SC-FDMA-IDMA: A Hybrid Multiple Access

Scheme for LTE Uplink

Xingzhong Xiong1,2，Zhongqiang Luo1
1
Artificial Intelligence of Key Laboratory of Sichuan Province
Sichuan University of Science & Engineering
2
National Key Lab. of Science and Technology on Communication
University of Electronic Science and Technology of China
E-mail: {[email protected], [email protected] }
Abstract-According to OFDM-IDMA suffers a peak-to-average operation can be avoided in the OFDM-IDMA scheme.
power ratio (PAPR) problem, in this paper, we propose a hybrid OFDM-IDMA inherits merits of OFDM and IDMA, which
multiple access scheme for LTE uplink, which combines IDMA has some advantages over pure OFDM or pure IDMA.
and SC-FDMA (Single Carrier Frequency Division Multiple However, OFDM-IDMA is similar to other multi-carrier
Access), named as SC-FDMA-IDMA. This hybrid multiple
systems. OFDM-IDMA suffers a peak-to-average power
access scheme inherits many attractive features of SC-FDMA
and IDMA. We discuss the performance comparison of ratio (PAPR) problem. The transmitted signal of
SC-FDMA-IDMA and OFDM-IDMA. Moreover, the BER OFDM-IDMA has high PAPR, which may degrade the
performance of different subcarrier mapping schemes for efficiency of power amplifier of the transmitters.
SC-FDMA-IDMA is also investigated. Simulation results show SC-FDMA is a multiple access technique that utilizes
that SC-FDMA-IDMA performs better than OFDM-IDMA in single carrier modulation and orthogonal frequency
PAPR for LTE uplink, and its BER performance is close to multiplexing [9]. It is sometimes referred as DFT-spread or
OFDM-IDMA. DFT-pre-coded OFDMA. SC-FDMA is similar to OFDMA
Keywords- IDMA; PAPR; hybrid multiple access; OFDM-IDMA; system in performance. Essentially, SC-FDMA has the same
SC-FDMA-IDMA
overall structure as an OFDMA system. One prominent
I. INTRODUCTION advantage over OFDMA is that the SC-FDMA signal has
lower PAPR because of its inherent single carrier structure.
As a special case of CDMA system, IDMA relies on Therefore, we consider SC-FDMA instead of OFDM to
interleaving as the only means to distinguish the signals from combine with IDMA technique. In this paper, we propose a
different users [1]. IDMA adopts a very simple chip-by-chip hybrid multiple access schemes for LTE uplink, which
iterative multi-user detection (MUD) strategy. The combines IDMA and SC-FDMA. We refer to this scheme as
computational complexity of the MUD in IDMA systems is a SC-FDMA-IDMA. This hybrid multiple access scheme
linear function of the number of users. The MUD algorithm inherits many attractive features of SC-FDMA and IDMA.
in IDMA systems is much simpler than that used in CDMA On the one hand, SC-FDMA has lower PAPR because of its
systems. IDMA aims at mitigating the multiple-access inherent single carrier structure. On the other hand, IDMA
interference (MAI) and inter-symbol interference (ISI) with has a simple and effective iterative MUD, thus IDMA has the
low computation complexity. IDMA significantly simple treatment of inter-symbol interference, multiple
outperforms conventional CDMA in the terms of spectral access interference, and effective mitigation of cross-cell
efficiency, receiver complexity and the combination of interference.
coding gain and diversity gain. Owing to its meritorious In this paper, we investigate the basic principle of
properties, IDMA has proposed for numerous applications, SC-FDMA-IDMA and OFDM-IDMA, and give the
such as next-generation cellular uplink systems as well as for performance comparison of these two systems. Furthermore,
time-hopping UMB systems. Thus IDMA has been we discuss different subcarrier mapping for
extensively studied [2]-[4]. SC-FDMA-IDMA, i.e., localized subcarrier mapping,
The idea of a hybrid communications scheme combining interleaved subcarrier mapping. Numerical results show that
OFDM and IDMA has been proposed and studied. The SC-FDMA-IDMA performs better than OFDM-IDMA for
OFDM-IDMA scheme, which employs IDMA instead of uplink in PAPR. Moreover, the BER performance of
CDMA in OFDM-CDMA, inherits many attractive features SC-FDMA-IDMA approaches OFDM-IDMA.
of well-studied OFDM-CDMA scheme, such as the simple The rest of paper is organized as follows. In section II, the
treatment of inter-symbol interference (ISI) and effective system model for SC-FDMA-IDMA and OFDM-IDMA is
mitigation of cross-cell interference (CCI) [2-8]. illustrated. Signal detection algorithm and performance
Furthermore, IDMA allows a simple and effective turbo-type evaluation for SC-FDMA-IDMA system are described in
iterative multiuser detection (MUD) algorithm applicable to Section III. Simulation results and discussions are presented
system with large number of users, which is crucial for in Section IV. Section V draws the conclusion.
system to achieve high throughput. The chip by chip (CBC)
detection algorithm has linear complexity with the number of II. SYSTEM MODEL
users. Due to random interleavers are employed to
The uplink transimitter and receiver structures of
distinguish signals from different users, the spreading

