Content uploaded by Robert Daniels
Author content
All content in this area was uploaded by Robert Daniels on Sep 19, 2019
Content may be subject to copyright.
60 GHz Wireless
Communications:
Emerging Requirements
and Design
Recommendations
Robert C. Daniels and Robert W. Heath, Jr.,
The University of Texas at Austin
1. Introduction
T
he growth of wireless communications is spurred by the consumer
desire for untethered access to information and entertainment. While
contemporary unlicensed systems support light and moderate levels
of wireless data traffic, as seen in Bluetooth and wireless local area
networks (WLANs), current technology is unable to supply data rates com-
parable to wired standards like gigabit Ethernet and high-definition multime-
dia interface (HDMI). Fortunately, as illustrated in Figure 1, an abundance of
widely available spectrum surrounding the 60 GHz (60G) operating frequen-
cy has the ability to support these high-rate, unlicensed wireless communi-
cations†. While unlicensed spectrum around 2.5 GHz and 5 GHz is also
available internationally, the amount of available 60G bandwidth is an order
of magnitude above that available at 2.5 GHz and 5 GHz.
Abstract: Multiple GHz of internation-
ally available, unlicensed spectrum
surrounding the 60 GHz carrier fre-
quency has the ability to accommo-
date high-throughput wireless
communications. While the size and
availability of this free spectrum make
it very attractive for wireless applica-
tions, 60 GHz implementations must
overcome many challenges. For
example, the high attenuation and
directional nature of the 60 GHz wire-
less channel as well as limited gain
amplifiers and excessive phase noise
in 60 GHz transceivers are explicit
implementation difficulties. The chal-
lenges associated with this channel
motivate commercial deployment of
short-range wireless local area net-
works, wireless personal area net-
works, and vehicular networks. In this
paper we detail design tradeoffs for
algorithms in the 60 GHz physical
layer including modulation, equaliza-
tion, and space-time processing. The
discussion is enhanced by considering
the limitations in circuit design, char-
acteristics of the effective wireless
channel (including antennas), and
performance requirements to support
current and next generation 60 GHz
wireless communication applications.
© IMAGE SOURCE & PHOTO F/X2
Digital Object Identifier 10.1109/MVT.2008.915320
SEPTEMBER 2007 | IEEE VEHICULAR TECHNOLOGY MAGAZINE 1556-6072/07/$25.00©2007IEEE |||
41
†The actual regulations for radiating devices vary per national entity in terms of allowable output
power, legal supported applications, antenna gain limitations, and other related parameters. For a
more specific treatment of 60G regulatory guidelines, see [1].
42
||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | SEPTEMBER 2007
Unfortunately, 60G systems exhibit several challenges
that have made them difficult to deploy. The 60G wireless
channel shows 20 to 40 dB increased free space path loss
and suffers from 15 up to 30 dB/km atmospheric absorp-
tion depending on the atmospheric conditions. Multipath
effects, except for indoor reflections, are vastly reduced
at 60G making non-line-of-sight (NLOS) communication
very difficult. Furthermore, millimeter-wave transceivers
are presented with new design challenges such as
increased phase noise, limited amplifier gain, and the
need for transmission line modeling of circuit compo-
nents. Consequently, the design of 60G systems is not
straightforward.
There are several surveys of 60G wireless communica-
tion that summarize 60G design tradeoffs. The current liter-
ature, however, lacks a detailed discussion of modulation,
equalization, and algorithm design at the physical layer.
For example [2], [3], [4] provide early surveys of 60G com-
munications, but focus more on channel characterization
or the supported applications. More recently, [1], [5]–[7]
have presented summaries on CMOS implementation [5],
amplifier design [6], and broad surveys of 60G wireless [1],
[7]. While [1] and [7] are similar in flavor to this article, nei-
ther provide a comprehensive overview of physical layer
algorithm design in 60G systems.
This paper provides two main contributions. First, this
paper summarizes the wireless applications that are well
suited for 60G, listing the advantages and drawbacks of a
60G implementation. Second, this article analyzes trade-
offs in modulation, equalization, and general physical
layer algorithm selection. This tradeoff is reinforced with
a preview of the physical layer for the IEEE 802.15.3c stan-
dard in development. To this end, Section II surveys 60G
channel characteristics and Section III summarizes
millimeter-wave circuit design issues. Building on the
background presented in these sections, Section IV will
discuss modulation and signal processing techniques to
compensate for and leverage the characteristics of the
wireless channel and transceiver design. Section V sug-
gests the most suitable applications to be supported at
60G. Next, Section VI summarizes the proposals for the
IEEE 802.15.3c 60G standard. Section VII describes future
technologies that may change the landscape of the 60G
market. Finally, Section VIII provides a brief summary of
the previous sections.
2. 60G Wireless Channel Characterization
Wireless channel characteristics must be well understood
to determine the appropriate system design at 60G.
Large-scale effects suggest range limitations for line-of-
sight (LOS) communication. Material propagation of 60G
radiation characterizes the performance indoors. The
extent of multipath determines the complexity of receiver
structures for equalization as well as the ability to com-
municate in NLOS scenarios. Multipath can be considered
in several contexts. In the context of transmit and receive
antennas, the severity of multipath depends both on
antenna directivity and polarization. How the multipath
changes with time determines the time coherence of the
channel while the spatial richness of the multipath sug-
gests whether spatial processing techniques are worth-
while. In this section, the properties of the 60G channel
and its implications are summarized.
2.1 Large Scale and Outdoor Channel Effects
Path loss is a measure of average radio wave energy
decay as a function of distance, frequency, and other
location specific parameters. Simple calculations using
the Friis free-space path loss formula show a 20–40 dB
power penalty for 60G versus unlicensed communica-
tion at operating frequencies below 6 GHz. Loss due to
atmospheric absorption can account for an additional
7–15.5 dB/km power loss in the received signal at 57–64
GHz carrier frequencies [3]. Although water vapor does
not contribute greatly to signal attenuation at normal
concentrations, rain droplets that form when the
atmosphere becomes saturated can further attenuate
the signal. For example, given a rainfall rate of 50
mm/hour, different models predict between 8–18
dB/km additional atmospheric attenuation [3]. Precipi-
tation also decreases cross-polarization discrimination
(XPD), reducing the benefits of the extra dimension
afforded by antenna polarization [8]. The erosive
effects on signal energy described above suggest that
FIGURE 1
International unlicensed spectrum around 60 GHz.
