Computer Communications 32 (2009) 159–168
Contents lists available at ScienceDirect
Computer Communications
journal homepage: www.elsevier.com/locate/comcom
An efficient initial ranging algorithm for WiMAX (802.16e) OFDMA q
Hisham A. Mahmoud a,*, Huseyin Arslan a, Mehmet Kemal Ozdemir b
a
b
Electrical Engineering Department, University of South Florida, 4202 E. Fowler Ave., ENB-118, Tampa, FL 33620, USA
Logus Broadband Wireless Solutions, 3803 Corporex Park Pl. Suite 700, Tampa, FL 33619, USA
a r t i c l e
i n f o
Article history:
Received 17 March 2008
Received in revised form 23 September
2008
Accepted 25 September 2008
Available online 10 October 2008
Keywords:
Initial ranging
OFDM
OFDMA
WiMAX
Synchronization
a b s t r a c t
Ranging is one of the most important processes in the mobile WiMAX standard. Power adjustment, timing offset estimation, and synchronization between a base station (BS) and all users within a cell are done
through the ranging process referred to as initial ranging. In this paper, we discuss the details of initial
ranging together with some of the proposed algorithms, as well as a novel algorithm to carry out a successful ranging process. Performance curves and computational complexity comparisons are presented.
The system performance is evaluated for both additive white Gaussian noise (AWGN) channels and practical multipath fading channels with multiuser interference. It is shown that the proposed algorithm
offers a reduced complexity ranging method that can be employed in practical WiMAX-based BSs.
Ó 2008 Elsevier B.V. All rights reserved.
1. Introduction
WiMAX, as a relatively new technology, has received the attention of researchers and wireless companies. This new wireless
technology promises to deliver both high data rates and long-range
coverage. With the approval of the mobile WiMAX standard
(IEEE802.16e-2005) at the beginning of the year 2006, this technology became even more exciting. Unlike WiFi [1,2], which is designed for indoor applications and wireless local area network
(WLAN), WiMAX is optimized for outdoor applications and wireless metropolitan area network (WMAN).
One of the exciting aspects of WiMAX is that its medium access
control (MAC) layer supports more than one physical layer (PHY)
mode [3]. This feature not only enables companies to differentiate
their products from each other, but also makes WiMAX an adaptive
technology that can satisfy different needs depending on the application. One of the most promising PHY modes supported by WiMAX standard is orthogonal frequency division multiple access
(OFDMA) PHY mode which enables a WiMAX base station (BS) to
support multiple fixed or mobile users at the same time. In this
mode, a BS system utilizes the channel by dividing available subcarriers into subchannels that can be assigned to multiple users
in a sophisticated and adaptive way. As a matter of fact, users
q
This work has been partially presented in the IEEE Military Communications
Conference, MILCOM’06, October 23–25, 2006, Washington, USA.
* Corresponding author. Tel.: +1 813 974 5005.
E-mail addresses: hmahmoud@mail.usf.edu (H.A. Mahmoud), arslan@eng.usf.
edu (H. Arslan), kemal@loguswireless.com (M.K. Ozdemir).
0140-3664/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.comcom.2008.09.032
can be assigned to different bandwidths, different time durations,
and different modulation orders based on various parameters such
as user carrier-to-interference-plus-noise ratio (CINR) and the
available bandwidth.
One of the problems that faces WiMAX systems, or OFDMA systems in general, is timing and synchronization between a BS and
subscriber stations (SS) within the cell. While other orthogonal frequency division multiplexing (OFDM) receivers can easily synchronize to the received signal [4,5], this is not the case for OFDMA
receivers. At an OFDMA receiver, where multiple users arrive at
the same time, if users are not synchronized with the receiver, they
will interfere with each other, and therefore the BS will not be able
to recover individual signals of each user. Hence, for OFDMA PHY
mode to work properly, all users should arrive at the BS at the
same time with a considerably high timing accuracy. This can be
achieved if all users are synchronized with the BS before the communication link is established. The standard states that for a user
to join the channel, first, the round trip delay (RTD) between the
SS and the BS must be known to the SS [3]. This delay estimation
is used by the SS to synchronize its signal such that it arrives at
the BS in its allocated time. The process in which this delay is estimated is called initial ranging, and this process is mandated for all
SSs that desire to synchronize to the channel initially. Other types
of ranging processes (e.g. periodic ranging and bandwidth request)
exist, but only initial ranging is considered in this paper.
As stated in the IEEE802.16e-2005 standard [3], the MAC layer
at the BS defines a ranging channel as a group of six (or more) subchannels, where a subchannel is a group of subcarriers that are
chosen according to a randomization formula. In addition, users
160
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
are allowed to collide in this ranging channel. Any SS that attempts
to establish a communication link is required to carry out a successful initial ranging process with the BS over the ranging channel. Once a SS senses a BS, for network entry, it first scans for a
downlink (DL) channel and synchronizes itself with the BS. Then,
the SS shall acquire transmit parameters, which are included in
the uplink channel descriptor (UCD), uplink (UL)-MAP, and DLMAP. Using acquired parameters, the SS initiates the initial ranging
process by sending a ranging code over the UL frame. At the receiver side, the BS is required to detect different received ranging
codes and estimate the timing offset and the power for each user
that bears an initial ranging code. The BS then broadcasts the detected ranging codes with adjustment instructions for the timing
and power level. The status notifications of either a successful
ranging process or retransmission are also broadcasted.
In initial ranging, the SS chooses one of the available ranging
codes randomly and transmits it twice over two consecutive OFDM
symbols with BPSK modulation. The SS should transmit the ranging code during the UL frame as long as there is a ranging opportunity. UL-MAP shows if a ranging opportunity is available through
the next UL frame. Another option is to send two consecutive ranging codes over four OFDM symbols to increase the probability of
code detection [3]. In this paper, ranging over two symbols is
considered.
Synchronization for multiuser OFDM systems has been discussed in the literature. The use of filters matched to the intended
user’s subcarriers (in our case, the ranging channel subcarriers)
and then use of cyclic prefix (CP) redundancy to estimate the timing
offset is proposed in [5]. However, in WiMAX standard, the ranging
subcarriers are not necessarily adjacent, which makes the filtering
process inapplicable. In [6], it is proposed to synchronize users to
the BS one at a time, with the assumption that other users are already synchronized. This method cannot be used for OFDMA with
multiple users colliding in the ranging channel. Finally, in [7,8] it
is proposed to use a bank of correlators (corresponding to number
of ranging codes) in time or in frequency domain to detect received
ranging codes. The disadvantage of this approach is that the computational complexity increases as the number of codes increases.
