See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332736318

Performance Analysis of Code-Domain NOMA in 5G Communication Systems

Conference Paper · January 2019

CITATIONS READS
2 851

3 authors, including:

Zeyad Elsaraf Qasim Ahmed

University of Huddersfield University of Huddersfield
2 PUBLICATIONS 3 CITATIONS 50 PUBLICATIONS 296 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Non-Orthogonal Multiple Access for implementation in 5G communication systems. View project

All content following this page was uploaded by Zeyad Elsaraf on 29 April 2019.

The user has requested enhancement of the downloaded file.

 Performance Analysis of Code-Domain NOMA in
5G Communication Systems
Zeyad Elsaraf, Faheem Khan, and Qasim Ahmed
Department of Engineering and Technology
University of Huddersfield
Huddersfield, HD1 3DH, UK
[email protected]

Abstract—Today’s wireless communication networks transmit main classes: Power Domain (PD-NOMA), and Code Domain
their signals based on the Orthogonal Multiple Access (OMA) (CD-NOMA).
principle. As the number of users increases, OMA based PD-NOMA explores a new dimension to be exploited
approaches may fail to meet the stringent requirements
emerging in the 5th Generation of wireless communications for increasing user capacity where the available transmitting
for very high spectral efficiency and massive connectivity. power at the base station is divided up between the users. A
Non-Orthogonal Multiple Access (NOMA) emerges as a solution user’s power allocation factor is determined via Channel State
to improve upon spectral efficiency and user capacity without Information (CSI). That is, users with low CSI are assigned
sacrificing system performance. This paper aims to demonstrate more power relative to users with higher CSI. Segmenting
the validity of NOMA as an optimal choice for 5G by comparing
it with OMA. Three Code-Domain NOMA (CD-NOMA) the available power levels in this manner allows a system to
schemes are examined and compared with an established promptly serve users with poor channel conditions or users
OMA technique, Orthogonal Frequency Division Multiplexing located at the cell edge as opposed to OMA, where users
(OFDM). The chosen schemes for CD-NOMA are: Low Density with higher CSI are favoured and users with poor channel
Spreading CDMA (LDS-CDMA), Low Density Spreading OFDM conditions have to wait for access according to the time slot
(LDS-OFDM), and Sparse Coding Multiple Access (SCMA).
The performance of each scheme is evaluated by computing assigned to them. At the transmitter side, users are allocated
its Bit error rate (BER) and Outage Probability (OP) and different power levels and their signals are then superposed
simulating them against different values of Signal-to-Noise-Ratio and sent through to each user in the system. The difference
(SNR) over an AWGN channel. It is observed in this paper that, in power between each user is used to perform Successive
while having varying performance levels, every NOMA scheme Interference Cancellation (SIC) at the receiver side [2]. Signals
outperforms OFDM, thereby proving NOMA to be a prime
candidate for implementation in future 5G communication with higher power levels are subtracted from the received
technologies. signal, leaving only that user’s low power signal. It is entirely
possible to achieve Multiple User Detection (MUD) for an
Index Terms—5G, orthogonal multiple access, non-orthogonal increasing number of users occupying the same frequency
multiple access, code domain NOMA, spectral efficiency subcarrier with more sophisticated SIC methods, although this
requires additional processing power at the receiver(s). [3]-[4]
I. I NTRODUCTION In conventional CDMA, users can share a common channel
W ireless connectivity spreads across the globe, a chal- simultaneously. User separation is done by assigning codes,
