Professional Documents
Culture Documents
An Overview of OTFS For Internet of Things Concepts Benefits and Challenges
An Overview of OTFS For Internet of Things Concepts Benefits and Challenges
Abstract—The Internet of Things (IoT) is envisioned to connect robustness against multipath fading [5]. In particular, it has
everything, spanning from terrestrial to nonterrestrial termi- been widely used in both the fourth-generation (4G) and the
nals, where reliable communication is expected to be allowed fifth-generation (5G) cellular systems. To further improve the
in both time-invariant and time-variant wireless channels. Since
classic orthogonal frequency-division multiplexing (OFDM) mod- spectral efficiency and system performance of OFDM, a vari-
ulation, which has been widely used in both the fourth-generation ety of innovative multicarrier waveforms, such as generalized
(4G) and the fifth-generation (5G) cellular systems, is sensitive frequency-division multiplexing (GFDM), filter bank multi-
to high Doppler effect, it is challenging to satisfy the ever- carrier (FBMC), and unified filter multicarrier (UFMC) have
growing demands of future IoT. To circumvent this issue, the been investigated in [6]–[13], which exhibit their respective
orthogonal time–frequency space (OTFS) scheme is proposed,
which modulates the information bits in both the delay and benefits in the presence of different communication scenarios.
the Doppler domains, and exhibits beneficial advantages in both However, the aforementioned waveforms are mainly designed
static and high-mobility wireless channel scenarios. In this article, for the static or low-mobility channel, and would suffer severe
we present a comprehensive overview of OTFS for IoT, includ- performance loss in the context of high-mobility and complex
ing the current transceiver design, the potential benefits, the time-variant channels due to the seriously negative impact of
challenge issues, as well as future design guidelines.
the high Doppler effect [14]–[16]. Pursuing the objective of
Index Terms—Internet of Things (IoT), orthogonal frequency- meeting the requirements of future IoT, novel waveforms are
division multiplexing (OFDM), orthogonal time–frequency space urgently to be developed.
(OTFS), time-variant channel, transceiver design.
Recently, orthogonal time–frequency space (OTFS) mod-
ulation, which modulates the information bits in the delay-
I. I NTRODUCTION Doppler (DD) domain rather than in the conventional time–
NTERNET of Things (IoT) aims to connect everything frequency (TF) domain, constitutes an attractive option for the
I at any time and place, where the user terminals include
smart home, wearable devices, unmanned aerial vehicles, high
high-mobility scenarios. Specifically, owing to its ability of
transforming a time-variant channel into a two-dimensional
speed-trains, low-Earth-orbit satellites, and so on [1]–[4]. As (2-D) quasi-time-invariant channel, it is more robust to high
a result, the wireless channels in future IoT may varied from Doppler spread. Hence, after it was first published in [18],
static channel to high-mobility channel, or from time-invariant extensive research has been fuelled in [19] and [91].
channel to fast time-variant channel, where reliable communi- Specifically, the original concept of OTFS was proposed
cation becomes challenging. Accordingly, designing efficient without guard interval (GI) [18], where the ideal waveform
air-interface techniques plays a vital role for future IoT. was employed for perfect signal recovery. Hence, the spec-
Specifically, for the sake of obtaining reliable communi- tral efficiency can be significantly improved in contrast to the
cation in the context of wireless channel variations, a wide classic OFDM associated with cyclic prefix (CP) occupying
range of advanced wireless air interface techniques have up to 20% of frame length. In general, the information bits
been proposed in decades. Especially, orthogonal frequency- are first mapped in the DD domain by means of employ-
division multiplexing (OFDM) was proposed as a classic ing amplitude-phase modulation (APM). Then, a combination
multicarrier waveform for broadband wireless communica- of the inverse symplectic finite Fourier transform (ISFFT) as
tions, due to its low complexity implementation and strong well as transmit windowing is applied. After employing the
Heisenberg transform, the TF modulated signal is converted
Manuscript received July 19, 2021; revised October 8, 2021; accepted to the time-domain signal for transmission over the wire-
November 29, 2021. Date of publication December 6, 2021; date of current
version May 9, 2022. This work was supported in part by the National Key less channel. The corresponding reverse operations are carried
Research and Development Program of China under Grant 2020YFB1807100; out for signal detection at the receiver. The wide-ranging
in part by the National Science Foundation of China under Grant 62001179; studies of this OTFS scheme, including the comparison of
and in part by the Fundamental Research Funds for the Central Universities
under Grant 2020kfyXJJS111. (Corresponding author: Tao Jiang.) performance [20]–[24], the effects of nonlinear power ampli-
Lixia Xiao is with the Research Center of 6G Mobile Communications fier [25], the effects of channel estimation [26]–[28], the
and Wuhan National Laboratory for Optoelectronics, Huazhong University design of low-complexity receiver [29]–[36], as well as the
of Science and Technology, Wuhan 430074, China (e-mail: lixiaxiao@
hust.edu.cn). extension of multi-input and multi-output (MIMO) [37]–[39]
Shuo Li, Ying Qian, Da Chen, and Tao Jiang are with the Research Center have been investigated.
