991. Australia Council during 1995 and 1996.

1060 PROCEEDINGS OF THE IEEE, VOL. 85, NO. 7, JULY 1997