As lenge for communication systems to accommodate the
incoming wave of new users with limited available resources
or spreading signatures, to each user uniquely. However, as a
result of this channel sharing, ISI in CDMA-based systems is
presents itself. To address this issue, novel communication unavoidable. CD-NOMA mitigates this limitation by utilizing
techniques that allow for multiple users to access the same spreading codes with low density signatures (LDS) and inter-
bandwidth have been developed for the fifth generation of leave sequences. In CD-NOMA, signals are spread using LDS
mobile communications (5G). The currently utilized Multiple (LDS-CDMA) which are comprised of sparse spreading codes
Access (MA) technique, Orthogonal Frequency Division Mul- each containing a small number of non-zero elements. The
tiplexing (OFDM), may no longer satisfy the requirements of sparsity of the codes allows for the generation of more unique
5G since it relies heavily on allotting different frequency bands codewords for signal transmission which, in turn, allows for
to users and stacking them orthogonally, which limits a sys- more users to be non-orthogonally superimposed on a chip.
tem’s user capacity by its bandwidth. Non-orthogonal multiple Unlike PD-NOMA, by utilizing Message Passing Algorithms
access (NOMA), however, explores a different approach in (MPA), discussed in [5], user separation at the receiver can
regards to increasing the user capacity and spectral efficiency be carried out even when the received users’ power levels are
of a system by allowing a number of users to occupy the comparable. Another advantage of LDS-CDMA is its ability
same frequency band with little inter-user interference (ISI) to achieve overloading, that is when the number of users in
[1]. NOMA techniques can be roughly categorized into two a system exceeds the processing gain. While the number of
users in an overloaded system requires reduced sparsity of The effective received signature can be denoted as
spreading codewords, it was proven in [6] that the number of
spreading codes can be increased by up to 300% in a noiseless H = AGS (3)
environment. Overloaded spreading codes are generated in
accordance to the Welch Bound Equality [7] in order to reduce where A, G, and S represent the users transmit gain, the
ISI. corresponding channel gain, and the spreading signature of
The LDS-OFDM system can be understood as a system each user. From (1) and (2), an expression for the received
which utilises LDS for multiple access and OFDM for mul- signal of each user can be written as
ticarrier modulation mapping. Due to its orthogonal mapping
and sparse spreading, LDS-OFDM benefits from frequency K
X
diversity as well being able to achieve overloading. This allows yk = hk xk + vk (4)
a system’s user capacity to rise as well as reduce the ISI k=1
that would usually accompany it. However, this convenience
B. LDS-OFDM
comes at the price of high, sometimes unaffordable, receiver
complexity. [8] Much like in LDS-CDMA, signal spreading is carried out
SCMA further optimizes the sparse spreading in LDS- by spreading, zero-padding, and interleaving. Each users’
CDMA by combining the LDS spreader with QAM mapping generated chip is transmitted over a subcarrier belonging to
to directly map a set of bits to a complex sparse vector to the OFDM mapper where the superposed signal is modulated
generate codewords [9]. SCMA codewords are sparse and (Fig.1). Users that are using the same subcarrier are superim-
allow for overloading much like LDS. Codebooks containing posed. Let the set of OFDM data symbols for the k th user,
multidimensionally mapped codewords replace modulation sharing the subcarrier n = [1, ..., N ], be represented as
mapping and spreading, allowing SCMA to benefit from
multidimensional and shaping gains as opposed to code rep- Dn|k = (k, i) : ski,n 6= 0