978-1-4244-6252-0/11/$26.00 ©2011 IEEE

SC-FDMA-IDMA and OFDM-IDMA systems with (DFDMA). The case of N = Q ⋅ M for the distributed
K simultaneous users are shown in Fig.1. We will build mode with equidistance between occupied subcarriers is
SC-FDMA-IDMA system model based on OFDM-IDMA. referred to as Interleaved FDMA (IFDMA). For LFDMA, the
frequency samples after subcarrier mapping Yk ,l can be
d1 c1 x1 X1 Y1 y1 described as follows:
π1
n ⎧ X k ,l , 0 ≤ l ≤ M − 1
Yk ,l = ⎨ (2)
dK cK xK XK YK yK ⎩0 , M ≤ l ≤ N − 1
πK
For DFDMA, Yk ,l can be described as follows:
R
⎧⎪ X , l = Q ⋅ m ( 0 ≤ m ≤ M − 1)
eESE ( c1 ( j ) ) eESE ( x1 ( j ) ) Yk ,l = ⎨ k ,l Q (3)
dˆ 1 π1−1
Ŷ ⎪⎩0 , otherwise
eDEC ( c1 ( j ) ) eDEC ( x1 ( j ) )
π1 r Where, 0 ≤ l ≤ N − 1 , N = Q ⋅ M and 1 ≤ Q < Q .
X̂
For IFDMA, Yk ,l can be described as follows:
eESE ( cK ( j ) ) eESE ( xK ( j ) )
dˆ K πK−1
⎧⎪ X k ,l Q , l = Q ⋅ m ( 0 ≤ m ≤ M − 1)
eDEC ( cK ( j ) ) eDEC ( xK ( j ) ) Yk ,l = ⎨ (4)
πK ⎪⎩0 , otherwise
Fig.1. Uplink transmitter and receiver structures of SC-FDMA-IDMA
and OFDM-IDMA (except the dashed frame models) y k represents the transmitted time domain channel symbol
In this SC-FDMA-IDMA system, the input data of user that are obtained from the inverse DFT (IDFT) of Yk .
k is encoded based on a low-rate code C, generating 1 N −1

sequence ck = [ck (1),..., ck ( j ),..., ck ( J )]T , where J is the yk , n =

N
∑Y
l =0
k ,l e j 2π nl N , n = 0,… , N − 1 (5)
block length. Then ck is permutated by an interleaver We can adopt IFFT instead of IDFT. Let the discrete length
T
πk , yielding xk = ⎡⎣ xk (1) , xk ( j ) , xk ( J ) ⎤⎦ , For of the guard interval is Lg . The sampled sequence with
simplicity, the symbol mapping process is not shown in the cyclic extended guard interval results in
Fig.1. With regard to a complex sequence complex sequence 1 N −1
xk by using quadrature phase shift keying (QPSK), each
yk , n = ∑
N l =0
Yk ,l e j 2π nl N , n = − Lg ,… , N − 1 (6)