In this paper, a new algorithm for OFDMA initial ranging is proposed. The proposed algorithm is examined using theoretical analysis and computer simulations over additive white Gaussian noise
(AWGN) and dispersive channels in the presence of multiuser
interference. A performance and complexity comparison between
the proposed algorithm and prior algorithms in practical system
conditions are presented. It is demonstrated that the proposed
algorithm offers a significant reduction in computational complexity while carrying out a successful initial ranging process. The
reduction in complexity becomes especially important for certain
practical WiMAX implementations. For example, the proposed
algorithm can be attractive to low-cost OFDMA-based femtocell
BS implementations where the number of users per BS is limited.
The remainder of this paper is organized as follows. The system
model is introduced in Section 2. Current and proposed ranging
algorithms are discussed in detail in Sections 3 and 4, respectively.
In Section 5, the computational complexity of the proposed algorithm is calculated and compared to current ranging algorithms.
Simulation results and discussions demonstrating the performance
of the proposed ranging algorithm compared to other algorithms
are presented in Section 6. Finally, the conclusions are outlined
in Section 7.
After assigning DC and guard subcarriers, the remaining subcarriers, N d , are grouped into Q subchannels. Each subchannel has
N Q ¼ N d =Q subcarriers, where Q is chosen such that N d is an integer multiple of Q. Each user in the UL is assigned one or more subchannels. The BS defines a group of six subchannels (or more) for
ranging. Note that the subcarriers assigned to each subchannel
are chosen randomly and thus they are not necessarily adjacent.
The BS broadcasts all the ranging information (i.e. ranging opportunities, ranging channels, ranging codes and so on) in the UL-MAP.
One ranging time slot spans two OFDMA symbol duration. The
kth ranging user signal in frequency domain is denoted as
ðkÞ
ðkÞ
ðkÞ
ðkÞ
cp ¼ ½cp ð1Þ; cp ð2Þ; . . . ; cp ðLÞT , where p is the index of the randomly chosen ranging code and L is the size of the ranging code.
The signal is then extended to N t by inserting N t L zeros, which
T
ðkÞ
ðkÞ
results in XpðkÞ ¼ ½X ðkÞ
p ð1Þ; X p ð2Þ; . . . ; X p ðN t Þ . Note that
X ðkÞ
p ðmÞ
¼
(
ðkÞ
cp ðnÞ if m ¼ ir ðnÞ;
0
otherwise;
ð1Þ
where ir ðnÞ is the index of the nth subcarrier within the ranging
channel subcarriers set ir ¼ ½ir ð1Þ; ir ð2Þ; . . . ; ir ðLÞT . The vector XpðkÞ is
then fed to an Nt -point inverse discrete Fourier transform (IDFT).
The resulting signal in time domain is extended over two OFDMA
ðkÞ
symbols by repeating xp twice and adding the cyclic prefix with
ðkÞ
no phase discontinuity, where xp is the time representation of
ðkÞ
Xp . Note that the BS receiver uses an observation window of
ðN t þ N CP Þ to acquire OFDMA symbols, where N CP is the size of the
CP.
Finally,
the
transmitted
signal
is
denoted
as
ðkÞ
ðkÞ
ðkÞ
ðkÞ
sp ¼ ½sp ð1Þ; sp ð2Þ; . . . ; sp ð2Nt þ 2N CP ÞT , where
ðkÞ
ðkÞ
sðkÞ
p ¼ ½xp ðN t N CP þ 1Þ; . . . ; xp ðN t Þ;
ðkÞ
xðkÞ
p ð1Þ; . . . ; xp ðN t Þ;
ð2Þ
ðkÞ
xðkÞ
p ð1Þ; . . . ; xp ðN t Þ;
T
ðkÞ
xðkÞ
p ð1Þ; . . . ; xp ðN CP Þ :
ðkÞ
The transmitted signal sp is then received by the BS after being corrupted by the communication channel. In the following analysis, we
assume that the channel is non-dispersive and that the received signals from multiple users are only affected by complex AWGN. However, we will evaluate the proposed ranging algorithm performance
for both AWGN and dispersive channels. A similar approach has
been used in [9] to analyze the proposed algorithm.
All users other than the users performing initial ranging are assumed to be already synchronized to the BS. This is a valid assumption as synchronizing to the BS is mandatory before a SS can
establish the communication link. Hence, it is guaranteed that
there is no interference from synchronized user signals to the ranging channel. Note that this is not the case for ranging users as their
unsynchronized signals can cause interference to synchronized
users. As this situation is unavoidable, initial ranging SS is required
by the standard to start the ranging process with minimum possible power level. Then, as long as the SS fails to get a response from
the BS, the power is increased incrementally until a response is detected. If the maximum power level is reached and the SS still cannot get a response from the BS, the user starts from the minimum
power level and the process is repeated. This shows how important
it is for a BS to detect ranging users with the lowest signal levels
possible.
3. Existing ranging algorithms
2. System model
The system model is based on the IEEE802.16e-2005 standard
[3]. The UL of an OFDMA system with Nt subcarriers is considered.
To detect ranging codes at the BS, one approach would be to
cross-correlate the received signal with all possible ranging codes
in time domain [7]. To reduce the high computational complexity
of this process, one can instead auto-correlate the received signal
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
with its delayed replica to exploit the repetition in the ranging
code. However, for this approach to work properly, the system
needs to extract other non-ranging user signals from the received
signal as the ranging users are frequency-multiplexed with those
synchronized users. Not only does such a process increase the complexity and delay of the algorithm, but it is also affected by the performance of non-ranging users signal estimator. In addition, the
codes used for ranging are modulated in frequency domain and
performing the correlation in time domain weakens the auto-correlation/cross-correlation properties of used codes.
Another approach to detect ranging codes is to perform the
cross-correlation on the frequency-domain signal at the output of
the discrete Fourier transform, (DFT) [8]. In this case, a complete
OFDMA ranging symbol in the observation window results in a correct ranging code in the frequency domain even if a timing offset
exists. The effect of the timing offset is translated into a linear
phase shift (also called a phase rotation) in the frequency domain.
To estimate the timing offset of the ranging code, the system applies all possible linear phase shifts (corresponding to possible timing offsets) to the signal. Then, the received signal is correlated
with all ranging codes. A threshold is set to detect the existence
of a ranging code in the current observation window and its timing
offset. One advantage of performing the cross-correlation process
in frequency domain is that there is no interference from nonranging users to the ranging channel. Thus, the probability of a
missed detection or a false alarm – due to multiuser interference
– is reduced. Another advantage of this approach is that operating
in frequency domain utilizes the auto-correlation/cross-correlation
properties of used ranging codes.
