Content uploaded by Kaveh Pahlavan
Author content
All content in this area was uploaded by Kaveh Pahlavan on Oct 01, 2014
Content may be subject to copyright.
Trends in Local Wireless Networks
The authors’ vision
of
the future, in which a ubiquitous local
wireless computing environment leads to
a
fusion
of
communications and computation, must overcome significant
technical obstacles before becoming
a
reality.
Kaveh Pahlavan, Thomas
H.
Probed, and Mitchell
E.
Chase
KAVEHPAHLAVXNis the
director of the Center for
Wreless
Information Network
Studies
at the Worcester
Polytechnic Intstitute.
THOMS
H.
PROBERTis
the
president of the Enterprise
Computing Institute, Inc.
MITCHELL E. CHASE
is
a
research scientist with the
Enterprise Computing
Insitute, Inc.
1
hfbls
is
considered
by
IEEE
802
LAN
standards
as
the lowest data rate for
a
LAN.
n
June 1985 the lead author of this article
published an article entitled “Wireless
Office Information Networks” in this
magazine
[
11.
The article examined spread
spectrum, standard radio and infrared
(IR) technologies for intra-office wire-
less networking. This article was published in a
timely manner. In May of that same year the FCC
released the ISM (industrial, scientific, and medi-
cal) bands for spread spectrum local
communica-
tions. Although ISM bands are not restricted to
any specific application, wireless local area net-
works (LANs) were one of the most prominent
applications that were envisioned by the
rule
mak-
ers in the FCC [2,3]. Since 1985, many small startup
companies, as well as small groups in larger com-
panies, have started to develop wireless
LANs.
In
this article we provide a sequel to the previ-
ously mentioned article by providing an overview
of the past and present of the wireless
LAN
indus-
try,
as well as a perspective of the future directions
that encompass a vision for a ubiquitous local
wireless computing environment that leads to a
fusion of communications and computation.
Historical
Trends
n
the late 1970s IBM Laboratories in Rusch-
I
likon, Switzerland published the results of their
experimental work
on
the design of a wireless
indoor network using diffused IR technology that
was envisioned to be used
on
manufacturing
floors [4]. Around the same time, another experi-
ment at Hewlett-Packard (HP) laboratories [5]
examined the use
of
direct sequence spread spec-
trum technology for wireless inter-terminal com-
munications. The data rates experimented by both
methods were around
100
Kb/s
[
11
and none of
the projects were turned into a commercial prod-
uct. The diffused IR design could not provide a
reliable link to meet the project goal of
1
Mbh,
and the spread spectrum project had to wait for
the approval of a commercial band. However,
these works initiated further research in high-
speed wireless indoor communications. HP labo-
ratories and others developed directed-beam IR
networks [6,7]; Motorola’s Codex worked on a
wireless
LAN
using ordinary radio modems at
1.7
GHz and petitioned the FCC for that band [2];
and GTE Laboratories worked on a fiber optic
high-speed LAN with wireless drops
[8],
none of
which turned into a commercial product.
Motivated by HP’s petition, in 1981 the FCC
began exploring the feasibility of regulating a
band for the commercial application of spread spec-
trum technology that led to the adoption of the
ISMbandsreleasedinMay 1985 [2,3]. Development
of commercialwireless LANproducts entered a more
serious phase after the announcement of the ISM
bands. Motorola’s petition for a
1.7
GHz band
was not granted. However, it revealed very encour-
aging market evaluations for wireless LAN prod-
ucts. Later
on,
Motorola was able to secure licensed
bandsat 18 to 19 GHzforwirelessLAN applications
that pointed at the availability of bands at higher
fre-
quencies for wireless local communications. The
release of the ISM bands, results
of
the market
evaluations, and timely publication of some
papers
[l,
5,
9, 101 prompted a significant interest
in the industry for the design of numerous wire-
less LAN products, mostly operating in the ISM
bands. By 1990, wireless LAN products using
direct sequence spread spectrum (DSSS) in the ISM
bands [ll], licensed radio at 18-19 GHz [12], and
IR technology appeared in the market. These
productswere the first that could be called wireless
LANs because they were operating at high
speeds1 and they could communicate with at least
one standard LAN software product, such as
Novell. In that year, theIEEE802.11 committeewas
formed as an independent standards group with-
in IEEE 802 to follow up the work of IEEE
802.4L,which had started earlier
as
a branch
of
802.4.
In May 1991 the first IEEE wireless LAN work-
shopdealtwith
thisevolvingtechnology[13].
In 1992,
WINFORUM, an alliance among the major com-
puter and communication companies to obtain
bands from the FCC for the
so
called data-PCS,
was initiated by Apple Computer in the United States,
and the HIPERLAN standards activities in the
European Community (EC) were initiated under
ETSI direction. In 1993 the EC announced bands
at 5.2 and 17.1 GHz for HIPERLAN, and the
FCC announced plans for the release of unli-
censed PCS bands that can be used for wireless
local data communications. Both IEEE 802.11
and HIPERLAN are expecting to complete their
standards in 1995. WINFORUM is working with
the FCC to develop a “Spectrum Etiquette” for
the PCS bands [14].
88
0163-6804/95/$04.00 1995
0
IEEE
IEEE
Communications Magazine March
1995
-
~~
-
Trends in Applications
uring the development of the first-generation
D
wireless LANs at the end of the last decade,
it was thought that the savings from the installa-
tion andrelocationcostsofwired LANsjustified the
additional cost of wireless equipment. It was
expected that a market close to a billion dollars
would evolve around workstations adopting wire-
less LANs.
