2. Comparative Analysis of Waveforms for Fifth
Generation Mobile Networks
Shashank Tiwari, Sourav Chatterjee and Suvra Sekhar Das
G S Sanyal School of Telecommunications,
Indian Institute of Technology Kharagpur, India.
Abstract—New requirements for fifth generation mobile net-
works (5G) have triggered the search of the waveforms beyond
Orthogonal Frequency Division Multiplexing (OFDM). In recent
years, many waveforms that are primarily the different form
of filtered multi-carrier transmission techniques, have been pro-
posed for 5G systems. Each has its own merits and demerits. The
results are presented in comparison to OFDM and different op-
erating condition. Further their system parameters widely differ.
In this article, we aim to provide the comparative performance
analysis of these waveforms in a unified manner whereby one can
readily get a relative performance of the schemes on a common
platform. The performance metrics used in the study are out of
band (OoB) leakage power, peak to average power ratio (PAPR),
bit error rate (BER) performance over a wireless channel, effect
of carrier frequency offset (CFO) and spectral efficiency. The
results are expected to help in selecting appropriate waveform
for next generation mobile communication system.
I. INTRODUCTION
Fourth generation (4G) communication systems [1] pro-
vide high spectral efficiency connections to users. The next
generation (5G) wireless technologies [2]–[4] have to serve
some new applications such as Machine Type Communication
(MTC) [5], Cognitive Radio (CR) [6], Great Service in crowd,
Tactile Internet [7] etc. whose requirements are very different
than that of 4G applications. In MTC, the number of devices
to be communicated is enormous hence it requires relaxed
synchronisation and low-cost device [8]. Tactile Internet re-
quire very low latency (1 millisecond) and very high reliability
(99.9999%) [7]. Cognitive radio applications which exploit
spectrum in opportunistic manner, need very low out of band
(OoB) leakage (-65dB) [9] and flexible operational bandwidth.
Orthogonal Frequency Division Multiplexing with a cyclic
prefix (OFDM-CP) is de-facto waveform for 4G due to its low
complexity and capability to combat frequency selective fading
channel (FSFC). However, the use of rectangular pulse shape
in time domain makes OFDM-CP vulnerable in case of fre-
quency and time offset and generates significant OoB leakage.
These limitations in OFDM-CP have led researchers to find ap-
propriate waveforms for 5G. Based on the above requirements,
many existing and new waveforms have been considered
for 5G, like Filter Bank Multi-Carrier (FBMC) [10]–[13],
Generalised Frequency Devision Multiplexing (GFDM) [14],
Unified Filtered Multi-Carrier (UFMC) [15], Block Inverse
Fourier Transform precoded GFDM (BIDFT-GFDM) [16],
DFT Precoded GFDM [17], Frequency Shift offset Quadrature
Amplitude Modulation GFDM (Fs-oQAM-GFDM) [18] etc.
In FBMC, each sub-carrier is linearly pulse shaped to
provide small OoB leakage and sound frequency localization.
To achieve maximum bandwidth efficiency, sub-carrier over-
lapping in frequency is allowed. The FBMC schemes have two
primary forms, namely Cosine Multi-tone (CMT) [13] which
uses vestigial side-band modulation and Staggered Multi-tone
(SMT) [12] which uses offset QAM modulation to eliminate
Inter Carrier Interference (ICI) due to sub-carrier overlapping
in frequency. However, CMT and SMT have similar structure
and same performance over the wireless channel with or
without transceiver impairments [19]. Hence, performance
analysis of one applies to the other. As FBMC systems do
not use Cyclic Prefix (CP), inter-symbol interference (ISI)
is induced when employed in FSFC. Due to linear filtering,
the signal spreads beyond its duration. Therefore, FBMC has
efficiency limitation like OFDM for the small symbol period.
It has higher complexity as compared to OFDM as well [20].
In UFMC, sub-carriers are divided into subgroups, and each
subgroup of sub-carriers is pulse shaped which broadens the
bandwidth of the filter, and hence filter length becomes small
which makes it suitable for low latency applications [21].
UFMC transceiver is around two times more complex than
OFDM transceiver [22].
GFDM is a flexible multi-carrier as well as a multi-time-slot
waveform that uses cyclic pulse shapes for each sub-carrier to
limit the signal length which enables the use of CP to avoid
inter-block interference (IBI) but induces self-interference. The
complexity of GFDM transceiver is four to six times higher
than OFDM [23].