The main disadvantage of previously mentioned algorithms is
the high computational complexity. If the total number of codes
is K and the maximum RTD considered has an integer value of
smax samples, then K smax cross-correlation operations are needed
for every OFDMA symbol. In the wireless IEEE standard 802.16a
[10] and 802.16a/b [11] considered in [7] and in [8], respectively,
the number of long ranging codes used for initial ranging is only
16. However, for IEEE802.16e [3], the number of ranging codes is
256. These codes are divided into three categories: initial ranging
codes, periodic ranging codes, and bandwidth-request codes. Initially, the BS is expected to assign more codes for initial ranging
as users within the cell start entering the network. If we assume
that 128 codes would be assigned for initial ranging, then for OFDMA systems based on IEEE802.16e-2005, the computational complexity for the ranging process would be eight times as much as
the complexity of 802.16a/b. Other ranging algorithms were proposed in the literature (i.e. [12,13]). However, those algorithms
are not applicable to the current standard in its present state.
Hence, they are not considered in this work.
4. Proposed algorithm
In the proposed algorithm, we concentrate more on the trade off
between computational complexity and performance of the ranging process so that the algorithm can be realized in practical systems. We choose an observation window with one OFDMA
symbol size and apply the algorithm in frequency domain for the
advantages of this method that is mentioned earlier in Section 3.
However, instead of directly cross-correlating with every possible
code and every possible phase offset for every OFDMA symbol,
we break the initial ranging process into three main steps. The first
step is to find OFDMA symbols containing ranging codes. This step
allows the system to find out which symbols it should process further and which ones the system should just drop so that no additional computations are performed on empty OFDMA symbols.
Energy detectors, which are discussed in more detail in Section
161
4.1, are used to detect OFDMA symbols containing ranging codes
in the ranging channel. In the second step, the algorithm finds
how many codes there are within detected OFDMA symbols from
step one and determine the timing offsets (or linear phase shifts)
for each code. Again, in this step the system further reduces the
computational complexity by first finding the timing offset of the
codes before performing the cross-correlation with all possible
codes for every possible linear phase shift. Section 4.2 discusses
the details of this step. The last step in the algorithm is the identification of multiuser codes by cross-correlating detected codes
with all possible ranging codes after removal of any timing offsets.
This step is covered in Section 4.3. Using the proposed approach,
the computational complexity is greatly reduced while the performance is still acceptable as it is shown in Sections 5 and 6.
4.1. Energy detector
If there is a ranging opportunity in the next UL frame, the ranging channel will be available through the entire UL subframe duration. The BS samples the received signal and groups it into N t þ N CP
samples. The CP is removed and the remaining N t samples are fed
to the DFT unit. The ranging channel contains noise and energy
from ranging users. The received unsynchronized signal of a ranging user consists of two ranging symbols with no phase discontinuity. The OFDMA symbols processed by the BS can be only one of
three possible cases: (a) empty symbols containing only noise,
(b) symbols containing incomplete parts of the ranging signal,
which cause interference to subcarriers other than the ranging
subcarriers and thus interfere with synchronized users, and (c)
symbols that are entirely filled with the ranging signal, referred
to as symbols with complete ranging signal. We are interested in
detecting the third kind, as it contains the required information
to detect the user ranging code. For every ranging user signal, at
least one symbol with complete ranging signal should be received
by the BS since the ranging signal spans two OFDMA symbols.
Empty symbols should be ignored since they contain no information. The information included in OFDMA symbols with incomplete
ranging signal (i.e. the ranging code, its timing offset, and signal
power) can be extracted from the symbols with complete ranging
signal. Therefore, a missed detection of symbols with incomplete
signal should not affect the performance of the algorithm. In order
to detect symbols with a complete ranging signal we use an energy
detector in frequency domain. The energy detector measures the
energy within the ranging channel. This method has two advantages: (a) since the energy is measured in the frequency domain and
since the ranging subcarriers are not adjacent, the likelihood of a
pulse of noise triggering the energy detector by mistake is low,
(b) the energy is already measured to obtain the noise variance
of the channel and can also be used later to measure the signal
power of ranging users. Therefore, no additional computational
complexity is required for this step.
The measured energy in the ranging channel is
Eg ¼
L1
X
jYðmÞj2 ;
ð3Þ
n¼0
where m ¼ ir ðnÞ and Y is the N t vector at the output of the DFT unit
at the receiver side.
After measuring the energy within the ranging channel, a
threshold g1 is used to decide if the OFDMA symbol contains a
ranging code or not. To find the best value for g1 , the probability
of a false alarm P fa and the probability of missed detection Pmd
are calculated. The probability of a false alarm is defined as the
probability of noise energy in empty OFDMA symbols exceeding
g1 . In the same manner, the probability of missed detection is defined as the probability of the energy of an OFDMA symbol not
162
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
n¼0
W 2R ðmÞ þ
n¼0
L1
X
W 2I ðmÞ;
1
where erfc is the complementary error function. For an OFDMA
symbol containing a complete ranging signal, if a user k signal has
a timing offset of sk samples,1 then the OFDMA symbol with complete ranging signal contains a copy of the ranging code which is
cyclically shifted by sk samples. The OFDMA symbol in the frequency
domain has a linear phase shift of 2pnsk =Nt , where n is the subcarrier index. In this case,
ð6Þ
n¼0
where /m ðsk Þ ¼ 2pmsk =N t . Since BPSK modulation is used,
ðkÞ
cp ðmÞ ¼ 1. Thus,
Eg ¼
L1 n
X
1 þ W 2R ðmÞ þ W 2I ðmÞ þ 2cpðkÞ ðmÞ cos½/m ðsk ÞW R ðmÞ
n¼0
o
þ 2cðkÞ
p ðmÞ sin½/m ðsk ÞW I ðmÞ :
ð7Þ
Using the central limit theorem, the distribution of this energy can
be approximated as a normally distributed random variable with
mean l2 ¼ L þ LN 0 and variance r22 ¼ LN 20 þ 2LN 0 as shown in
Appendix A. Hence, the probability of a missed detection becomes
Pmd
0
1
Bg l C
¼ 1 0:5 erfc@q1ffiffiffiffiffiffiffiffiffiffiffiffi2 A:
2pr22
0.5
0.3
ð5Þ
2
L1
X
ðkÞ
cp ðmÞej/m ðsk Þ þ WðmÞ ;
0.6
0.4
n¼0
Bg l C
Pfa ¼ 0:5 erfc@q1ffiffiffiffiffiffiffiffiffiffiffiffi1 A;
2pr21
Eg ¼
0.7
ð4Þ
where W R ðmÞ and W I ðmÞ are the real and imaginary parts of WðmÞ,
respectively. The energy in this case is the sum of 2L samples of the
square of normally distributed random variables with zero mean
and N 0 =2 variance. Hence, the measured energy can be described
as a random variable with Chi-square distribution having a mean
l1 ¼ LN0 and variance r21 ¼ LN20 . Using the central limit theorem,
if 2L is large enough, the energy distribution can be approximated
as a normally distributed variable with the same mean and variance. Based on the standard [3], L or the ranging code length is
equal to 144 which is large enough to validate the above approximation. The probability of a false alarm then becomes [14]
0
0.8
0.2
0.1
0
0
5
10
15
20
25
30
Normalized Threshold η1 (2/LN 0 )
35
40
Fig. 1. P fa and P md for different noise levels.