In
reality, old buildings were already
wired and the cost of wiring in new construction
was
so
low that it would be done during construc-
tion. Meanwhile, the twisted pair LAN technolo-
gy dominated the more expensive coaxial cable
LANs, which reduced the installation costs signif-
icantly. As a result, the first-generation wireless
LANs
did not closely meet the predicted market.
The first generation of wireless LANs were
designed to operate with workstations and had an
electronic power consumption
of
approximately
20
Watts. These devices are not suitable for bat-
tery operated portable computers. The next-gen-
eration wireless LANs are developing around the
lap-top, palm-top, and pen-pad computers in a
PCMCIAcard operating with small batteries. Wire-
less
connection is the natural medium for the per-
sonal portable computing devices that are growing
in popularity. In this environment researchers are
thinking of new concepts such as ad hoc network-
ing, nomadic access and mobile computing that leads
to the fusion of the computer and communica-
tions in a ubiquitous computing environment.
Today, first-generation wireless LANs are
marketed mostly as LAN extension. LAN extension
refers to applications in which the coverage of a
wireline LAN is extended to areas with wiring
difficulties. Examples of buildings with wiring dif-
ficulties are buildings with large open areas such
as manufacturing floors, stock exchange halls, or
warehouses; historical buildings where drilling
holes for wiring is prohibited; and small offices
such as a branch of a real estate agency where
maintenance of wireline LANs is not economical-
ly attractive. Recently, wireless LAN manufactur-
ers were also marketing their products aggressively
for connecting LANs located in two different
buildings. Here, the wireless LAN is a LAN inter-
connect device that is easy and inexpensive to install.
Figure
1
shows wireless LANs used for LAN
extension, cross-building interconnect, and nomadic
access. A nomadic access provides a wireless con-
nection between a hub and a personal portable com-
puter. This access
is
particularly useful for a person
returning to the office from traveling who needs
to transfer large data files between his personal
portable computer and the backbone information
network. Another situation related to the same users
is ad hoc networking in which a group of portable
users, for example
in
a classroom or a meeting, intend
to set up a network among themselves in an unpre-
dicted situation. Figure
2
shows an ad-hoc network
among several lap-tops. Figure
3
depicts a ubiqui-
tous computing environment in which a large dis-
tributed information and computing base is
available to the pen-pad users in a local area through
a high-speed wireless link. Typical applications
include a wireless campus or a wireless battlefield
inwhichauser(astudent oraso1dier)canmovewith-
in a local area with continuous access to the back-
bone distributed computational facilities.
W
Figure
1.
Applications for wireless
LANs.
W
Figure
2.
Ad
hoc
networking.
Trends in Frequency
A dm in istra tion
major difference between awireline and a radio
A
service is that the transmission medium for radio
communications is regulated and needs frequency
administration. All new services and products
must go through an exploratory phase to examine
various technologies against the available market.
In
a wireline environment, innovative ideas and new
technologies can be examined against the market
immediately.
To
examine the market for an inno-
vative idea or a new concept using radio technolo-
gy, one first needs to convince the frequency
administration organization.
On
one hand, wireless
LANs need a relatively large bandwidth for high-
speed transmission, and the current market needs
more time to grow to a reasonable size.
On
the
other hand, it is a new technology that is expected
IEEE
Communications Magazine March
1995
89
Trends in Spread Spectrum
Figure
3.
Fusion
of
computers and communications
into
a computational
cloud.
to bring revolutionary applications to the market.
This situation leaves the frequency administra-
tion agencies in a difficult position to justify the
assignment of adequate bands to this application
while other traditional voice-oriented services
may use the same bands to stimulate a larger mar-
ket. Two steps are taken to resolve this situation:
one is to resort to higher frequencies where larger
bandwidths are available; the other is to release
unlicensed bands. Raising the frequency of oper-
ation increases the size and power consumption
and limits the coverage of the signal in indoor
areas. The size and power consumption will reduce
in time as the electronic design technologies
advance. Limitations on coverage is not impor-
tant in many applications and can be solved easily
by reducing the size of the cells. In fact, in some
applications confinement of the signal to a small
area is a desirable feature for wireless LANs
[15].
An
unlicensed band with minimum regulation
provides a ground for exploration
of
different
wireless technologies against the market in a mul-
tivendor environment. This approach particularly
suits listen-before-talk random access methods com-
monly used in wireless
LANs.
Today, all activities
around wireless
LANs
are focused on unlicensed
operation. The major criticism of the unlicensed
bands is that they provide a self-defeating pur-
pose. In an unlicensed band a successful market
generates high utilization by devices from differ-
ent manufacturers possibly using different tech-
nologies. In the contention access methodsused in
data communication applications, high utilization
causes instability in the system
[15].
Any sort of
traffic control to stabilize the system requires a
hub base network which is an extremely chal-
lenging task in a multivendor environment. This
situation pushes the trend toward integration with
other cellular networks.
pread spectrum wireless LANs have been
S
more successful in the market than any other
wireless
LAN
technology, and most firms involved
in wireless
LANs
are using this technology. These
spread spectrum wireless LANs are developed in
the ISM bands that were the first unlicensed
bands in which a high-speed wireless LAN could
beimplemented. These bands allow unlicensed trans-
missions of up to
1
W
of spread spectrum signal
with a minimal spreading factor specified for direct
sequence and frequency hopping. Operatingin these
bands is secondary to other users that already exist-
ed in these bands, such as microwave ovens at 2.4
GHz or amateur radios at
900
MHz. These bands
can be used for any application, but wireless
LANs were one of the primary applications con-
sidered in the rule making [2,3].