AN ABUNDANCE OF WIDELY AVAILABLE
SPECTRUM SURROUNDING THE 60 GHZ
OPERATING FREQUENCY HAS THE ABILITY TO
SUPPORT HIGH-RATE, UNLICENSED WIRELESS
COMMUNICATIONS FOR UNTETHERED CONSUMER
INFORMATION AND ENTERTAINMENT
60G systems will have to compensate for lower
received signal energy, and that outdoor systems may
be infeasible for communication over distances beyond
a few kilometers.
2.2 Material Attenuation
The attenuation experienced by 60G electromagnetic
fields through indoor materials is summarized and com-
pared to 2.5 GHz in Table 1. Although not always drastic
or necessarily consistent (see glass attenuation), there is
a basic trend of higher material attenuation observed at
60G. This effect, combined with the increased path loss
contains signals to a single room for 60G indoor wireless
propagation [9]. This allows for increased security and
decreased interference from adjacent networks while at
the same time complicating multi-room coverage.
2.3 Multipath Contributions
Multiple paths in the 60G propagation channel create
multipath interference. The delay and attenuation asso-
ciated with each multipath component corresponds to
path length differences due to surface reflections, scat-
tering by small objects, and propagation through differ-
ent mediums. The scattering effect is drastically
reduced since scattering occurs when objects are simi-
lar in dimension to the operating wavelength. Transmis-
sion through most objects is also reduced as a
consequence of the limited ability to penetrate solid
substances [9]. Although these traditional sources of
multipath are diminished, reflection effects are ampli-
fied. Reflection occurs on objects larger than the dimen-
sions of operating wavelength implying that objects
traditionally acting as scattering objects now become
reflectors at 60G. The root mean square (RMS) delay
spread, TRMS , quantifies the temporal spreading effect
(in seconds) of the wireless channel. Here, we repro-
duce the approximated measurements from [10] in
Table 2 for indoor scenarios. As a rule of thumb, a sys-
tem experiencing an RMS delay spread of TRMS , will suf-
fer |fsTRMS −1|symbols of temporal dispersion or
intersymbol interference (ISI) in the wireless channel
given fsas the data symbol frequency. Because 60G sys-
tems will likely operate on very large bandwidth chan-
nels (on the order of GHz) even traditionally
small RMS delays spreads, as in Table 2, pro-
vide appreciable ISI. Hence, as discussed below,
60G wireless communication systems indoors
or in vehicles must employ antenna techniques
and/or complex receiver equalization to elimi-
nate the severe ISI.
2.4 Impact of Antenna Parameters
on Multipath
The selection of transmit and receive antennas
influences the observed multipath at the receiv-
er. Directive antennas can be used to reduce multipath
contributions by limiting the azimuth and elevation angu-
lar range from which radiation is emitted and captured. In
[11] the RMS delay spread was reduced from 18 ns to 1 ns
by switching from an omnidirectional antenna to one
with a 5◦3-dB-beamwidth at the receiver. The polariza-
tion of antennas can also influence propagation. In [12] it
was observed that circularly polarized signals reduce
RMS delay spread values by a factor of 2 over linearly
polarized signals. Consequently, many 60G systems
employ highly directional antennas with circular polariza-
tion to limit the digital equalization of ISI resulting from
excessive multipath. This forces the transmitter and
receiver to somehow “point” to each other in order to
communicate, resulting in undesired additions to medium
access control (MAC) and physical layer functionality.
2.5 Spatial Characteristic of Multipath
The spatial characteristic of multipath is often measured
using angular spread, which is defined with respect to a
mean angle of arrival or departure. An angular spread of
value 1 indicates an observed channel that varies heavily
with azimuth angle at the receiver while 0 indicates a
channel that does not vary with azimuth angle. In [13] the
Material Loss at 60 GHz Loss at 2.5 GHz
Drywall 2.4 (dB/cm) 2.1 (dB/cm)
Whiteboard 5.0 (dB/cm) 0.3 (dB/cm)
Glass 11.3 (dB/cm) 20.0 (dB/cm)
Mesh Glass 31.9 (dB/cm) 24.1 (dB/cm)
TABLE 1
Office material attenuation at 60 GHz
and 2.5 GHz [9].
Dimensions (m3) Wall Material TRMS,90%(ns)TRMS ,10%(ns)
24.5 ×11.2 ×4.5 wood 40 45
30 ×21 ×6 rock wool 30 35
43 ×41 ×7 concrete 40 60
33.5 ×32.2 ×3.1 concrete 40 70
44.7 ×2.4 ×3.1 metal 60 80
9.9 ×8.7 ×3.2 metal 40 45
12.9 ×8.9 ×4.0 wood 15 25
11.3 ×7.3 ×3.1 concrete 25 30
TABLE 2
Approximate TRMS for indoor scenarios [10]. TRMS <TRMS,x%
for x%of the time.
WIRELESS CHANNEL CHARACTERISTICS MUST
BE WELL UNDERSTOOD TO DETERMINE THE
APPROPRIATE SYSTEM DESIGN AT 60G. THIS
INCLUDES PATH LOSS, MATERIAL ATTENUATION,
MULTI-PATH EFFECTS, ANTENNAS AND SPATIAL
AND TEMPORAL CHANGES.
SEPTEMBER 2007 | IEEE VEHICULAR TECHNOLOGY MAGAZINE |||
43
44
||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | SEPTEMBER 2007
authors characterized spatial properties of angular
spread. It was discovered that the angular spread typical-
ly measures from 0.3 to 0.8 in indoor environments, sug-
gesting that space-time processing techniques are viable.
Outdoor scenarios, however, have a reduced angular
spread of 0.1 to 0.5. Reflections contribute to the majority
of received multipath and the material of the reflective
element (i.e., wall) is very critical in determining the
angular spread. These angular spread statistics suggest
that, while space-time processing techniques may not be
helpful in outdoor channels, the spatially rich nature of
indoor channels allows 60G systems to take advantage of
recent advances in array processing and multiple-input-
multiple-output processing (MIMO).
