Antennas are most often used to increase the range of WLAN (wireless LAN)
systems, but proper antenna selection can also enhance the security of your
WLAN. A properly chosen and positioned
antenna can reduce the signal leaking out of
your workspace, and make interception extremely difficult.
In this article, I analyze the signal of different
antenna designs, and how the positioning of the user's antenna
makes a difference in signal reception.
The 2.4-GHz ISM Band
Wireless networking uses radio frequencies originally set aside for
unlicensed "Industrial Scientific and Medical" (ISM) use. There are
three of these bands, at 902-928 MHz, 2400-2483.5 MHz, and 5725-5850
MHz.
Unlike all other parts of the radio spectrum, you do not need a
license to operate a transmitter in the ISM bands. But you must be
prepared to accept interference from other users of the bands, and,
to prevent anarchy, you must obey Federal Communications Commission
rules governing the use of ISM spectrum.
The IEEE 802.11b specification sets up 11 channels within the 2.4-GHz
band, centered between 2.412 and 2.462 GHz. The wireless LAN hops
between these channels in a manner designed to reduce interference
and increase data integrity.
802.11b has a number of software security features built into its
protocols, but each one of these security features has now been
broken. If somebody wants to intercept the data travelling over your
WLAN, and if they can hear the radio signals from it, you cannot stop
them from listening. The best you can do is encrypt the data going to
your WLAN using secure high-level software protocols, like SSH, which
are often inconvenient to implement and regarded by many WLAN users
as an unnecessary inconvenience.
Good antenna deployment on a WLAN can reduce stray RF radiation,
making your signal up to 100 times lower outside of the work area, and
much harder to surreptitiously intercept. A good antenna will also
make your WLAN less susceptible to stray signals from other WLANs,
telephones, and microwave ovens, which all use the same 2.4-GHz ISM
spectrum.
The Ubiquitous Dipole Antenna
The most common WLAN antenna is the Dipole antenna. Simple to design,
it is standard equipment on most access points.
This is a D-Link DI-714 802.11b Wireless Router, DSL Firewall, and
Bridge. It is fitted with two (removable) dipole antennas. The dipole
in the foreground has had its protective black plastic cover removed,
so that you can see its construction (magnified view here).
The dipole has a (white) radiating element just one inch long. This
performs an equivalent function to the "rabbit ears" antennas on
television sets. It is much smaller because the WLAN frequencies are
in the 2,400-MHz microwave spectrum instead of the 100-MHz TV
spectrum. As the frequency gets higher, the wavelength, and the
antennas, become smaller.
I have used the Numerical Electromagnetic Code (NEC) Finite-Element
Antenna Simulator software from the Lawrence Livermore Laboratories
to calculate the theoretical radiation pattern of this dipole antenna.
(A radiation pattern is a diagram that allows us to visualize in what
directions the energy will radiate from an antenna.) The NEC
software, which was originally written in Fortran, is available for Linux,
Windows and MacOS.
You can see that the dipole radiant energy is concentrated into a
region that looks like a donut, with the dipole vertically through
the "hole" of the "donut." If an antenna radiates in all directions
equally we say it is an "isotropic radiator." All practical antennas
concentrate their energy into some region of the isotropic sphere.
I guess somebody should have designed a better way of visualizing 3-D
radiation patterns, but it is often tough to get our minds around
them. Consequently, we usually split the 3D donut (the red lines)
into two perpendicular planes, called Azimuth and Elevation. We can
then visualize these polar plots, rotate them conceptually in our
mind, and allegedly form a 3D overview more easily. Azimuth and
Elevation patterns are supplied for every commercial antenna, and
once you have a grasp of how the polar plots combine (in your mind)
to form the "red line" 3D pattern, it will become very easy to
choose an antenna that is optimum for your application.
 |
When the RF energy is concentrated, such as into the yellow region of
this cone, we say that the antenna has a "gain" over an isotropic
radiator. The gain is measured in Decibels, a logarithmic measure.
Gain over isotropic is written as "dBi." The isotropic gain is
roughly equal to the inverse ratio of the areas of the yellow
(coverage) region and the total surface area of the isotropic sphere.
The gain of a dipole is roughly 2.1 dBi.
The dipole radiates equally in all directions around its axis (the
Azimuth), but does not radiate along the length of the wire (above
and below). Hence the donut pattern.