At the time of rule making it was thought that
spread spectrum technology had two interesting
features for wireless
LAN
applications: It allowed
different systems in a multivendor environment
to operate simultaneously using CDMA, and its
anti-multipath nature provided a reliable trans-
mission at high data rates. It was thought that
other access methods such as TDMA or FDMA
require explicit cooperation among those sharing
the channel. This is not practical in a multivendor
environment. Therefore, CDMA was thought to
be the only choice [2,3].
Shortly after the announcement
of
the
ISM
bands, the efficiency of CDMA for wireless
PBX
operation was shown
[lo].
However, it was point-
ed out that to avoid the near-far problem, power con-
trol is essential for proper operation in a CDMA
environment
[IO,
161. Power control, however,
requires communication among all terminals and
a central unit and it cannot be implemented
in
a
multivendor environment unless all manufactur-
ers agree to certain principles that were not envi-
sioned in the rule making for ISM bands. In a
single-vendor environment, itwas noticed thatwire-
less LAN users need the entire capacity of the
channel for quick transmission of data bursts gen-
erated by each terminal, and CDMA is not suit-
able for
thispurpose[ll].Asaresult,
the firstwireless
LANs
in ISM bands that appeared in the market used
spread spectrum without CDMA.
Usingspreadspectrumwithout
CDMA, it can
be
argued that the maximum supportable data rate
(bandwidth) is sacrificed to gain transmission
reliability.
But
data rate is the most important
technical feature appealing to a wireless LAN
user. The transmission reliability
of
spread spectrum
is due
to
the anti-multipath and anti-interference
nature of this technology. Other techniques such,
as decision-feedback equalization (DFE), multi-
carrier transmission, and sectored antennas
[
151,
are
also anti-multipath, but do not sacrifice band-
width. These could be adopted as well.
So
far as
the resistance to interference is considered, the
low spreading factors used in the
ISM
band
wireless
LANs
are insufficient to providevery much
resistance. Another related important issue is the
reliability of the delivered packets of data.
In
a
wireless LAN, the delivery is checked with feed-
back acknowledgment in the communication soft-
ware; this reduces theneed for an extremelyreliable
transmission media. Due to the channel fading
90
IEEE
Communications Magazine March
1995
Data rate (Mbis)
-
Mobility
Range (ft.)
Detectability
Wavelength/frequency
Modulation technique
____~-
Radiated power
Access method
1-3
__I_--
1-4
10
5-10
2-20
Stationary/ Stationary Stationaryimobile Mobile
mobile with
LOS
-..
50-200
80 40-
1
30
100-800
100-300
Little
~~
Negligible Some Little
.__I_.__-.
~
E,=800
-
900
nm
18
GHz
or
ISM ISM bands
GFSK
-
__-
FS/QPSK
QPSK
25
mW
ilW
..
.
...-
OOK
___
I
Table
1.
Comparison
of
wireless
LAN
technologies
condition or interference, lost packets are simply
retransmitted. Therefore, the reliability of trans-
mission is not as necessary as it is in the case of
real-time voice transmission in which there is no
acknowledgment mechanism.
From the above discussion we conclude that
the most practical reason for using spread spec-
trum in the ISM bands for multivendor wireless
LAN applications is the availability of this band
to host a high-speed, unlicensed data link. The
main problem with this technique is the reduction
in the maximum supportable data rate for a given
bandwidth.
To
compensate for the loss in the
maximum data rate in the ISM bands, some com-
panies have adopted multiamplitude and multiphase
modulation. Other companies have adopted sin-
gle channel CDMA, in which each transmitter
uses several orthogonal codes simultaneously in
the same channels. Since all the codes are modulated
over the same carrier, the received power for
each code is the same. This is equivalent to hav-
ing perfect power control in place. Using this method,
some companies have been able to achieve data rates
on the order
of
20
Mbls
in
an ISM band. The
problem with this approach is the complexity of
the design of the receiver and the transmitter.
The fact that without power control and in a
multivendor environment spread spectrum is not
appealing for wireless LAN applications does not
imply that spread spectrum is not suitable for
wireless data communications. Indeed, in a multi-
media environment where various information
sources have different requirements, CDMA is a
promising technique. However, an efficient com-
munication in that environment requires power
control, which restricts the independence of the
designs in a multivendor environment. Although the
current trends in spread spectrum wireless LANs
in the ISM bands have their own problems, spread
spectrum technology will play an important role
in the future multimedia wireless communication
industry.
Trends in Products
ireless LANs are designed for a small
W
number
of
users, usually operating in indoor
areas. The range of coverage is small, which
leaves many options open for the transmission
Reservation
ALOHA,
CSMA
CSMA
technology. Technologies used in the existing
wireless LAN products are divided into five cate-
gories: diffused IR (DFIR), directed beam IR
(DBIR), standard radio (RF), direct sequence spread
spectrum
(DSSS),
and frequency hopping spread
spectrum (FHSS). In each category several products
with different specifications are available in the
market. Table
1
shows a comparison among vari-
ous
features of the existing products in each of
these categories. These technologies have evolved
around the availability of the channel and the
suitability of the transmission technique to pro-
vide a high data rate link in the wireless media.