2.6 Time Variation in the Channel
The Doppler frequency is proportional to the transmitted
signal frequency and represents the maximum frequency
difference between received signals and transmitted
signals due to mobility of the transmitter, receiver, or
objects in the channel. Hence, increasing the carrier fre-
quency in wireless systems causes proportionally magni-
fied Doppler effects. It follows that the Doppler frequency
for 60G is 10 times that of a 6 GHz system under the same
mobility conditions. Since there is also a proportional
relationship between the Doppler effect and the time
varying nature of the wireless channel, channel informa-
tion must be estimated 10 times more frequently. The
transmitter uses this information in adapting the modula-
tion scheme or transmission mode for increased reliability
and throughput, while the receiver uses this information
in channel equalization. As a result, 60G systems can
expect increased overhead to obtain the channel at the
transmitter through feedback or at the receiver through
training/pilot symbols. This can be particularly troubling
for outdoor and inter-vehicle communications that must
operate under high mobility.
3. Circuit Design Issues
Initially, 60G transceivers were constructed using hetero-
junction transistors in non-standard technologies such as
GaAs and InGaAs due to their superior noise characteris-
tics and power handling capabilities at higher frequen-
cies. For the commercial market that will access the
unlicensed spectrum, such technologies are ill suited due
to limited digital integration, high relative cost, and low
power efficiency. As a result, recently, there is a focus on
silicon-based analog RF and digital circuits. Building a
60G transceiver presents unique challenges for amplifier
design and phase noise reduction, while the silicon tech-
nology scaling presents new considerations for analog
circuit design.
3.1 Amplifier Challenges
It is well known that as the frequency of operation for
transistors in an amplifier circuit increases, the available
effective gain per transistor decreases. Although the
severity of this trend depends on the technology
implemented, it can universally be stated that the design
of 60 GHz power and low noise amplifiers is a gain-limited
challenge. For low-noise amplifiers the limited gain caus-
es more significant noise contributions. Hence, receiver
noise-figure specifications are much more difficult to
accommodate at 60 GHz. For power amplifiers, transis-
tors need to increase in size to provide enough power to
drive radiating devices such as transmit antennas.
Increasing transistor size, however, also decreases the
gain such that the performance of the amplifier is com-
promised. Ideal designs have yet to surface in silicon-
based circuits, largely because the inherently low gain
characteristic of silicon transistors only exacerbates the
already difficult problem.
3.2 Transceiver Architecture Considerations
Standard wireless receivers typically use heterodyne
architectures, as in Figure 2(a), or homodyne architec-
tures, as in Figure 2(b). Heterodyne receivers, after a
coarse filtering operation, translate the modulated signal
to an intermediate frequency where a higher precision
channel-select filter (to remove out-of-band components)
is easier to design. The translation from a higher fre-
quency to a lower frequency is accomplished by mixing
the signal with a local oscillator and then low pass filter-
ing to remove imaged components. The homodyne
FIGURE 2
60 GHz Transceiver Architectures.
THE 60G WIRELESS CHANNEL SHOWS 20 TO 40
DB INCREASED FREE SPACE PATH LOSS, AND
SUFFERS FROM 15 TO 30 DB/KM ATMOSPHERIC
ABSORPTION. THIS, ALONG WITH REDUCED
MULTIPATH AFFECTING NON LINE-OF-SIGHT
COMMUNICATIONS, MEANS THE DESIGN OF 60G
SYSTEMS IS NOT STRAIGHTFORWARD.
receiver removes the intermediate frequency conversion
and converts the received modulated signal directly to
baseband. Direct conversion receivers are popular due to
integration ability in CMOS.
Heterodyne and homodyne receivers, when applied to
60G systems, experience aggravated phase noise effects.
Oscillator signals, like those used for mixing at the trans-
mitter and receiver, are typically derived from a refer-
ence oscillator. A multiplier circuit scales this reference
to the desired oscillator signal frequency. Reference
oscillator phase noise amounts to variability of the oscil-
lator’s signal frequency with time. The phase noise effect
is random and there currently exists no way to correct
the problem in traditional transceivers, aside from fre-
quency stabilization, which is largely unsuccessful [14].
It is possible to include better reference oscillators, but
this significantly increases the analog design cost.
Assuming the same reference oscillator, 60 GHz signals
suffer from phase noise that is 10 times greater when
compared to unlicensed wireless systems below 6 GHz.
This effect leads to degraded signal quality and has
proven to be a major limiting factor for the production of
low-cost 60 GHz radios [15].
3.3 Millimeter-wave Silicon Analog Circuits
Current bipolar transistor technology such as 120 nm
silicon-germanium BiCMOS is capable of 60G transceiver
design [16]. Bipolar technologies have superior gain and
noise properties compared to CMOS. Even with the
impressive performance and integration displayed by
bipolar silicon technology, however, the increased power
efficiency, the increased integration, and reduced cost of
a CMOS solution is attractive. Chew et al previewed
CMOS implementations of 60G radios [17]. At that time,
180 nm technology was on the verge of becoming 60 GHz
capable. 135 nm and 90 nm CMOS technology has subse-
quently surfaced (and soon 65 nm), encouraging research
into the design of 60G CMOS transceiver components for
evaluating the potential of 60 GHz CMOS radios [5]. Here
we highlight some of the issues with the design of 60G
CMOS radios:
■
In general, parasitic elements of tran-
sistors contribute to reduced high-fre-
quency performance. At 60 GHz,
where performance optimization
needs to be finely tuned, the role of
parasitics is magnified. In particular,
resistive parasitics due to gate resis-
tance necessitate finger widths be
optimized during layout to reduce this
parasitic effect [5].
■
Design rules for CMOS at higher fre-
quencies must incorporate microwave
techniques to account for traveling
wave delay since the operating wave-
length is on the order of circuit dimensions. As a
result, passive elements are often realized using trans-
mission line techniques.