Normalized measured energy
jWðmÞj2 ¼
L1
X
0.9
400
300
200
100
0
0
2
Symbols
Eg ¼
L1
X
1
Probability
exceeding g1 while containing a complete ranging signal. Note that
the case of an OFDMA symbol containing incomplete ranging signal is ignored in the calculation of P md as missing this symbol does
not affect the algorithm performance. In fact, detecting an incomplete ranging signal can additionally provide correct ranging information if the timing offset is relatively small. If the current OFDMA
symbol contains no ranging code, then YðmÞ ¼ WðmÞ, where W is a
vector of complex AWGN samples with zero mean and N0 =2 variance. From (3),
4
6
OFDMA symbol
8
S5
S4
S3
S3
S2
S1
0
2
10
12
S5
S4
S2
S1
4
6
Time
8
10
12
Fig. 2. Normalized measured energy on the ranging channel at SNR = 10 dB.
users are shown in time domain. Also, the figure shows the normalized measured energy on the ranging channel for every OFDMA
symbol and for an SNR of 10 dB. Note that in the above analysis, it
is assumed that only one complete ranging signal exists in the received symbol. This assumption is based on worst case scenario,
since P md gets even lower if more than one complete signal exists
in the received symbol as shown in Fig. 2.
4.2. Timing offset estimation
ð8Þ
Fig. 1 shows Pfa and P md against normalized g1 (normalized by L
and N 0 ) for different signal-to-noise ratio (SNR) levels. Note that P fa
does not change as the SNR changes since g1 is normalized and
since Pfa depends only on N 0 . In this work, N 0 is assumed to be
available for choosing the best value of g1 . This is a valid assumption since the BS needs to estimate the noise level for calculations
of different users SNR. An example of a received UL frame with
ranging codes is shown in Fig. 2. The received symbols of ranging
1
We only consider timing offset that is integer multiple of the sampling time. The
non-integer part of the delay is insignificant to the OFDM system performance and is
usually incorporated as part of the communication channel.
In the previous step, the system detects OFDMA symbols containing one or more complete ranging signals. The next step is to
identify how many codes are in each symbol and estimate the timing offset for each of these codes. Since the proposed algorithm is
applied in frequency domain, timing offset would be directly translated into a linear phase shift. In this case, what is estimated is
actually the linear phase shift for each code in the current OFDMA
symbol. This can be done by cross-correlating the ranging channel
of the current OFDMA symbol with all possible codes after applying all possible linear phase shifts to the symbol. The correlator
output is followed by a threshold detector to detect different ranging codes and their phase shifts within the current symbol. However, this is computationally complex. We intended to reduce the
complexity of this operation by exploiting the fact that timing off-
163
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
sets only affect the phase of the frequency domain signal. Since
BPSK modulation is used, the signal ideally should not have an
imaginary part. If there are K ranging users within the current OFDMA symbol, each user has a timing offset sk samples, where
0 < sk < smax and smax is the maximum RTD between the SS and
the BS within the current cell. Then
j/m ðsk Þ
cðkÞ
þ WðmÞ;
p ðmÞe
0.8
Normalized energy level
YðmÞ ¼
K1
X
1
ð9Þ
k¼0
where m ¼ ir ð0Þ; ir ð1Þ; . . . ; ir ðL 1Þ. As shown in (9), if there were
no timing offsets (i.e. sk ¼ 0 for k ¼ 0; 1; . . . ; K 1), the whole energy of the ranging user will be only in the real part of the signal
and the imaginary part will contain only noise. By applying all possible linear phase shifts and taking the energy of the real part of the
signal, we have
Er ðuÞ ¼
L1
X
n¼0
¼
k¼0
i2
^
cðkÞ
p ðmÞ cos½/m ðsk uÞ þ W R ðmÞ ;
~ r ðuÞ ¼ ½Er ðuÞ r =r ;
E
ð10Þ
ð11Þ
where
r ¼
1
K
Er ðuÞ:
ð12Þ
u¼0
As shown in Appendix B,
EfEr ðuÞg
LðK þ 1 þ N0 Þ=2 for u ¼ sk ;
LðK þ N0 Þ=2
ð13Þ
for u–sk :
From (12) and (13), we get
r ¼
0.2
0
−0.2
R YðmÞej/m ðuÞ
^
^ R ðmÞ ¼
where u ¼ 0; 1; . . . ; smax , WðmÞ
¼ WðmÞej/m ðuÞ and W
^
RfWðmÞg.
User timing offset sk , is modeled as a uniformly distributed random variable between 0 and smax . Since m is chosen randomly from the available subcarriers, the phase ½2pmðsk uÞ=N t
can be approximated as a uniformly distributed random variable
^ and W have the same distribution,
between p and p. Note that W
average and variance. For sk ¼ u, cos½/m ðsk uÞ ¼ 1 resulting in a
peak in the measured energy Er ðuÞ. Therefore, a threshold g2 is
set to detect those peaks and obtain an estimate of sk for all values
of k. The measured real energy is normalized with respect to the
average real energy of the symbol r such that
K1
X
0.4
2
"
L1 X
K1
X
n¼0
0.6
KEfEr ðuÞju ¼ sk g þ ðsmax KÞEfEr ðuÞju–sk g
smax
ð14Þ
The above equation is valid given that K < smax , which is a reasonable assumption since – depending on the cell radius – smax can go
up to N t =2, while the number of ranging users within the same OFDMA symbol is usually much lower than this value. Fig. 3 shows
e r ðuÞ j u–sk g for different values of K and
e r ðuÞ j u ¼ sk g and Ef E
Ef E
for SNR = 10 dB. For simulated results, we use the system setup presented in Section 6.1. Assuming all ranging users are received with
equal power, the signal-to-interference-plus noise ratio (SINR) levels for a given number of ranging users per OFDMA symbol and for
an SNR level of 10 dB are presented in Table 1, where SINR is defined as the ratio of the power of one user signal to the power of
the remaining users signals plus noise power. Fig. 3 shows that as
the number of ranging users increases, the probability of a missed
detection increases since the difference between the two means decreases. In addition, the variance of the measured energy also increases with the number of ranging users, since they act as
interference noise for u–sk as shown in (10). Note that since K is
2
3
4
5
6
7
8
Ranging users per OFDMA symbol
9
10
e r ðuÞ j u ¼ sk g and Ef E
e r ðuÞ j u–sk g for different numbers of ranging users.
