|
Product
Focus
We recently
established a relationship with
Global Mesh Technologies and in
doing so, added their exciting
new product: CAMMS™, Communications
and Multimedia Management System.
CAMMS is communications software
which enables data, images, floorplans,
maps, multi-user whiteboard and VoIP
phone communication between wired
and wireless users in a mobile, deployed
scenario. It perfectly supports the “First
Responder” and
Public Safely sector.
About
the company and the product –
Global Mesh Technologies (GMT) develops
communications software and hardware
solutions that enable data sharing
in a variety of wireless and mobile
environments. CAMMS software
(Command Anywhere Media Management
System) enables users to collaborate
and securely share data, voice
and video.
When used
in conjunction with COTS mesh networking
hardware, CAMMS can
be used to establish and control
an ad hoc, self-forming, self-healing
mobile mesh network without the need
for servers or any fixed infrastructure.
CAMMS can also be used as
a stand-alone web-based communications
tool where Internet access is available.
CAMMS unique
features include: mesh formation,
IP camera discovery, interactive
whiteboarding, file sharing, TerraViewer
mapping, GPS-AVL, IM, video conferencing,
Internet sharing, Web operation,
VoIP phone connectivity and more.
All features are accessible via
a single, intuitive GUI ensuring
ease of use for non-technical users.
Please
contact Connect802 Sales at 925.552.0802
to learn more about CAMMS. We look
forward to hearing from you!
Essential Wi-Fi
Before 802.11n, each 802.11 physical
layer operated in a single frequency
band. 802.11b and 802.11g operate
exclusively in the 2.4 GHz ISM
band, while 802.11a operates only
in the 5 GHz UNII band. But 802.11n
is capable of operating in either
of those two bands, which leads
to the question: which frequency
band should you choose for your
802.11n network?
Backwards compatibility is the
primary reason to choose the
2.4 GHz frequency band. Only
the 2.4 GHz band can provide
connectivity to 802.11b/g
clients and 802.11n clients
at the same time, since b/g
clients don’t operate
in the 5 GHz frequency band. However,
the performance of the 802.11n clients
will be hampered when they are forced
to operate in the same frequency
band as 802.11b/g clients. Because
802.11b/g clients transmit at lower
data rates than 802.11n clients,
they take up much more airtime to
transmit the same amount of data
than an 802.11n client would. Imagine
a fast sports-car (802.11n) stopped
at a four-way intersection while
a parade (802.11b/g) creeps past.
When 802.11n devices operate on the
same channel as 802.11b/g devices,
the 802.11n devices must use so-called
protection mechanisms to avoid interfering
with the 802.11b/g devices. These
protection mechanisms add overhead
and decrease 802.11n performance.
If it is desired to support 802.11n
and 802.11b/g clients on the same
wireless network, the best performance
will probably be realized by purchasing
dual-radio access points and leaving
the 802.11b/g devices in the 2.4
GHz spectrum, while giving the 5
GHz spectrum to the 802.11n devices.
This will allow the 802.11n devices
to achieve their full performance
potential.
For raw performance, the 5 GHz
frequency band is the clear winner.
The 5 GHz band has much less
noise and interference than
the cluttered 2.4 GHz band,
meaning that the RF characteristics
of the channels will be conducive
to higher data rates, all
other things being equal.
The 5 GHz band also has more
non-overlapping channels
available--up to 24 channels,
vs. the 3 channels available
in the 2.4 GHz band with
802.11b/g. This means that
a much higher density of
access points can be supported
without increased contention
between access points who
are on the same channel.
A higher density of access
points results in higher maximum
user density or more bandwidth
per user. Because of the increased
number of available channels,
the 5 GHz spectrum is much
more compatible with 802.11n’s
40-MHz channels.
40-MHz
channels roughly double the throughput
of the WLAN compared to 20-MHz
channels, all other things being
equal. The maximum data rate of
an 802.11n network is 289 Mbps
with 20 MHz channels and 600 Mbps
with 40-MHz channels! Real-world
equipment is currently not capable
of achieving maximum 802.11n data
rates, so a more typical scenario
would be 144 Mbps with 20 MHz channels
vs. 300 Mbps with 40-MHz channels.
Clearly, 40-MHz channels are
desirable.
The problem is that 40-MHz channels
are twice as wide as 20-MHz channels.