dimension of xk is denoted by xkRe or xkIm . After the IDMA Consider a K user multiple access system over quasi-static
flat-fading channels. We assume that the channel gains hk are
signal processing, xk is passed to SC-FDMA modulation.
independent, identically distributed and perfectly known at
Xk represents the M samples of the DFT of xk . We transmitter and receiver. The received signal in a multiple
assume that the size of DFT is equal to the frame length, access channel can be written as
namely M = J . K
R = ∑ hk y k + n . (7)
M −1
X k ,v = ∑ xk ,m e− j 2π mv M , v = 0,… , M − 1
k =1
(1)
Where y k is user- k ’s transmit signal and n is an additive
m =0

Then, Xk is implemented by subcarrier mapping, white Gaussian noise (AWGN) with variance σ2 . We
achieving Yk ( l )( l = 0,1, , N −1, N = Q ⋅ M ) , where Q is assume that y k are independent, Gaussian distributed. We
spreading factor. Conventional subcarrier mapping method consider the AWGN channel, i.e. hk = 1, ∀k .
are distributed mode and localized mode. In the localized The received data is first implemented by FFT. Then it
subcarrier mapping mode, the modulation symbols are implements subcarrier de-mapping and IDFT. For the sake of
assigned to M adjacent subcarriers. In the distributed mode, attaining the best possible performance, the structure of
the symbols are equally spaced across the entire channel iterative multiple user detection of IDMA is shown in Fig.1.
bandwidth. In both modes, the IDFT in the transmitter This structure consists of a Soft-In-Soft-Out (SISO)
assigns zero amplitude to the N − M unoccupied Elementary Signal Estimator (ESE) and a bank of
subcarriers. We refer to the localized subcarrier mapping K individual SISO Decoding (DEC), where the soft
mode of SC-FDMA as Localized FDMA (LFDMA) and information exchanged between the receiver components is
distributed subcarrier mapping mode as Distributed FDMA constituted by the extrinsic Log-Likelihood Ratios (LLRs).
In the process of iteration, the ESE generates the extrinsic
output information eESE ( xk ( j ) ) and deinterleaves it in following part, we focus on the detection of xkRe ( j ) . The
order to create the stream eESE ( c ( j ) ) , which is forward as
k xkIm ( j ) can be obtained the same as xkRe ( j ) .
a priori information to the SISO DEC. In the feedback loop, Step (i): Estimation of Interference Mean and Variance
the SISO DEC computes the extrinsic information stream
of eDEC ( ck ( j ) ) , which is interleaved to generate the k
(
E ( r Re ( j ) ) = ∑ hkRe E ( xkRe ( j ) ) − hkIm E ( xkIm ( j ) ) (16) )
eDEC ( xk ( j ) ) as a priori information for ESE. var ( r Re ( j ) ) = ∑ ( hkRe ) var ( x Re ( j ) )
2
stream of
The ESE employs Soft Interference Cancellation (SIC) on a k
(17)
subcarrier-by-subcarrier basis. The DECs carry out APP + ∑(h Im
k ) var ( x ( j ) ) + σ
Im
k
2

decoding using the output of the ESE as input. The detailed k

iterative multiple user detection algorithm of the receiver
will be described in next section.
k
(
Ψ ( j ) = ∑hkRe hkIm var ( xkRe ( j ) ) − var ( xkIm ( j ) ) ) (18)

III. SIGNAL DETECTION AND PERFORMANCE

( )
var Re( hkζk ( j) ) =( hkRe ) var( rRe ( j) ) +( hkIm) var( rIm ( j) )
2 2

EVATUATION
+2hkRehkImΨ( j) − hk var( xkRe ( j) )
4
A. Signal detection
We adopt the low complexity ESE algorithm derived in
[1][7]. This algorithm is derived through Gaussian (19)
approximation based on the assumption that the interfering Step (ii): LLR Generation and Combining
symbols are independent of each other. This assumption is eESE ( xkRe ( j ) )
reasonable when the coded block is long enough beacuse of
the user-specific random interleavers. In receiver, it
implements inverse SC-FDMA before IDMA detection. = 2 hk i
2 (
Re ( hk r ( j ) ) − E Re ( hk ζ k ( j ) ) ) (20)