Fig. 3. Ef E
Table 1
SINR vs number of users per OFDMA symbol.
Number SINR of users (dB)
Number SINR of users (dB)
1
2
3
4
5
6
7
8
9
10
10.0
0.4
3.2
4.9
6.1
7.1
7.9
8.5
9.1
9.6
usually very small compared to smax , then from (14),
e r ðuÞ j u–sk g 0, as shown in Fig. 3.
r EfEr ðuÞ j u–sk g and Ef E
4.3. Code detector
In the previous two steps of the proposed algorithm, OFDMA
symbols containing ranging codes are detected and their timing
offsets are estimated. The last stage is to identify which codes
are transmitted out of the available ranging codes, P, where P is
the number of codes assigned by the BS for initial ranging. In this
step, the linear phase shift corresponding to each detected ranging
code is removed and then a cross-correlation with all possible
ranging codes is performed. The correlator output is
EcðvÞ ðiÞ ¼
:
1
L1
X
R YðmÞej/m ðs^v Þ cðiÞ
p ðnÞ;
ð15Þ
n¼0
^v is the estimated timing offset of the
where i ¼ 0; 1; . . . ; P 1 and s
ðvÞ
vth user. The vector c is calculated for every detected ranging
^ 1, and K
^ is the total number of decode, v, where v ¼ 0; 1; . . . ; K
ðvÞ
tected ranging codes in the second step. c is then compared to a
threshold, g3 , to identify which ranging code is within the current
OFDMA symbol. g3 is normalized by the root mean square value
ðvÞ
of c . Since this step should not be reached unless at least one
ranging code exists in the current OFDMA symbol, the code with
maximum correlation could just be chosen for every value of v.
However, a threshold is used so that the algorithm can detect multiple codes with the same timing offset which are interpreted as a
^ is upsingle code in the second step of the algorithm. In this case, K
dated to reflect the increase in the number of detected ranging
codes. In addition, if two users happen to use the same ranging code
and both their signals are received in the same symbol but with different timing offsets, the system declares a collision and both codes
are dropped. The corresponding users then have to retry in the next
available ranging opportunity.
164
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
Algorithm 2:
5. Computational complexity
For practical applications, the computational complexity of an
algorithm is important. In this section, we evaluate the complexity
of the proposed algorithm and compare it to other existing
algorithms.
The proposed algorithm is compared with the proposed algorithms in [7] and [8], which are referred to as Algorithms 1 and
2, respectively. For Algorithm 1, an observation window with the
OFDMA symbol size is used. A bank of correlators, equal to the
number of available ranging codes P, is used to separate ranging
codes in the received signal. If the maximum possible delay is
smax , then ðsmax þ 1ÞP cross-correlation operations are performed
for every OFDMA symbol with ranging opportunity. Assuming
the current UL consists of N UL OFDMA symbols, where N UL is required by the standard to be an integer multiple of 3, then the total
number of correlation operations performed would be
N UL ðsmax þ 1ÞP. The same number applies for threshold comparison
operations. Algorithm 2, on the other hand, performs the cross-correlation in frequency domain. Thus, to perform the cross-correlation at every possible timing offset, a linear phase shift of /m ðuÞ
where u ¼ 0; 1; . . . ; smax , is applied to the frequency domain signal
prior to the cross-correlators bank. As a result, Algorithm 2 has
an addition of smax þ 1 linear phase shifts added to its computational complexity. However, since the correlation is performed in
the frequency domain, and since BPSK modulation is used, the correlations are done on real-signals unlike Algorithm 1 which has to
perform complex-signal correlations.
In the proposed algorithm, the energy of the ranging channel for
every OFDMA symbol is calculated. As a result, N UL energy calculation operations and threshold comparisons are needed. The energy
calculation is already performed for N 0 estimation. Assuming that
every ranging symbol triggers the energy detector over two OFDMA symbols, then at most 2K OFDMA symbols will reach the next
stage of the algorithm, where K is the total number of ranging users
within the current UL frame. Of course, if one or more codes collide, then less OFDMA symbols will reach the next stage. In the second stage, all possible linear phase shifts are applied to the OFDMA
symbol (i.e. from 0 to smax ). The energy of the real part of the signal
is measured and compared to a threshold. Thus, ð1 þ smax Þ linear
phase shifts, real energy calculation, and comparison operations
are performed. A maximum of K ranging codes, within 2K ranging
OFDMA symbols, will reach the third and last stage. A cross-correlation with all possible codes is performed. As a result, 2KP correlations and comparison operations (CMPs) are performed at this
stage.
Table 2 shows the number of additions (ADDs) and multiplications (MULs) needed for each operation. Using Table 2, the computational complexity of the algorithms under investigation is
Algorithm 1:
NUL Pðsmax þ 1Þð4L 2ÞADD þ 4LNUL Pðsmax þ 1ÞMUL
þ NUL Pðsmax þ 1ÞCMP:
ðsmax þ 1Þð2L þ NUL PL N UL PÞADD þ Lðsmax þ 1Þð4 þ N UL PÞMUL
þ NUL Pðsmax þ 1ÞCMP:
Proposed Algorithm:
½NUL ð4L 2Þ þ 2Kðsmax þ 1Þð3L 1Þ þ 2KPðL 1ÞADD
þ ½2LðKP þ 2Þ þ 10KLðsmax þ 1ÞMUL
þ ½NUL þ 2Kðsmax þ P þ 1ÞCMP:
Finally, the complexity of all three algorithms is calculated for a
practical system.2 We assume NUL ¼ 12 OFDMA symbols, P ¼ 256
codes(maximum), L ¼ 144 bits(standard [3]), and smax ¼ 512 samples for Nt ¼ 1024. The number of cycles needed by each algorithm
is as follows:
Algorithm 1 : 3:629 109 cycles;
Algorithm 2 : 9:088 108 cycles;
Proposed Algorithm : 2:963 106 cycles for K ¼ 1;
5:909 107 cycles for K ¼ 20:
The difference in computational complexity is evident. The
complexity of the proposed algorithm is a few orders of magnitude
lower than the complexity of Algorithms 1 and 2. While both of
Algorithms 1 and 2 maintain fixed complexity regardless of the
number of ranging users K, the proposed algorithm can update to
K, which gives it the lowest limit in computational complexity.
Note that the above results are optimistic as we assume there
are no false alarms in the first and second stages of the proposed
algorithm. A complexity comparison for a typical WiMAX system
operating in a practical wireless environment is presented in the
following section.