This means that only a single
40-MHz channel is usable in the
2.4 GHz frequency band. More
40-MHz channels than that would
not fit in the band without overlapping.
Even worse, if a 40-MHz 802.11n
channel is used in the 2.4 GHz
frequency band, it covers enough
of the band that no 802.11b/g
devices can use any other channels
in the band. To put this in perspective,
imagine that you have an 802.11b/g
network with access points on
channels 1, 6, and 11. Now you
assign an 802.11n access point
to use 40-MHz channels. It must
use channel 6, which is in the
middle of the band, in order
to avoid spilling its signal
outside the upper and lower bounds
of the band. The 802.11n access
point will interfere with the
802.11b/g APs on channels 1 and
11. In other words, when an 802.11n
access point uses 40-MHz channels
in the 2.4 GHz band, only one
channel is available! This dramatically
limits the potential performance
of the network. In most cases,
this limitation is unacceptable,
and Connect802 does not recommend
using 40-MHz 802.11n channels
in the 2.4 GHz band. When 802.11n
devices use 20-MHz channels in
the 2.4 GHz band, the number
of available channels is the
same as with 802.11b/g, and this
is the recommended configuration.
Even though 20-MHz channels limit
the data rate of the individual
access points, they increase
the aggregate throughput of the
network as a whole because they
allow for more simultaneously-usable
channels.
Because the 5 GHz UNII band has
so many more available channels,
it is practical to use 40-MHz
802.11n channels there. Even
though each channel is twice
as wide, this only reduces the
number of available channels
from 24 down to 12, a more-than-acceptable
number for most installations.
Since 40-MHz channels approximately
double the throughput of the
network compared to 20-MHz channels,
it is recommended to use them
in the 5 GHz band.
In summary, our recommendations
for configuring basic 802.11n
options are as follows:
- Configure 802.11n access points
to use the 5
GHz UNII spectrum with 40-MHz
channels.
- If backwards compatibility
is desired,
use separate access points
or dual-mode 802.11n access
points to provide 802.11b/g
access in the 2.4 GHz band.
- If it is absolutely necessary,
802.11n
access points can be deployed in the
2.4 GHz band, with backwards compatibility
for 802.11b/g devices, but 802.11n
performance will be limited
in a mixed 802.11b/g/n
environment.
- 40-MHz channels should almost
never
be used in the 2.4 GHz band, unless
there is only a single access point
in the network, and no other access
points are active within range of
that access point—even
access points from other
networks.
Technology and Engineering
802.11n
increases the maximum possible
data rate for an 802.11 network
from 54 Mbps to 600 Mbps. In this
article, we’ll examine some of the ways
in which this dramatic performance
improvement is achieved. It is well
beyond the scope of this discussion
to detail the inner workings of the
802.11n standard. Nonetheless, a
reasonable working perspective can
be obtained if we’re willing
to “hand wave” (i.e.
accept without a detailed explanation)
over some of the underlying technology.
The basic behaviors and protocols
of 802.11n are identical to its 802.11b/g/a
predecessors. A client device listens
for beacon frames from an access
point to discover its presence. The
client can send probe requests to
confirm that an access point is configured
with a specific SSID (Service Set
IDentifer; the network name.) The
client associates and authenticates
to the access point.
There are three main areas in which
802.11n differs from 802.11b/g/a:
Internal Engineering Improvements
A number of subtle (but significant)
improvements have been made to the
core operation of the protocol and
the structure of transmitted data.
These improvements raise the data
rate from the 802.11g/a 54 Mbps to
roughly 75 Mbps for 802.11n. Moreover,
the actual TCP/IP throughput increased
from 50% to closer to 70%. This means
that while an 802.11g/a connection
at 54 Mbps would provide roughly
27 Mbps TCP/IP throughput, a 75 Mbps
802.11n connection will provide close
to 52 Mbps TCP/IP throughput.
40 MHz Channels
802.11b/g/a transmits data in 20
MHz wide channels. Some manufacturers
offer channel bonding options to
allow the use of two separate channels
for the transmission of a single
data stream. Channel bonding (sometimes
called turbo mode) is vendor-proprietary.
802.11n allows the use of a true
40 MHz wide channel (as opposed to
two separate, independent channels
used in parallel.) The data rate
in a 40 MHz channel is twice that
of a 20 MHz channel. This option
would allow the data rate for 802.11n
to rise from 75 Mbps to 150 Mbps.
