An IEEE 802.11
WLAN Primer
This newly published standard for wireless networking has created a whole new
set of design elements for
developing access points and other devices. Here, these
elements are discussed in terms of the three physical and single MAC layer implementations.
By Bruce Tuch
The new IEEE 802.11 standard for wireless LANs (WLANs) promises to be a very significant
milestone in the evolution of wireless networking technology. Just finalized, the
standard will promote interoperability among products from different vendors, and
further boost an already burgeoning industry.
The 802.11
standard is limited in scope to the physical (PHY) and medium access
control (MAC) network layers. The PHY layer corresponds directly to the lowest layer
defined by the International Standards Organization in its 7-layer open system interconnect
(OSI) network model. The MAC layer corresponds to the lower half of the second layer
of that same model with logical link control (LLC) functions making up the upper
half of OSI Layer 2.
The standard actually specifies a choice of three different PHY
layers, any of
which can underlie a single MAC layer (Figure 1). Members of the 802.11 working group
felt that a choice of PHY layer implementations was necessary so that systems designers
can choose a technology that matches the price, performance, and operations profile
of a specific application. These choices are exactly analogous to choices such as
10Base-T, 10Base-2, and 100Base-T in the extremely successful Ethernet arena. Moreover,
enterprise LANs will regularly mix wired Ethernet and wireless
nodes with no logical
distinction between the two.
Figure 1
Infrared & RF
Specifically, the standard provides for an optical-based PHY layer that uses infrared
light to transmit data, and two RF-based PHY layers that leverage different types
of spread-spectrum radio communications. The infrared PHY layer will typically be
limited in range and most practically implemented within a single room. The RF-based
PHY layers, meanwhile, can be used to cover significant areas and indeed entire campuses
when deployed in cellular-like configurations.
The infrared PHY layer provides for peak data rates of 1 Mbps with an optional
2-Mbps rate, and relies on pulse position modulation (PPM). The RF PHY layers include
direct-sequence spread spectrum (DSSS) and frequency-hopping spread spectrum (FHSS)
choices. As the names imply, both DSSS and FHSS artificially spread the transmission
band so that the
transmitted signal can be accurately received and decoded in the
face of noise. The two RF PHY layers, however, approach the spreading task in significantly
different ways.
FHSS systems essentially use conventional narrowband data transmission techniques,
but regularly change the frequency at which they transmit. The systems hop at a fixed
time interval around a spread or wideband using different center frequencies in a
predetermined sequence. The hopping phenomena allows the FHSS system to avoid
narrowband
noise in portions of the transmission band (Figure 2). DSSS systems, meanwhile, artificially
broaden the bandwidth needed to transmit a signal by modulating the data stream with
a spreading code. The receiver can detect error-free data even if noise persists
in portions of the transmission band (Figure 3).
Figure 2
Figure 3
In 802.11, the DSSS PHY layer defines peak data rates of both 1 Mbps and 2 Mbps.
The former uses differential binary phase shift keying (DBPSK) and the latter uses
differential quadrature phase shift keying (DQPSK). The standard defines the FHSS
PHY layer to operate at 2 Mbps with fallback to 1 Mbps in extremely noisy environments,
and allows for optional 2-Mbps operation. The PHY layer uses 2- or 4-level Gaussian
frequency shift keying (GFSK) modulation.
Both DSSS and FHSS WLANs will operate in the same frequency band, and neither
require site licenses nor permits
throughout the US, Europe, and Asia. The IEEE 802.11
standard specifies that the WLANs operate in the 2.4-GHz band that regulatory agencies
around the world have set aside for spread-spectrum usage.
RF PHY-layer channelization
One of the key advantages of RF PHY layers is the ability to have a number of
distinct channels. The channelization allows WLAN users to colocate channels in the
same or adjacent areas to boost aggregate throughput, or to deploy a
cellular-like
array of channels that support roaming clients. In the case of DSSS, different channels
simply use different frequency bands. In the case of FHSS, the hopping sequence used
differentiates one channel from the next, but all channels operate in the same wide
frequency band.
Obviously, the DSSS and FHSS PHY layers use the allocated RF spectrum in the 2.4-GHz
band differently. Moreover, different regions with slightly different regulations
throughout the world affect exact channelization
schemes for both PHY layer types.