The IRproducts aredesigned tooperate in theopti-
cal frequencies that are not regulated by the
FCC. The spread spectrum LANs are designed to
operate in the ISM bands. The RFproducts are either
implemented in the
18-19
GHz
licensed bands or
in the ISM bands using very low power. The
ISM
bands allow non-spread spectrum transmission
devices if the power is very low.
Data rate is an essential ingredient of local
communication networks and the most important
aspect for marketing and sale of the product. The
higher the data rate, the more likely the impact of
the product in the market. Another important
feature affecting the market forwireless LANs is the
mobility of the terminal, which is a function of
the power consumption and the size of the prod-
uct, except for DBIR, in which the terminal must
stay stationary to keep the radiation pattern
of
the device effective. If the technology can be
implemented with small batteries and light weight,
it would be suitable for mobile applications and
can be used for personal portable computers. Power
consumption is a function of the electronic imple-
mentation of the device, and some technologies
can be implemented either with or without bat-
tery operation for mobile and stationary applica-
tions, respectively. Most wireless LANs use aversion
of
CSMA, and some use reservation slotted
ALOHA or token ring.
Current Trends in Standards
lthough none of the standards for wireless
A
LANs are completed, there are numerous
wireless LAN products on the market. Wireless
LANs are stand-alone products that can be man-
-
Spread
spectrum
technology
willplay
an
important
role
in
the
fiture
multimedia
wireless
com-
munication
industry.
IEEE
Communications Magazine March
1995
91
-
The future
of
this industry
is toward
interopera
bil-
ity; however;
a discovery
period,
in
which the
technology
and the
market
settles itselj
is
being
planned.
ufactured without a widely accepted standard. As
the penetration of wireless LANs in the market
grows and the standards are completed, this situ-
ation will change.
Currently,
all
standard activities forwireless
LANs
use unlicensed bands, and there are
two
approach-
es to regulate an unlicensed band. One approach
is to develop a standard that will allow different ven-
dors to communicate with one another using a set
of interoperable rules. This approach is taken by
IEEE 802.11 and ETSI’s RES10, HIPERLAN.
The second approach is to provide
a
minimum
set of rules or “Spectrum Etiquette” [14] that
allow terminals designed by different vendors to use
a fair share of the available channel frequency-
time resources and coexist in the same band. The
second approach does not preclude the first
approach and it is pursued by the WINFORUM.
In a coexisting environment a vendor can inter-
operate with another vendor by using the same
protocol and transmission scheme. The future of this
industry is toward interoperability. However, a
discovery period in which the technology and the
market settles itself is being planned.
An
analogy
existed in the development of voiceband modems:
at the beginning there was no standard, and as
the industry evolved CCITT adopted successful
modems as the standard [17].
The three major activities related to wireless
LANs are IEEE 802.11, HIPERLAN, and WIN-
FORUM. IEEE 802.11 is developing a standard
for DSSS, FHSS, and DFIR technologies and uses
the ISM bands as the radio channel. The HIPER-
LAN standard is concerned with the recently
released 5.2 and 17.1 GHz bands in the EC. It is
expected that HIPERLAN will adopt non-spread
spectrum modulation techniques. To achieve high
data rates in multipath fading, HIPERLAN may
resort to techniques such as adaptive equalization
[18-201, sectored antenna [21] or multicarrier
modulation [22-241. The WINFORUM goal is to
obtain parts of the PCS band for unlicensed data
and voice applications and develop a “spectrum
etiquette” for them.
IEEE
802.1
1
and
ISM
Bands
IEEE802.11 focuseson the physical andmedia access
protocol (MAC) layers for peer-to-peer and cen-
tralized topologies accommodating DFIR, DSSS
and FHSS. Both spread spectrum systems oper-
ate in the 2400-2483.5 MHz ISM band. This band
is
selected over the 902-928 MHz and 5725-5850
MHz ISM bands because it is widely available in
most leading countries. Figure 4 shows the map
of four major geographic areas with their position
toward the ISM bands at 2.4-2.5 GHz. In this
band, more than
80
MHz of bandwidth is avail-
able that is suited to high-speed data communica-
tion. The implementation in this band is also
more cost effective as compared with implemen-
tation in frequencies that are a few GHz higher.
IEEE 802.11 supports DSSS with BPSK and
QPSK modulation for data rates of
1
and 2 Mbh,
respectively; FHSS with GFSK modulation and
twohoppingpatternswithdataratesof
1
and2Mbls;
and DFIR with
OOK
modulation with a data rate
of
1
Mbls. For
DSSS
the band is divided into five
overlapping 26 MHz sub-bands centered at 2412,
2442,2470,2427 and 2457 MHz, with the last
two
overlapping the first three. This setup provides
five orders of frequency selectivity for the user,
which is very effective in improving the transmis-
sion reliability in the presence of interference or
severe frequency selective multipath fading. For
FHSS, the channel is divided into 79 subbands,
each with
1
MHz bandwidth, and three patterns
of 22 hops are left as options for the users. A
minimum hop rate
of
2.5
hops/s is assigned to
provide an opportunity for slow frequency hop-
ping in which each packet can be sent in one hop
and, if it is destroyed, the following packet can be
sent from another hop for which the channel con-
dition would be different. This approach will pro-
vide a very effective time-frequency diversity that
takes advantage of a retransmission scheme to
provide a robust transmission. The standard sup-
ports CSMA and intends to provide interoper-
ability among all users.