■
It may occur that CMOS power amplifiers are unable to
accommodate the desired output power. This results
in three choices: a power combining circuit, an exter-
nal power amplifier, or using an antenna array to spa-
tially distribute the signal.
4. Digital Modulation at 60G
Modulation for 60G digital wireless communication must
consider all of the facets of 60 GHz wireless systems
including the channel characteristics, antenna configura-
tions, circuit limitations, and the nature of the data traffic.
Unfortunately, there is not and cannot be a clear choice
for the best modulation scheme at 60G. OFDM, constant
envelope modulation, and linear single carrier modula-
tion are potentially applicable since all offer advantages
to the 60G wireless system, but in different situations.
Like most designs, the modulation choice results from a
consideration of design tradeoffs as detailed in this sec-
tion and summarized in Table 3.
4.1 Orthogonal Frequency Division Multiplexing
Orthogonal frequency division multiplexing (OFDM) is a
digital modulation scheme that has nice equalization
properties. By representing the information bearing ele-
ments in the frequency domain (with an IFFT at the
transmitter and a cyclic prefix with FFT operation at the
receiver) the OFDM system is able to send information
over approximately frequency flat components of a fre-
Modulation
Scheme Advantages Disadvantages
CE/CPM Maximum power efficiency Poor spectral efficiency;
high complexity receiver
designs
OFDM High spectral efficiency; Sensitive to phase noise;
simple transmitter and poor power efficiency;
receiver designs; flexible added complexity due
frequency allocation to coding and IFFT
Linear SC Simple transmitter and Non-ideal spectral
receiver; performance efficiency; non-ideal
vs. efficiency compromise power efficiency
TABLE 3
Modulation Strategies Summary.
BUILDING A 60G TRANSCEIVER PRESENTS
UNIQUE CHALLENGES FOR AMPLIFIER DESIGN
AND PHASE NOISE REDUCTION, WHILE THE
SILICON TECHNOLOGY SCALING PRESENTS NEW
CONSIDERATIONS FOR ANALOG CIRCUIT DESIGN.
SEPTEMBER 2007 | IEEE VEHICULAR TECHNOLOGY MAGAZINE |||
45
46
||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | SEPTEMBER 2007
quency selective channel. For severely frequency selec-
tive channels (as with the large bandwidth channels for
high data rate applications in 60G) the total complexity
of OFDM to eliminate ISI is less than half that of single
carrier time-domain equalization [4]. OFDM also offers
high spectral efficiency when adapting the power and
signal constellation over each frequency flat component
of the frequency selective channel. Finally, by selectively
activating or nulling subcarriers, OFDM is easily adapted
to different bandwidths, making its application very
desirable.
OFDM, however, suffers from many drawbacks that
complicate its application at 60 GHz. The time domain
transmitted signal is observed to have large “peaks” when
compared to average power values over the OFDM sym-
bol (i.e., a large peak to average power ratio (PAPR)). To
maintain linearity, ordinary transmitters must limit out-
put power levels leading to less efficiency in terms of
power consumed (resulting in shorter battery life) and
power transmitted (resulting in shorter range). To enable
operation at a higher average power, more expensive
power amplifiers are employed, increasing hardware
costs in OFDM systems. OFDM systems are also severely
susceptible to frequency offsets culminating in higher
design costs for synchronization. Consequently, phase
noise will trouble OFDM implementations to a higher
degree. Finally, although OFDM enables simple equaliza-
tion, it will require significant error control coding to pro-
tect against frequency selective fades. While OFDM is
desirable for its equalization properties and spectral effi-
ciency, 60G OFDM transceivers are difficult to design
given the phase noise and amplifier limitations at 60 GHz.
Additionally, OFDM does not allow for simple
coding/decoding methods for reducing complexity in
large bandwidth 60G frequency selective channels.
4.2 Constant Envelope Modulation
Constant envelope modulation (CEM) transmits a signal
with the information-bearing element contained entirely
within its phase. CEM is ideal in terms of power efficiency
since its constant magnitude baseband signal does not
suffer from nonlinear distortion. CEM signals can operate
in the nonlinear saturation region of the power amplifier
at the transmitter. One specific and popular CEM imple-
mentation is continuous phase modulation (CPM) where
the phase is a continuous function of time, creating a
bandwidth efficient signal. CEM and CPM systems have
their own selection of drawbacks. From a capacity per-
spective, CEM and CPM systems have a lower achievable
throughput, especially for high SNR [18]. In terms of
implementation, optimal non-differential receiver and
equalization structures for CPM systems can be highly
complex since CPM signals are differentially encoded at
the transmitter. Frequency domain equalization for CPM
signals is successful in reducing receiver complexity.
Even with this reduced complexity, however, as the con-
stellation size grows, equalization quickly becomes com-
plex [19]. CEM systems that operate with differential or
symbol-by-symbol detectors are less complex, but will be
highly sensitive to phase noise effects and multipath.
Consequently, although CEM eases amplifier design at
60G by its linearity independence, the susceptibility of
higher-order CEM receivers to phase noise, the subopti-
mal spectral efficiency of CEM signals, and the complexity
of CPM receivers suggest that only low-order CEM (such
as PSK) or CPM techniques (such as MSK) will find any
application at 60G.
4.3 Linear Single Carrier Modulation
Single carrier (SC) systems implemented with linear
modulation (e.g., QAM and PAM constellations) offer a
good compromise for many of the qualities discussed in
this section. SC systems, while not necessarily constant
envelope, display better PAPR values than OFDM. With
ISI equalization, linear SC receivers do not have to
account for memory, making its complexity much less
than CPM. Various equalization algorithms can be imple-
mented, although the complexity of time-domain equal-
ization will exceed that of OFDM for severely frequency
selective channels. For severely frequency selective
channels (like the 60 GHz indoor channel), it is best
served to use frequency domain equalization (FDE) for
SC systems such that the complexity and performance is
comparable to OFDM. Finally, because linear SC systems
usually send information using the amplitude and phase
of the symbol, they are more robust to phase noise than
CEM constellations of the same order. The choice of lin-
ear modulated single carrier systems offers a good com-
promise for power efficiency, spectral efficiency, and
receiver complexity while at the same time displaying
phase noise tolerance that is desirable for 60G systems.