Multiple-Input / Multiple-Output
(MIMO)
By an
ingenious combination of RF engineering
and mathematics 802.11n allows
the parallel transmission of
more than one bit stream in the
same frequency band through the
use of multiple transmit and multiple
receive radios and antennas in
the same access point. This is
called multiple-input/multiple
output or MIMO (pronounced “my-moe”.)
Up to four radios can be used (i.e.
with up to four separate antennas.)
A common configuration is to find
that the access point uses three
radios but the client device only
uses two. This is referred to as
a “2 X 3 MIMO system.” A
2 X 3 MIMO system is limited to
two separate bit streams. Environmental
characteristics determine whether
or not one or more bit streams
can actually be turned on. If two
streams are available then the
bit rate would rise from 150 Mbps
to 300 Mbps (based on a 40 MHz
channel.) Three streams would yield
450 Mbps and four streams would
result in the maximum 802.11n 600
Mbps data rate (with which the
TCP/IP throughput would be roughly
420 Mbps.) At the time of this
writing, most real-world equipment
is only capable of supporting up
to two streams. A few devices can
support three streams. No known
devices can currently support four
streams, although they will surely
be released in the future.
Getting to 75 Mbps: The Internal
Engineering Improvements
You’ve been lied to. We’ll,
maybe not an outright lie but a great
misrepresentation of reality nonetheless.
The lie is that you can get 600 Mbps
with 802.11n. You can’t. To
even suggest that you can actually
achieve connectivity rates of 300
Mbps in a practical, commercial environment
verges on falsehood. Perhaps someday
in the future these fantasy data
rates will enter the realm of practical
reality – but not today. What
you can reasonably expect from 802.11n
is something close to a consistent
75 Mbps data rate (with proper RF
signal coverage and an appropriately
low level of environmental noise)
with data rates up to 150 Mbps in
the typical best case. This initial
increase in speed, up from 802.11g/a
54 Mbps, and the reduction in overhead
yielding a 70% TCP/IP throughput
rate (up from 50%) is the result
of the following internal engineering
improvements.
- In both 802.11b/g/a and 802.11n,
each data frame begins with a special
sequence of bits called the preamble.
The preamble sequence has been optimized
in 802.11n
- Frames
are made up of bits and transmitted
bits are represented by electromagnetic
symbols. A guard interval of
time is inserted between successive
symbols so they’re clearly
differentiated. 802.11n allows
the use of a guard interval that
is half as long as 802.11a or 802.11g.
- A short delay must follow the
transmission of a frame so that
stations can recognize when no
one is transmitting and so the
start of the subsequent frame is
clearly defined. 802.11n includes
a mode whereby a shorter interframe
gap time can be used.
- Each data frame must be acknowledged
by the transmission of a special
ACK frame so the transmitting
station knows that the frame has
been successfully received. 802.11b/g/a
requires each individual frame
to receive its own ACK. 802.11n
provides for frame aggregation
whereby multiple frames can receive
a single ACK for an entire sequence.
- A number of mathematically-based
error correction mechanisms are
used to minimize the probability
of data corruption. The error correction
mechanisms in 802.11n have been
optimized to significantly reduce
inherent bit overhead.
- Data is transmitted using a set
of adjacent, narrowband sub-carriers
in a mechanism called Orthogonal
Frequency Division Multiplexing
(OFDM.) Some of the OFDM sub-carriers
must be dedicated to the process
of transmitting and receiving the
signal, leaving the rest available
for carrying data. 802.11n introduces
less overhead in the OFDM symbol
allowing it to carry more data.
- Guard bands are small frequency
gaps that are introduced to separate
adjacent signals and keep them
from interfering with each other.
Frames are preceded with a special
synchronizing and initializing
bit stream called the preamble.
802.11n uses narrower guard bands
and has optimized the preamble.
It’s these engineering improvements
(which are included in the current
802.11n Draft 2.0 document) that
are going to absolutely bring you
up from 54 Mbps. It’s important
to note that each improvement provides
a small increase in performance.
Also, not all improvements can be
put into practice in all environments.
For example, the reduced interframe
gap is only available when no legacy
802.11b/g/a devices are present.