of 6G Mobile Communications, School of Cyber Science and Engineering, To further simplify the implementation, the concept of
Huazhong University of Science and Technology, Wuhan 430074, China
(e-mail: tao.jiang@ieee.org). OFDM-based OTFS was conceived in [40] as well as its
Digital Object Identifier 10.1109/JIOT.2021.3132606 MIMO arrangement was developed in [41], where only
2327-4662
c 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7597
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7598 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
nicating over this multipath channel. Observed from Fig. 3, Hc = hi li . (10)
i=1
due to the multipath delay, there are lP symbols consid-
ered as ISI to the next frame signal. According to (4), the
K received symbols r = [r1 , . . . , rK ]T and lP ISI symbols B. Time-Variant Multipath Channel
rISI = [rK+1 , . . . , rK+lP ]T can be expressed as In many wireless systems, the transmitters or receivers
are moving, where the emitted wave is subjected to the
r = Hs + n Doppler effect, resulting in frequency shifts and variant chan-
rISI = HISI s + n̂ (5) nel. Fig. 2(b) characterizes the model of the time-variant
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7599
where rISI λi
tv denotes the ISI for the next-frame signal and
is defined by
⎡ j2π λi ⎤
e
⎢ ej2π λi 2 ⎥
λi ⎢ ⎥
=⎢ .. ⎥. (16)
⎣ . ⎦
ej2π λi K .
According to (9) and (10), by adding CP at the transmitter
and removing CP at the receiver, the K received symbols can
be obtained by
P
cp λi
rtv = hi li s + n. (17)
i=1
C. OFDM System
In this section, the system models of the OFDM technique
communicating over the time-invariant and the time-variant
channels are introduced. By analyzing the limitation of OFDM
Fig. 4. Received signal of time-variant multipath channel.
communicating over the time-variant channel, the motivation
of developing OTFS is introduced.
channels. Observed from Fig. 2(b), the Doppler frequency of 1) OFDM Over the Time-Invariant Channel: For an OFDM
the line-of-sight (LOS) path and that of the ith reflected path system with K subcarriers, the inverse fast Fourier transform
can be obtained by (IFFT) operation is first employed for an APM symbol vector
v s. To mitigate the effect of ISI, a block of CP with length
υlos = υmax = fc of C is inserted. Then, the signal with length of (K + C)
c
υi = υmax cos(θi ) (11) is transmitted over the wireless channel. At the receiver, by
removing CP and performing the fast Fourier transform (FFT)
where fc represents the carrier frequency and θi denotes the operation, the received signal can be formulated by
direction of arrival. According to [15], the CIR of the time-
YOFDM = FHc FH s + n (18)
variant dispersive multipath channel can be defined by
H
where F and F are the FFT and the IFFT matrices, respec-
P
h(τ, υ) = hi δ(τ − τi )δ(υ − υi ) (12) tively, and Hc is defined by (10). Moreover, the value of
i=1 FHc FH can be further expressed by
⎡ ⎤
where τi and υi are defined by (1) and (11), respectively. H1
⎢ .. ⎥
According to [15], the received signal over the time-variant FHc FH = D =⎣ . ⎦ (19)
channel can be expressed as Diagonal matrix HK
r(t) = h(τ, υ)s(t − τ )ej2π υ(t−τ ) dτ dυ with the elements of the diagonal matrix being obtained by
P [H1 , . . . , HK ] = FFT([h1 , . . . , hP ]) (20)
j2π υi (t−τi )
= hi s(t − τi )e . (13) where FFT(·) denotes the FFT operation. Therefore, the kth
i=1 received subcarrier can be obtained by
Assuming that T is the sampling interval and f is the
frequency spacing, we have t = kT, τi = li T, and υi = λi f , YOFDM
k = Hk sk + nk (21)
the discrete form of r(t) can be expressed as where sk is the transmitted symbol at the kth subcarrier and
P nk denotes the AWGN noise at the kth subcarrier. Hence, low-
r(k) = hi s(k − li )ej2π λi (k−li ) (14) complexity one-tap equalization can be directly applied.