(5)
etition in LDS. These gains are offset by a complex design
procedure for SCMA codebooks as each multidimensional where ski,n denotes the ith row of the spreading signature
layer is designed using Euclidean geometry. However, SCMA matrix s at the nth subcarrier for the k th user. Let bk =
enjoys a moderate receiver complexity since the codebooks [b1 , b2 , ..., bK ] be the set of user data; a transmitted symbol
are transparent between the transmitter and receiver. This can be presented as
paper is organised as follows: Section II presents the NOMA
techniques’ system models. Section III uses the bit error rate n=N,k=K
X
and outage probability to evaluate each scheme’s performance xkn = bk ski,n (6)
when transmitting over an AWGN channel (OFDM is used (i,k)∈Dn|k
as a base for comparison). Finally, Section IV concludes the
paper’s findings and suggests further future directions for As established in (4), the received signal can be denoted by
NOMA research.
n=N,k=K
X
II. S YSTEM M ODEL yn = hkn xkn + vn (7)
A. LDS-CDMA (i,k)∈Dn|k

Consider a CDMA system with K users. Let y, H, x, v

denote the superposed transmitted signal, effective received C. SCMA
signature, transmitted symbols, and noise vectors respectively.
As shown in Fig.1, the LDS spreading in this paper is divided The SCMA encoding process takes the complex layered
into three stages: Spreading, Zero-padding, and Interleaving. codeword ith column of the j th predefined codebook, design
Spreading is done with a randomly generated Hadamard method discussed in [10], and uses it to spread the signal of
matrix, where the k th user is spread with the codewords in the k th user. Let the set of predefined codewords in each code
the nth row. Zero-padding and Interleaving are designed to book be C(i,j) and the user data set be bk = [bk , ..., bK ], the
further increase the sparsity of the spread codeword(s) while spread data set can be defined as
maintaining the processing gain. The transmitted signal can
K
be represented as X
xk = C(i,j) bk (8)
K
X i,j,k=1
y= hk xk + v (1)
k=1 Received signal can then be denoted as
which can be further generalised as K
X
yk = xk hk + vk (9)
y = Hx + v (2) i,j,k=1
 Fig. 1. LDS CDMA/OFDM System Model [2]

TABLE I TABLE III

S IMULATION PARAMETERS FOR OFDM S IMULATION PARAMETERS FOR LDS-OFDM

Number of Users 16 Number of Users 3

Symbols Per Frame 53 Bits Per Signal 2
FFT Length 64 Symbols Per Frame 8
Cyclic Prefix Length 16 Hadamard Matrix Size 8x8
Channel Model AWGN Channel Model AWGN
Modulation 16-QAM Mapping BPSK
Transmit Antennas 1 Modulation OFDM
Cyclic Prefix Length 16
TABLE II FFT Length 19
S IMULATION PARAMETERS FOR LDS-CDMA Number of Symbols 6
Number of Users 16
Bits Per Signal 2
Hadamard Matrix Size 8x8 TABLE IV
S IMULATION PARAMETERS FOR SCMA
Channel Model AWGN
Modulation 4-QAM Number of Users 6
Codebooks 6
Codewords Per Book 4
III. P ERFORMANCE ANALYSIS UNDER AWGN C HANNEL Bits Per Signal 2
In this section, the performance of CD-NOMA techniques Channel Model AWGN
mentioned in Section II is analysed. OFDM is chosen as the
OMA technique to be used as a base for comparison and is
simulated in MATLAB using the predefined operator in the
system library. The BER of each technique is measured over at the output for OFDM. As fig.2 shows, OFDM achieves
a range of SNR values while transmitting over an AWGN a BER performance of about 0.01 at approximately 18 dBs
Channel. The simulation of each technique is carried out of signal to noise power while every CD-NOMA technique
according to its respective simulation parameters. In order achieves the same or lower error rate while requiring much
to ensure result accuracy, each technique runs one value of less signal power. LDS-CDMA at 0.01 BER with around 11
SNR for 5000 iterations. The bit/symbol error is computed for dBs, LDS-OFDM at 0.005 BER with 10 dBs, and SCMA
each iteration then averaged over the total number of iterations at approximately 0 BER with less than 6 dBs. The outage
before moving on to the next SNR value. The total average performance of OFDM, as shown in fig.3, remains inoperable
error is then normalised to produce the error rate. until the 15 dB mark unlike CD-NOMA which achieves vastly
The SNR values range from 0 to a maximum of 20 dBs superior performance while in outage. With LDS-CDMA and
(with +1 increment). A system is considered to be in outage LDS-OFDM arriving at 0.45% and 0.3% OP respectively with
if even one user does not receive 50% of its message for CD- about 11 dBs and SCMA at approximatley 0% OP at less than
NOMA and more than or equal to 10.6 erroneous symbols 5 dBs.
 Fig. 2. BER for CD-NOMA Vs OMA Fig. 3. Outage Performance for CD-NOMA Vs OMA