OFDM or CEM, however, may be preferred over linear SC
modulation if maximal spectral or power efficiency is
desired, respectively.
4.4 Space-Time Processing
Space-time processing is used in conjunction with the
modulation techniques discussed above to improve
performance [25]. The measured spatial properties of the
60 GHz indoor channel and the small dimensions of 60G
THERE IS NO CLEAR CHOICE FOR THE BEST
MODULATION SCHEME AT 60G. OFDM,
CONSTANT ENVELOPE MODULATION, AND
LINEAR SINGLE CARRIER MODULATION ALL
OFFER ADVANTAGES BUT IN DIFFERENT
SITUATIONS.
SEPTEMBER 2007 | IEEE VEHICULAR TECHNOLOGY MAGAZINE |||
47
antennas suggest that space-time processing techniques
are suitable for 60G indoor systems. Space-time process-
ing falls into two distinct categories: diversity and multi-
plexing. Diversity techniques, such as digital
beamforming (not to be confused with beam steering dis-
cussed later in the paper) and space-time coding, exploit
spatially selective fading to improve the received signal
energy. Spatial multiplexing takes advantage of the inde-
pendent nature of spatially uncorrelated channels to
increase effective throughput. An advanced space-time
processing solution could enable high spectral efficien-
cies for 60G systems as well as reduce the fading margin
in the link budget. Unfortunately, because 60G channels
provide unparalleled bandwidths, complex digital space-
time techniques on gigabit systems will likely have to wait
until digital technology scales accordingly. Simple diversi-
ty techniques such as delay diversity, antenna selection,
and transmit/receive combining, however, may be
applied in first generation 60G systems.
5. Potential Applications
Despite the availability of 60G bandwidth, it is currently
not utilized to a sizable measure of its full potential. While
60G systems have attracted mobile broadband and cellu-
lar systems with the potential of high data rates, the
range restrictions at 60G limit such an application. Conse-
quently, the interest in 60G for cellular data delivery has
waned. Fixed wireless access, wireless local area net-
works, wireless personal area networks, portable multi-
media streaming, and vehicular networks, however, are
all applications that will likely find application at 60 GHz
due to the bandwidth offered. To what extent and in what
capacity these applications are utilized will be deter-
mined by the ability of designers to overcome 60G chan-
nel and transceiver circuit limitations.
5.1 Mobile Broadband
The emergence of current EV-DO and UMTS networks and
next generation 3GPP-LTE, WiMax and iBurst networks
have provided a competitive market for mobile broad-
band data delivery. Mobile broadband services are
expected to provide data outdoors and indoors, reliably,
to users at distances up to several kilometers. 60G sys-
tems, however, cannot accommodate this range except
for line-of-sight (LOS) scenarios with highly directive
antennas due to losses attributed to the high carrier fre-
quency, atmospheric, and precipitative effects. This was
observed by IEEE 802.16d, a precursor to WiMax, which
considered mobile broadband delivery up to 66 GHz, but
due to the short range observed, its sequel (IEEE 802.16e,
of which WiMax is based) focused on operating frequen-
cies to below 11 GHz. Although the large bandwidth
afforded 60G systems is desirable for mobile broadband,
the range restrictions do not allow 60G to compete with
lower frequency systems.
5.2 Fixed Wireless Access
60G systems exist for supporting FWA installations of
gigabit data delivery along an LOS path. Transmission
ranges on the order of a kilometer are possible if very
high-gain antennas are deployed. This application, while
viable, does not press the commercial market and indi-
vidual proprietors are already providing solutions. Physi-
cal layer designs must include expensive transceivers and
microwave components to provide reliability and maxi-
mize range, resulting in bulky equipment. Furthermore, to
provide robust communications, modulation schemes
need to be simple and utilize low order constellations.
BridgeWave Communications, for example, provides 60G
wireless back-haul products with 0.6◦beamwidth anten-
nas that deliver 1.25 Gbps, reliably, using BFSK modula-
tion up to 2.38 km. High data rate FWA products will
continue to find 60G very attractive except for scenarios
where LOS operation cannot be accommodated or when
longer range is needed.
5.3 Wireless Local Area Networks
Presently, wireless local area networks (WLANs), which
largely carry computer network data and Internet traffic,
have been the most popular wireless application of unli-
censed spectrum. Next generation devices that operate
under the IEEE 802.11n standard have the ability to oper-
ate at hundreds of megabits per second. The key feature
of 60G technology for WLANs is the capability to provide
gigabits per second of throughput, i.e. gigabit Ethernet.
Due to the increased attenuation of 60G signals and the
decreased propagation through materials, it will take con-
siderable effort to develop 60G WLANs that communicate
over NLOS links. The lack of 60G communication between
rooms indoors will necessitate 60G repeaters for typical
WLAN installations. Some vendors have proposed hybrid
2.5/5/60 GHz WLAN solutions that use lower frequencies
for normal operation and 60G when there exists a short-
range LOS path for high-speed operation. Therefore,
while 60G will likely find applications in WLAN at some
point in the future, it is not expected to immediately drive
60G technology to the consumer market due to range and
NLOS propagation restrictions. Future WLAN designers,
as in the Very High Throughput study group of IEEE
802.11 [28], still consider millimeter-wave and 60G WLAN
promising due to the large unlicensed bandwidth offered.
5.4 Wireless Personal Area Networks
Increasing the data-rate of legacy wireless personal area
networks (WPANs) such as Bluetooth provides the main
driving force for 60G solutions in the commercial arena.