Since ISI is only present in the first Lg samples of received (

var Re ( hk ζ k ( j ) ) )
sequenc R ( n ) , n = − Lg ,… , N − 1 , these Lg samples are Where Ψ ( j ) is the covariance of r
Re
( j) and r Im ( j ) .
removed before FFT operation. After FFT operatin,
N −1
E ( i ) and var ( i ) denote the expectation and variance.
Yˆv = ∑ Rn e − j 2π nv N , v = 0,… , N − 1 (8) The DEC carry out APP algorithm. The extrinsic LLRs for
n =0
{ x ( j )} and { x ( j )}
Re
k
Im
k will be used in the ESE to
Then, it implements subcarrier de-mapping, yielding X̂ . updata the interference mean and variance in the next
After IDFT operation , producing r
M −1
iteration.
1
r (v) =
M
∑ Xˆ
m=0
m e j 2π mv M , v = 0,… , M − 1 (9) E ( xkRe ( j ) ) = tanh eDEC ( xkRe ( j ) ) 2( ) (21)

var ( xkRe ( j ) ) = 1 − E ( xkRe ( j ) ) ( )

2
In the process of IDMA detection, we reformulate the (22)
algorithm for QPSK signaling as following.
r ( j ) = ∑ hk xk ( j ) + n ( j ) (10)
B. Performance evaluation
PAPR is a performance measurement that is indicative of
k
the power amplifier efficiency of the transmitter. PAPR is
r ( j ) = hk xk ( j ) + ζ k ( j ) (11) specific to the transmitter. Therefore, we can only take into
Where account the structure of transmitter for SC-FDMA-IDMA and
OFDM-IDMA. As far as OFDM-IDMA is concerned, the
ζ k ( j ) ≡ r ( j ) − hk xk ( j ) = ∑ hk ′ xk ′ + n ( j ) (12) core structure of transmitter is based on OFDM signal
k ′≠ k
structure. With respect to SC-FDMA-IDMA, the kernel
xk ( j ) = xkRe ( j ) + ix Im
k ( j) (13) structure of transmitter is on the basis of SC-FDMA signal
structure. In other words, the different parts decide the PAPR
hk = hkRe + ihkIm (14) performance for both systems. According to the transmitter of
The received complex signal can be expressed by: both systems, it only adds the encoding and interleaver
r ( j ) = ∑ ( hkRe xkRe ( j ) − hkIm xkIm ( j ) ) function in front of OFDM and SC-FDMA signal processing.
However, these models don’t change the signal transmission
k
(15) structure for OFDM-IDMA and SC-FDMA-IDMA. We can
+i ∑ ( h x Re Im
k k ( j ) + hkIm xkRe ( j ) ) + n ( j ) know that a salient advantage of SC-FDMA signal over
k OFDM signal is lower PAPR because of its inherent single
The IDMA detection algorithm is as following. In the carrier structure [9] [10]. In the case of OFDM-IDMA, due to
the IFFT operation, the time-domain signal is a weighted sum
of symbols, which has a high PAPR. For SC-FDMA-IDMA, mapping for SC-FDMA-IDMA. The different parameters are
the additive DFT model transforms the time-frequency marked in Fig.2. From Fig.2, we can know that the
domain structure of signal, which brings out lower PAPR. In performance drastically improves with the increasing of the
[9], it expresses the PAPR without pulse shaping with symbol iterative number. Therefore, we can know that the number of
rate sampling as follows: iterations directly affect the performance of
1 N -1 (23) SC-FDMA-IDMA.
∑
2 2
PAPR = max
X X n
N n
n = 0,1,…, N −1 n =0
The BER performances of SC-FDMA-IDMA in different
According to (23), we can count the PAPR of these users are shown in Fig.3. Where the spreading gain is 16, and
systems. The amplitude of single carrier modulated signal the frame length is 128, the iterative number is 10. We use
does not have a Gaussian distribution unlike OFDM signals. interleaved subcarrier mapping for SC-FDMA-IDMA.
And it is difficult to derive analytically the exact form of the According to Fig.3, we can know that the performance
distribution. Therefore, we resort to numerical analysis to becomes worse with the increasing of user numbers.
investigate the PAPR properties. Specifically, we numerically
analyze the complementary cumulative distribution function
0
10