BIDFT-GFDM and DFT precoded GFDM have been pro-
posed to enhance the performance of GFDM system. These
schemes reduce the PAPR of GFDM significantly and increase
reliability by exploiting frequency diversity in FSFC. Fs-
oQAM-GFDM uses offset QAM instead of QAM modulation
to force orthogonality among sub-carriers in GFDM. Fs-
oQAM-GFDM is shown to reduce OoB significantly at the
expense of spectral efficiency [18].
Although proposed during and before 4G [24], [25], Single
Carrier Frequency Division Multiplexing (SCFDMA) [26] and
Single Carrier Frequency Domain Equalization (SCFDE) [27]
has low peak to average power ratio (PAPR) and is effective
in some 5G scenarios.
To find the appropriate waveform for 5G, waveforms should
be compared in the unified framework as well as all perfor-
mance parameters should be compared. General treatment for
3. all the design aspects of multicarrier waveforms is discussed
in [20]. The complexity and MIMO capabilities of OFDM
and FBMC have also been compared in [20]. The Time-
Frequency Localization of FBMC and OFDM have been
compared in [28]. OoB leakage and Area Spectral efficiency
of FBMC, SCFDE-CP and OFDM-CP have been compared
in [29]. However, they have not compared other performance
parameter such as Bit Error Rate(BER), PAPR, etc. Authors
in [21] have compared the OoB leakage of OFDM, FBMC,
and UFMC but have not compared other relevant parameters.
In our knowledge, all the waveforms mentioned above have
not been compared altogether in one framework.
Hence, the objective of this work is to compare the perfor-
mance of waveforms in a single framework. We have consid-
ered orthogonal waveforms such as ODFM, SCFDMA, and
SCFDE as well as non-orthogonal waveforms such as GFDM,
FBMC, BIDFT-GFDM, DFT precoded GFDM, and oQAM-
GFDM. Performance metrics such as BER in FSFC, effect
of CFO in FSFC, PAPR, OoB leakage and Spectral Efficiency
considering water-filing power allocation are considered in this
work.
The rest of the paper is organised as follows. The system
model is developed in Section II. Results are given in Sec-
tion III. Section IV has the conclusion.
In this work vectors are represented by small bold letters,
matrices are represented by bold capital letters, and scalars are
represented as standard small letters. ‘∗’ represents convolu-
tion operation and IN represents identity matrix with order N.
WN represents normalized IDFT matrix of size N × N and
j =
√
−1.
II. CONTENDING WAVEFORMS : SYSTEM MODEL
In this section system model for considered waveform is
described. The waveforms are divided into two groups (i)
QAM Modulated Waveforms and (ii) Offset-QAM Modulated
Waveforms.
A. QAM Modulated Waveforms
Transmitted signal of QAM Modulated waveform can be
given as,
x(K+L)×1 = B(K+L)×KdK×1, (1)
where, B is modulation matrix and d is complex modulated
data symbols. Input data size is K which is described in Table I
and output size is K + L where L is excess sample points.
For CP based waveform, L is equal to CP length,LCP and for
filter based waveforms, it is Lfilt - 1 where Lfilt is the filter
length.
Waveform Type K=
Single Carrier Number of Time Slots
M
Multi Carrier with one time slot Number of sub-carrier N
Block Based Multi-carrier N × M
TABLE I: Size of data samples for different waveforms
1) Waveforms with CP: In this work we are considering
QAM modulated waveforms which are used with CP based
multi-carrier techniques. OFDM-CP, SCFDE-CP, SCFDMA-
CP, GFDM, BIDFT-GFDM and DFT-GFDM falls into this
category. Using (1) transmitted signal for waveforms with CP
can be given as,
x(K+LCP )×1 = B(K+LCP )×KdK×1. (2)
Waveform Reduced Modulation Matrix ˜B
OFDM-CP WN
SCFDE-CP IN
SCFDMA-CP WN PcPm refer [17]
GFDM A refer [30]
DFT-GFDM APcPm refer [17]
BIDFTM-GFDM AM FbM refer [16]
TABLE II: Reduced Modulation Matrix ˜B for different wave-
forms with CP
The modulation matrix B can be written as B = [ ˜B(K −
LCP + 1 : K) ˜B] where ˜B is reduced modulation matrix
of size K × K. Reduced modulation matrix for different
waveforms is given in Table II.
