26-04-2012, 10:09 AM
IEEE 802.11n Development: History, Process, and Technology
IEEE 802.11n Development History, Process, and Technology.pdf (Size: 110.52 KB / Downloads: 37)
ABSTRACT
This article provides insight into the IEEE
802.11n standard amendment development process,
beginning with a general overview of the
IEEE 802.11 process. Development of requirements
and usage models in the study group and
task group is discussed. The lengthy proposal
down selection process used by 802.11n is
described and critiqued. We also discuss the
expected time to develop a standard from a market
perspective. An overview of the physical
layer technology used to achieve the 600 Mb/s
data rate is presented. We outline the medium
access layer features employed to enhance usable
throughput to over 400 Mb/s. The added robustness
afforded by techniques in the standard and
issues with backward compatibility with legacy
IEEE 802.11a/g devices are addressed.
INTRODUCTION
In 2002 discussions began in the IEEE 802.11
Working Group (WG) to extend the data rates
of the physical layer beyond those of IEEE
802.11a/g in order to address higher throughput
wired applications that would benefit from the
flexibility of wireless connectivity. The WG proceeded
through the typical steps in developing a
standard. The high throughput study group (HT
SG) was formed with great interest, with participants
at the meetings numbering well over 100.
The group introduced new antenna technology,
such as multiple-input multiple-output (MIMO).
Subsequently the IEEE 802.11n Task Group
(TGn) began to develop an amendment to the
IEEE 802.11 standard (i.e., IEEE 802.11n). Initially
there was large participation in TGn. Often
votes on TGn proposals caused other task groups
to temporarily recess their meetings, and garnered
on the order of 250 votes. As the technology
and the draft matured, interest in TGn has
declined to a few core participants resolving
comments. Now the excitement in TGn has shifted
from the standards body to the marketplace,
where numerous draft IEEE 802.11n products
are becoming available to the consumer. These
new products enhance basic networking in the
home and office. Also, new types of products are
beginning to become available, such as IEEE
802.11n-based wireless multimedia and gaming
systems.
As described in the subsequent section, the
standard amendment process seems straightforward
and benign — for IEEE 802.11n it has
been everything but so. The process has turned
out to be as challenging as the technology itself.
The history that brought us to the current phase
of the process is described in the sections on
study group and task group activity. The proposal
process specific to IEEE 802.11n and its associated
issues is outlined. We describe how
compromise was finally reached, leading to the
first draft of the amendment. In the next section
of the article market expectations regarding the
time to complete the IEEE 802.11n standard are
discussed. Following that we provide an overview
of the physical layer (PHY) and medium access
layer (MAC) enhancements in IEEE 802.11n.
Lastly, we highlight some lessons learned and
propose changes to the process that may reduce
the duration of amendment development.
As a note, in the remainder of the article the
notation 802.11 will be used as a simplification
of IEEE 802.11 with the same intended meaning.
802.11 PROCESS TO AMEND THE STANDARD
The 802.11 WG follows five steps to amend the
802.11 standard:
• Initial discussion of new ideas in the Wireless
Next Generation Standing Committee
(WNG SC)
• Formation of an SG to formulate the purpose
and scope of the amendment
• Creation of a TG to develop a draft of the
amendment that addresses the purpose and
scope
• Approval of the draft by the WG, and opening
of a Sponsor Ballot pool for review of
the draft by the IEEE Standards Association
(IEEE-SA)
• Ratification of the draft by the IEEE-SA
Standards Board
To elaborate on the five steps above, the
standard amendment process begins in the WG
with presentations of a new concept to the WNG
IEEE STANDARDS IN COMMUNICATIONS
AND NETWORKING
Eldad Perahia, Intel Corporation
IEEE 802.11n Development: History,
Process, and Technology
IEEE Communications Magazine • July 2008 49
SC. If there is broad interest for the new idea,
the WNG SC participants vote to request the
WG to create a new SG. Presentations can all be
made in one 802.11 meeting or span multiple
meetings. The WG then votes on whether to create
a new SG, which requires 75 percent
approval. Unlike other organizations where each
company gets a single vote, in 802.11 each individual
participant who has achieved voting member
status gets a vote (status achieved through
meeting attendance).
With a passing vote, a new SG is created. The
SG prepares a document called the Project
Authorization Request (PAR), which contains
the purpose and scope of the amendment. These
will become the guiding requirements for the
TG. The SG must also address five criteria
demonstrating the need for the new amendment.