WINFORUM
and
PCS
Bands
An important issue that was not addressed in the
ISM bands was the enforcement of a time limit
for the air time of a terminal. Suppose that a
wireless
LAN
and another device operate
in
the ISM
bands near one another. Further assume that the
other device serves an application that constantly
radiates a signal in the band. The wireless LAN is
a data communication device and it communi-
cates with bursts of information. In this situation,
all information bursts generated from the wire-
less LAN will suffer from the interference caused
by the other device, while the wireless LAN only
produces bursts of interference to the device.
This is not fair to the wireless LAN because it is
not getting a fair share of the available resources.
A fair method to allow several devices to operate
in the same band is to restrict their frequency-
time share of the channel according to their
transmission power. This issue is addressed by
WINFORUM.
WINFORUM was initiated by Apple Com-
puter to form an alliance in the industry to obtain
frequency bands for the
so
called data-PCS.
Today, WINFORUM has approximately
40
mem-
bers from leading information technology compa-
nies, and its objective is to obtain and effectively
employ radio spectrum for unlicensed user-provided
voice and data personal communication services
referred to as User-PCS. Currently, WINFORUM
intends to foster technical advances in applications
such as wireless
LANs
and wireless PBX services.
Originally, WINFORUM was looking for a
40
MHz
unlicensed band for the data-PCS, and the FCC
has shown some indications to provide them with
a 20 MHz band divided into two 10 MHz sepa-
rate sub-bands for voice and data applications.
The technical innovation initiated by the WIN-
FORUM is the development of the so-called
“spectrum etiquette” that is a means to provide
fair access to an unlicensed band to widely differ-
ent applications and devices [14]. The etiquette
does not intend to preclude any common air
interface standards or access technologies. Spec-
trum etiquette demands listen-before-talk (LBT),
which means a device may not transmit if the
spectrum it will occupy is already in use within its
range. The power is limited to keep the range
short. That allows operation in densely populated
office areas. The power and connection time
is
related to the occupied bandwidth to equalize the
IEEE
Communications Magazine March
1995
-
-__
Unit
fd
Kingdom
MPT
1349
2
445-2 475 GHz
FHSS
or
DSSS
Peak
ElRP
36
dBm
(will
be
supcrwded
by
European allocation)
CEPT
T/R10-01
2 400-2
500
GHz
FHSS
or
DSSS
Peak ElRP 20
dBm
United States
FCC
Part
15.247
2.400-2.4835
GHz
FHSSor
DSSS
Peak
ElRP
36
dBm
Japan
FHSS or
DSSS
Peak
power density
10
mW/Mhz
2.400-2.500 GHZ
tlRP
Equivalent
isotropically
radiated power
i
II_
-
figure
47
Worldwide wireless
LAN
spectrum allocation in the
2400-2500
MHz
band.
Trends in Research
interference and provide a fair access to frequen-
cy-time resources. In May 1993 WINFORUM
filed its spectrum etiquette with the FCC.
In theview of WINFORUM, there are two class-
es of information type generated from the asyn-
chronous and the isochronous transmissions. The
asynchronous transmission, typified by wireless-
LAN-like applications is bursty, begins transmission
within milliseconds, uses short bursts that contain
large amounts of data, and releases the link
quickly. The isochronous transmission, typified by
voice services such as wireless PBX, exhibits long
holding times, periodic transmissions, and flexi-
ble link access times that may be extended up to a
second. The asynchronous sub-bands may range from
50
KHz
to
10
MHz, while the isochronous sub-bands
may be divided into 1.25 MHz segments. The two
types are technically contrasting and cannot share
the same spectrum.
HIPERLAN and
5.2
and
17.1
GHz Bands
Europeans are approaching wireless LAN devel-
opment from a different angle. They intend to
establish a standard first and then develop the
products based on the standard. ETSI has asked the
Sub-Technical Committee RES10 to develop a
standard
for
High Performance Radio Local Area
Networks (HIPERLAN). The committee has
secured two bands at5.12-5.30
GHz
and 17.1-17.3
GHz
for the development of the HIPERLAN to oper-
ateat aminimumuseful
bitrateof20Mb/sforpoint-
to-point data communications with a range of
50
m.
They expect that at this rate and range they can pro-
vide 500-1000 Mb/s for a standard building floor
of approximately 100 meters square that is com-
parablewithFDDI[25].
RESlOischarteredtodefine
a radio transmission technique that includes type
of modulation, coding, and channel access, as
well as the specific protocols. The first meeting of
the HIPERLAN took placed in December 1991.
They secured the bands in 1992-93 and they expect
to receive
ETC
approval this year.
esearch today is directed toward enabling the
R
average person to access vast compute power
and an enormous amount of information through
a “cloud” comprised of a synergistic fusion of
communications and distributed computers. The
student, nurse, business person, or firefighter
dealing with a hazardous materials incident will
be able to tap into an enormous quantity of infor-
mation and computing power assembled to bene-
fit all.
From the Past to the Present
In theview of many, computers and communications
are perceived as complementary technologies:
related, but different, and not integrated. This is
borne out by the focus of research during the past
20 years.