Generally, the 802.11 spec defines thirteen DSSS channels that are used to carry
a spread 22-MHz signal. Channels overlap with a new center frequency located at 5-MHz
intervals. For deployment in the US, the standard defines eleven independent DSSS
channels in the ISM bandwidth allocated by the FCC. Throughout much of Europe and
much of Asia which follows the lead of European regulatory agencies, DSSS implementations
can leverage thirteen channels. In Japan,
however, the allocated bandwidth only supports
a single channel. In the US and Europe, the channel definition ensures that three
frequency-isolated channels are available for colocation.
FHSS systems, meanwhile, are mandated to use seventy-nine hops, or center frequencies,
in the US and Europe, and twenty-three hops in Japan. Typically, FHSS systems dwell
at each hop for 20 msecs. The spec defines seventy-eight different hopping sequences,
and each independent hopping sequence is defined as a
channel. Practically, however,
only a few channels can be effectively deployed in close proximity to one another.
WLAN topologies
Regardless of the type of PHY layer chosen, IEEE 802.11 supports three basic topologies
for WLANs ı the independent basic service set (IBSS), the basic service set (BSS),
and the extended service set (ESS). In fact, the MAC layer implements the support
for IBSS, BSS, and ESS configurations.
IBSS configurations are
also referred to as an independent configuration or an
adhoc network. Logic-ally, an IBSS configuration is analogous to a peer-to-peer office
network in which no single node is required to function as a server. IBSS WLANs include
a number of nodes or wireless stations that communicate directly with one another
on an adhoc, peer-to-peer basis. Generally, IBSS implementations cover a limited
area and arenıt connected to any larger network.
BSS configurations rely on an access point (AP) that
acts as the logical server
for a single WLAN cell or channel. Communications between node A and node B actually
flow from node A to the AP, and then from the AP to node B. At first, it may seem
that the AP adds an unnecessary layer of complexity and overhead to the WLAN, but
the AP enables quite a few features of 802.11 that will be described later in this
article. Moreover, an AP is necessary to perform a bridging function and connect
multiple WLAN cells or channels, and to connect WLAN cells to a
wired enterprise
LAN. In fact, ESS WLAN configurations consist of multiple BSS cells that can be linked
by either wired or wireless backbones. IEEE 802.11 supports ESS configurations in
which multiple cells use the same channel, and configurations in which multiple cells
use different channels to boost aggregate throughput.
Key MAC-layer features
To understand the full extent of capabilities of BSS and ESS WLANs, you must first
examine the details of
the 802.11 MAC. First and foremost, be assured that the 802.11
MAC was developed to work seamlessly with standard Ethernet to ensure that wireless
and wired nodes on an enterprise LAN are logistically indistinguishable. The 802.11
MAC is necessarily different from the wired Ethernet MAC, but any such differences
are masked by an AP that connects a WLAN channel to a LAN backbone. (For more information
on the AP, see Communication Systems Design, July 1996 p.18, "Wired-Wireless
Internetworking").
For starters, 802.11 defines both a frame format and MAC scheme that differs from
standard Ethernet. Figure 3 depicts the details of the 802.11 frame format. In fact,
this robust frame format enables a number of compelling features such as fast acknowledge,
handling hidden stations, power management, and data security. The WLAN standard
uses a carrier sense multiple access with collision avoidance (CSMA/CA) MAC scheme,
whereas standard Ethernet uses a carrier sense multiple access with
collision detection
(CSMA/CD) scheme.
Frame receipt acknowledgment provides one good example of differences between
the 802.11 MAC and the wired Ethernet MAC, along with the advantage the difference
affords in a wireless system. Most LANs rely on a receiving node to send an acknowledge
message to verify that it received an incoming data frame. In Ethernet and most wired
LANs, however, the acknowledge message is handled above the MAC layer.
The 802.11 standard specifies that the MAC layer
handle acknowledgment and resend
lost frames resulting in more efficient usage of the available bandwidth and quicker
acknowledgment. The 802.11 frame format relies on an interframe spacing of 50 ms.
The standard requires that the receiving station send an acknowledgment 10 ms after
the end of each frame, providing the CRC check is correct. The 10-ms limit ensures
that the receiving station can take immediate control of the airwaves rather than
competing with other nodes for medium access, as would be
required if it waited past
the 50-ms interframe spacing. LANs that handle acknowledgment in layers above the
MAC canıt meet the strict timing requirements and, therefore, essentially compete
for medium access and send a standard frame to convey each acknowledgment. The MAC-layer
implementation eliminates the latencies of medium access and allows the acknowledgment
to use some of the interframe spacing time period in which no other activity would
occur in any case.