The IEEE 802.15.3c international standard will provide reg-
ulation for the short-range data networks (≈10 m) over
60G wireless channels. Although this standard is still in
the development phase the task group has set out to pro-
vide, in its first phase, 2 Gbps mandatory and 3 Gbps
48
||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | SEPTEMBER 2007
optional wireless data transfers to support cable replace-
ment of USB, IEEE 1394, gigabit Ethernet, and multimedia
delivery. In the United States, ultrawideband (UWB) has
the ability to provide the hundreds of megabits per
second needed for next generation PANs in frequencies
ranging from 3–10 GHz [26]. Unfortunately UWB communi-
cation suffers from two major problems. The first problem
with UWB communication is that UWB spectrum is not
commensurate worldwide. Europe, for example, only pro-
vides a fraction of the available UWB spectrum in the Unit-
ed States. Thus, while UWB is a viable technology for
WPANs it does not provide the worldwide market that
made WLANs so successful. Second, UWB does not supply
high enough data rates to support cable replacement for
some next generation applications. UWB under the WiMe-
dia multi-band OFDM standard only supports 480 Mb/s at
its highest rate. Given the restrictions placed on UWB in
terms of bandwidth and power this is likely within a factor
of two of the upper limit. In order to achieve multi-gigabit-
per-second wireless communication, it seems that 60 GHz
is the only foreseeable solution in the near future.
5.5 Multimedia Streaming
Cable replacement for high rate multimedia streams
such as the high-definition multimedia interface (HDMI)
provides an important potential application of 60G wire-
less technology. Current wireless HDMI products radiate
in the 2.5 GHz and 5 GHz unlicensed spectrum where
bandwidth is limited. As a result these systems imple-
ment either lossy or lossless compression, which signifi-
cantly adds component and design cost, digital
processing complexity, as well as product size. Off the
shelf, retail wireless HDMI products currently available
exceed 100 US dollars for a bulky set-top box. If 60G
transceivers are able to achieve in excess of 5 Gbps it
will enable cost-effective wireless HDMI solutions with
tight integration to support even the most demanding
1080p HDTV streams.
5.6 Vehicular Applications
60G vehicular applications are partitioned into three class-
es: Intra-vehicle wireless networks, inter-vehicle wireless
networks, and vehicular radar. Intra-vehicle networks are a
subset of WPANs that exist entirely within a vehicle. Of this
subset, broadband communication within an automobile
or aircraft [20] has garnered interest for removing wired
connections of vehicular devices (e.g., wires between con-
sole dashboard DVD player and backseat displays) as well
as multimedia connectivity of portable devices inside the
vehicle (e.g., MP3 players, laptops, etc.). The 60 GHz carri-
er frequency is especially attractive for intra-vehicle com-
munications due to its containment within the vehicle and
decreased ability to penetrate and interfere with other
vehicular networks. Inter-vehicle networks are significantly
different than intra-vehicle networks mainly because they
operate in an outdoor propagation environment. Automo-
bile communication for delivery of traffic information and
range extension of mobile broadband networks has also
received significant interest [21]. Unfortunately, realization
of inter-vehicle communication at 60G is more challenging
than other 60G applications due to the high Doppler, which
increases overhead for maintaining links, and low range,
which limits the distance between automobiles of a con-
nected network. Therefore, while the large amount of
bandwidth available is attractive for inter-vehicle net-
works, it is unlikely to be an early application of 60G tech-
nology. The last class of vehicular networks, vehicular
radar [22], have already been deployed at millimeter-wave
frequencies albeit not within the aforementioned unli-
censed 60G spectrum. Nevertheless, 60G has attracted
interest for adaptive cruise control in cars as well as auto-
motive localization [23]. 60G radar is particularly effective
at resolving small objects due to the millimeter wavelength
of 60G signals.
6. IEEE 802.15.3c Standard
The IEEE 802.15.3c standard, when completed, will pro-
vide the first widespread international physical layer
framework to support consumer 60G WPANs. In
September of 2007 Task Group 3c in the IEEE 802.15
working group narrowed down the selection of its 60G
physical layer into two proposals [27]. Many contribut-
ing bodies including Matsushita, Philips, Korea Universi-
ty, the Electronics and Telecommunications Research
Institute (ETRI), the Georgia Electronic Design Center
(GEDC), Decawave, and Astrin Radio provided support
for Proposal #1, spearheaded by the National Institute of
Information and Communications (NICT) in Japan. Pro-
posal #2, led by Tensorcom, includes contributions from
France Telecom, Innovations for High Performance (IHP)
Microelectronics, LG Electronics, Matsushita, Sony,
Intel, Sibeam, and Samsung Electronics. These two pro-
posals, while containing notable differences, still offer a
clear view of the yet-to-be-finalized physical layer stan-
dard. This section provides a general summary of the
two proposals.
6.1 Spectrum Occupancy
60G unlicensed spectrum, as noted in Figure 1, is incon-
sistent internationally. To encompass all available unli-
THE 60 GHZ CARRIER FREQUENCY IS ESPECIALLY
ATTRACTIVE FOR INTRA-VEHICLE COMMUNICATIONS
DUE TO ITS CONTAINMENT WITHIN THE VEHICLE
AND DECREASED ABILITY TO PENETRATE AND
INTERFERE WITH OTHER VEHICULAR NETWORKS.
censed frequencies both proposals divide nearly 9 GHz of
spectrum starting at 57.24 GHz and ending at 65.88 GHz
into four 2.16 GHz channels. The Nyquist bandwidth of
both proposals is 1.632 GHz, leaving roughly 250 MHz of
guard bandwidth on each side to prevent spectral leak-
age. Proposal #1 also allows for a half-rate channel, which
operates at the same center frequency, but with half the
Nyquist bandwidth.
6.2 Transmission Modes
Both proposals share three basic transmission modes:
Common Mode, Single Carrier (SC) Mode, and OFDM
Mode. The Common Mode is used for channel scanning
as well as providing a low rate communication mode.
Channel scanning consists of the transmission and recep-
tion of beacons to negotiate a link between two 60G
devices. Both the SC and OFDM Mode occur only after the
Common Mode operation is complete and a 60G wireless
link is established. The SC Mode and OFDM Mode are fur-
ther classified according to the throughput delivered:
Low Rate, Medium Rate, High Rate. Therefore, there are a
large number of potential modes, allowing providers to
consider the tradeoffs discussed in Section IV.