(CCDF) of PAPR for OFDM-IDMA and SC-FDMA-IDMA.

user=1
In theory, we can get that SC-FDMA-IDMA has lower PAPR user=8
performance than OFDM-IDMA because of their single user=16
carrier structure from SC-FDMA. This is similar to the 10
-1

relationship of SC-FDMA and OFDM. In the case of

BER
subcarrier mapping scheme, we can acquire that the IFDMA
signal maintains the input time symbols each sample whereas
LFDMA have more complicated time samples because of the 10
-2

complex-weighted sum of the input symbols. The numerical

analysis of PAPR shows that IFDMA has lower PAPR than
LFDMA in SC-FDMA system [9]. According to the above
analysis, we can know that the PAPR of subcarrier mapping 10
-3

scheme is fit for SC-FDMA-IDMA. Next we will give the 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Eb/No(dB)
simulation results and discuss the performances of
SC-FDMA-IDMA. Fig.3 BER Performance of SC-FDMA-IDMA with different user
numbers
IV. SIMULATION AND DISCUSSION 0
10
To demonstrate the effectiveness of
SC-FDMA-IDMA, we conduct simulation experiments to -1
10
evaluate the performance of SC-FDMA-IDMA. For
making a comparison, OFDM-IDMA is also illustrated. -2
10
0
10
BER

-3
10
SC-FDMA-IDMA it=1
-4
SC-FDMA-IDMA it=3
10
-1
10
SC-FDMA-IDMA it=10
-5
OFDM-IDMA it=1
10
BER

OFDM-IDMA it=3
it=1 -6
OFDM-IDMA it=10
-2 10
10 it=3 0 1 2 3 4 5 6 7 8 9 10
Eb/No(dB)
it=5
Fig.4 BER Performance comparison between OFDM-IDMA and
it=10 SC-FDMA-IDMA with different iterative numbers
-3
10
0 1 2 3 4 5 6 7 8 9 10 Next, we consider the BER performances of
Eb/No(dB) OFDM-IDMA and SC-FDMA-IDMA with different
Fig.2 BER Performance of SC-FDMA-IDMA with different iterations iterations. We assume that the user number is 8, the frame
length is 256. The BER performance is illustrated in Fig.4.
Fig.2 shows the BER performance of SC-FDMA-IDMA From Fig.4, it is obvious that OFDM-IDMA performs better
in the CBC detection with different iterative numbers, where than SC-FDMA-IDMA, but the performance of
the number of user is 8, the frame length is 256, the length of SC-FDMA-IDMA is near to OFDM-IDMA. However, in
repetition code is 8, and the spreading factor is 2,where it terms of uplink communication, the multicarrier schemes
denotes iterations. We employ the interleaved subcarrier suffer higher PAPR problem, which may degrade the
efficiency of power amplifier of the transmitters. Then we SC-FDMA-IDMA and OFDM-IDMA. We investigated the
will talk about the PAPR performance of these both systems. basic principle of the two systems, and build the system
Fig.5 shows the PAPR performance comparison of simulation models. The similarity and difference between
OFDM-IDMA and SC-FDMA-IDMA. The frame length is SC-FDMA-IDMA and OFDM-IDMA are also illuminated.
128, the spreading gain is 8, and we generate 10000 We evaluated the performances in several scenarios, such as
uniformly random data blocks to acquire the CCDF of PAPR. iterative number, different user number, PAPR and different
It is obvious that SC-FDMA-IDMA has lower PAPR than subcarrier mapping. Simulation results show that
OFDM-IDMA, which is accord with the above analysis. At SC-FDMA-IDMA performs better than OFDM-IDMA in
the same time, we give the PAPR performance of different PAPR for uplink, and its BER performance approaches that
subcarrier mapping, as shown in Fig.4. We can see that of OFDM-IDMA.
interleaved subcarrier mapping has lower PAPR comparing
with localized subcarrier mapping. ACKNOWLEDGMENT
0
10 This work is fully supported by Nature Science Founding
of China [Project No. 60872030], China Postdoctoral
Science Foundation funded project [Project
-1
10 NO.20100471672], and Sichuan Province Science
Foundation for Youths [No. 2011JQ0034].
CCDF