When a wireless server is
used, the bandwidth efficiency becomes doubly important
in a BSS or ESS configuration because many frames must be sent twice ı first from
node A to an AP, and then from the AP to node B. In such a case, the AP must send
an acknowledgment to node A, and node B sends an acknowledgment to the AP.
Hidden station provisions
Some common phenomena found in WLANs prompted other 802.11 features. For example,
in a BSS or ESS configuration, the WLAN can suffer
from collisions caused by a hidden-station
provision. In such a case, station A can communicate with the AP with no problem,
and station B can communicate with the AP with no problem, but station A and B are
physically separated by sufficient distance to prevent direct communications. Since
A and B donıt directly communicate in an ESS configuration, the hidden-station problem
doesnıt affect the actual communications, but rather the competition for medium access.
The collision avoidance scheme
adopted in 802.11 requires a station to avoid transmitting
while other nodes are actively transmitting. Yet node A would not be able to detect
that node B was transmitting to the AP in the hidden-station example.
The 802.11 includes an optional request to send (RTS)/clear to send (CTS) provision
to protect against hidden-station interference. All 802.11 receivers must support
RTS/CTS, but support is optional in transmitters.
The option to use RTS in the transmitter is exercised by the
designer of the 802.11
equipment, or, in some cases, by a WLAN administrator or user. For example, a network
interface card (NIC) designer might choose not to support the RTS feature to minimize
cost, especially since the phenomena never occurs in many installations. Alternatively,
the NIC designer may implement the feature, but allow the user to selectively enable
or disable RTS.
Fragmentation
The MAC also supports a concept called fragmentation
that provides for flexibility
in transmitter/receiver design, and can be useful in environments with RF interference.
An 802.11 transmitter can optionally break messages into smaller fragments for sequential
transmission. A receiver can more reliably receive the shorter data bursts because
the shorter duration of each fragment transmission reduces errors due to signal fading
or noise. Moreover, the smaller fragments have a better chance of escaping burst
interference such as that from a microwave
source.
The 802.11 standard mandates that all receivers support fragmentation, but leaves
such support optional on transmitters. Designers (and potentially end users) can
determine when or if fragmentation is used. For example, designers could add the
ability for the end user to enable fragmentation when transmission errors become
a problem. Designers can also use fragmentation as a price/performance trade-off.
By assuming full-time fragmentation in a transmitter, the receiver can be designed
with
less expensive components resulting in lower receiver sensitivity. Fragmentation,
however, incurs overhead on every fragment rather than every frame, thereby reducing
aggregate throughout of the WLAN and the realizable peak throughput rate achieved
between stations.
Power management
The subject of roaming suggests that, by nature, many WLAN clients will be portable
systems ı both standard portable notebooks and application-specific systems, such
as
handheld message pads. The fact that portable systems are typically battery powered
led to another area of specialization in the 802.11 MAC specification. Specifically,
features were added to the MAC that could maximize battery life in portable clients
via power-management schemes.
To support clients that periodically enter sleep mode, the 802.11 specifies that
APs include buffers to queue messages. Sleeping clients are required to awaken periodically
and retrieve any messages. The APs are
permitted to dump unread messages after a
specified time, and the messages go unretrieved.
The power management scheme is relatively easy to implement in a BSS or ESS configuration
because the AP is always present. Moreover, the AP isnıt battery powered and never
enters sleep mode. In an IBSS configuration, however, no full-time AP exists and
all systems may desire to enter sleep mode.
The 802.11 does provide for power savings in an IBSS configuration. Essentially,
all clients in an IBSS
configuration must awaken each time a beacon is sent. The
clients randomly alternate the task of transmitting a beacon (Figure 4). Immediately
after the transmission of each beacon, a short time period, called an announcement
traffic information message (ATIM) window, commences. During the ATIM window, any
station can indicate the need to transfer data to another station during the ensuing
data-transmission window. Clients with no incoming or outgoing frames pending can
reenter sleep mode during the
data-transmission window.