6. Simulation results
6.1. System setup
An OFDMA system model based on [3] is used with the following parameters: N t ¼ 1024, N CP ¼ 128 samples, N UL ¼ 12 OFDMA
symbols, L ¼ 144 bits. The system is assumed to be operating at
2 GHz with a bandwidth of 10 MHz and a sampling frequency of
11.2 MHz. The maximum RTD considered is 45.71 lS (smax ¼ 512
samples) allowing for a cell radius of 6.86 Km. The total number
of ranging codes is 256 codes. In the considered system, all codes
are assigned to initial ranging, i.e. P ¼ 256. The ranging channel
is made up of six subchannels and spanning 144 subcarriers per
OFDMA symbol. Ranging users choose two consecutive symbols
randomly to send their ranging code during the UL frame with
equal probability. In each simulation, 10,000 UL frames or
120,000 OFDMA symbols are used to evaluate the system
performance.
6.2. Channel model
Table 2
Proposed algorithm’s computational complexity.
Operation
Correlation,
Correlation,
Energy Cal.,
Energy Cal.,
Phase shift
complex
real
complex
real
ADD
MUL
2ðL 1Þ þ 2L
L1
2ðL 1Þ þ 2L
L1
2L
4L
L
4L
L
4L
The performance of the proposed ranging algorithm is evaluated in both an AWGN channel and an outdoor dispersive channel.
For the dispersive channel, we use one of the channel environments defined by ETSI for the evaluation of the UMTS radio inter-
2
Using CPU cycle counts based on XilinX DSP48 slice.
165
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
face proposals [15]. The time-varying channel impulse response for
these models can be described by
hðs; tÞ ¼
X
ai ðtÞdðs si Þ:
1
ð16Þ
i
Table 3
Characteristics of the ETSI ‘‘Vehicular A” channel environment.
Tap
Relative delay (nS)
Average power (dB)
1
2
3
4
5
6
0
310
7104
1090
1730
2510
0.0
1.0
9.0
10.0
15.0
20.0
1
(b) AWGN channel, with timing error
2
Imaginary
Imaginary
(a) AWGN channel, no timing offset
2
0
−1
1
0
−1
−2
0
2
−2
0
2
Real
Real
(c) Dispersive channel, no timing offset (d) Dispersive channel, with timing error
2
2
1
1
Imaginary
Imaginary
−2
−2
0
−1
−2
−2
0
−1
0
Real
2
−2
−2
0
Real
2
Fig. 4. The effect of synchronization error on the constellation points of a typical
BPSK-modulated OFDM symbol received over AWGN and dispersive channels.
0.8
Probability
This equation defines the impulse response of a tapped-delay channel with every tap i having a delay of si and gain of ai ðtÞ. In this paper, we consider the ‘‘Vehicular A” channel environment [15]. The
taps’ relative delays and average powers are shown in Table 3.
The channel taps ai ðtÞ are complex independent Rayleigh-fading
variables. Different user channels are assumed independent and
the channels are also independent between UL frames. The multipath fading channel is expected to degrade the performance of
the ranging algorithms under investigation. For the proposed algorithm, only the second and third steps are significantly affected by
the multipath channel. The impact of channel models on the third
step, where the signal is correlated with all possible codes, has already been discussed in [8] where Algorithm 2 is presented. Therefore, we are mainly interested in examining the effect of a
dispersive channel on the second step where we detect ranging
codes within the OFDMA symbol and estimate their timing offsets.
As discussed earlier, the subcarriers of an OFDM symbol with a
timing error exhibit a linear phase shift. Therefore, by applying the
appropriate linear phase shift to the received symbols, this effect is
compensated. Fig. 4 shows that effect on the constellation points of
a typical BPSK-modulated OFDM symbol received over AWGN
channel with SNR ¼ 10 dB and over the dispersive channel referred
to earlier. As seen in the figure, the linear phase shift causes the
symbol energy to be distributed equally between the real and
imaginary parts of the signal. On the other hand, at the correct tim-
0.6
0.4
0.2
0
0
0.2
0.4
0.6
0.8
1
e r ðuÞ j> g j u ¼ sk g and Prfj E
e r ðuÞ j> g j u–sk g for different values of
Fig. 5. Prfj E
2
2
g2 over non-dispersive and dispersive channels.
ing offset, the energy is more concentrated into a single axis. The
effect of the dispersive channel is that at the correct timing offset,
the signal is more noisy and there is also a random gain and phase
shift. However, the correct timing offset is still detectable compared to the same signal with wrong timing offset. The real part
of the signal contains 50% of the total symbol energy when a timing
offset is present regardless of the channel model . In the above
example and with no timing offset, the real part of the signal holds
99% and 69% of the total symbol energy over AWGN and multipath
fading channels, respectively. To further illustrate this effect, computer simulations were used to calculate the probabilities
e r ðuÞ j> g j u–sk g for different
e r ðuÞ j> g j u ¼ sk g and Prfj E
Prfj E
2
2
values of g2 over non-dispersive and dispersive channels. The ree r ðuÞ j as to avoid
sults are shown in Fig. 5. Note that we consider j E
negative values that can result from phase shifts introduced by the
e r ðuÞ j> g j u ¼ sk g is significantly
channel. As expected, the Prfj E
2
e r ðuÞ j> g j u–sk g for a given threshold level
higher than the Prfj E
2
e r ðuÞ j>
g2 . The figure shows that for both channels, Prfj E
g2 j u–sk g exhibits the same behavior where it decays at a fast rate
e r ðuÞ j> g j u ¼ sk g on the other hand, deas g2 increases. Prfj E
2
creases as g2 increases only for dispersive channels.
6.3. Proposed algorithm performance
The system performance is evaluated at an SNR = 10 dB over
both AWGN and ‘‘Vehicular A” dispersive channels. The proposed
algorithm performance is compared to that of Algorithm 2. The
performance is investigated for different numbers of ranging users
per UL frame. To be able to fairly evaluate the system performance
for different numbers of ranging users, it is assumed that all ranging signals are received with equal powers. However, both the proposed algorithm and Algorithm 2 can operate when different users
have different power levels. For the results presented in this paper,
the SINR corresponding to a given number of users per OFDMA
symbol is shown in Table 1. Based on our analysis and the simulation results shown in Figs. 1 and 5, the thresholds (g1 ; g2 ; g3 ) for the
proposed algorithm are chosen to be (3, 0.2, 6) for AWGN channels
and (3, 0.1, 6) for dispersive channels. Figs. 6 and 7 show the probability of missed detection and the probability of false alarm,
respectively, for both the proposed algorithm and Algorithm 2,
and for different numbers of ranging users per UL frame. Over
AWGN channels, both algorithm performances are almost identical
with zero probability of false alarm. Over dispersive channels, the
proposed algorithm suffers from a higher probability of missed
166
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
0.35
10
Proposed alg., dispersive
Alg. 2, dispersive
Proposed alg., AWGN
Alg. 2, AWGN
8
7
0.25
Standard deviation
Probability of missed detection
0.3
Proposed alg., dispersive
Alg. 2, dispersive
Proposed alg., AWGN
Alg. 2, AWGN
9
0.2
0.15
0.1
6
5
4
3
2
1
0.05
0
0
1
2
3
4
5
6
7
8
Number of ranging users per UL frame
9
−1
10
1
2
Fig. 6. Probability of missed detection.