These criteria include:
• Broad market potential
• Compatibility with the family of IEEE 802
standards
• Distinct identity from other IEEE 802 standards
• Technical feasibility
• Economic feasibility
The SG then votes to approve the PAR and five
criteria, and request the WG to create a TG.
This step also requires 75 percent approval. Typically
the WG briefly discusses the PAR and five
criteria, and often requests modifications from
the SG. Eventually the WG votes to approve the
PAR and five criteria, which again requires 75
percent approval. The WG then requests the
IEEE 802 Executive Committee (EC) and after
that the IEEE-SA Standards Board to approve
the PAR and formation of a new TG. The SG
typically lasts six months to a year. On formation
of the TG, the SG dissolves. If either the WG or
the EC do not approve the PAR and five criteria,
a TG is not created.
The primary goal of the TG is to create a
draft amendment. The TG can either “design by
committee” or issue a call for proposals. With
the “design by committee” approach, individuals
present submissions on new features. With a 75
percent vote by the TG members, the new feature
is adopted as part of the draft. On the other
hand, with a call for proposals, typically groups
of individuals or companies form proposal teams
and create a proposal that on acceptance would
become the initial draft of the amendment. This
approach requires a down selection procedure
since typically numerous proposals are submitted
for consideration. Details of the down selection
procedure are decided by the TG, but a confirmation
vote by the TG of the winning proposal
is required. The confirmation vote requires 75
percent approval for the final proposal to
become the initial draft.
After a draft is created, a letter ballot pool is
formed from all the voting members of the WG.
The members review the draft, and a letter ballot
vote occurs on whether the draft is acceptable
for submission to the IEEE-SA as a Sponsor
Ballot. As part of the voting process, the members
create comments regarding their issues with
the draft. If there is not 75 percent approval, the
TG goes back to work on a new draft addressing
all the comments. This is the comment resolution
phase of the TG. If the vote exceeds 75 percent
approval but with voters still generating
many new comments, the TG works on a revision
of the specific sections of the draft that
addresses the new comments. Subsequent votes
are termed recirculation votes. Finally, when no
new no votes and comments are received from
the WG, the draft is submitted to IEEE-SA for a
Sponsor Ballot.
The Sponsor Ballot pool is made up of members
of the IEEE-SA. Any member of the IEEE
may join the IEEE-SA as an addition to annual
IEEE membership. This process provides a
broad review of the draft, beyond just the participants
of 802.11.
The Sponsor Ballot process is similar to the
letter ballot process. The Sponsor Ballot vote
occurs, and the TG receives comments and generates
updates to the draft that address the comments.
When the Sponsor Ballot pool finally
approves the draft amendment with no new no
votes or comments, the draft is ratified by the
IEEE-SA Standards Board. Optimistically, creation
of a new amendment, starting with TG formation
to final ratification, takes two to three
years.
WNG SC AND SG ACTIVITY
In January 2002 a presentation was given to
WNG SC expressing interest in a high-data-rate
extension to 802.11a [4]. The interest was in part
based on increasing data rates in wired Ethernet
and wireless products emerging on the market
with proprietary extensions to 802.11a/g. Subsequently,
other presentations were given calling
for greater than 100 Mb/s data rates by spatial
multiplexing and/or doubling the bandwidth in
addition to improving MAC efficiency. Claims
were made that new markets and applications
would require higher throughput (e.g., wireless
home entertainment). It is important to note
that a presentation was made describing a
MIMO prototype and actual measurements [2].
The goal of standards development is to foster
new commercially successful markets by interoperable
products, not to perform research. Proof
of feasibility of new technology by prototyping is
a reasonable starting point for standards development.
For market success, the goal should be
that by the time standard development is complete
(or mature), manufacturers are capable of
low-cost silicon implementation of the system.
The new High Throughput (HT) SG was
formed in September 2002. Beyond development
of the PAR and five criteria, work also began in
HT SG on usage model development, channel
model development, and selection procedures.