Figure 4
The power-management capabilities built into 802.11 apply primarily to NICs. System-level
power management in a client remains the domain of the system designer. Obviously,
however, the NIC must be able to awaken independently of the client system, and the
NIC must be able to awaken the client when it learns of pending incoming frames.
Privacy
One final area of
differentiation between 802.11 and either wired LANs or existing
WLAN implementations centers on data security. The standard defines a mechanism through
which the WLANs can achieve wired equivalent privacy (WEP). The optional WEP mechanism
is especially important because RF transmissions ı even spread-spectrum transmissions
ı can be intercepted more easily than wired transmission.
The 802.11 embeds the WEP mechanism within the MAC that covers station-to-station
transmission. The standard specifies usage
of the RC4 security algorithm from RSA.
The scheme relies on a 40-bit key to encrypt the payload of data frames. The working
group chose the RC4 algorithm in part because the US Government does not restrict
the export of products using the RC4 encryption method. In contrast, other algorithms
such as DES can only be exported in a few specific applications. Moreover, tests
by members of 802.11 prove RC4 offers security that matches or exceeds the privacy
achievable by standard wired Ethernet.
Follow-on efforts in IEEE 802.11
While vendors are rushing to bring 802.11-compliant products to market, the committee
members and companies behind the standard can also look forward to follow-up efforts
that provide continuous improvement. For example, 802.11 addresses roaming, provided
that all APs in an ESS installation were manufactured by the same vendor. The standard
does not ensure that clients can roam among APs from different vendors.
To
address multivendor roaming, Aironet Corp, Digital Ocean, and Lucent Technologies
have collaborated to develop the inter-access point protocol (IAPP) specification.
The IAPP will extend the 802.11 multivendor interoperability benefits with comprehensive
roaming protocols. There are two transfer protocols defined within the IAPP, one
for single logical LANs and one for crossing router boundaries using the user datagram
protocol/internet protocol (UDP/IP). Nodes that roam across router boundaries also
require mobile IP software. Several other companies, including IBM, have voiced support
for the IAPP as a necessary step towards true multivendor interoperability.
Faster data rates
While 802.11 will offer aggregate throughput comparable to that of
10-Mbps wired Ethernet, some applications, such as multimedia, demand faster peak-data
rates. Moreover, faster peak rates will allow more nodes to effectively connect to
a WLAN via a single channel. The
next step in the evolution of 802.11 is most likely
a standard for higher data rates, expected in the 10 Mbps and above range. In the
last IEEE meeting, the 802 committee approved a draft project authorization request
(PAR) for 20 Mbps in the 5.2-GHz band. In the 2.4-GHz band, there have also been
discussions for extending the current standards for higher speeds.
Meanwhile, Bell Laboratories has developed a DSSS-base WLAN scheme that can extend
peak data rates from 2 Mbps to 10 Mbps.
Moreover, the scheme ensures backwards compatibility
with IEEE 802. Called direct sequence/pulse position modulation (DS/PPM), the scheme
varies the timing of data symbols, and then that variance is used to convey additional
bits of data in each symbol. This PPM technique boosts normal 2-Mbps DS rates to
8 Mbps. DS/PPM also uses quadrature amplitude modulation (QAM), whereas 802.11
DS WLANs use phase shift keying (PSK) modulation. Combining QAM and PPM yields the
10-Mbps DS/PPM rate.
The
DS/PPM scheme uses the same frequency bands and channels as those defined
for 2-Mbps DS WLANs in 802.11. This new technologyıs implementation will allow 2-Mbps
and 10-Mbps devices to operate seamlessly in the same channel and a
10-Mbps device to automatically fall back to the 2-Mbps rate in order to communicate
with 802.11 nodes.
Overall, the 802.11 standard is a robust and rich, 7-year effort created by all
the leading scientists in the WLAN field. In addition to providing high
performance
and robust systems, it also promises multivendor interoperability amongst products
with the same PHY layers. This means that customers are free to mix and match vendors
to meet their requirements for each given application. Furthermore, standardization
also delivers lower cost-components, which will translate into lower prices for users.
Bruce Tuch is director of engineering of the wireless communications and networking
division at Lucent Technologies. He holds patents in the
areas of data communications
networks and RF circuit designs. He received a BSEE degree from the State University
of New York at Stony Brook, and an MSEE degree from the Eindhoven University of Technology,
Eindhoven, The Netherlands.
Puzzled by a network processing design issue?