If a dipole antenna is placed in the center of a single floor of a
multistory building, most of its energy will be radiated along the
length of that floor, with some significant fraction sent to the
floors above and below the access point.
We can reduce the amount of energy radiating to the floors above and
below our target work area by squashing the donut. This type of
antenna, termed a "Slotted Waveguide" or a "Co-linear," has a number
of dipoles, one above the other, radiating their signals in phase so
that the energy is concentrated along the axis of the dipole(s).
This is the (calculated) radiation pattern of a 6-dipole slotted
waveguide antenna that I built. Its gain is 8.6 dBi. Every 6-dB
increase in gain means that the range over which the signal
propagates has been doubled over that of a single dipole. This gives
a stronger signal for users within the main lobes of the radiation
pattern. At the same time the sidelobe energy going above and below
the main donut has been attenuated to 12 dBi, or 20 dB lower in
strength than the main signal.
This means that the energy radiated to floors above and below our
desired work area has been reduced just as if they were 10 times
further away from the access point than our users. It has become many
times harder for someone not in our primary coverage area to
intercept our WLAN signal and compromise its security.
Directional Antennas
The dipole is an "omnidirectional" antenna, because it radiates its
energy in all Azimuth directions equally. Directional antennas
concentrate their energy into a cone, known as a "beam."
This is the radiation pattern of a Biquad. An antenna such as this
could be placed in the corner of a work area, concentrating the
energy into the work area and radiating very little outside the
building (to the back and sides of the antenna).
The gain of this antenna is 11.3 dBi. I have only shown the Azimuth
plot, as the elevation is essentially identical. Notice that all
sidelobes are at least 20 dB lower in level, we normally say "20 dB
down," from the main lobe signal strength. In general, it is much
easier to get high gains from directional rather than from
omnidirectional antennas.
Horizontal and Vertical Polarization
The dipole transmits a vertically polarized signal. This means that
the electrical component of the energy, the so-called "E-field," is
parallel to the dipole element and perpendicular to the floor. By
turning the dipole 90 degrees (so its axis is horizontal) it will
radiate a horizontally polarized signal, where the E-field vector is
parallel to the ground.
In my experience, horizontally polarized antennas generally propagate
better within a building, probably due to reflections from the floor
and ceiling.
When the WLAN signal hits an object, such as a metal cabinet or pole,
it is reflected, and its polarization is scattered. Inside any work
area there will be a mixture of vertically and horizontally polarized
signals.
PCMCIA Cards Have Terrible Inbuilt Antennas
And this leads us nicely into the real world. The designers of the
antennas for PCMCIA cards face a real problem. It is not easy to form
antennas onto the small circuit board inside the bulbous plastic
cover that sticks of the end of the PCMCIA card. I won't go into the
technology here, but below is plotted a typical sensitivity
measurement for a laptop equipped with a PCMCIA WLAN card. The
effective gain of this antenna is low, less than 0 dBi (typically -4
dBi) and it is very directional.
You can see that the sensitivity varies greatly with Azimuth, and is
quite unlike a well-behaved antenna. If you look at the blue trace
you can see that the antenna is 6 dB more sensitive (twice the range)
for signals coming in at 165 degrees compared with those at 330
degrees. There are also deep nulls, from which directions no signal
at all is received.
The red trace shows sensitivity to vertically polarized signals. It
is significantly lower than for horizontal polarization. I always
recommend the use of horizontally polarized access point antennas
when a significant number of PCMCIA-equipped workstations will be in
use.
It should now be obvious why you have to jiggle and wiggle and rotate
your laptop to get a decent WLAN signal. There has to be a better
way...
Zoom-Air 4105
Zoom-Air has a
PCMCIA card it ships with a PCI adapter that contains a very handy
SMA-RP socket for an external antenna. Here is
a picture of it in my Vaio laptop. The card is equipped with a
standard (crummy) antenna, but you have the option of merely screwing
a dipole onto it if you are in a weak signal area. This is not an
ideal solution, but it is much handier than having to carry around
cables and other things that get lost in my laptop case. Another
advantage — they do sell spare antennas to replace the ones you
lose.
Compex WLU-11
The Compex WLU-
11 is a self-contained USB WLAN client. It does not have any
PCMCIA card internally (unlike most other USB clients). It has two
easily removed shielding panels, below which are the pins ready to
solder an SMA connector. This device offers one of the easiest ways
to get a working 802.11b Prism-II reference test bed. The dipole
antenna is easily desoldered, and this device is a WLAN hacker's
paradise. Oh — it also is a very nice WLAN client for those of us who
don't care to rip things apart.