Let, h = [h1, h2, · · · hL]T
be L length channel impulse
response vector, where, hr, for 1 ≤ r ≤ L, represents the
complex baseband channel coefficient of rth
path [31], which
we assume is zero mean circular symmetric complex Gaussian
(ZMCSC). We also assume that channel coefficients related
to different paths are uncorrelated. We consider, Ncp ≥ L.
Received vector of length NCP + K + L − 1 is given by,
ycp = h ∗ x + νcp, (3)
where νcp is AWGN vector of length K + Ncp + L − 1 with
elemental variance σ2
ν . The first Ncp samples and last L − 1
samples of ycp are removed at the receiver i.e. y = [ycp(Ncp+
1 : Ncp + K)]. Use of CP converts linear channel convolution
to circular channel convolution when Ncp ≥ L [1]. The K
length received vector after removal of CP can be written as,
y = H ˜Bd + ν, (4)
where H is circulant convolution matrix of size K × K
which can also be written as H = WKSWH
K where
S = diag{S0, S1, · · · SK−1} is a diagonal matrix which
holds frequency coefficients of channel impulse response h.
Received signal y is first equalized by MMSE frequency
domain equalization [32]. The channel equalized signal yF DE
can be given as,
yF DE = WK(S + SNR−1
IK)−1
SWH
Ky, (5)
where, SNR =
σ2
d
σ2
ν
where σ2
d is the elemental variance of
the complex modulated symbol d. yF DE is further equalized
whose output can be given as,
˜d = CyF DE, (6)
where, C = ˜B−1
for OFDM, SCFDE, SCFDMA, GFDM and
DFT-GFDM and C = FH
bM D−1
bM for BIDFT-GFDM refer (
4. [16]).
2) Waveform without CP but having OFDM-like implemen-
tation: UFMC falls into this category. Transmitted signal can
be given as [21],
x(K+Lfilt−1)×1 =
B
i=1
FiVidi, (7)
where Vi is IDFT matrix of size K×Ni, Fi is Toeplitz matrix
of size (K + Lfilt − 1) × K which holds filter coefficients,
di is data vector of size Ni × 1 for ith
resource block, B is
number of resource blocks and Ni is number of sub-carriers in
one block and K is the total number of subcarriers. The filter
length Lfilt depends on of the length of the channel impulse
response and is taken to be same as the OFDM CP length
here. At the receiver, at first, zeros are padded to make the
resultant received signal of length 2K. Then a 2K point DFT
is taken, and the output on even alternate indices are taken
out. It is seen that the even subcarrier indices contain the data
multiplied by the filter coefficient and channel response, all
in the frequency domain. Mathematically, the output can be
written as,
Y (k) = H(k)Fi(k)Xi(k) + interference + W(k) (8)
For k even where i is the index of the resource block in which
the subcarrier k is. The interference term can be neglected
following the results in [33]. For odd k, only ICI and ISI exist.
Channel equalisation is done in the frequency domain(zero-
forcing method) and the output obtained is the frequency
domain data. The process is described in [15].
B. oQAM Modulated Waveform
FBMC and FS-oQAM-GFDM fall into this category.For Fs-
oQAM Modulated waveform, data vector d is divided into real
and imaginary parts i.e. dK×1 = di
+jdq
. FBMC transmitted
signal can be given as [34],
x[n] =
K−1
k=0
∞
m=−∞
dk,mg(n − m
K
2
)e
j2πk(n−
Lfilt−1
2
)
K jm+k
,
(9)
where g(n) is the prototype filter of length Lfilt and dk,n
is real data symbol obtained from d in offset QAM format
as described in [34]. Transmitted vector can be written as,
x = [x(0) x(1) · · · x(Lfilt + N
2 − 1)]. The filter length is a
multiple of the number of subcarriers. At the receiver, the data
at the k − th subcarrier for the m − th symbol is recovered
as,
ˆdk,m =
∞
z=−∞
g(z − m
K
2
)e−
j2πk(z−
Lfilt−1
2
)
K j−(m+k)
x(z)
(10)
The receiver processing is considered as in [35].