The work was then continued in TGn. A committee
was formed to define various marketbased
usage models that were used to define
network simulation scenarios for the performance
evaluation of the proposals. The usage
models were to be relevant to the expected uses
of the technology. Furthermore, they were to
require higher throughput than was available
with 802.11a/g. The components of the usage
models included applications, environments, and
use cases [7]. The applications included various
forms of video and audio, Internet and local file
The Sponsor Ballot
pool is made up of
members of the
IEEE-SA. Any
member of the IEEE
may join the IEEE-SA
as an addition to
annual IEEE
membership. This
process provides a
broad review of the
draft, beyond just
the participants
of 802.11.
50 IEEE Communications Magazine • July 2008
transfer, and VoIP. Requirements in terms of
offered load, maximum packet loss rate, maximum
delay, and network protocol were captured
for each application. The main environments
included home, office, and hot spots. Use cases
were collected that gave a description of how
someone uses the application in a particular
environment (i.e., multiple TVs running throughout
the home getting their content from a single
remotely located media server). The various use
cases were merged together to create a small
number of realistic usage models, but each capable
of stressing the system. Each usage model
contained an access point (AP) and a defined
number of stations running a mix of applications
based on the use cases. The environment thus
dictated the channel model. The usage models
were then converted into simulation scenarios.
With MIMO one of the primary physical
layer candidate technologies, new standardized
channel models were required in order to benchmark
different proposals. A channel model ad
hoc committee was created to develop indoor
MIMO channel models. The channel modeling
effort incorporated a literature search for existing
models and measurements, new measurements,
and the development of new models [1].
TGN PROPOSAL PROCESS
TGn officially began in September 2003. TGn
decided to proceed with a call for proposals
rather than a design by committee approach.
The first order of business was to complete the
creation of the selection procedure. As a first
step in the selection criteria, functional requirements
and comparison criteria were defined [6].
The Functional Requirements document was
created containing a list of mandatory features,
performance, and behavior [8]. The Comparison
Criteria document defined the simulation results
that were required of a complete proposal [9]. A
complete proposal was one that did not violate
the PAR and met all the functional requirements
addressing the comparison criteria.
Once the usage models, channel models,
functional requirements, and comparison criteria
were adopted, the task group issued a call for
proposals in May 2004. The selection criteria
called for a series of down selection votes to one
proposal. After each down selection vote, the
proposal with the least number of votes was
eliminated. The final proposal was required to
pass a confirmation vote by 75 percent. If the
confirmation vote failed, the last three proposals
would be brought back and the process restarted.
As can be seen, only 25 percent of the group
can force this process to repeat forever (or at
least until the defined duration of the TG
expires).
In 802.11n five complete proposals were submitted
for consideration along with a large number
of partial proposals. Three of the proposals
were created by individual companies. The other
two proposals were each created by a team of
companies: TGn Sync (started by Intel, Cisco,
Agere, and Sony) and WWiSE (started by
Broadcom, Conexant, Texas Instruments, and
Airgo Networks). Many other companies were
involved in the proposal process, and most ended
up joining one of these two proposal teams. The
first round of proposal presentations was in Sept
2004, two years after the start of the study group.
Partial proposals were given time to present, but
were required to merge with complete proposals
for further consideration.
After a series of down selection votes, in
March 2005 TGn had its final down selection
vote and first failed confirmation vote of the
TGn Sync proposal. In May 2005, the second
failed confirmation vote took place and selection
procedure reset to the last three proposals. The
down selection process naturally creates a contentious
environment. Companies expend a
tremendous amount of resources developing
technology and a proposal. A long drawn out
down selection process increases the tension
between camps, and makes compromise more
and more difficult. An attempt was made to create
a joint proposal between the three proposals,
but the effort was unsuccessful due to the acrimony
and lack of trust between the participants.
The basic features and technologies in the various
proposals were actually the same. All proposals
included MIMO, 40 MHz bandwidth
channels, frame aggregation, and enhanced
block acknowledgment. For the most part, the
difference between the proposals was the implementation
details of these features. But due to
the process, it became impossible to negotiate
within the IEEE standards development environment.
In light of the stagnation of the proposal
selection process, a group of silicon providers
(started by Intel and Broadcom) went outside
the IEEE and formed the Enhanced Wireless
Consortium (EWC) special interest group to
craft a basic interoperable specification such that
they could start implementing interoperable
devices. The specification picked pieces from
each of the top two proposals to create the first
draft. Ultimately the EWC felt it was beneficial
to the industry for the specification to become
an IEEE standard. Since EWC had no formal
standing in the IEEE, EWC was required to
convince others to adopt the EWC specification
as the TGn joint proposal. This required passing
the confirmation vote with 75 percent. Negotiations
with more companies to garner support
resulted in many new optional features to the
EWC specification. The final EWC specification
was adopted as a joint proposal and submitted
for confirmation in TGn, where it passed unanimously
in January 2006.
