e market. The first ex-
tension project, IEEE 802.11a, started in September 1997.
It added an OFDM(Orthogonal Frequency Division Mul-
tiplexing) PHY that supports up to 54 Mb/s data rate.
Since IEEE 802.11 a operates in the 5 GHz band, com-
munication with plain IEEE 802.11 devices is impossible.
This lack of interoperability led to the formation of IEEE
802.11 g, which introduced the benefits of OFDM to the
2.4 GHz band. As IEEE 802.11 g’s extended rate PHY
provides DSSS-compatible signaling, an easy migration
from IEEE 802.11 to IEEE 802.11 g devices became
possible. While IEEE 802.11b uses only DSSS technol-
ogy, IEEE 802.11g uses DSSS, OFDM, or both at the 2.4
GHz ISM band to provide high data rates of up to 54
Mb/s. combined use of both DSSS and OFDM is
achieved through the provision of four different physical
layers. These layers coexist during a frame exchange, so
the sender and receiver have the option to select and use
one of the four layers as long as they both support it. The
four different physical layers defined in the IEEE 82.11g
standard are ERP-DSSS/CCK, ERP-OFDM, ERP-DSSS/
PBCC and DSSS-OFDM. From the above four physical
layers, the first two are mandatory and the other two are
optional [1].
As the first project whose targeted data rate is meas-
ured on top of the MAC layer, IEEE 802.11n provides
user experiences comparable to the well known Fast
Ethernet (IEEE 802.3u). Far beyond the minimum re-
quirements that were derived from its wired paragon’s
maximum data rate of 100 Mb/s, IEEE 802.11n delivers
up to 600 Mb/s. Its most prominent feature is MIMO
(Multiple-Input Multiple-Output) capability. A flexible
MIMO concept allows for arrays of up to four antennas
that enable spatial multiplexing or beam forming. Its
most debated innovation is the usage of optional 40 MHz
channels. Although this feature was already being used
as a proprietary extension to IEEE 802.11a and IEEE
802.11g chipsets, it caused an extensive discussion on
neighbor friendly behavior. Especially for the 2.4 GHz
band, concerns were raised that 40 MHz operation would
severely affect the performance of existing IEEE 802.11,
Bluetooth (IEEE 802.15.1), ZigBee(IEEE 802.15.4), and
other devices. The development of a compromise, which
disallows 40 MHz canalizations for devices that cannot
detect 20 MHz-only devices, prevented ratification of
IEEE 802.11n until September 2009. As a consequence
of 20/40 MHz operation and various antenna configura-
tions, IEEE 802.11n defines a total of 76 different MCSs.
Figure 1 provides an overview of the IEEE 802.11 PHY
amendments and their dependencies [2].
Copyright © 2013 SciRes. CN
H. C. LEE
12
A key element to the IEEE 802.11 success is its simple
MAC operation based on the DCF protocol. This scheme
has proven to be robust and adaptive to varying condi-
tions, able to cover most needs sufficiently well. Follow-
ing the trends visible from the wired Ethernet, IEEE
802.11’s success is mainly based on over provisioning of
its capacity. The available data rate was sufficient to
cover the original best effort applications, so complex
resource scheduling and management algorithms were
unnecessary. However, this may change in the future.
Because of the growing popularity of IEEE 802.11,
Wireless LANs are expected to reach their capacity lim-
its. Moreover, applications like voice and video stream-
ing pose different demands for quality of service. There-
fore, traffic differentiation and network management
might become inevitable. Figure 2 shows IEEE 802.11
MAC layer amendments
3. FER Analysis
3.1. FER of Fixed Wireless Channel
In IEEE 802.11a/g wireless LAN, fixed wireless channel
isassumed to be Rayleigh fading channel. The probability
of bit error is upper bound by
Figure 1. The IEEE 802.11 PHY layer amendments and
their dependenc ie s[2].
Figure 2. The IEEE 802.11 MAC layer amendments[2].
1
free
b
dd
PB
k
dd
P (1)
where
f
ree
d
B is the free distance of the convolutional
code, d is the total number of information bit ones on
all weight d paths, d is the probability of selecting a
weight d output sequence as the transmitted code se-
quence, and k is the number of information bits per clock
cycle. Because the weight structure is generally accepted
that the first five terms in equation (1) dominate, equa-
tion (1) can be rewritten as
P
4
1free
free
d
b
dd
PB
k
dd
P (2)
The probability of selecting the incorrect path when d
is odd.

1
2
1
ddi
i
dd
i
d
Pp
i

p
(3)
where p is the probability of channel bit error. The prob-
ability of selecting the incorrect path when d is even.
 