FS-oQAM-GFDM signal can be expressed as in [18],
˜x = ˜A(i)
di
+ ˜A(q)
dq
, (11)
where ˜A(i)
= WKA(i)
, ˜A(q)
= WKA(q)
and A(i)
, A(q)
are
K×K modulation matrix corresponding to real and imaginary
values of d. Finally the CP is added to obtain the transmitted
signal, given as, x = [˜x(K − LCP + 1 : K) ˜x]. Since it
is a block based waveform, like OFDM and GFDM, wireless
channel is equalized using one tap frequency domain equalizer.
Channel equalized output is further equalized using Matched
Filter. Estimated values of the data can be given as in [18],
ˆd = ℜ ˜A(i)
H
y + jℜ ˜A(q)
H
y (12)
III. RESULTS
In this section, we present the performance comparison
of the contending waveforms. We have done Monte Carlo
simulations to compare the performance of waveforms. For
evaluation, we have considered the channel model, given in
[30]. We have considered uncoded system and use of linear
receiver structures. The parameters of considered waveforms
are provided in Table III. It is assumed that the sub-carrier
bandwidth is comparable to the coherence bandwidth of the
channel for FSFC. SNR loss due to CP is also considered for
FSFC.
Eb
/N 0
in dB
0 5 10 15 20
BER
10 -4
10 -3
10 -2
10 -1
BIDFT-GFDM
SCFDE
IFDMA-GFDM
LFDMA-GFDM
SCFDMA-IFDMA
SCFDMA-LFDMA
FS-oQAM-GFDM
GFDM
OFDM
FBMC
UFMC
14.9 15 15.1
0.01
0.011
0.012
0.013
0.014
0.015
0.016
Fig. 1: Comparison of BER for Ideal Channel and 16QAM
Modulation
Fig. 1 shows the comparison of BER vs Eb
N0
of considered
waveforms for 16 QAM modulation. It is seen from the
figure that, BIDFT-GFDM outperforms other waveforms. This
happens because in case of BIDFT-GFDM waveform, signal
is spread in entire operational bandwidth and hence exploits
frequency diversity. SCFDE waveform which also spreads
signal in entire bandwidth performs worse than BIDFT-GFDM
because GFDM systems use lesser CP length than OFDM
system and hence have spectral efficiency gains. This gain
manifest better BER performance of GFDM based systems
than OFDM based systems. IFDMA-GFDM performs better
than LFDMA-GFDM because IFDMA-GFDM spreads signal
in non-contiguous manner and LFDMA-GFDM spreads signal
locally and hence IFDMA-GFDM exploits more frequency
diversity. IFDMA-SCFDMA performs better than LFDMA-
SCFDMA due to the same reason.
5. Parameters Attributes
Number of Subcarriers N 64
Number of Timeslots M (for GFDM) 5
Mapping 16 QAM
Pulse shape (GFDM) RRC with ROF =0.1
Pulse shape (FS-oQAM-GFDM) Root Raised Mayer with ROF =1 [18]
Pulse shape (FBMC) Phy-Dyas Filter [36]
Pulse shape (UFMC) Equi-Ripple 120 dB attenuation FIR
Number of Resource Blocks (UFMC) 4
Block Based Pulse Shape for OFDM, SCFDMA and SCFDE RRC with ROF = 0.35 and filter length = 17
CP length NCP 16
Channel Length Nch 16
Power delay profile [10− α
5 ]T
, where α = 0, 1 · · · Nch − 1
Sub-carrier Bandwidth 3.9 KHz
RMS delay Spread 4.3 µ sec
Coherence Bandwidth 4.7 KHz
TABLE III: Simulation Parameters of Comparison of Waveforms
UFMC performs better than vanilla forms of GFDM,
FBMC, OFDM, and oQAM-GFDM. FS-oqam-GFDM has
similar performance as GFDM and better performance than
FBMC and OFDM. FBMC performs better than OFDM at
low SNR values and worse at high SNR values.