i=1 2) OFDM in Time-Variant Channel: According to (17), the
received signal over time-variant channel can be expressed as
which is shown in Fig. 4. P
According to (4)–(8), the K received symbols can be
λi li
formulated in matrix form as FHc F = F
tv H
hi FH = D . (22)
i=1 Diagonal matrix
P
λi
rtv = hi s+n
li
Since the equivalent channel matrix FHtv H
c F is no longer a
i=1 diagonal matrix, the received signal at the kth subcarrier will
P
suffer from inter channel interference (ICI) as
λi
tv =
rISI hi li − li s + n̂ (15)
i=1 YOFDM,tv
k = Hk sk + nk + NkICI (23)
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7600 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
where NkICI denotes the ICI at the kth subcarrier. By employ- symbols as
ing low-complexity one-tap equalizer, OFDM will suffer ⎡ ⎤
from serious ICI, resulting in significant degradation of bit x11 x12 ··· x1N
⎢ x21 x22 ··· x2N ⎥
error ratio (BER) performance. Hence, how to improve the ⎢ ⎥
X = ⎢. .. .. .. ⎥. (24)
performance of OFDM in the presence of severe ICI has ⎣ .. . . . ⎦
attracted a lot attentions [15]–[17]. Recently, an emerging xM1 xM2 ··· xMN .
OTFS design paradigm, which modulates the signal in the DD
domain instead of the conventional TF domain, constitutes an After operating the 2-D ISFFT, the DD symbols are trans-
attractive option for fast time-varying channel scenarios. Next, formed to a 2-D TF symbols as
the concept of OTFS for IoT network will be introduced.
1 M−1
N−1 nk ml
j2π N −M
Umn = √ xkl e . (25)
MN l=0 k=0
III. OTFS C ONCEPT FOR I OT N ETWORK
A. Concept of OTFS Next, Heisenberg transform and a 2-D windowing are
The concept of OTFS was first introduced in [18], and its applied for the delay-domain symbols as
corresponding system model is portrayed in Fig. 5. As shown
in Fig. 5, the information bits are conveyed by M × N classic M−1
N−1
f (t−nT)
s(t) = Umn gtx (t − nT)ej2π m . (26)
APM symbols, where M is the number of subcarriers and N is
n=0 m=0
the size of time slots. Then, the unit of ISFFT is employed to
transform the DD domain signal into TF domain signal. After After that, the signal s(t) will be transmit-
operating Heisenberg transform, the signal s(t) will be trans- ted over the time-variant channel via (13) as
mitted over the wireless channel. At the receiver, the Wigner r(t) = h(τ, υ)s(t − τ )ej2π υ(t−τ ) dτ dυ.
transform is first employed, SFFT is then carried out for signal At the receiver, Wigner transform is first employed by
recovery.
To be more specific, a block of information bits with length
Ymn = ∗
grx (t − τ )r(t)e−j2π v(t−τ )
dt τ =nT,v=m f . (27)
of MN log2 (L) is first mapped to MN L-APM DD domain
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7601
For the sake of signal recovery, the corresponding SFFT C. Matrix-Form Representation
transform is performed by For easy understanding, the matrix-form representation of
the above-mentioned OTFS concept is introduced. According
1 M−1
N−1 −j2π nk ml to (25), the 2-D TF symbol matrix can be reformulated by
N −M
ykl = √ Ymn e . (28)
NM n=0 m=0 U = FM XFH
N (29)
where FM and FH
denote the M-point FFT matrix and the
N
Finally, low-complexity signal detector can be designed N-point IFFT matrix, respectively. Then, the signal of (26)
for OTFS received signal, which will be introduced in can be represented by
Section V.