-2
10
REFERENCES
[1] Li Ping, L.Liu, K. Y .Wu, and W. K. Leung,
“Interleaved-division multiple-access”, IEEE Trans.
-3
10
Wireless Commun. Vol. 4, Apr. 2006，pp.938-947
OFDM-IDMA
SC-FDMA-IDMA(LFDMA) [2] Xingzhong Xiong, Jianhao Hu, and Ling Xiang. “An
-4
SC-FDMA-IDMA(IFDMA) Efficient Uplink Transmission Technique for IDMA
10
3 4 5 6 7 8 9 10 11 Based on Time-Reversal”. AEÜ International Journal
PAPR[dB]
of Electronic and Communication,2010,64(2),
Fig.5 PAPR Performance comparison between OFDM-IDMA and pp.125-132.
SC-FDMA-IDMA [3] Hongxia Bie, ZhiSong Bie, “A hybrid multiple access
10
0
scheme: OFDMA-IDMA”, IEEE, 2006, pp.4244-4266.
[4] Rong Zhang and Lajor Hanzo, “Iteratively Detected
Multi-Carrier Interleave Division Multiple
Access”,MICROCOLL’07，May. 2007, pp.1-5
[5] Jun Tong, Qinghua Guo and Li Ping, “Analysis and
design of OFDM-IDMA system”, Eur. Trans.
BER

-1
10
Telecomms. 2, June, 2008, pp.561-569
IFDMA user=1
[6] Li Ping, QingHua Guo and Jun Tong, “The
IFDMA user=4
IFDMA user=8 OFDM-IDMA Approach to Wireless Communication
LFDMA user=1 System”, IEEE Wireless Communication, June, 2007,
LFDMA user=4 pp.18-24.
-2
LFDMA user=8 [7] Jun Tong, Qinghua Guo and Li Ping, “Performance
10
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Analysis of OFDM-IDMA Systems with Peak-Power
Eb/No(dB)
Limitation”, IEEE Wireless Communication, 2008.
Fig.6 BER Performance of conventional sub-carrier mapping for [8] Rong Zhang and Lajios Hanzo, “Three Design Aspects
SC-FDMA-IDMA of Multicarrier Interleaved Division Multiple Access”,
Fig.6 shows the performance of interleaved subcarrier IEEE Transactions on Vehicular Technology, VOL. 57,
mapping and localized subcarrier mapping for NO.6, November 2008, pp.3607-3617
SC-FDMA-IDMA. The spreading gain is 8, Q =2 [9] Hyung G.Myung and David J.Goodman,Single Carrier
FDMA:A New Air Interface for Long Term
and J =256. From Fig.6, we can know that both subcarrier
Evolution,John Wiley & Sons Ltd,England,2008.
mapping schemes essentially have the same performance for
[10] Wang Yong, Liu Guang-yi, and Zhang Jian-hua,
AWGN channel.
“Performance Evaluation on OFDMA and SC-FDMA
V. CONCLUTION in 3GPP LTE”, ZTE Communications, 2007, pp.39-42

SC-FDMA-IDMA is a new hybrid multiple access scheme.

In this paper, we compare the performances of