Probability of false alarm
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
1
2
3
4
5
6
7
8
Number of ranging users per UL frame
10
9
10
Fig. 7. Probability of false alarm.
detection compared to Algorithm 2. The difference between the
two algorithms increases as the number of users increases. On
the other hand, the proposed algorithm shows a much lower probability of false alarm than Algorithm 2 especially for low number of
ranging users. Even with up to eight ranging users per UL frame,
the proposed algorithm is able to detect more than 75% of received
codes with a false alarm probability less than 0.2%.
An important measure of the ranging algorithm quality for WiMAX systems is the timing estimation accuracy. For dispersive
channels, the timing error definition is ambiguous [5]. In this paper, we consider the timing error relative to the channel tap with
maximum average power, or first tap in the considered dispersive
channel. The standard deviation for the timing estimator errors for
both the proposed algorithm and Algorithm 2 is shown in Fig. 8.
Both algorithms show zero timing errors for AWGN channels. Over
dispersive channels, the proposed algorithm timing estimator
shows a higher error standard deviation than that of Algorithm 2.
Overall, both algorithms show an accurate timing estimation capability with an average standard deviation of 5 and 3 for the proposed algorithm and Algorithm 2, respectively. A more important
measure of ranging accuracy is the probability of the timing error
falling outside a given interval. For example, if we assume the tim-
ing errors to be a normally distributed random variable with zero
mean [5], the probability of the timing error to exceed 32 samples
(25% of the CP) is less than 1% for both algorithms.
Next, we consider the computational complexity of the proposed algorithm. For both algorithms under consideration, correlating the received signal with all possible ranging codes
constitutes the majority of the algorithm computations. As such,
the number of times this function is called by the ranging algorithm is used as a measure of how complex the algorithms are.
In this regard, Algorithm 2 has a fixed computation complexity
regardless of the number of ranging opportunities or the number
of ranging users within the received UL frame as shown in Section
5. The reduction of computational complexity gained by using the
proposed algorithm over Algorithm 2 is evaluated and the results
are shown in Fig. 9. The computational complexity are normalized
by the number of detected codes ð1 P md Þ to take into consideration the different probabilities of missed detection between both
methods. The proposed algorithm reduces the computational complexity by over 98% over AWGN channels. Over dispersive channels, the complexity reduction ranges from 96% to 80%
depending on the number of ranging users per UL frame.
100%
Dispersive
AWGN
98%
Computational complexity reduction
Proposed alg., dispersive
Alg. 2, dispersive
Proposed alg., AWGN
Alg. 2, AWGN
0.04
9
Fig. 8. Standard deviation of timing errors.
0.05
0.045
3
4
5
6
7
8
Number of ranging users per UL frame
96%
94%
92%
90%
88%
86%
84%
82%
80%
1
2
3
4
5
6
7
8
9
Number of ranging users per UL frame
10
Fig. 9. Computational complexity reduction using the proposed algorithm.
167
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
7. Conclusions
0.35
Probability of missed detection, Nt=512
Probability of missed detection, Nt=1024
0.3
Probability of false alarm, N =512
t
Probability of false alarm, Nt=1024
Probability
0.25
0.2
0.15
0.1
0.05
0
1
2
3
4
5
6
7
8
Number of ranging users per UL frame
9
10
Fig. 10. Probabilities of missed detection and false alarm for N t ¼ 1024 and
N t ¼ 512.
A novel algorithm for OFDMA initial ranging process based on
IEEE802.16e-2005 standard is proposed. The proposed algorithm
performs multiuser code detection and timing offset estimation
for ranging users. The algorithm is divided into three stages. In
the first stage, the system detects OFDMA symbols carrying ranging users. The second stage is responsible for detection of codes
and estimation of timing offsets for each user within current OFDMA symbol. Finally, the last stage identifies user ranging codes. The
proposed algorithm performance is evaluated for both AWGN
channels and dispersive channels. A complexity comparison between the proposed algorithm and other existing algorithms is carried out as well. The results show that proposed algorithm reduces
the computational complexity by 80–96% depending on the number of ranging users while maintaining the timing error standard
deviation under 5% of the guard interval. Simulation results
showed that the proposed algorithm can perform well with as high
as 10 users per ranging channel in a given UL frame. Hence, it is believed that the proposed algorithm can be realized in practical mobile WiMAX BSs.
Appendix A
10
9
Proposed alg., Nt=1024
8
Proposed alg., N =512
From (7), the expected value of Eg is
Alg. 2, Nt=1024
"
t
Alg. 2, Nt=512
7
Standard deviation
E½Eg ¼ E L þ
6
þ2
5
W 2R ðmÞ þ
n¼0
L1
X
L1
X
W 2I ðmÞ
n¼0
cðkÞ
p ðmÞ cos½/m ðsk ÞW R ðmÞ
n¼0
4
þ2
3
L1
X
cðkÞ
p ðmÞ sin½/m ð k ÞW I ðmÞ
s
n¼0
2
#
:
ðA:1Þ
Note that
1
E½cðkÞ
p ðmÞ ¼ 0;
0
—1
L1
X
ðA:2Þ
E½W R ðmÞ ¼ E½W I ðmÞ ¼ 0;
1
2
3
4
5
6
7
8
Number of ranging users per UL frame
9
10
Fig. 11. Standard deviation of timing errors for N t ¼ 1024 and N t ¼ 512.
and since m values are randomly chosen between 0 and Nt 1,
/m ðsk Þ can be approximated as a uniformly distributed random variable between p and p. Thus
E½cos½/m ðsk Þ ¼ E½sin½/m ðsk Þ ¼ 0:
Finally, another set of system parameters from the WiMAX
system profiles is considered to verify the performance of the proposed algorithm. The new set of system parameters are: N t ¼ 512,
N CP ¼ 32 samples. In accordance to the standard, the signal bandwidth, in this case, is 5 MHz and the sampling frequency is
5.6 MHz. For fairness of comparison, the maximum RTD, the number of OFDMA symbols per UL frame, and the number of frames
used to evaluate the simulation results remain constant. The probabilities of missed detection and false alarm for the first and second
sets of system parameters are compared in Fig. 10. In both system
profiles, the performances are almost identical. This is due to the
fact that WiMAX standard maintains fixed subcarrier spacing
regardless of the signal bandwidth or FFT size. As a result, the ranging channel, which consists of a fixed number of subcarriers, has a
fixed bandwidth for different signal bandwidths. In Fig. 11, the
standard deviations of timing errors for both cases as well as for
both considered ranging algorithms are compared. Note that as
the bandwidth is reduced, the sampling time is increased which
leads to a reduction in timing errors (in samples). In this case,
the timing estimation performances of the proposed algorithm
and Algorithm 2 are almost equal.