A pure 802.11n environment is called
a Greenfield deployment, implying
that 802.11b/g/a is disallowed by
specific configuration.
Can You Use 40 MHz Channels?
Yes; and
you can mix 20 MHz and 40 MHz channels
in the same environment – but
watch out! First of all, forget 40
MHz channels in the 2.4 GHz frequency
band. Across the 802.11b/g 2.4 GHz
band there’s only enough bandwidth
for a single, non-overlapping 40
MHz channel. So, if you’re
in a residential environment with
only one access point you have a
chance to use the single 40 MHz channel
in the 2.4 GHz band. If you’re
in a corporate enterprise, school,
or other larger space you’ll
need to move to 5.8 GHz where you
can actually implement multiple 40
MHz channels.
Another
consideration is the fact that
if you mix 40 MHz and 20 MHz channels
you’re going to add
some overhead for protection. Protection
is the mechanism by which, prior
to transmitting, a station takes
steps to prevent other stations from
trying to transmit at the same time.
The desired transmission bandwidth
is “protected” against
conflicting transmissions. When some
users are operating with 20 MHz channels
then any 40 MHz transmitter must
undertake the protection process
on both of the adjacent 20 MHz channels
that it wants to use as a single
40 MHz channel. It’s a small
amount of additional overhead but
it’s there nonetheless.
Finally,
the default configuration for 802.11n
is to use 20 MHz channels. Unless
an IT department is in charge of
the wireless clients in the environment
there’s little hope of assuring
that everyone is configured to actually
use a 40 MHz channel.
Be absolutely sure that you quiz
your hardware equipment manufacturer
or vendor to find out the nuances
of 40 MHz channel use for a particular
brand of equipment.
If you
do implement 40 MHz channels then
you can expect your throughput
to roughly double – from 75
Mbps to 150 Mbps. Tests have confirmed
that this environment produces TCP/IP
throughput consistent with, and in
some cases slightly better than,
the 70% expectation. You’re
looking at between 105 Mbps and 125
Mbps.
The Potential for Multiple, Simultaneous
Bit Streams
Multiple-Input
/ Multiple Output (MIMO) is typically
the first thing you hear about
802.11n. It should be the last
thing you consider with regard
to throughput and capacity. MIMO
gives you the “icing on
the cake”, not the cake.
As previously
discussed, the multiple antennas
in a MIMO system create separate
spatial streams of bits between
a single transmit antenna and a
single receive antenna. Two streams
means twice the throughput, three
streams means three times the throughput,
and four streams means four times
the throughput – it’s
just that simple.
The exact
way that streams work is well beyond
the scope of this article but,
suffice it to say, it’s
like Arthur C. Clarke said in 1961, “Any
sufficiently advanced technology
is indistinguishable from magic.”
What you
need to know about streams is that
there is a very high probability
that they can’t be established
in a line-of-sight environment. That
means that if your client wireless
device is in sight of the access
point you won’t get multiple
streams. This flies in the face of
conventional Wi-Fi design where we
put an access point inside the conference
room to support 20 people sitting
around the conference table. With
the access point in the conference
room the people in the room will
not be able to establish spatial
streams to the access point 12 feet
away on the ceiling.
Spatial
streams need multipath reflections
in order to work. That means that
you’ll have a much higher probability
of establishing multiple spatial
streams to an access point in the
next room, or far away in a large
auditorium. It’s really counterintuitive – but
it’s true.
The bottom
line is that while you can expect
a performance increase to roughly
75 Mbps from 802.11n’s
basic engineering improvements, and
you have the option of doubling the
throughput using a 40 MHz channel,
you have no way to guarantee that
a user in a particular location will
absolutely be able to establish multiple
spatial streams. If you can get multiple
streams then you’ve gotten
the “icing on the cake” – but
the “cake” caps out at
roughly 75 Mbps.
Remember,
too that with 20 MHz channels,
even if you can establish two spatial
streams your actual TCP/IP throughput
is going to be between 105 and 125
Mbps – not 150 Mbps.
A preliminary
802.11n design should be based
on RF predictive CAD modeling and
simulation to identify the most
suitable access point locations based
on RF signal coverage. Only by comparing
the preliminary model to the environment
during on-site throughput and multipath
analysis can any assessment be made
regarding the possibility of achieving
multiple spatial channels. In no
case will 100% of an 802.11n access
point’s coverage area be able
to receive multiple spatial channels;
it’s always a subset area.