E
b
/N
o
in dB
10 12 14 16 18 20 22 24
BER
10 -2
BIDFT-GFDM
SCFDE
IFDMA-GFDM
LFDMA-GFDM
SCFDMA-IFDMA
SCFDMA-LFDMA
UFMC
FS-oQAM-GFDM
GFDM
OFDM
FBMC
Fig. 2: Comparison of BER for five percent CFO and 16QAM
Modulation
Next, we will see the effect of CFO on the BER performance
of considered waveforms. We have considered five percent
CFO value i.e. the frequency offset is five percent of sub-
carrier bandwidth. Fig. 2 shows the comparison of BER vs
Eb
N0
of considered waveforms for 16 QAM modulation and
five percent CFO. It is observed that FBMC outperforms all
other waveforms in this case. GFDM performs better than
precoded OFDM, precoded GFDM, OFDM, and UFMC. Pre-
coded GFDM performs better than precoded OFDM. IFDMA-
GFDM performs better than OFDM, which has comparable
performance to LFDMA-GFDM. LFDMA-GFDM performs
better than BIDFT-GFDM. Fs-oQAM-GFDM performs worst
among all the waveforms. SCFDMA-IFDMA performs similar
to UFMC and better than SCFDMA-LFDMA, which performs
better than SCFDE. It is also observed that Precoded GFDM
waveforms and precoded OFDM waveforms show degradation
of two order (100) in the case of CFO when compared with
the ideal case (Fig. 1). Whereas, OFDM shows degradation
one order (10) and GFDM and FBMC demonstrates the
deterioration of lesser than one order. It can be inferred
that the frequency spreading has the adverse effect on BER
performance in the presence of CFO. BIDFT-GFDM as well
SCFDE show the highest degradation as these waveforms
are spread over the entire frequency range. Among DFT
spreading waveforms, IFDMA based waveforms degrade more
than LFDMA waveforms as IFDMA based waveforms spread
signal in a non-contiguous manner whereas LFDMA based
waveforms spread signal in a contiguous manner. FBMC and
GFDM are rather resilient towards CFO due to the use of very
well localised pulse shapes in the frequency domain. UFMC,
which filters a group of sub-carriers, is less resistant towards
CFO than FBMC and GFDM.
SNR in dB
0 2 4 6 8 10 12 14 16 18
S.E.inbits/sec/Hz
0.5
1
1.5
2
2.5
3 GFDM
OFDM
SCFDMA-LFDMA
SCFDMA-IFDMA
SCFDE
IFDMA-GFDM
LFDMA-GFDM
BIDFT-GFDM
FBMC
FS-oQAM-GFDM
UFMC
8.9 9 9.1
1.4
1.45
1.5
1.55
1.6
1.65
Fig. 3: Comparison of Spectral Efficiency for Ideal Channel
using Water Filling Algorithm
Fig. 3 shows the spectral efficiency of considered waveforms
using optimum per symbol power derived from water-filling
algorithm [31]. We assume perfect channel knowledge both
at the transmitter as well as at receiver. It can be seen
that UFMC has the highest spectral efficiency. GFDM has
higher spectral efficiency than FBMC, OFDM and precoded
OFDM and precoded GFDM waveforms. FBMC has higher
6. PARAMETER OFDM FBMC GFDM BIDFT-GFDM LFDMA-GFDM IFDMA-GFDM FS-oQAM-GFDM SCFDMA-LFDMA SCFDMA-IFDMA SCFDE UFMC
Out-of-band emission Very High Very Low High High High High Low Very High Very High Very High Low
BER Performance Moderate Poor Moderate Very Good Good Good Moderate Good Good Very Good Moderate
PAPR performance Poor Very Poor Poor Very Good Good Very Good Poor Good Good Good Poor
Resilience to Carrier Frequency Offset Poor Very Good Very Good Very Poor Very Poor Very Poor Very Poor Very Poor Very Poor Very Poor Good
Spectral Efficiency Good Moderate Good Poor Poor Poor Very Poor Poor Poor Poor Very Good
TABLE IV: A Comprehensive Comparison of the Different Performance Parameters of the Waveforms Considered
spectral efficiency than OFDM and comparable capacity to
GFDM at low SNR. At high SNR values, FBMC has lower
spectral efficiency than OFDM. OFDM has higher spectral
efficiency than precoded OFDM waveforms and precoded
GFDM waveforms. LFDMA based schemes are found to better
than IFDMA based schemes for both GFDM and SCFDMA.
SCFDE has higher spectral efficiency than oQAM modulated
GFDM waveforms. It can be concluded that SNR gain due to
lesser CP length in GFDM manifests higher spectral efficiency
in GFDM based waveforms than OFDM based waveforms.
It can also be concluded that waveforms that are exploiting
frequency diversity provide lesser spectral efficiency than
those who are not exploiting frequency diversity. FBMC and
UFMC waveforms have spectral efficiency gains because they
do not use CP. However, in the case of FBMC, sub-carriers
become non-orthogonal in FSFC, which degrades its SINR and
hence spectral efficiency. FS-oQAM-GFDM uses one null sub-
symbol to enhance OoB performance that degrades its spectral
efficiency by NM−N
NM , which in given scenario is 0.8.