ðA:3Þ
Therefore, the expected value of the last two terms in (A.1) goes to
zero for large values of L. Then
"
#
"
#
L1
L1
X
X
2
2
E½Eg ¼ L þ E
W R ðmÞ þ E
W I ðmÞ ¼ L þ LN0 :
n¼0
ðA:4Þ
n¼0
Next, we calculate E½E2g from (7). Note that
2
2
E4 2
E
L1
X
E4
L1
X
n¼0
cðkÞ
p ðmÞ cos ½/m ðsk ÞW R ðmÞ
n¼0
"
!2 3
5 ¼ 1 LN2 þ 1 L2 N 2 ;
0
2 0 4
3
!2
W 2R ðmÞ
L1
X
n¼0
W 2R ðmÞ
!
L1
X
n¼0
!#
W 2I ðmÞ
ðA:5Þ
5 ¼ LN 0 ;
ðA:6Þ
¼
ðA:7Þ
1 2 2
L N0 :
2
Then
E½E2g ¼ L2 þ LN 20 þ L2 N20 þ 2LN0 þ 2L2 N0 :
ðA:8Þ
168
H.A. Mahmoud et al. / Computer Communications 32 (2009) 159–168
Appendix B
References
From (10)
Er ðuÞ ¼
(
L1
K1
X
X
n¼0
þ
cðkÞ
p ðmÞ cos½/m ðsk uÞ
k¼0
L1
X
n¼0
¼
)2
"
^ R ðmÞ
2W
K 1
X
þ
L1
X
cðkÞ
p ðmÞ cos½/m ð k
s uÞ
k¼0
L1 X
K1
X
cos2 ½/m ðsk uÞ þ
n¼0 k¼0
^ 2 ðmÞ
W
R
n¼0
#
L1 X
K1
X
n¼0 k¼0
K1 n
o
X
ðvÞ
cðkÞ
p ðmÞc p ðmÞ cos½/m ðsk uÞ cos½/m ðsv uÞ
v¼0
v–k
þ
L1
X
n¼0
þ
L1
X
n¼0
^ 2 ðmÞ
W
R
(
^ R ðmÞ cos½/ ðuÞ
2W
m
K1
X
cðkÞ
p ðmÞ cos½/m ð k
)
s uÞ :
k¼0
ðB:1Þ
The expected value of the second and last terms in (B.1) is zero.
Therefore
E½Er ðuÞ ¼ E
E½Er ðuÞ ¼
"
L1 X
K1
X
(
n¼0 k¼0
1
LK
2
1
LK
2
#
cos2 ½/m ðsk uÞ þ E
"
L1
X
n¼0
#
^ 2 ðmÞ :
W
R
ðB:2Þ
þ 12 L þ 12 LN0
for u ¼ sk for only one value of k;
þ 12 LN 0
for u–sk for all values of k:
ðB:3Þ
The assumption that u ¼ sk for only one value of k implies that
in the same symbol and for a give timing offset u, only one code
exist.
[1] Supplement
to
IEEE
Standard
for
Information
TechnologyTelecommunications and Information Exchange Between Systems – Local
and Metropolitan Area Networks – Specific Requirements – Part II: Wireless
LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications:
Higher-speed Physical Layer Extension In The 2.4 GHz Band, IEEE Std 802.11b1999 Std., 2000.
[2] http://www.wi-fi.org.
[3] IEEE Standard for Local and metropolitan area networks Part 16: Air Interface
for Fixed and Mobile Broadband Wireless Access Systems Amendment 2:
Physical and Medium Access Control Layers for Combined Fixed and Mobile
Operation in Licensed Bands and Corrigendum 1, IEEE Std 802.16e-2005 and
IEEE Std 802.16-2004/Cor 1-2005 (Amendment and Corrigendum to IEEE Std
802.16-2004) Std., 2006.
[4] T.M. Schmidl, D.C. Cox, Robust frequency and timing synchronization for
OFDM, IEEE Trans. Commun. 45 (12) (1997) 1613–1621.
[5] J.J. van de Beek, P.O. Borjesson, M.-L. Boucheret, D. Landstrom, J. Arenas, P.
Odling, C. Ostberg, M. Wahlqvist, S.K. Wilson, A time and frequency
synchronization scheme for multiuser OFDM, IEEE J. Select. Areas Commun.
17 (11) (1999) 1900–1914.
[6] M. Morelli, Timing and frequency synchronization for the uplink of an OFDMA
system, IEEE Trans. Commun. 52 (2) (2004) 296–306.
[7] X. Fu, H. Minn, Initial uplink synchronization and power control (ranging
process) for OFDMA systems, in: Proc. IEEE Global Telecommun. Conf.
(GLOBECOM), vol. 6, Nov. 29–Dec. 3, 2004, pp. 3999–4003.
[8] J. Krinock, M. Singh, M. Paff, V. Tien, A. Lonkar, L. Fung, C.-C. Lee, Comments on
OFDMA ranging scheme described in IEEE 802.16ab-01/01r1, IEEE 802.16abc01/24, Aug. 2001.
[9] J.J. van de Beek, M. Sandell, P.O. Borjesson, ML estimation of time and
frequency offset in OFDM systems, IEEE Trans. Signal Process. 45 (7) (1997)
1800–1805.
[10] Broadband Wireless Access: IEEE MAN standard, IEEE LAN/MAN standards
committee 802.16a, 2003.
[11] Air Interface for Fixed Broadband Wireless Access Systems, IEEE LAN/MAN
standards committee IEEE 802.16ab-01/01r1, July 2001.
[12] D.H. Lee, OFDMA uplink ranging for IEEE 802.16e using modified generalized
chirp-like polyphase sequences, in: The First IEEE and IFIP International
Conference in Central Asia on Internet, Sep. 26–29, 2005.
[13] H. Minn, X. Fu, A new ranging method for OFDMA systems, in: Proc. IEEE
Global Telecommun. Conf. (GLOBECOM), vol. 3, Nov. 28–Dec. 2, 2005.
[14] J.G. Proakis, Digital Communications, fourth ed., McGraw-Hill, New York, 2001.
[15] ETSI, Overall requirements on the radio interface (s) of the UMTS, Tech. Rep.
ETR/SMG-21.02, Tech. Rep., 1997.