Interestingly, more optional features ended
up in the EWC/joint proposal than were in either
the TGn Sync or WWiSE proposals. Such is the
nature of compromise necessary to achieve the
75 percent confirmation vote. For example, TGn
Sync had implicit beamforming and WWiSE had
no beamforming. Yet the joint proposal contained
implicit beamforming, explicit beamforming,
and antenna selection.
In hindsight, there are a number of ways the
down selection process could have been streamlined.
The top two proposals (TGn Sync and
WWiSE) received the largest number of votes at
each down selection. Furthermore, the top proposal
(TGn Sync) received the largest number of
votes at each down selection vote. TGn could
With MIMO being
one of the primary
physical layer
candidate
technologies, new
standardized channel
models were
required in order to
benchmark different
proposals. A channel
model ad hoc
committee was
created to develop
indoor MIMO
channel models.
IEEE Communications Magazine • July 2008 51
have held just two down selection votes, one to
reduce the number of proposals to two and a
final down selection vote to select a winner.
Considering the basic technology was the same
between proposals, the down selection winner
could have been converted to a first draft of the
standard amendment, bypassing confirmation
votes. In essence, the first draft would be selected
based on the proposal receiving greater than
50 percent vote rather than 75 percent. Such an
approach would completely change the dynamics
of the proposal process. To achieve the extra 25
percent for confirmation, proposal teams (and
EWC) were required to incorporate features
from various companies that had little general
support and in many cases were no more than
research ideas.
The letter ballot vote on the draft requires 75
percent, so a super-majority vote is still required
to approve the draft. One may argue that the
first letter ballot would be guaranteed to fail if
the winning proposal only achieved 50 percent
approval. On the other hand, the TGn joint proposal
achieved unanimous support in the confirmation
vote, but still failed the first letter ballot
and generated thousands of comments in large
part due to all the extra features. Once these
features are in the draft, a 75 percent vote is
required to remove them, which is almost impossible.
Therefore, a great deal of time is required
to fix all the extraneous features and address
their associated letter ballot comments. A draft
based on a proposal with only 50 percent support
may also fail the letter ballot, but would be
guaranteed to have far fewer comments due to
the smaller number of optional features. Furthermore,
new features would then be required
to achieve 75 percent support, resulting in higher-
quality additions to the draft.
MARKET EXPECTATIONS AND TIMESCALES
As the technology of 802.11a/b/g matured, silicon
and system providers were looking for new
technology to incorporate into new products to
refresh product lines. In 2004 Atheros developed
a proprietary 40 MHz mode built on 802.11g. In
2005 and 2006 the market saw the first wave of
proprietary MIMO-based wireless LAN products.
These were typically called “pre-n.” Interoperability
between different products was only
guaranteed by falling back to 802.11a/g operation.
Looking back at the history of TGn, the initial
schedule put forth by HT SG called for completing
the PAR in November 2002, completing
the first draft in July 2003, completing the second
draft in September 2003, going to sponsor
ballot in March 2004, and final approval in July
2004. Obviously this was a bit optimistic, but
even an additional year or two would have met
the needs of manufacturers with an IEEE standard,
rather than having to produce proprietary
and non-interoperable modes of operation.
Turning back to the down selection process
of TGn, the issue with incorporating so many
“pet features” as part of a compromise is the
time it takes to thoroughly review, edit, check,
and test each feature. This is illustrated by the
number of comments TGn received in its first
three letter ballots. Six thousand unique comments
were received in the letter ballot for draft
1.0. Three thousand unique comments were
received in letter ballot for draft 2.0. Nine hundred
comments were received in letter ballot for
draft 3.0. With comment resolution for draft 3.0
expected to be completed in March 2008, comment
resolution has thus far taken two years.
The current projected completion date is June
2009, approaching eight years after the first presentation
in WNG SC.
With the exception of coexistence between 40
and 20 MHz channel bandwidth modes of operation,
the basic functionality of the PHY and
MAC layer stabilized between drafts 1.0 and 2.0.