An
earlier direction of U.S. government spend-
ing for computer research was directed toward
supercomputers applied to complex problems,
the so-called “grand challenges.’’ The object was
to apply massive computing power to solve com-
plex tasks. Wireless networks were not a factor in
these programs. Over the past decade, the super-
computer market has been eroded by the emergence
of cost effective high-performance desktop work-
stations and personal computers [26]. This is par-
tially responsible for an expansion in research
direction to the “national challenges,” which includes
wireless networks.
Yielding to the demands of businesses, health-
care, digital libraries, and other interests and
applications, the
U.S.
government has broadened
its focus from the “grand challenges” to the
“national challenges.” The expanded goal now
includes funding to provide computational power to
the average person in a meaningful way through
the National Information Infrastructure (NII)
[27]. Support for the range of applications envisioned
for the NI1 requires a fusion
of
computers and
-
Research is
directed
toward
enabling the
average
person to
access an
enormous
amount
of
information
through a
fusion
of
communica-
tions and
distributed
computers.
IEEE
Communications Magazine March
1995
93
~~
-
In an
increasing
number
of
applications,
the network
is providing
computation
resources
through
the commu-
nications
path
way.
Funded
by
ARPA
under
contract
#DABT63-91-C-
0016.
networks with wireless access. The network can
no longer be thought of as simply a pathway for
the transmission of files from one computer to anoth-
er. The network must meld with the computer,
enabling a new paradigm referred to as a “cloud.”
Fusion
of
Computer and
Communications
In the simplest sense, networks provide the ability
for computers to communicate. However, the
integration of computers and networks can pro-
vide more than file transfers. In an increasing
number of applications, the network is providing
computation resources through the communications
pathway. One need only look at the burgeoning
Internet to see this trend. This coupling has made
the Internet more than a vehicle for electronic
mail. It has enabled information and provided
computational power independent of geographi-
cal location. This trend promises to fuel a huge
demand for widely distributed computers seamlessly
networked together.
The ultimate challenge is to exploit wasted
computing cycles and provide location-independent
access to the computational resources. If the
number of personal computers is
50
million and
each is capable of
2
MIPS, then the combined
computing power is
1OI4
MIPSwith, perhaps,
95
per-
cent of that idle. When one considers worksta-
tions, minicomputers and mainframes, the combined
computing power is immense. The difficulty is
how to put this incredible resource to productive
work and provide ubiquitous local wireless access to
the cloud anytime and from anywhere.
Impact
of
the
Cloud
To
illustrate how the fusion of computers and
wireless
LANs
will impact the future, we present
three applications under development or envisioned:
a
military application, the military concept applied
to the office or hospital, and a community emergency
services response.
Digitalization
of
the Battlefield
-
Command,
control, communications, and intelligence
(
C31)
are a difficult task in the hostile environment of a
battlefield. The field commander needs to know
the location and situation of the troops under his
command, and have all available intelligence as
well, in order
to
make the appropriate decisions. The
lives of the troops and the success of the mission
depend
on
the commander and his C3I.
If
the
commander along with all support and communi-
cations is at one site, the battalion is vulnerable.
To reduce the battalion’s exposure and assure the
integrity of the chain of command, the integrated
wireless digital battlefield has been proposed.
In the digitized battlefield, each soldier carries
a backpack computer connected through a wire-
less LAN to all others in the battalion. Through
the network, each soldier’s physical condition and
location can be monitored through the use of a
personal network
[28],
and instructions can be given
accurately and without delay. Visual data can be
relayed to the commander. More importantly, the
backpack computers form a distributed, fault-tol-
erant computer, survivable after the
loss
of one
or more nodes. The wireless
LAN
enables the unen-
cumbered movement of troops and machines, as well
as providing a communication infrastructure. The
commander, or his chain of command, can moni-
tor the health of the troops, monitor movements,
gather intelligence, communicate with neighboring
commanders or superiors, and give orders.
The wireless digital battlefield provides a sig-
nificant strategic advantage. The enemy is no longer
able to destroy the effectiveness of a battalion by
striking a central command site. The C31 function
is more difficult tocompromise. Command isspread
over a larger area, with each soldier a part
of
the
distributed C3I structure.
Jacquard
-
The Jacquard project2
[29]
fuses
computers and communications into a scalable,
real-time, distributed, fault-tolerant architecture.
Jacquard provides for both shared-memory and
message-passing distributed-processor paradigms
with wireless access. It is essentially a scalable
local cloud connectable to a global cloud. Pro-
cesses on the system can be run on processors
with available compute cycles, thus putting wast-
ed resources to use. If a machine fails, the pro-
cess can be automaticaIly restarted on another
machine under control of the operating system.
The user need not be aware of the intricacies of
this computational backbone, only that his mis-
sion-critical process will run on the available
cycles, thus increasing resource utilization. Consider
the battleship captain needing to compute missile
trajectories to launch a counter-attack after suf-
fering damage to the weapons control. The
Jacquard architecture will still allow the ship to
compute and launch a retaliatory strike by utiliz-
ing the remaining processors distributed through-
out the ship. The captain keeps informed of the
situation through his personal digital assistant
(PDA), connected to the ship’s Jacquard framework,
regardless of where he roams on the ship.
In the office or campus environment, an indi-
vidual is able to access the Jacquard cloud through
a PDA with ubiquitous wireless communications.
The PDA is of limited processing capability to
conserve batteypower, and has auser-friendlygraph-
ical interface
[30,31].