Frequency in MHz
0 1 2 3 4 5 6 7 8 9 10
PowerSpectralDensityindBw/Hz
-80
-70
-60
-50
-40
-30
-20
-10
0
SCFDE
OFDM
SCFDMA
GFDM
BIDFT-GFDM
DFTs-GFDM
Fs-oQAM-GFDM
UFMC
FBMC
8 9
-30
-28
Fig. 4: Comparison of Out of Band Leakage
Fig. 4 shows one-sided power spectral density plot of con-
sidered waveforms. In a 20 MHz system, 128 sub-carriers are
considered out of which 64 sub-carriers are switched off (32
on each edge) to observe OoB characteristics of waveforms.
Power spectral density is averaged over 104
transmitted sym-
bols for each considered waveform. FBMC has the lowest stop
band attenuation and narrowest transition band. FS-oQAM-
GFDM and UFMC have 58dB and 63 dB more stop-band
attenuation than OFDM respectively. FS-oQAM-GFDM has
the narrowest transition band whereas UFMC has very large
transition band. GFDM has 5dB more stop band attenuation
and narrower transition band than OFDM. Precoded GFDM
has quite similar stop-band attenuation to GFDM and Precoded
OFDM also has quite similar stop-band attenuation to OFDM,
which shows that DFT-based precoding has a little effect on
OoB characteristics in OFDM and GFDM. SCFDE has the
worst stop-band attenuation.
PAPRo in dB
2 4 6 8 10 12 14
Pr(PAPR>PAPRo)
10 -4
10 -3
10 -2
10 -1
10 0
OFDM
SCFDE
SCFDMA-IFDMA
SCFDMA-LFDMA
FBMC
UFMC
FS-oQAM-GFDM
BIDFTM-GFDM
GFDM
GFDM-IFDMA
GFDM-LFDMA
Fig. 5: Comparison of PAPR
Fig 5 shows complementary cumulative distribution func-
tion (CCDF) plot of PAPR of all waveforms. 105
trans-
mitted Frames were generated, where each frame has four
transmitted signal blocks. We compare the PAPR value that
is exceeded with probability less than 0.01% (Pr{PAPR >
PAPRo = 10−4
}). It is observed that BIDFM-GFDM and
IFDMA-GFDM have lowest PAPR that is lesser than OFDM
by 8.5 dB. SCFDE, SCFDMA-IFDMA, GFDM-LFDMA and
SCFDMA-LFDMA have comparable PAPR, which is smaller
than OFDM by approximately 3.5 dB. Block based pulse
shapes used in SCFDE and SCFDMA reduces the PAPR
gain in these systems [26].FS-oQAM-GFDM and UFMC have
quite similar PAPR to OFDM. GFDM has 0.7 dB whereas
FBMC, which has highest PAPR, has 3 dB worse PAPR than
OFDM. A review of the critical performance parameters for
the different waveforms is given in Table IV.
IV. CONCLUSIONS
Based on the detail performance analysis of the contending
waveforms, the following conclusions can be made.
In the case of the ideal channel, Precoded GFDM has BER
gain of more than two order (10−2
) better than plain GFDM,
OFDM, UFMC and FBMC. Regarding SNR, it is nearly 8 dB
better than them. Single carrier based schemes are relatively
poorer than precoded GFDM because of CP loss. In the case of
CFO, FBMC and plain GFDM is performing best. FBMC and
plain GFDM achieve around 3 dB SNR gain over OFDM in the
presence of CFO. UFMC provides highest spectral efficiency
7. REQUIREMENT OFDM FBMC GFDM BIDFT-GFDM LFDMA-GFDM IFDMA-GFDM FS-oQAM-GFDM SCFDMA-LFDMA SCFDMA-IFDMA SCFDE UFMC
High Spectral Efficiency PR PR PR NR NR NR NR NR NR NR HR
Good Spectrum Isolation NR HR NR NR NR NR PR NR NR NR PR
High Reliability(low error) NR NR NR HR PR PR NR PR PR HR NR
Low Latency Applications NR NR PR PR PR PR PR NR NR NR HR
High CFO Applications NR HR HR NR NR NR NR NR NR NR PR
Low cost Power Amplifier NR NR NR HR PR HR NR PR PR PR NR
HR-Highly Recommended, PR-Partially Recommended, NR-Not Recommended
TABLE V: Waveform Recommendations for Different Application Requirements
that is followed by GFDM, OFDM and lastly by FBMC at
high SNR. FBMC has lowest OoB leakage which is followed
by FS-oQAM-GFDM. PAPR of precoded GFDM is better than
rest by several dB whereas FBMC has highest PAPR.