M FM XFN = Gtx XFN
S = Gtx FH H H
(30)
with
B. From TF to DD Domain
According to [18], a Fourier dual relationship of a grid in the Gtx = diag gtx [0], gtx [T/M], . . . , gtx [(M − 1)T/M] (31)
DD plane and a reciprocal grid in the TF is depicted in Fig. 6, where gtx represents the transmit window function. Based on
where the paraments T and f denote the symbol period and the property of Kronecker product,1 the vector presentation
the subcarrier spacing, respectively. To be more specific, there of S can be presented by
are M points with a resolution of τ = 1/M f upon the delay T
axis of the DD domain grid, which corresponds to M points s = vec(S) = FH N ⊗ Gtx x (32)
having a resolution of f along with frequency axis of the TF
where ⊗ denotes the Kronecker product and x = vec(X).
domain grid. Similarly, there are N points with a resolution of
According to (14), the received signal can be formulated by
( f /N) on the Doppler axis of the DD domain grid, which
corresponds to N points occupying a resolution of T with the rtv = Htv s + n (33)
time axis of the TF domain grid. Moreover, the transformation
where Htv represents the channel matrix, which is associated
from the TF grid to the DD grid can be verified by SFFT,
with the type of prefix.
where an M-dimensional FFT is applied to the columns of
the matrix Umn , and an N-dimensional IFFT is applied to its 1 The Kronecker product obeys the following property: 1) (A⊗B)(C⊗D) =
rows. (AC) ⊗ (BD) and 2) C = AXB ⇒ vec(C) = (BT ⊗ A)vec(X).
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7602 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
TABLE I
C HARACTERISTICS OF VARIOUS WAVEFORMS FOR OTFS
At the receiver, the received signal of (27)–(28) can be IV. T RANSCEIVER D ESIGN FOR OTFS
represented in matrix form by In this section, the transceiver design of OTFS is introduced,
including its waveform design, its variant GI paradigms, as
M (FM Grx R)FN = Grx RFN .
Y = FH (34)
well as its IM extension.
By substituting (32) into (34), its final vector form can be
derived by A. Waveform Design for OTFS
In this section, various waveforms for OTFS are introduced.
y = (FN )T ⊗ Grx rtv
The characteristics of different waveforms are summarized in
= (FN ⊗ Grx )Htv FH N ⊗ Gtx x + (FN ⊗Grx )n (35) Table I.
Heff n̂ 1) Ideal Waveform: According to [31], for the case of ideal
waveform, the transmitting window function of gtx (t) and its
where Heff denotes the equivalent matrix of OTFS and is con- received window function of grx (t) satisfy the biorthogonal
junction with the designed window function as well as the property as
type of GI imposed, which will be detailed in the following
section. Agrx ,gtx (t, f )t=nT+(−τ ,τ ),f =m f +(−ν ,ν )
max max max max
= δ[n]δ[m]qτmax (t)qνmax (f ) (39)
D. OTFS for IoT Network where qa (x) = 1 for x ∈ (−a, a). Therefore, the received
Based on the analysis of matrix-form representation, an signal of (27) can be simplified by
example of an OTFS-based IoT network with U uplink users
Ymn = Hmn X (40)
is shown in Fig. 7. Observed from Fig. 7 that one base station
(BS) simultaneously serves multiple users for both stationary where Hmn = h(τ, v)ej2π vnT e−j2π(v+m f )τ dτ dv. The cor-
IoT terminals and mobile IoT terminals. The DD resources can responding received signal in the DD domain can be
be partitioned by users upon the Doppler domain. Specifically, expressed as
the DD resource composed of the uth user can be given as
1
N−1 M−1
⎧ ! ykl = xk l hw (41)
⎪
⎨ d ∈ S, if k ∈ u NNu , . . . , (u + 1) NNu − 1 NM k−k ,l−l
k =0 l =0
xu [k, l] = l ∈ {0, 1, . . . , M − 1} (36)
⎪
⎩ where hw is the sampling of the impulse response function as
0, otherwise.