The businessperson, stu-
dent, or engineer can access vast quantities of
data, run simulations, utilize decision support
software, and teleconference through their PDA.
Beyond ad hoc networks, the PDA provides
mobility to the user by maintaining connectivity
throughout the local area. Discussions in a con-
ference room, hallway, or associate’s office can
be more productive by testing hypothesis, run-
ning simulations, searching databases, or calcu-
lating projections quickly, utilizing the heretofore
wasted compute cycles of traditional computers
through wireless
LANs.
Emergency Service Response
-
The local fire
department is called to respond to a hazardous
material spill
on
a busy highway in the communi-
ty. Speed is critical to mitigate the situation and
to insure citizen safety and protect property.
Upon arrival at a safe distance from the spill, the
highly-trained response team dons protective cloth-
ing and attaches a PDA with wireless LAN. The
incident commander and his assistants monitor
the vital signs of the response team and observe
the situation through video, audio, and data
transmitted over the wireless
LANs.
As the team
leader nears the spill, he observes the placard on
94
IEEE
Communications
Magazine March
1995
~-
-~
-~
__
__
the trunk and on the leaking drums.
The chemical identification is entered into the
PDA and, through the
cloud,
identification, health
and fire hazards, recommendation for evacuation,
and containment and neutralization procedures are
relayed to the team. Additionally, the mixture of dif-
ferent chemicals is relayed to a network of computers
that quickly analyzes the resultant chemical com-
pound, which may not be contained in any reference
material, and suggests a course of action.
This represents avast improvement over the cur-
rent method of limited voice-only communica-
tions. Similar technologies can be effectively applied
to emergency medical services, law enforcement,
highway workers, and
so
on.
Technical Challenges
his vision of the future, the integration of com-
T
puters and communications into one entity or
cloud,
must overcome significant technical obsta-
cles before becoming a reality. The communications
technologist is faced with providing a ubiquitous
wireless
LAN
for connectivity anywhere, anytime.
The computer technologist is faced with the problem
of integrating the computers into a cohesive dis-
tributed network through the operating system
and software applications.
The communications problems to be overcome
include limited bandwidth, latency, dropout, lim-
ited available power, and secure communications.
In distributed computing applications, latency is
one of the biggest problems.
The computer component of the
cloud
has its
hurdles to overcome. The speed of the processor
is not the limiting factor. Density, power require-
ment, performance, and cost of memory are lim-
iting factors. Display technology is another area
where improvement must be made. Resolution,
power, size, and cost are areas for improvement
in PDA display technology.
The system issue of resource management must
be considered and may be intractable. Imagine
allocating resources on
10
different computers or
processors to run
48
different applications
or
pro-
cesses. There are on the order of
1048
different
allocation assignments or possibilities, each with
its own system-wide performance implications.
The complexities of efficient, fair, and equitable
allocation will be a challenge for the future.
The fusion of computers and communications
into a unified entity may appear a subtle distinc-
tion from the computers and networks popular
today. However, the impact of the fusion along
with wireless
LANs
will be profound. It will affect
everything from the way we do business to the
way we live. It will be the harbinger of an infor-
mation wave.
References
111
K.
Pahlavan, "Wireless office information networks," IEEE Com-
mun. Mag., vol. 23, no.
6,
June 1985, pp. 19-27.
121 M.
J.
Marcus, "Regulatory policy considerations for radio local area
networks," IEEE Commun. Mag., vol. 25, no. 7, July 1987, pp. 95-99.
[31 M. J. Marcus, "Regulatory policy considerations for radio local area
networks," Proc. IEEE Workshop on Wireless LANs, Worcester,
MA, May 1991, pp. 42-48.
141
F.
R.
Gfeller and U. Bapst, "Wireless in-house data communication via
diffuwinfraredradiation,"Pnx
/E€€,,vd.
67, Nov. 1979,pp. 1474-1486.
[SI P. Ferert, "Application of spread spectrum radio to wireless ter-
minal communications," Proc. IEEE NTC'80. Houston, TX, 1980,
pp. 244-248.
161
C.
S.
Yen and
R.
D. Crawford, "The use of directed beams in wire
less computer communications," Proc.
IEEE
GLOBECOM
'85,
Dec.
1985. pp. 1181-1184.
I71 Y. Nakata et al.. "In-house wireless communication systems using
infrared radiations," Proc. Int. Conf. Comp. Commun., Sydney
Australia, 1984.
pp.
333-338.
[81
R.
Mednick, "Office information network: an integrated LAN,"
Proc. IEEE GLOBECOM
'85,
New Orleans, W. 1985, pp. 15.2.1-5.
[91 T.
S.
Chu and M.
J.
Cans, "High
speed
infrared
local
wireless commu-
nication,"
I€€€
Commun. Mag., vol. 25, no.
7,
July 1987, pp. 4-1
0.
[lo]
B.
Ramamurthi and M. Kavehrad, "Direct-sequence spread spec-
trum with DPSK modulation and diversity for indoor wireless
communications,"
IEEE
Trans. Commun., vol. COM- 35, Feb.
11 11
B.
Tuch "An ISM band spread spectrum local area network Wave
LAN," Proc.
IEEE
Workshop on Wireless LANs, Worcester, MA,
May 1991, pp. 103-111.
[12lT.
Freeburg,"Anewtechnologyforhighspeedwirelesslocalarea
net-
works," Proc.