Therefore, we can say that, for high spectral efficiency
requirement one may use UFMC, for high-reliability require-
ment one may choose BIDFT-GFDM, for low PAPR re-
quirement one may choose BIDFT-GFDM or IFDMA-GFDM,
for flexibility requirement one may choose GFDM, for CFO
resilient requirement one may choose GFDM or FBMC while
for low OoB leakage requirement one may choose FS-oQAM-
GFDM or FBMC. These conclusions can be followed by Table
V.
REFERENCES
[1] S. Sesia et al., LTE - The UMTS Long Term Evolution: From Theory to
Practice. John Wiley & Sons ( Hoboken, NJ, USA), 2011.
[2] A. Osseiran et al., “Scenarios for 5G mobile and wireless commu-
nications: the vision of the METIS project,” IEEE Communications
Magazine, vol. 52, no. 5, pp. 26–35, May 2014.
[3] J. Andrews et al., “What Will 5g Be?” IEEE Journal on Selected Areas
in Communications, vol. 32, no. 6, pp. 1065–1082, Jun. 2014.
[4] L. Gavrilovska et al., “Visions Towards 5g: Technical Requirements and
Potential Enablers,” Wireless Personal Communications, pp. 1–27, May
2015.
[5] H. Shariatmadari et al., “Machine-type communications: current status
and future perspectives toward 5g systems,” IEEE Communications
Magazine, vol. 53, no. 9, pp. 10–17, Sep. 2015.
[6] X. Hong et al., “Cognitive radio in 5g: a perspective on energy-spectral
efficiency trade-off,” IEEE Communications Magazine, vol. 52, no. 7,
pp. 46–53, Jul. 2014.
[7] G. Fettweis, “The Tactile Internet: Applications and Challenges,” IEEE
Vehicular Technology Magazine, vol. 9, no. 1, pp. 64–70, Mar. 2014.
[8] G. Wunder et al., “5GNOW: Challenging the LTE Design Paradigms
of Orthogonality and Synchronicity,” in 2013 IEEE VTC Spring, Jun.
2013, pp. 1–5.
[9] R. Datta et al., “Improved ACLR by Cancellation Carrier Insertion in
GFDM Based Cognitive Radios,” in Vehicular Technology Conference
(VTC Spring), 2014 IEEE 79th, May 2014, pp. 1–5.
[10] M. Bellanger, “Physical layer for future broadband radio systems,” in
2010 IEEE Radio and Wireless Symposium, Jan. 2010, pp. 436–439.
[11] G. Cherubini et al., “Filtered multitone modulation for very high-speed
subscriber lines,” IEEE Journal On Selected Areas in Communications,
vol. 20, no. 5, pp. 1016–1028, 2002.
[12] B. Saltzberg, “Performance of an efficient parallel data transmission
system,” IEEE Transactions on Communication Technology, vol. 15,
no. 6, pp. 805–811, 1967.
[13] S. D. Sandberg et al., “Overlapped discrete multitone modulation for
high speed copper wire communications,” IEEE Journal On Selected
Areas In Communications, vol. 13, no. 9, pp. 1571–1585, 1995.
[14] N. Michailow et al., “Generalized Frequency Division Multiplexing:
Analysis of an alternative multi-carrier technique for next generation
cellular systems,” in 2012 International Symposium on Wireless Com-
munication Systems, Aug. 2012, pp. 171–175.
[15] V. Vakilian et al., “Universal-Filtered Multi-Carrier technique for wire-
less systems beyond LTE,” in 2013 IEEE Globecom Workshops, Dec.
2013, pp. 223–228.
[16] S. Tiwari et al., “Precoded generalised frequency division multiplexing
system to combat inter-carrier interference: performance analysis,” IET
Communications, vol. 9, no. 15, pp. 1829–1841, Oct 2015.