h (v, τ ) =
w
h τ , v w v − v , τ − τ e−j2π vτ dτ dv
Moreover, the DD resource can be divided along the delay
domain and the resource of the uth user can be obtained by M−1
N−1
⎧ w(ν, τ ) = 1 · e−j2π (νnT−τ m f)
. (42)
⎨ d ∈ S, if k ∈ {0, 1, . . . , N − 1}
⎪ ! n=0 m=0
xu [k, l] = l ∈ u NMu , . . . , (u + 1) NMu − 1 . (37)
⎪
⎩ By substituting (42) into (41), the received signal of ykl can
0, otherwise. finally be formulated by
" #
e−j2π (−q−κνi ) − 1
Each IoT terminal employs the allocated DD domain resources P Ni
and independently performs OTFS modulation to transmit ykl ≈ hi e−j2π νi τi
−j 2π
Ne N ( −q−κ ν )
i −N
data. According to (35), the received signal at the BS can i=1 q=−Ni
be formulated by × xk−kν +q ,l−lτ (43)
i N i M
U where Ni denotes the number of neighbor points affected by
y= Hueff xu + n (38) the inter Doppler interference (IDI). For the case of no frac-
u=1 tional Doppler, we have Ni = 0 and the received signal in DD
where Hueffand xu denote the equivalent channel matrix and the domain can be simplified by
transmitted signal of the uth user in the DD domain, respec-
P
tively. For signal recovery, the OTFS demodulation will be ykl = hi e−j2π νi τi xk−kν . (44)
i N, l−lτi M
carried out at the BS. i=1
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7603
TABLE II
PARAMENTS C ALCULATION OF OTFS
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7604 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
TABLE III
2) OTFS With One CP: For the case of OTFS with one C OMPARISONS FOR VARIOUS GI-BASED OTFS S YSTEMS
CP, N blocks signal are first vectorized, the last C symbols
of the vectorized signal are copied to the beginning, as shown
in Fig. 8. According to (17) and (35), the received signal of
OTFS with one CP can be formulated by
1-cp
y1-cp = Heff x + n̂
" P #
H
λi
= (FN ⊗ Grx ) hi li FN ⊗ Gtx x + n̂. (51)
i=1
1-cp
Heff with
3) OTFS With N CP: The structure of OTFS with N CP
P
λi ,n
scheme is similar to that of OFDM, where the last C symbols H̃n = hi lMi , n = (1, 2, . . . , N). (56)
are copied to the beginning at each time slot. According to (17) i=1
and (35), the received signal yN−cp and equivalent channel
N−cp
matrix Heff are obtained by 6) Examples of OTFS With Various GIs: As an exam-
ple, let us consider an OTFS transmission associated with
N−cp
yN−cp = Heff x + n̂ M = N = 2 communicating over the time-variant channel. For
⎡ ⎤
H1 the sake of simplicity, the rectangular waveform is employed
⎢ H2 ⎥ having Gtx = Grx = I2 . We are interested in comparing
⎢ ⎥ H
= (FN ⊗ Grx )⎢ .. ⎥ FN ⊗ Gtx x + n̂ the sparsity of the equivalent channel matrix with respect to
⎣ . ⎦
different types of GI, which is capable of evaluating the attain-
HN able BER performance. The specific examples are shown in
N−cp Fig. 9, where the equivalent channel matrices are calculated
Heff
by (49) and (51)–(55). Observed from Fig. 9 that the spar-
(52)
sity of one CP-based OTFS is similar to that of N CP-based
with OTFS, and both of them are lower than that of one ZS-based
P OTFS. Although the sparsity of N ZS-based OTFS is the low-
λi ,n
Hn = hi lMi , n = (1, 2, . . . , N) est, it is obvious that it cannot provide full diversity due to the
i=1 existence of amounts of zeros. Explicitly, the comparisons for
⎡ ⎤
ej2π λi [(n−1)N+1] various GI-based OTFS systems are presented in Table III.
λi ,n ⎢ .. ⎥
=⎣ . ⎦. (53)
ej2π λi (nN) C. OTFS With Index Modulation
To further exploit the benefits of multidimensional modu-
4) OTFS With One ZS: As shown in Fig. 8, for case of one
lation, OTFS with IM (OTFS-IM) was proposed in [81]. In
ZS-aided OTFS system, one GI is added at the end of vector-
the OTFS-IM scheme, the subcarriers at each time slot are
ized signal, during which no signal is transmitted. According
divided into G groups. Each group is composed of k subcar-
to (15) and (35), the received signal y1−zs and the equivalent
riers, only ka out of k subcarriers are activated to transmit
channel matrix H1−zs
eff are given by
information and the activated indices are considered as an
y1−zs = H1−zs
eff x + n̂
additional means to convey information.