Beyond draft 2.0, the vast majority of the comments
and most of the time spent creating resolutions
have been on the numerous optional
features. The nature of the down selection process
that results in adding many options to garner
75 percent confirmation vote greatly extends
the comment resolution process.
Having realized the market demand for standards-
based interoperable products, the Wi-Fi
Alliance considered certifying devices based on
an 802.11n draft due to the slow development of
the 802.11n amendment. Wi-Fi certified products
provide consumers with assurance of interoperability
of core functionality that was not
guaranteed by proprietary modes in pre-n products.
The Wi-Fi Alliance developed a certification
program based on a subset of features in
802.11n draft 2.0 and began certification of
802.11n draft 2.0 devices in June 2007. Part of
the rationale was that the core functionality of
the draft was stable, with draft 2.0 being a major
enhancement over draft 1.0. Considering the
current timeline for 802.11n, early release by
IEEE of the 802.11n draft 2.0 amendment and
Wi-Fi certification of draft 2.0 was a significant
step forward in speeding up the adoption of
802.11n-based WLAN technology. Additional
optional features may be certified as the draft
matures or reaches final approval.
Early release of a draft of the standard and
draft-based certification changed the dynamics in
TGn. Participants in TGn whose employer produced
draft 2.0 certified products are now motivated
to maintain interoperability between their
current certified products and later final standard-
based products. It is now unlikely that any
changes will occur to the core features certified
based on draft 2.0. The focus of the comment
resolution process has shifted to refining the
optional features that were not part of the draft
2.0 based certification.
OVERVIEW OF 11N ENHANCEMENTS
The key requirement that drove most of the
development in 802.11n is the capability of at
least 100 Mb/s MAC throughput. Considering
that the typical throughput of 802.11a/g is 25
Mb/s (with a 54 Mb/s PHY data rate), this
requirement dictated at least a fourfold increase
in throughput. Defining the requirement as
MAC throughput rather than PHY data rate
forced developers to consider the difficult prob-
It is now unlikely
that any changes will
occur to the core
features certified
based on draft 2.0.
The focus of the
comment resolution
process has shifted
to refining the
optional features
that were not part of
the draft 2.0 based
certification.
52 IEEE Communications Magazine • July 2008
lem of improving MAC efficiency. Figure 1
demonstrates the achievable throughput when
the PHY data rates are increased with an
unmodified 802.11e-based MAC. The inability to
achieve a throughput of 100 Mb/s necessitated
substantial improvements in MAC efficiency
when designing the 802.11n MAC.
Two basic concepts are employed in 802.11n
to increase the PHY data rates: MIMO and 40
MHz bandwidth channels. Increasing from a single
spatial stream and one transmit antenna to
four spatial streams and four antennas increases
the data rate by a factor of four. (The term spatial
stream is defined in the 802.11n standard [3]
as one of several bitstreams that are transmitted
over multiple spatial dimensions created by the
use of multiple antennas at both ends of a communications
link.) However, due to the inherent
increased cost associated with increasing the
number of antennas, modes that use three and
four spatial streams are optional, as indicated in
Fig. 2. And to allow for handheld devices, the
two spatial streams mode is only mandatory in
an access point (AP). As shown in Fig. 2, 40
MHz bandwidth channel operation is optional in
the standard due to concerns regarding interoperability
between 20 and 40 MHz bandwidth
devices, the permissibility of the use of 40 MHz
bandwidth channels in the various regulatory
domains, and spectral efficiency. However, the
40 MHz bandwidth channel mode has become a
core feature due to the low cost of doubling the
data rate from doubling the bandwidth. Almost
all 802.11n products on the market feature a 40
MHz mode of operation. Other minor modifications
were also made to the 802.11a/g waveform
to increase the data rate. The highest encoder
rate in 802.11a/g is 3/4. This was increased to 5/6
in 802.11n for an 11 percent increase in data
rate. With the improvement in radio frequency
(RF) technology, it was demonstrated that two
extra frequency subcarriers could be squeezed
into the guard band on each side of the spectral
waveform and still meet the transmit spectral
mask. This increased the data rate by 8 percent
over 802.11a/g. Lastly, the waveform in 802.11a/g
and mandatory operation in 802.11n contains an
800 ns guard interval between each orthogonal
frequency-division multiplexing (OFDM) symbol.
An optional mode was defined with a 400 ns
guard interval between each OFDM symbol to
increase the data rates by another 11 percent.