[17] S. S. Das et al., “Discrete Fourier transform spreading-based generalised
frequency division multiplexing,” Electronics Letters, vol. 51, no. 10, pp.
789–791, 2015.
[18] I. Gaspar et al., “Frequency-shift Offset-QAM for GFDM,” IEEE
Communications Letters, vol. PP, no. 99, pp. 1–1, 2015.
[19] B. Farhang-Boroujeny et al., “Cosine Modulated and Offset QAM Filter
Bank Multicarrier Techniques: A Continuous-Time Prospect,” EURASIP
Journal on Advances in Signal Processing, vol. 2010, pp. 1–17, 2010.
[20] A. Sahin et al., “A survey on multicarrier communications: Prototype
filters, lattice structures, and implementation aspects,” IEEE Communi-
cations Surveys Tutorials, vol. 16, no. 3, pp. 1312–1338, Third 2014.
[21] F. Schaich et al., “Waveform Contenders for 5G - Suitability for
Short Packet and Low Latency Transmissions,” in 2014 IEEE Vehicular
Technology Conference (VTC Spring), May 2014 (Seoul, Korea ), pp.
1–5.
[22] T. Wild et al., “A Reduced Complexity Transmitter for UF-OFDM,” in
2015 IEEE 81st Vehicular Technology Conference (VTC Spring), May
2015, pp. 1–6.
[23] A. Farhang et al., “Low-Complexity Modem Design for GFDM,” IEEE
Transactions on Signal Processing, vol. 64, no. 6, pp. 1507–1518, Mar.
2016.
[24] A. Ghosh et al., “LTE-advanced: next-generation wireless broadband
technology [Invited Paper],” IEEE Wireless Communications, vol. 17,
no. 3, pp. 10–22, Jun. 2010.
[25] L. Verma et al., “Wifi on steroids: 802.11ac and 802.11ad,” IEEE
Wireless Communications, vol. 20, no. 6, pp. 30–35, Dec. 2013.
[26] S. Slimane, “Peak-to-average power ratio reduction of OFDM signals
using broadband pulse shaping,” in 2002 IEEE Vehicular Technology
Conference VTC Fall, vol. 2, 2002, pp. 889–893 vol.2.
[27] D. Falconer et al., “Frequency domain equalization for single-carrier
broadband wireless systems,” IEEE Communications Magazine, vol. 40,
no. 4, pp. 58–66, Apr. 2002.
[28] C. Boyd et al., “On the time-frequency localisation of 5g candidate
waveforms,” in 2015 IEEE International workshop on Signal Processing
advances in Wireless Communications, June 2015, pp. 101–105.
[29] P. Banelli et al., “Modulation Formats and Waveforms for 5g Networks:
Who Will Be the Heir of OFDM?: An overview of alternative modula-
tion schemes for improved spectral efficiency,” IEEE Signal Processing
Magazine, vol. 31, no. 6, pp. 80–93, Nov. 2014.
[30] N. Michailow et al., “Generalized frequency division multiplexing for
5th generation cellular networks,” IEEE Transactions on Communica-
tions, vol. 62, no. 9, pp. 3045–3061, 2014.
[31] J. G. Proakis et al., Digital Communication. McGraw Hill (New York,
USA), 2008, ch. 14.
[32] A. Czylwik, “Comparison between adaptive OFDM and single carrier
modulation with frequency domain equalization,” in Vehicular Technol-
ogy Conference, 1997, IEEE 47th, vol. 2, May 1997, pp. 865–869 vol.2.
[33] X. Wang et al., “Pilot-aided channel estimation for universal filtered
multi-carrier,” in VTC Fall, 2015 IEEE 82nd, Sept 2015, pp. 1–5.
[34] P. Siohan et al., “Analysis and design of OFDM/OQAM systems based
on filterbank theory,” IEEE Transactions on Signal Processing, vol. 50,
no. 5, pp. 1170–1183, May 2002.
[35] D. Waldhauser et al., “MMSE subcarrier equalization for filter bank
based multicarrier systems,” in IEEE 9th Workshop on Signal Processing
Advances in Wireless Communications, 2008. SPAWC 2008, Jul. 2008,
pp. 525–529.
[36] A. Viholainen et al., “Prototype filter design for filter bank based
multicarrier transmission,” in Signal Processing Conference, 2009 17th
European, Aug. 2009, pp. 1359–1363.
View publication statsView publication stats