6.3 Modulation
The modulation for each mode varies between Proposal
#1 and Proposal #2, although not significantly. Rather
than providing an exhaustive list of the proposed modula-
tion in each transmission mode and each proposal, this
section will summarize the common modulation trends
that will likely represent the combined proposal. The
reader is encouraged to view the proposals available on
the Task Group 3c web site [27] for more detail. The mod-
ulation selection across both proposals is briefly summa-
rized in Table 4. Common Mode transmission utilizes
power efficient MSK for reliable, low complexity commu-
nication in the presence of channel impairments, such as
the intrinsically large path loss, and hardware impair-
ments, such as nonlinear power amplifiers. The Common
Mode also implements a low-complexity Golay code for
spreading the signal by a factor of 32 in Proposal #1 and
64 in Proposal #2. This spreading gain allows the
Common Mode to be tolerant to interference from other
networks as well as operate in low SNR environments.
For communication beyond the base rate of approxi-
mately 50 Mbps provided by the Common Mode, the SC
Mode is preferred due to its lower complexity, tolerance
to phase noise, and tolerance of nonlinear power ampli-
fiers. Forward error correction and spreading gain are
varied to provide quantized levels of physical layer
throughput depending on the channel and hardware
impairments observed. The mandatory rates required in
SC Mode will offer between 2–3 Gbps. OFDM Mode, which
utilizes a size 512 FFT, will likely be relegated to an
optional mode in the final standard. Although OFDM suf-
fers from sensitivity to phase noise and nonlinear power
amplifiers, the spectral efficiency and ISI immunity of
OFDM encouraged both proposals to include this mode.
Given the modes provided in the proposals future 60G
systems can achieve up to 6 Gbps throughput as trans-
ceiver design improves. Both OFDM and SC modes use a
unified frame format, meaning that FDE is supported for
both SC Mode and OFDM Mode.
7. Developing Technologies
The future of 60G systems will be shaped by emerging
technologies to enable advanced techniques in the wire-
less transceiver. In this section we discuss two examples
of emerging technologies that will have a large impact on
the progression of 60G wireless design.
7.1 Steerable Beam Antennas
As previously discussed, high-gain antennas may be
implemented to overcome fading margins in the link bud-
get. As a result of the inherent “pointing” problems asso-
ciated with these high gain antennas, many designers
have proposed steerable beam antennas [24]. Such anten-
na configurations discover the best direction to point
through a searching algorithm and then either mechani-
cally alter the configuration of the antenna (as in micro
electrical mechanical systems (MEMS)), or use several
antenna elements to construct a phased array with opti-
mized radiation patterns. While such configurations have
been demonstrated, it is yet to be observed in a practical
commercial transceiver with a small form-factor required
for WPAN or WLAN devices. If such technologies become
practical, it will help enable 60G systems that do not rely
on LOS links.
7.2 Self-Heterodyne Transceivers
To bypass the phase noise problem it was observed that
by transmitting the local oscillator signal with the modu-
lated signal, mixing could be accomplished by multiplying
the received local oscillator signal and the received mod-
ulated signal. Since they both suffer from the same phase
Transmission Mandatory Optional
Mode Components Components
Common CPM; large N/A
spreading gain
SC: Low Rate CPM; large CPM or CEM;
spreading gain variable
spreading gain
SC: Med Rate N/A QPSK or 8-QAM
SC: High Rate QPSK 8-PSK, 8-QAM,
or 16-QAM
OFDM: Low Rate N/A BPSK or QPSK
OFDM: Med Rate N/A QPSK or 16-QAM
OFDM: High Rate N/A 16-QAM or
64-QAM
TABLE 4
IEEE 802.15.3c Modulation Preview.
SEPTEMBER 2007 | IEEE VEHICULAR TECHNOLOGY MAGAZINE |||
49
50
||| IEEE VEHICULAR TECHNOLOGY MAGAZINE | SEPTEMBER 2007
noise effects, the noise is canceled out. This design
(Figure 2(c)) is referred to as a self-heterodyne receiver.
This technique allows for cheap reference oscillators
with poor phase noise performance. Unfortunately, this
design has four major disadvantages. First, the local oscil-
lator that is transmitted takes up bandwidth. Second, the
local oscillator and the modulated signal have to be sepa-
rated in the frequency domain by at least the bandwidth
of the modulated signal. Third, the local oscillator that is
transmitted is very sensitive to interference and multi-
path effects. As a result, there is a different source of sig-
nal degradation corresponding to mixing with a distorted
local oscillator. Finally, and perhaps most importantly,
the local oscillator signal received over-the-air is much
weaker than local oscillator signals supplied on typical
heterodyne receivers. Hence, RF self-product terms
become much more significant in the mixer output of a
self-heterodyne receiver, causing increased received sig-
nal distortion. Despite these drawbacks, self-heterodyne
architectures have received interest [14], especially for
robustness to phase noise.
8. Conclusion
The large quantity of unlicensed bandwidth internation-
ally available at 60 GHz provides system designers with
a multitude of options for wireless communications
application. Specific challenges for 60G design include
heavy large-scale attenuation due to the increased
carrier frequency, substantial atmospheric absorption
for communication over large distances, increased
phase noise for transceivers, and limited gain ampli-
fiers. The application-space for 60G systems must con-
sider these challenges. While mobile broadband
systems may not be feasible, there are still promising
applications for WLANs, WPANs, FWA, vehicular net-
works, and multimedia delivery to take advantage of the
large bandwidth over short distances. The modulation
choice offered, as displayed in the IEEE 802.15.3c pro-
posals, will depend on a tradeoff between digital com-
plexity, power efficiency, phase noise sensitivity, and
spectral efficiency. The best implementation will depend
on which tradeoff best suits the design given the band-
width offered, the antenna parameters, and transceiver
characteristics. Developing technologies such as steer-
able antennas and self-heterodyne transceivers could
help shift these tradeoffs.
References
[1]S.K. Yong and C.-C. Chong, “An overview of multigigabit wireless through
millimeter wave technology: Potentials and technical challenges,” EURASIP
Journal on Wireless Communications and Networking, vol. 2007, 2007.
[2] M. Chelouche and A. Plattner, “Mobile broadband systems (MBS): Trends and
impact on 60 GHz band MMIC development,” Electronics & Communication Engi-
neering Journal, vol. 5, no. 3, pp. 187–197, 1993.