TABLE V
I MPLEMENTATION F EASIBILITY C OMPARISON

OFDM LDS-CDMA LDS-OFDM SCMA

Encoding Complexity Low Low Average Very High
Decoding Complexity Low Average Average Average
Low-SNR Performance Very Low Average High Very High
High-SNR Performance Very High High High Very High
ISI Very Low Average Low Low
Receiver Complexity Low Low Very High Average
Overall Feasibility in Large Networks High Average Low Average

IV. C ONCLUSION R EFERENCES

[1] Z. Chen, Z. Ding, X. Dai, and R. Zhang, “A mathematical proof of the
This paper presented an overview of PD and CD superiority of NOMA compared to conventional OMA,” IEEE Trans.
NOMA, experimental simulations, and performance evalua- Commun., (submitted) Available on-line at arXiv:1612.01069.
tions. NOMA superposes multiple users in the power domain, [2] H. Sadia, M. Zeeshan and S. A. Sheikh, ”Performance analysis of down-
link power domain NOMA under fading channels,” 2018 ELEKTRO,
optimising the usage of available bandwidth by allowing Mikulov, 2018, pp. 1-6.
subcarriers to accommodate multiple users, as opposed to [3] S. M. Raizul Islamet al., ”Power-Domain Non-Orthogonal Multiple
dedicated frequency bands in OMA. The simulations have Access (NOMA) in 5G Systems: Potentials and Challenges,” IEEE
Communications Surveys and Tutorials, vol.PP, no.99, 25 Oct. 2016,
showcased the performance of NOMA and OMA scheme(s) in pp. 1-1.
terms of their bit error rate and outage probability at a range [4] Z. Ding, X. Lei, G. K. Karagiannidis, R. Schober, J. Yuan, V. K.
of SNR values while transmitting over an AWGN channel. Bhargava, ”A survey on non-orthogonal multiple access for 5G net-
works: Research challenges and future trends,”, “IEEE J. Select. Areas
NOMA was revealed to be superior in terms of bit error as well Commun.”, 35(10) pp. 2181-2195, Oct. 2017.
as outage performance over OMA. Among the tested NOMA [5] Anass Benjebbour et al., ”Novel Low-Density Signature for Syn-
schemes, SCMA was shown to have the lowest bit error chronous CDMA Systems Over AWGN Channel,“ in IEEE Transactions
on Signal Processing, vol. 56, no. 4, pp. 1616-1626, April 2008.
rate and outage probability at high interference channels and [6] K. E. Ahmed and M. M. Farag, ”Enhanced Overloaded CDMA Intercon-
with low transmit power. Despite its highly complex design nect (OCI) Bus Architecture for On-Chip Communication,” 2015 IEEE
procedure for generating sparse spreading codewords, SCMA 23rd Annual Symposium on High-Performance Interconnects, Santa
Clara, CA, 2015, pp. 78-87.
far outperforms other CD-NOMA schemes, making it the [7] L. Welch, ”Lower bounds on the maximum cross correlation of signals
most likely candidate for focus in future research. Promising (Corresp.),” in IEEE Transactions on Information Theory, vol. 20, no.
future directions for NOMA include: investigating receiver 3, pp. 397-399, May 1974.
[8] R. Hoshyar, R. Razavi, and M. Al-Imari, “LDS-OFDM: an efficient
complexity in NOMA relative to OMA in order to improve on multiple access technique,” in Proc. IEEE 71st Vehicular Technology
its implementation feasibility, combining the NOMA principle Conf. (VTC 2010-Spring), pp. 1–5.
with MIMO in larger networks, investigating the efficacy of [9] H. Nikopour and H. Baligh, “Sparse code multiple access,” in Personal
Indoor and Mobile Radio Communications (PIMRC), 2013 IEEE 24th
NOMA from an energy efficiency standpoint, and applying International Symposium on. IEEE, 2013, pp. 332–336.
the co-op transmission scheme to NOMA in an attempt to [10] Taherzadeh, M., Nikopour, H., Bayesteh, A., and Baligh, H. “SCMA
increase diversity for each user which, in turn, may lead to a Codebook Design,” In Proc. IEEE Vehicular Technology Conference
(VTC Fall), 2014, pp. 1-5.
better outage performance.

View publication stats