Good Antennas and Bad Antennas
WLANs operate at a frequency of 2.4-2.4835 GHz. These are microwave
frequencies, and many antennas that work well at lower frequencies
are just not suitable for WLAN deployment. On the other hand, a 3-foot
piece of 4x2 inch aluminum rectangular tube with a few slots cut
in its sides makes a very high gain omnidirectional antenna.
Antennas for WLAN deployment must be chosen carefully. I tend to
favor simple antennas. There is less that can go wrong. There are
parabolic grid and reflector antennas from
Hyperlink, Andrews/Conifer,
Wincomm, and Telex.
In the omnidirectional antenna category there is a standout. Although
very expensive, the Andrews Waveguide
Omnidirectional offers high omnidirectional gain as well the
horizontal polarization that will easily penetrate cubicle walls.
Most of the manufacturers listed above offer vertical polarization
omnidirectionals, many at a much lower cost.Agere and Cisco
offer fully integrated systems, ideal for less demanding
deployments.
The SWR Specification
We have seen how the radiation patterns of an antenna are the most
important factor to consider when choosing an antenna. There is one
other important specification, the Standing Wave Ratio (SWR). This is
a measure of the amount of energy absorbed and radiated by an antenna
compared to the amount it reflects back to the transmitter. An SWR
value of 1:1 is perfect (no reflected energy), while a WLAN antenna
should have an SWR less than 1.5:1.
If its SWR is greater than 1.5 (1.5:1) over any region of the WLAN frequency
range, do not buy it. In addition to signal attenuation, the reflected energy
causes radiation from the coaxial cable and spurious sidelobes that can alert
a sniffer.
This is a plot of energy reflected by my ultra-high-gain dish antenna
as a function of frequency.
The graph above shows reflected energy from the 30-dBi gain parabolic
dish I built as a reference antenna. The sweep frequency range is
quite large, but the reflected energy between the two markers
designating the 2.4-GHz WLAN band is essentially zero, with an SWR
well under 1.25
The graph was produced with my HP 8690 Microwave Sweep Generator, a
directional coupler, an HP 18-GHz RF detector and my HP 54522 Digital
Scope. There is significant reflected energy at the 2402-MHz marker,
indicating that the antenna is not performing well on WLAN Channel 1.
In fact, ultra-high-gain antennas are not allowed to radiate on
channels 1 and 2 or 10 and 11, and I have computer optimized the
performance of the dish feed for channels 3 through 9. Throughout
this range it has virtually no reflected energy, a unity SWR, and has
sidelobes 40 dB down on its main beam.
That means that the sidelobes radiate less than 1/100th the distance
of the main beam, confining almost the entire signal within the path
to the target. The horizontal beamwidth is about 3 degrees.
Be fussy when buying antennas. If you shop around, it is possible to
find systems that generate very little spurious energy. Insist on
getting a full set of specifications and polar plots so that you can
properly evaluate them.
FCC Radiated Power Rules
The FCC foresaw that unlicensed 802.11b WLANs would be used to link
networks at distances of several miles (although I am sure they had
no idea of the ultimate potential of the open networks that are
springing up). Because 802.11b uses the 2.4-GHz ISM band,
transmitters and receivers have to coexist with each other,
accepting interference from Microwave Ovens and Portable Telephones
in addition to interference from other wireless LANs.
On balance, they decided to encourage the use of high-gain antennas
for point-to-point communication, as high-gain directional antennas
have tightly controlled beams that do not spray radiation over such
a large area.
The FCC had already decided to place a limit of +36 dBm (4 watts)
Effective Isotropic Radiated Power (EIRP) on Multi-Point WLAN links,
and a maximum power of +30 dBm (1 watt) at the WLAN tranmitter's
connector. So they also defined that if any antenna used in point-to-
point links has a gain higher than 6 dBi, the transmitter power must
be reduced so that the "peak output power of the intentional
radiator" is reduced by 1 dB for every 3 dB of antenna gain beyond 6
dBi.