[3] F. Giannetti, M. Luise, and R. Reggiannini, “Mobile and personal communications
in 60 GHz band: A survey,” Wirelesss Personal Comunications, vol. 10, pp. 207–243,
1999.
[4] P. Smulders, “Exploiting the 60 GHz band for local wireless multimedia access:
Prospects and future directions,” IEEE Communications Magazine, vol. 40, no. 1,
pp. 140–147, 2002.
[5] C. Doan, S. Emami, D. Sobel, A. Niknejad, and R. Brodersen, “Design considera-
tions for 60 GHz CMOS radios,” IEEE Communications Magazine, vol. 42, no. 12,
pp. 132–140, 2004.
[6] M. Karkkainen, M. Varonen, P. Kangaslahti, and K. Halonen, “Integrated amplifi-
er circuits for 60 GHz broadband telecommunication,” Analog Integrated Circuits
and Signal Processing, vol. 42, pp. 37–46, 2005.
[7] N. Guo, R.C. Qiu, S.S. Mo, and K. Takahashi, “60 GHz millimeter-wave radio: Prin-
ciple, technology, and new results,” EURASIP Journal on Wireless Communica-
tions and Networking, vol. 2007, 2007.
[8] D. Rogers, “Propagation considerations for satellite broadcasting at frequencies
above 10 GHz,” IEEE J. Select. Areas Commun., vol. 3, no. 1, pp. 100–110, 1985.
[9] C. Anderson and T. Rappaport, “In-building wideband partition loss measure-
ments at 2.5 and 60 GHz,” IEEE Trans. Wireless Commun., vol. 3, no. 3, pp. 922–928,
2004.
[10]P. Smulders and A. Wagemans, “Wideband indoor radio propagation measure-
ments at 58 GHz,” Electronics Letters, vol. 28, no. 13, pp. 1270–1272, 1992.
[11]T. Manabe, Y. Miura, and T. Ihara, “Effects of antenna directivity and polariza-
tion on indoor multipath propagation characteristics at 60 GHz,” IEEE J. Select.
Areas Commun., vol. 14, no. 3, pp. 441–448, 1996.
[12]T. Manabe, K. Sato, H. Masuzawa, K. Taira, T. Ihara, Y. Kasashima, and K.
Yamaki, “Polarization dependence of multipath propagation and high-speed
transmission characteristics of indoor millimeter-wave channel at 60 GHz,” IEEE
Trans. Veh. Technol., vol. 44, no. 2, pp. 268–274, 1995.
[13]H. Xu, V. Kukshya, and T. Rappaport, “Spatial and temporal characteristics of
60-GHz indoor channels,” IEEE J. Select. Areas Commun., vol. 20, no. 3, pp.
620–630, 2002.
[14] Y. Shoji, K. Hamaguchi, and H. Ogawa, “Millimeter-wave remote self-heterodyne
system for extremely stable and low-cost broad-band signal transmission,” IEEE
Trans. Microwave Theory and Techniques, vol. 50, no. 6, pp. 1458–1468, 2002.
[15]D. Cabric, M.S.W. Chen, D.A. Sobel, S. Wang, J. Yang, and R.W. Brodersen,
“Novel radio architectures for UWB, 60 GHz, and cognitive wireless sys-
tems,” EURASIP Journal on Wireless Communications and Networking,
vol. 2006, 2006.
[16]S. Reynolds, B. Floyd, U. Pfeiffer, T. Beukema, J. Grzyb, C. Haymes, B. Gaucher,
and M. Soyuer, “A silicon 60 GHz receiver and transmitter chipset for broad-
band communications,” IEEE J. Solid-State Circuits, vol. 31, no. 12, pp. 2820–2830,
2006.
[17]K. Chew, S.-F. Chu, and C. Leung, “Driving CMOS into the wireless communica-
tions arena with technology scaling,” in IEEE Conference on Custom Integrated
Circuits, San Diego, CA, 2001, pp. 571–574.
[18]K. Yu and A. Goldsmith, “Linear models and capacity bounds for continuous
phase modulation,” in proceedings of IEEE International Conference on Communi-
cations, vol. 2, 2002, pp. 722–726.
[19]J. Tan and G. Stuber, “Frequency-domain equalization for continuous phase
modulation,” IEEE Trans. Wireless Commun., vol. 4, no. 5, pp. 2479–2490, 2005.
[20]M. Peter, W. Keusgen, A. Kortke, and M. Schirrmacher, “Measurement and
analysis of the 60 GHz in-vehicular broadband radio channel,” in proceedings of
IEEE Vehicular Technology Conference, Sept., 2007.
[21]H. Bischl and W. Schäfer, “The 60 GHz mobile-to-mobile radio channel—fading
statistics and estimated packet error rates,” in proceedings of IEEE Vehicular
Technology Conference, June, 1994.
[22]J. Ruoskanen, P. Eskelinen, and H. Heikkila, “Millimeter wave radar with clutter
measurements,” IEEE AES Magazine, vol. 18, pp. 19–23, 2003.
[23]H.-J. Fischer, “Digital Beacon vehicle communications at 61 GHz for interactive
dynamic traffic management,” 8th International Conference on Automotive Elec-
tronics, London, 1991.
[24]K. Chang, M. Li, T.-Y. Yun, and C.T. Rodenbeck, “Novel low-cost beam-steering
techniques,” IEEE Trans. Ant. Prop., vol. 50, no. 5, pp. 618–627, 2002.
[25]R.W. Heath, Jr. et al. “Advanced wireless: Space-time communication,” Univer-
sity of Texas Course Text, 2007.
[26]J. del Prado Pavon, N. Sai Shankar, V. Gaddam, K. Challapali, and C.-T. Chou,
“The MBOA-WiMedia specification for ultra wideband distributed networks,”
IEEE Communications Magazine, vol. 44, no. 6, pp. 128–134, 2006.
[27]“IEEE 802.15 WPAN Millimeter Wave Alternative PHY Task Group 3c,” [Online]:
http://www.ieee802.org/15/pub/TG3c.html, Sept., 2007.
[28]“Status of Project IEEE 802.11 VHT Study Group,” [Online]:
http://grouper.ieee.org/groups/802/11/Reports/vht_update.htm, Sept., 2007.