Another functional requirement of 802.11n
was interoperability between 802.11a/g and
802.11n. The TG decided to meet this requirement
in the physical layer by defining a waveform
that was backward compatible with
802.11a and OFDM modes of 802.11g. The
preamble of the 802.11n mixed format waveform
begins with the preamble of the 802.11a/g
waveform. This includes the 802.11a/g short
training field, long training field, and signal
field. This allows 802.11a/g devices to detect
the 802.11n mixed format packet and decode
the signal field. Even though the 802.11a/g
devices will not be able to decode the remainder
of the 802.11n packet, they will be able to
properly defer their own transmission based on
the length specified in the signal field. The
remainder of the 802.11n Mixed format waveform
includes a second short training field,
additional long training fields, and additional
signal fields followed by the data. These new
fields are required for MIMO training and signaling
of the many new modes of operation. To
ensure backward compatibility between 20 MHz
bandwidth channel devices (including 802.11n
and 802.11a/g) and 40 MHz bandwidth channel
devices, the preamble of the 40 MHz waveform
is identical to the 20 MHz waveform and is
repeated on the two adjacent 20 MHz bandwidth
channels that form the 40 MHz bandwidth
channel. This allows 20 MHz bandwidth
devices on either adjacent channel to decode
the signal field and properly defer transmission.
The preamble in 802.11a has a length of 20 μs;
with the additional training and signal fields,
the 802.11n mixed format packet has a preamble
with a length of 36 μs for one spatial stream
up to 48 μs for four spatial streams.
n Figure 1. Throughput vs. PHY data rate assuming no MAC changes. Reproduced
with permission from [5].
PHY rate (Mb/s)
0
10.0
Throughput (Mb/s)
0.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
100 200 300 400
Legacy
20 MHz,
1 x 1
40 MHz, 1 x 1
40 MHz, 2x2
40 MHz, 3 x 3 40 MHz, 4 x 4
20 MHz, 2 x 2 20 MHz, 3 x 3
20 MHz, 4 x 4
500 600
40 MHz, 4 x 4
40 MHz, 3 x 3
40 MHz, 2 x 2
40 MHz, 1 x 1
20 MHz, 4 x 4
20 MHz, 3 x 3
20 MHz, 2 x 2
20 MHz, 1 x 1
Legacy
n Figure 2. Mandatory and optional 802.11n PHY features. Reproduced with
permission from [5].
1, 2 spatial streams
20 MHz; rate 5/6;
56 sub-carriers
Basic MIMO/SDM
Convolution code
Mixed format
Mandatory
3,4 spatial streams
40 MHz
Short GI
TxBF
STBC
LDPC code
Greenfield format
Optional
Throughput
enhancement
Robustness
enhancement
Interoperability
with legacy
Unfortunately, MIMO training and backward
compatibility increases the overhead,
which reduces efficiency. In environments free
from legacy devices (termed greenfield) backward
compatibility is not required. As illustrated
in Fig. 2, 802.11n includes an optional
greenfield format. By eliminating the components
of the preamble that support backward
compatibility, the greenfield format preamble is
12 μs shorter than the mixed format preamble.
This difference in efficiency becomes more pronounced
when the packet length is short, as in
the case of VoIP traffic. Therefore, the use of
the greenfield format is permitted even in the
presence of legacy devices with proper MAC
protection, although the overhead of the MAC
protection may reduce the efficiency gained
from the PHY.
Range was considered as a performance metric
in the PAR and comparison criteria. To
increase the data rate at a given range requires
enhanced robustness of the wireless link. 802.11n
defines implicit and explicit transmit beamforming
(TxBF) methods and space-time block coding
(STBC), which improves link performance
over MIMO with basic spatial-division multiplexing
(SDM). The standard also defines a new
optional low density parity check (LDPC) encoding
scheme, which provides better coding performance
over the basic convolutional code.
To break the 100 Mb/s throughput barrier,
frame aggregation was added to the 802.11n
MAC (as illustrated in Fig. 3) as the key
method of increasing efficiency. The issue is
that as the data rate increases, the time on air
of the data portion of the packet decreases.
However, the PHY and MAC overhead remain
constant. This results in diminishing returns
from the increase in PHY data rate, as illustrated
in Fig. 1. Frame aggregation increases the
length of the data portion of the packet to
increase overall efficiency.