This is a gift from the FCC. It allows point-to-point WLANs to
achieve an EIRP well in excess of +36 dBm, and the greater range that
results from the higher power. Here is a table of some typical values:
| Antenna Gain |
Max Antenna Input Power |
Attenuation |
EIRP(max) |
| +6 dBi |
+30 dBm |
0 dB |
36 dBm |
| +12 dBi |
+28 dBm |
2 dB |
40 dBm |
| +18 dBi |
+26 dBm |
4 dB |
44 dBm |
| +24 dBi |
+24 dBm |
6 dB |
48 dBm |
| +30 dBi |
+22 dBm |
8 dB |
52 dBm |
The output of unamplified Orinoco or Prism WLAN chipsets is less than
+22 dBm, typically about +18 dBm, or 60 milliwatts. If you do not
have an external power amplifier, or a higher power access point, you
can employ an antenna with up to 30 dBi of gain without fear of
retribution from the FCC. With a higher power transmitter you will
need to add coaxial attenuators (remember that the loss in your
coaxial cable can be included in this attenuation requirement).
Coaxial Cable
As you can see from the size of the dipole antenna, just one inch of
wire becomes an antenna at these microwave frequencies, and it is
critical that you use high-quality coaxial cable, with SMA, TNC, or N
series connectors. Thin coax, of the one-eighth-inch RG-174 variety,
has an attenuation (loss) of about 3 dB for every 4 feet of its
length. You lose half your signal in every 4 feet of cable. It is not
a very good idea to use this stuff at WLAN frequencies.
I personally use Times
Microwave LMR-195 coax for short runs (2-3 feet). It uses the
same SMA and N connectors as RG-58. For longer runs I use LMR-400,
similar in size to RG-8, which has a loss of 3 dB for every 40 feet
of cable length. Small quantities of these cables and connectors can
often
be found on eBay.
Diversity
Most access points have two antennas. One of these is used as the
primary transmitting and receiving port, while the other is
periodically checked (polled) to see if it is receiving a stronger
signal than the main antenna. This is called a "diversity" antenna
system. It can help to reduce variations in signal strength as you
vary the location of an access point and a client. While there is
nothing to stop you deploying two good antennas for each access
point, one good antenna is always superior to two ordinary ones.
Attenuators
Now that you have increased the signal strength in your work area way
beyond the needs of your users, the next step in security enhancement
is to attenuate the output from your access point. Attenuation of the
signal in your workspace will also attenuate the signal that leaks to
the outside world.
Fixed coaxial attenuators are available in a variety of sizes and
configurations. I suggest that you start with 3 dB attenuators, and
cascade them until you start to lose coverage within your workgroup.
Summary: Isn't It Fun?
I have always loved RF technology. As a kid, I have fond memories of
playing with surplus radio gear, and I built my first
transmitter before my first TV set, and long before my first
computer. Ah — those were the days...
Now, with the ready availability of numerical modeling, it is easy to
produce complex RF filters using PCB microstrips, and to design
antennas that you actually expect to work! This is a far cry from
the "cut-and-try" techniques of the past. And the technology is not
standing still. NEC was written in Fortran, for input from card decks
on (slow) mainframes. I was surprised to find that the 4000-element
NEC program that I am using on my PC is many times more capable than
the simulators used by the scientists at Lawrence Livermore only a
decade ago.
The tools may change, but the fundamental rules of electromagnetic
propagation and antenna theory do not. Put a signal in the air and
chances are that somebody will hear it. Our ancestors never dreamed
of desktop computers, let alone computers connected together through
the ether.
Wireless networking is an amazingly complex technology. Deployment of
a good WLAN need not be difficult, but it does require care and
planning. RF technology often seems esoteric, but nothing
beats the feeling of successfully sending a radio signal through that
ether...
Acknowledgments
A lot of people contributed to the preparation of this tutorial,
particularly David Jefferies, Paul Wade (W1GHZ), John Richey (Agere
Systems), and Darrel Emerson. Without their input it would have been
a lot shorter.
Trevor Marshall
is an engineering management consultant, with interests ranging from RF and Hardware
design to Linux internals, Internet infrastructure, MPEG, and Digital Video. He
started his career in the '70s, designing the Maplin Electronic Music Synthesizers.
When the Microcomputer came along, he got sidetracked into computer software,
programming the 2650, 8080, Z80, Z8000, 8048, 8306, 6805, 80x86, and Power PC
families. Along the way, he also picked up a little expertise in RF system design,
biomedical engineering, and the printing industry. His web site is www.trevormarshall.com.