IEEE
Workshop
on
Wireless LANs, Worcester, MA,
May
1991,
pp. 127-139.
1131 IEEE Workshop
on
Wireless IANs, Worcester Polytechnic Institute,
Worcester, MA. May 1991.
I141 D. G. Steer, "Coexistence and access etiquette in the United
States unlicensed PCS band,"
/€€E
Personal Commun., vol. 1,
no.
4,49 1994.
11 51 K. Pahlavan and A. H. Levesque, "Wireless data communications,"
Proc.
I€€€,
vol. 82. no.
9,
Sept. 1994, pp. 1338-1430.
[161 K.
Pahlavan,"Spreadspectrumforwirelesslocalnetworks."Proc.
IEEE
PCCC,
Feb. 1987.
I171
K.
Pahlavan and
J.
L. Holsinger, "Voice-band data communica-
tions, a historical review: 1919-1988," Invited paper,
IEEE
Com-
mun. Mag., Jan. 1988
I181
K.
Pahlavan,
S.
Howard, and T. Sexton, "Adaptive equalization of
indoor radio channel,"
/E€€
Trans. Commun.. vol. 41, Jan. 1993,
pp. 164-1 70.
1191
S.
W.
Wales, "Modulation and equalization techniques for HIPER-
IAN," Proc. of IEEE PIMRC'94, Sept. 1994, pp. 959-963.
I201 A. Nix et a/., "Modulation and equalization considerations for
high performance radio LANs (HIPERIAN)," Proc. of IEEE PIMRC'94,
Sept. 1994, pp. 964-968.
[211 G. Yang and
K.
Pahlavan, "Sector antenna and DFE modems for
high speed indoor radio communications."
lEEE
Trans. on Vehic.
Techno/., Nov. 1994.
[221 G. Yang and
K.
Pahlavan. "Performance analysis of multicarrier
systems in office environment using
3D
ray tracing," Proc.
IEEE
GLOBECOM, Dec. 1994.
I231 M. Aldinger, "Multicarrier COFDM scheme
in
high bit rate radio
local area net- works," Proc. of IEEE PIMRC'94, Sept. 1994, pp.
969-973.
I241 G. Yang and
K.
Pahlavan, "Performance analysis of multicarrier
modems in an office environment using 3D ray tracing." Jan.
1994.
(In
review.)
I251
8.
Bourin. "HIPERLAN markets and applications standardization
issues," Proc. of IEEE PIMRC'94, Sept. 1994, pp. 863-868.
1261 'The reinvention of supercomputing: Silicon Graphicsand thechang-
ing technology," International Data Corporation White Paper,
1994.
[271 "High performance computing and communications: Technology
for the National Information Infrastructure," Committee on Infor-
mation and Communication, National Science and Technology
Council, 1995.
I281 P. P. Carvey, "8odyLAN: A wireless body local area network,"
Wireless, Adaptive and Mobile Information Systems Program
Summary, Advanced Research Projects Agency, 1994.
[291
R.
P. LaRowe Jr., et. a/., "Jacquard: An architecture for shared
memory networks of workstations," Enterprise Computing Insti-
tute Technical Report, Nov. 1994 (submitted for publication).
I301A.SmailagicandD. P.Siewiorek,"TheCMU mobi1ecomputers:Anew
generationof computersytems,"F'roc. IEEECOMPCON 94, Feb. 1994.
1311
S.
Sheng, A. Chandrakasan, and R. W. Brodersen, "A portable
multimediaterminal,"/€EECommun.
Mag., vol. 30,
no.
12, Dec. 1992.
1987. pp. 224-236.
pp.
64-75.
-
The system
issue
of
resource
management
may be
intractable.
Imagine
allocating
resources
on
10
different
computers
or
processors
to
run
48
different
applications
or
processes.
Biographies
~VEH
PAHL~VAN
is
the Westin Hadden Professor of Electrical and Com-
puter Engineering and the director of the Center for Wireless Informa-
tion Network Studies at the Worcester Polytechnic Institute, Worcester,
Massachusetts. His recent research has been focused
on
indoor radio
propagation modeling and analysis of the multiple access and trans-
mission methods for wireless local networks. His previous research
background
is
on
modulation, coding and adaptive signal processing
for digital communication over voice-band and fading multipath radio
channels. He
is
a
senior member of the IEEE Communication Society.
THOMAS H.
PROBERT
is
the founding President of the Enterprise Comput-
ing Institute, Inc.
(ECI),
a private not-for-profit organization conduct-
ing a program of research, development, and education in advanced
computer and communications technology. He has been a principal
investigator
on
a number of ARPA-funded research projects. In addi-
tion, he has held academic appointments at numerous universities
and colleges. He holds M.S. and Ph.D. degrees in computer and infor-
mation science from the University of Massachusetts at Amherst.
MITCHELL E. CHNE
is
a research scientist with the Enterprise Computing
Institute Inc. (ECI), Hopkinton, Massachusetts. Prior to joining ECI, he
was involved with communication systems and network simulation at
the Center for High Performance Computing and Comdisco Systems.
He received a 8.E.
in
electrical engineering from City College
of
New
York, an M.S. in biomedical engineering from Iowa State University,
an M.B.A.fromNortheastern Univenity,andaPh.D. inelectricalandcom-
puter engineering from Worcester Polytechnic Institute.
IEEE
Communications Magazine March
1995
95
__~