Two forms of aggregation exist in the standard:
MAC protocol data unit aggregation (AMPDU)
and MAC service data unit aggregation
(A-MSDU). Logically, A-MSDU resides at the
top of the MAC and aggregates multiple MSDUs
into a single MPDU. Each MSDU is prepended
with a subframe header consisting of the destination
address, source address, and a length field
giving the length of the SDU in bytes. This is
then padded with 0 to 3 bytes to round the subframe
to a 32-bit word boundary. Multiple such
subframes are concatenated together to form a
single MPDU. An advantage of A-MSDU is that
it can be implemented in software. A-MPDU
resides at the bottom of the MAC and aggregates
multiple MPDUs. Each MPDU is prepended
with a header consisting of a length field,
8-bit CRC, and 8-bit signature field. These subframes
are similarly padded to 32-bit word
boundaries. Each subframe is concatenated
together. An advantage of A-MPDU is that if an
individual MPDU is corrupt, the receiver can
scan forward to the next MPDU by detecting the
signature field in the header of the next MPDU.
With A-MSDU, any bit error causes all the
aggregates to fail.
MAC throughput with frame aggregation
increases linearly with PHY data rate with traffic
conducive to aggregation, as illustrated in Fig. 4.
With a PHY data rate of 600 Mb/s, a MAC
throughput of over 400 Mb/s is now achievable
with 802.11n MAC enhancements.
When using block acknowledgment from
802.11e, a station transmits a burst of packets
before receiving an acknowledgment. A simple
increase in efficiency when not employing frame
IEEE Communications Magazine • July 2008 53
n Figure 3. Summary of 802.11n MAC enhancements. Reproduced with permission
from [5].
Management
plane
Data plane Control plane
Aggregation
Throughput and robustness
Enhanced block
Ack Capability
management
20/40 MHz BSS
40 MHz coexistence
Channel switching
Neighboring BSS
signaling
(Optional)
Protection
Phased coexistence operation (PCO)
Low power (handhelds)
Power save multi-poll (PSMP)
RIFS burst Reverse direction
protocol
Fast link
adaptation
TxBF control
n Figure 4. Throughput Vs PHY data rate with frame aggregation. Reproduced
with permission from [5].
PHY rate (Mb/s)
0
50.0
Throughput (Mb/s)
0.0
100.0
150.0
200.0
250.0
300.0
350.0
400.0
100 200
Legacy
20 MHz, 1 x 1
40 MHz, 1 x 1
40 MHz, 2 x 2
40 MHz, 3 x 3
40 MHz, 4 x 4
20 MHz, 2 x 2
20 MHz, 3 x 3
20 MHz, 4 x 4
300 400 500 600
Legacy
20 MHz, 1 x 1
20 MHz, 2 x 2
20 MHz, 3 x 3
20 MHz, 4 x 4
40 MHz, 1 x 1
40 MHz, 2 x 2
40 MHz, 3 x 3
40 MHz, 4 x 4
54 IEEE Communications Magazine • July 2008
aggregation is reducing the interframe spacing
(RIFS) between packets, which is possible since
the station no longer requires additional time to
switch between transmit and receive states.
Additional enhancements to the block acknowledgment
(BA) mechanism in 802.11e include
compressing the BA frame by eliminating support
for fragmentation. The reverse direction
protocol was incorporated, which allows a station
to share its transmit opportunity (TXOP)
with another station. This increases throughput
with traffic patterns that are highly asymmetric,
for example, when transferring a large file with
FTP operating over TCP. Time is borrowed during
the TXOP to send the short TCP Acknowledgment
in the reserve direction. Depending on
the usage model, TCP traffic throughput may
improve up to 40 percent.
Many new methods of control and management
were added to 802.11n, as illustrated in
Fig. 3. In order to more rapidly track changes in
the channel, fast link adaptation assists in the
selection of the optimal modulation and coding
scheme (MCS). Transmit beamforming may be
considered a PHY technique, but it requires a
great deal of control in the MAC for channel
sounding, calibration, and the exchange of channel
state information or beamforming weights.
Protection mechanisms had to be devised to
ensure that legacy 802.11a/g devices are not
harmed by the new modes of operation and vice
versa. These new modes, which may require protection,
include RIFS bursting and greenfield
format transmissions.