2
2
1
2
1
11
22
dd
di d
i
dd
i
d
d
Ppp p
d
i


 


p
(4)
To achieve data rates of 54 Mbps for wireless access,
the IEEE 802.11 a standard utilizes MQAM (6q
,
64M
) with convolutional coding at rate r = 3/4. We
obtain the approximate channel bit error probability for
the sub-channel for MQAM with a square constella-
tion as
th
i






3
3211
2
3
2
3211
2
1
41
3211
221 1
1
21
311
11
bi
ibI
i
i
bi
bI
i
i
qr
dqr M
id
bI
I
qr
dqr M
bI
I
e
M
pqr M
qc M
e
M
qr M
cq M






 
 




 





 





 

(5)
where 22.6 0.1c
is empirically obtained and d = 1
for HDD. i
is the ratio of direct-to-diffuse signal
power on the sub-channel.
th
i
has 0 in a pure
Rayleigh fading channel and ranges from 0 to 10 in a
composite Rayleigh/Ricean fading channel. bi
is the
ratio of received average energy per bit-to-noise power
spectral density on the sub-channel. The overall
p is
the average of the probability of bit error on each of the
N OFDM sub-channels [5, 6].
th
i
Copyright © 2013 SciRes. CN
H. C. LEE 13
1
1N
i
i
p
N
p (6)
Note that for either no channel fading or for all sub-
channels experiencing the same fading (that is, i

and for all
i
bb ), then i.
i

pp/N
bbo
is
the ratio of received average energy per bit-to-noise
power spectral density ,
E
is the ratio of direct-to-dif-
fuse signal power. Now, using equation (6) in equation (3)
or (4) and taking the result into equation (2), we obtain
the performance of 64 QAM with HDD over Ricean
fading channels. For basic access mechanism, a data
packet including the PHY header and the MAC header
needs retransmission if any one bit of them is corrupted.
We define a variable which is the probability that a
backoff occurs in a station due to bit errors in packets.
We further assume that bit errors randomly appear in the
packets. So frame error rate is represented by (7).
P
c
1(1 )
p
reamble ACKLPHYMACPL
hh
PP
cb 
  (7)
CSMA/CA is also used as the MAC scheme in IEEE
802.11n wireless LAN, and it has basic and RTS/CTS
access scheme. Although there is a successful RTS/CTS
transmission in the time slot, a frame has to be retrans-
mitted when there is a bit error in a payload. For conven-
ience, we define a variable which is the probability
that a backoff occurs in a station due to bit errors in
packets. We further assume that bit errors randomly ap-
pear in the packets and A-MSDU scheme is used. So
frame error rate is represented by (8).
P
e
1(1 )
L
P
e q (8)
where L is the aggregated MAC frame’s size. For a con-
volutional code with a coding rate kc/nc, the bit error rate,
denoted as q, can be approximated by

1
2
1()1( is odd)
free free
free
ddi
free i
bb free
d
ci
d
qqqd
i
k




(9)

()1
1
1
2
122
()(1)
22
( is even)
free free
free bb
cfree
free free
free
bb
free
c
free
ddi
di
qq
i
qkd
i
dd
dqq
d
k
d









where dfree is the maximum free distance of the convolu-
tional code and qb is the probability of a bit error for the
M-QAM[5].
2( 1)
s
b
M
q
qM
(10)
qs is the SER(Symbol Error Rate) under the Rician fading
channel.
2
min 2
2
min
()
(|| ||)
8
()
()
1
44 8
2
min
1
()
()
18
dH
d
K
sK
qe
d
K


(11)
K is the Rician factor may be interpreted as the average
SNR at the receive antenna in a SISO fading link. dmin is
the minimum distance of separation of the underlying
scalar constellation. H is MR MT channel transfer func-
tion and 2
|| ||
H
is the squared Frobenius norm of the
channel [6, 7].
3.2. FER of Mobile Wireless Channel
Mobile wireless channel is assumed to be flat fading
Rayleigh channel with Jake spectrum. The channel is in
fading states or inter-fading states by evaluating a certain
threshold value of received signal power level. If and
only if the whole frame is in inter-fading state, there is
the successful frame transmission. If any part of frame is
in fading duration, the frame is received in error. In the
fading channel fading margin is considered and defined
as ρ = Rreq/Rrms, Where Rreq is the required received
power level and Rrms is the mean received power. Gener-
ally, the fading duration and inter-fading duration can be
taken to be exponentially distributed for ρ<-10dB. With
the above assumptions, let be the frame duration,
then the frame error rate is given by (12).
Tpi
1(
f
i
Ti
FERP tTpi
Ti T
 
)
(12)
where, is inter-fading duration and
it
f
t is fading du-
ration. is the mean value of the random variable
and
Ti it
f
T is the mean value of the random variable
f
t.
is the probability that inter-fading duration
lasts longer than . Since exponential distribution is
()Pti Tpi
Tpi
assumed for ,
it()exp(iTpi
Pt TpiTi
)
. For Rayleigh fading
channel, the average fading duration is given by (13).
exp()1
2
Ti fd
(13)
iTTf
is 1
f
N, where
f
N is the level crossing rate,
which is given by 2exp( )fd

. is the maxi- df
mum Doppler frequency and evaluated as
.
is the
mobile speed and
is wavelength. Frame error rate
can be expressed by (14).
1 exp(2)d
F
ERf Tpi