$ To be more specific,
" P # the information bits of B = G g
H g=1 B are partitioned into G
λi g g
groups, each of which is split into two parts as Bg = B1 + B2 .
= (FN ⊗ Grx ) hi li FN ⊗ Gtx x + n̂. (54)
g ka
i=1 A block of information bits with a length of B1 = log2 (Ck )
n is the bino-
1−zs is used to select activated indices, where Cm
Heff
mial coefficient and · denotes the floor operator. The left
g
5) OTFS With N ZS: For the case of OTFS with N ZS, as blocks of information bits with a length of B2 = ka log2 L are
shown in Fig. 8, the ZS GI is padded to the ending at each mapped into L-ary APM symbol and transmitted by the acti-
time slot. The received signal yN−zs and the equivalent channel vated subcarriers. Hence, the nth n = (1, 2, . . . , N) column
matrix HN−zs
eff can be obtained by signal of (24) can be formulated by
yN−zs = HN−zs
eff x + n̂
⎡ ⎤ xn = [x1n , x2n , . . . , xMn ]
H̃1
⎢ ⎥ = x1n , . . . , xkn , x(k+1)n , . . . , x(2k)n , . . . , x(M−k+1)n , . . . , xMn
⎢ H̃2 ⎥ H ⎡ ⎤
= (FN ⊗ Grx )⎢
⎢ ..
⎥ F ⊗ Gtx x + n̂. (55)
⎥ N
⎣ . ⎦ ⎢ ⎥
=⎢ n n n n ⎥
⎣. . . , 0, s1 , . . . , ska , 0, . . . , , . . . , 0, s(M/k−1)ka +1 , . . . , sMka /k , 0, . . .⎦.
H̃N
The first group The last group
N−zs
Heff (57)
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7605
Fig. 9. Examples of different prefix/suffix for OTFS having M = N = P = 2, Gtx = Grx = I2 , l = [0, 1], and k = [0, 1].
The main difference of OTFS-IM and OTFS lies in the bit- single-tap frequency-domain channel equalization. According
to-symbol mapping. After completing the mapping by (57), to (35), it is obvious that they may not be directly
the conventional OTFS modulation will be invoked. At the applied to OTFS. For the sake of obtaining near-optimal
receiver, conventional OTFS demodulation is first employed. BER performance with low-complexity imposed, extensive
Then, the information bits can be recovered by corresponding researches have been investigated on OTFS signal detection.
bit-to-symbol demapping. Since the equivalent channel matrix varies from different GIs,
signal detectors can be separately designed by exploiting their
special structure, which are introduced in detail as follows.
D. Signal Detection for OTFS
Detectors for Ideal Waveform-Based OTFS: The concept of
1) Overview of OTFS Detectors: Conventional detectors OTFS was first proposed with ideal waveform in [18], where
designed for the classic OFDM system mainly rely on the signal can be recovered by exploiting the benefits of Turbo
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7606 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
for fractionally spaced sampling (FSS)-based OTFS system. pi,j snj ∝ p yk |xj = snj
k=1,k=i
By employing a novel probability clipping (PC) solu- ⎛ ⎞
tion, a low complexity matched filtering (MF)-based MP %
MN −yk − ukj − hkj snj
was proposed in [71], which exhibits reduced complexity = exp⎝ ⎠ (61)
owing to the MF operation. Moreover, a low complexity k=1,k=i
2σkj2
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7607
Fig. 11. Performance comparison of ML detector-based OTFS with different Fig. 13. Performance comparison of uncoded OTFS and OFDM with
GIs using (M, N, L) = (2, 2, 2). The number of paths in the DD domain is (M, N, L) = (16, 16, 2) at different moving speeds. The MP detector is
P = 2 for the velocity of 506 km/h. employed for both OTFS and OFDM systems. The number of paths in the
DD domain is P = 3 with τmax = (2/M f ) as well as υmax = (4/NT).