With the introduction of the 40 MHz bandwidth
channel came the complexity of managing
coexistence between 40 MHz bandwidth 802.11n
devices and 20 MHz bandwidth 802.11n and
802.11a/g devices. This becomes especially difficult
when operating in the 2.4 GHz band where
the channel numbering is incremented by 5
MHz, causing complicated partial overlapping
channel conditions between neighboring APs.
Rules were put in place mandating that an AP
scan for neighboring basic service sets (BSSs)
prior to establishing a 40 MHz BSS and preventing
the establishment of a 40 MHz BSS when
neighboring BSSs are detected in overlapping
channels. Furthermore, during the operation of
a 40 MHz BSS in the 2.4 GHz band, active 40
MHz bandwidth stations must periodically scan
overlapping channels. If conditions change disallowing
40 MHz operation (i.e., a new 20 MHz
BSS appears in an overlapping channel), the AP
must switch the BSS to 20 MHz bandwidth channel
operation.
With the increased interest in Wi-Fi enabled
handheld devices, Power Save Multi-Poll
(PSMP) was incorporated in the 802.11n MAC
to provide a minor improvement of channel utilization
and reduction in power consumption
when transmitting and receiving small amounts
of data periodically. These conditions arise with
multiple voice over IP (VoIP) sessions in the
BSS. Downlink transmissions are grouped
together, and uplink transmissions are scheduled.
The schedule for the downlink transmission
is provided at the start of the PSMP phase,
which allows for devices to power down their
receivers until needed.
SUMMARY AND LESSONS LEARNED
As indicated by the rate of Wi-Fi certification of
new wireless products, 802.11n is showing the
beginnings of being a resounding market success.
Over 100 devices were certified in the first few
months, three times as many as with 802.11b,
802.11a, or 802.11g. Consider that just 10 years
ago data rates were on the order of just a few
megabits per second. Now products are available
to the consumer capable of hundreds of megabits
per second and able to support wireless video
(e.g., two 20 Mb/s HTDV streams between adjacent
rooms). These devices include the latest
advances in wireless networking technology,
including MIMO, frame aggregation, and 20/40
MHz bandwidth channels.
That said, initial expectations in HT SG were
for a completed standard amendment a few
years ago. Fundamentally we need to avoid
lengthy adversarial processes. The participants in
IEEE 802.15.3a went through years of fighting
and political maneuvering before disbanding
without completing a standard. Fortunately, in
802.11n the technical issues did not cause as
wide a divide, and a few companies from both
sides of the fence were able to come together
and form EWC. This led to a broadly accepted
compromise and ended the contentious proposal
process. Mergers outside the often politically
charged standards body should be encouraged
earlier in the proposal process.
With a call for proposals and down selection
approach, changes need to be made to the proposal
process to guarantee conclusion and disallow
endless loops. As described, a possible
approach to speeding up the process is to reduce
the number of down selection votes and eliminate
the confirmation vote.
When initially proposed, many of the new
features in 802.11n were no more than new
research ideas, such as PSMP or coexistence of
20 and 40 MHz bandwidth channel devices. On
initial adoption there were no presentations containing
simulation results of these features within
an 802.11 system. Proposed features should be
more mature before being adopted into a draft
of the standard to shorten the standardization
process. This avoids continual improvements to
the feature during the development of the draft.
For the time to develop a standards amendment
to meet market needs, the scope of an
amendment should be narrowed. 802.11n tried
to address a wide scope of diverse environments.
For example, to keep pace with the increasing
data rates of Ethernet requires much higher data
rates emphasizing features that provide high
throughput, whereas serving VoIP handheld
devices in an outdoor Wi-Fi hotspot requires
features with a focus on low power and small
form factor. As witnessed, each new feature, be
it mandatory or optional (and especially new
research ideas), results in hundreds of letter ballot
comments that lengthen the letter ballot and
comment resolution phase. The number of new
features in an amendment should be limited
within a narrow scope to shorten the time to
market.
The 802.11 WG has formed a new study
group to investigate “very high throughput”
The 802.11 WG has
formed a new study
group to investigate
“very high
throughput”
potentially providing
throughput in the
order of giga-bits per
second. Hopefully
the lessons learned
from 802.11n
will result in
improvements of the
processes of future
task groups.
IEEE Communications Magazine • July 2008 55
potentially providing throughput on the order of
gigabits per second. Hopefully the lessons
learned from 802.11n will result in improvements
in the processes of future task groups.