  (14)
Copyright © 2013 SciRes. CN
H. C. LEE
14
Equation (14) shows that frame error rate is deter-
mined by fading margin, maximum Doppler frequency
and frame duration. Since fading margin and maximum
Doppler frequency are hard to dynamically control, the
only controllable parameter is frame duration to get re-
quired frame error rate. For the RTS/CTS access mode,
the frame duration
p
iTis
H
RTS CTSDATA ACK
TTTT T
 .
H
T is preamble transmission time + PLCP header trans-
mission time + MAC header transmission time.
D
ATAT
ACKT is
MSDU transmission time and is ACK frame
transmission time.
R
TS is RTS frame transmission time
and is CTS frame transmission time[6,7].
T
CTS
T
4. Numerical Results of FER over the Fading
Channel
4.1. FER Results with Fixed Stations
In the Figure 3, Pc(P, b
, K) shows FER(Frame Error
Rate) due to b
, the ratio of received average energy per
bit- to-noise power spectral density[6,7]. K means Rician
factor and P means payload size. And as expected, the
FER performance improves with K and the smaller pay-
load size is, the better performance is.
In the Figure 4, qs(ρ,K) shows SER(Symbol Error
Rate) and Pe(K,ρ,ns,P) shows FER(Frame Error Rate)
[6,7]. K means Rician factor and as expected, the FER
performance improves with K and the smaller subframe’
payload size is, the better performance is.
(a) IEEE 802.11a OFDM
(b) IEEE 802.11g ERP-OFDM
(c) 802.11g DSSS-OFDM
Figure 3. Frame error rate of IEEE 802.11a/g fixed LAN
over Rayleigh fading channel.
(a) SER
(b) FER
Figure 4. SER and FER of IEEE 802.11n fixed LAN over
Rician fading channel.
4.2. FER Results with Mobile Stations
In the Figures 5(a)-(c), the symbol fer (,
, P) shows
frame error rate of IEEE 802.11a/g. In the Figure 5(d),
the symbol fer (ns, ,
, P) shows frame error rate of
IEEE 802.11n with the horizontal parameter of sub-
frame’ payload size. In the Figure 5(e), the symbol fer (,
ns,
, P) shows frame error rate of IEEE 802.11n using
the number of subframes as the horizontal parameter. It
is generally identified that the higher mobile speed is, the
C
opyright © 2013 SciRes. CN
H. C. LEE
Copyright © 2013 SciRes. CN
15
higher FER is. In case of payload size, the same result
mentioned above is also acquired.
(e) IEEE 802.11n OFDM (58.5 Mbps, number of subframe)
Figure 5. Frame error rate of IEEE 802.11a/g/n mobile LAN.
5. Remarks
(a) IEEE 802.11a OFDM (54 Mbps)
This paper explored the FER performance of MAC layer
in the IEEE 802.11a/g/n wireless LAN under the er-
ror-prone channel. The fixed wireless channel was as-
sumed to be Rayleigh fading channel and the mobile
wireless channel was assumed to be flat fading Rayleigh
channel with Jake spectrum. The MAC protocol that they
are based upon is the same and employs a CSMA/CA
protocol with binary exponential back-off. IEEE 802.11
DCF is the de facto MAC protocol for wireless LAN
because of its simplicity and robustness.
REFERENCES
(b) IEEE 802.11g ERP-OFDM (54 Mbps)
[1] D. Vassis, G. Kormentzas, A. Rouskas and L. Maglo-
giannis, “The IEEE 802.11g Standard for High data rate
WLANs,” IEEE Network, 2005, pp. 21-26.
doi:10.1109/MNET.2005.1453395
[2] G. R. Hiertz, D. Denteneer, L. Stibor, Y. P. Zang, X. P.
Costa and B. Walke, “The IEEE 802.11 Universe,”
IEEE Communications Magazine, Jan. 2012, pp.
62-70.
[3] Y. Lin and V. W. S. Wong, “Frame Aggregation and
Optimal Frame Size Adaptation for IEEE 802.11n
WLANs,” in Proceedings of IEEE GLOBECOM, San
Francisco, CA, Nov. 2006
[4] D. Skordoulis, Q. Ni, H. Chen, A. P. Stephens, C. Liu and
A. Jamalipour, “IEEE 802.11n MAC Frame Aggregation
Mechanisms for Next-Generation High-Throughput
WLANs,” IEEE Wireless Communications, Vol. 15, Feb.
2008, pp. 40-47.
doi:10.1109/MWC.2008.4454703
(c) IEEE 802.11g DSSS-OFDM (54 Mbps)
[5] C.-H. Kao, “Performance of the IEEE 802.11a Wireless
LAN Standard over Frequency-selective, Slow, Ricean
Fading Channels,” Master’s Thesis, Sep. 2002.
[6] H. C. Lee, “A MAC Throughput in the Wireless LAN,”
Advanced Wireless LAN, Intech, 2012, pp. 23-62.
[7] H. C. Lee, “A MAC Throughput over Rayleigh Fading
Channel in The 802.11a/g/n-based Mobile LAN,” MESH
2011, Aug. 2011.
(d) IEEE 802.11n OFDM (58.5 Mbps, Payload size)