Ask the Expert
We've
run into an issue at one of our
facilities in La Courneuve, France.
At this building, we have 5 Access
Points on channels 10, 11, 12,
13, and 10. My question: How are
channels regulated outside
the United States, particularly
in France? Are we prohibited
from using channels 1- 9
in France?
In France, only channels 10-13 are
allowed. Given the assumption of
a minimum of 5 channels of separation,
I would conclude that only one channel
could be used at a time without interference
between them. It might be worthwhile
to set APs to channels 10 and 13
(only three channels of separation,
but the maximum available in France)
and run throughput tests to see if
performance improved relative to
setting them all to the same channel.
If channels 10 and 13 are viable,
then perhaps you could end up with
a little net throughput increase
relative to setting all the APs to
the same channel.
For example: You have 5 APs.
Assuming that all
APs are within range
of each other, then,
putting them all
on the same channel
means that only one
can talk at a time.
Average throughput
per AP should be about 4 Mbps
(20 Mbps usable channel
capacity divided
by five APs).
Now, let’s say that you use
10 and 13, so that three APs are
on 10 and two are on 13, in this
pattern: 10, 13, 10, 13, 10. You’re
going to lose some of your usable
channel capacity to interference
between the overlapping channels.
The question is, “How much,
and is the net effect worse or better
than putting them all on the same
channel?” In the U.S., the
answer is pretty much always, “Put
them on non-overlapping channels,” but
given that France doesn’t give
you but one non-overlapping channel,
that might not be the case. Let’s
say that your usable channel capacity
is reduced by interference from 20
Mbps to 10 Mbps. Now, the APs on
channel 10 get roughly 3.3 Mbps each
(10 Mbps usable channel capacity
divided by 3 APs) and the APs on
channel 13 get roughly 5 Mbps each
(10 Mbps usable channel capacity
divided by 2 APs). You can see that
this situation is not much better
than just putting all of the APs
on the same channel. On the other
hand, if interference only reduced
usable channel capacity to 12 Mbps,
then the 10, 13, 10, 13, 10 plan
would be slightly better than the
10, 10, 10, 10, 10 plan. Channel
10 would get about 4 Mbps per AP
and channel 13 would get about 6
Mbps per AP, for a net improvement
of 2 Mbps compared to the 10, 10,
10, 10, 10 plan. Only real-world
testing can confirm whether the 10,
13, 10, 13, 10 plan would be better
than 10, 10, 10, 10, 10.
I would speculate that using
channels 10, 11, 12, and 13 at
the same time would result in
much worse performance than simply
setting all APs to the same channel.
At this point, the APs are so
close together (in terms of their
frequency) that interference
is very likely, and it would
probably be better to set them
on the same channel and let them
share a single pool of bandwidth
using CSMA/CA. Remember in class
when we set the APs to channels
1 and 3? One AP seemed to be
cruising along at 20 Mbps, while
the other AP was having dropouts
and only achieving data rates
of a few hundred kbps. In my
experience, the situation only
gets worse the closer together
the channels are.
To Infinity... and Beyond!
Like all
wireless technologies, the current
802.11 standards (a, b, g, and n)
are full of compromises. They use
unlicensed frequency bands, so they’re
easy to deploy, but the output power
that unlicensed devices can use,
and therefore their range, is limited.
The nearly-finished 802.11y standard
could change all that.
802.11y operates in the 3.65
GHz frequency band. This
band requires licenses, but
the licenses are non-exclusive,
so they should be much easier to
get than licenses for exclusive
bands. The licenses are available
only in parts of the country
that don’t
have pre-existing ground-to-satellite
or radar users that overlap, which
excludes most of the eastern seaboard,
and most major cities.
The
up-side is that this band will
allow for up to 50 MHz of bandwidth
and up to 20 Watts EIRP from a
transmitter, which is five times
as much power as an omnidirectional
802.11b/g transmitter can use.
The increased output power can
be expected to increase range,
making 802.11 more viable for
outdoor point-to-point and
point-to-multipoint links—at
least in rural markets where
the 3.65 GHz licenses will
initially be made available.
At Connect802 we're your PAGE ONE resource for wireless networking!
|