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7608 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
V. MIMO-OTFS S YSTEMS
In this section, the benefits of MIMO-OTFS systems
are investigated, including vertical bell labs space-time
(VBLAST)-aided OTFS, space-time block code (STBC)-aided
OTFS, spatial modulation (SM)-aided OTFS, and MIMO-
aided OTFS-IM systems. We consider a broadband MIMO-
OTFS system employing nt transmit antennas (TAs) and nr
Fig. 14. BER performance comparison of LDPC-aided OTFS and OFDM receiver antennas (RAs). The system models of VBLAST-
with (M, N, L) = (16, 16, 2) at different moving speeds. The number of paths OTFS, STBC-OTFS, SM-OTFS, and VBLAST-OTFS-IM are
in the DD domain is P = 3 with τmax = (2/M f ), as well as υmax = (4/NT).
introduced as follows.
A. VBLAST-OTFS
The VBLAST system is a classic MIMO multiplexing
scheme, where each TA transmits information independently.
In VBLAST-OTFS, the throughput is RV-O = nt log2 L. The
information bit stream with a length of nt LMN is first par-
titioned into nt groups each. The ith group with a length of
LMN is mapped to MN APM symbols as Xi , which has the
same structure of (24). Next, the DD domain Xi symbol matrix
will be modulated by the OTFS technique based on (24)–(26).
Finally, the OTFS modulated signal will be vectorized and
transmitted by the ith TA. At the receiver, the corresponding
OTFS demodulation is applied based on (27) and (28) for each
RA. According to (35), the received signal using equivalent
channel matrix can be formulated as
Fig. 15. Performance comparison of OTFS-IM and OFDM-IM systems with ⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
(M, N, L) = (16, 16, 4) using MP and MMSE detectors. The number of paths Y1 H11 · · · H1nt x1 n1
in the DD domain is P = 4 with τmax = (3/M f ), as well as υmax = (3/NT). ⎢ .. ⎥ ⎢ .. .. .. ⎥⎢ .. ⎥ + ⎢ .. ⎥ (63)
⎣. ⎦ = ⎣ . . . ⎦⎣ . ⎦ ⎣ . ⎦
Ynr Hnr 1 ··· Hnr nt xnt n nr
to improve the performance. Hence, it is shown in Fig. 13 that where Hji denotes the equivalent channel matrix corresponding
the performance of OFDM at 256 km/h can be significantly to the jth RA and the ith TA obtained by (35), as well as
improved in contrast to that of OFDM at 64 km/h. Importantly, xi = vec(Xi ).
the OTFS system is capable of outperforming OFDM by 7 dB
at BER = 10−4 at the speed of 256 km/h.
To provide further insights, Fig. 14 compares the BER B. STBC-OTFS
performance of coded OTFS to that of the OFDM system in To further exploit the diversity of MIMO-OTFS, STBC-
conjunction with different moving speeds of 64 and 256 km/h. based OTFS was proposed in [39], where space-time coding
LDPC with an encoding ratio of 0.5 is employed. Similar to is employed before OTFS modulation.
Fig. 13, the MP detector is employed for both OTFS and In the STBC-OTFS system, the information bit stream with
OFDM systems. It was shown that the performance of one a length of nt LMN is first partitioned into nt groups each.
CP-based OTFS system relying on the rectangular waveform The ith group with a length of LMN is mapped to MN APM
is also close to that of ideal waveform-based counterpart at symbols as Xi , which has the same structure of (24). Then, the
both aforementioned speeds. The coded OTFS system is capa- DD domain Xi symbol matrix assigned to different antennas
ble of outperforming the corresponding OFDM counterpart by will be encoded by the STBC encoder as
around 2 dB at BER = 10−4 at the speed of 256 km/h. ⎡ ⎤
X̃11 · · · X̃T̃1
3) Performance Comparison of OTFS-IM and OFDM-IM: ⎢ . .. .. ⎥
Fig. 15 characterizes the performance comparison of OTFS- X̃ = ⎢⎣ .. . . ⎦.
⎥ (64)
IM and OFDM-IM systems in the context of (k, ka ) = (2, 1). X̃nt · · · X̃nt .
1 T̃
The parameters of M = N = 16, P = 4, and L = 4 are
employed. To make a fair comparison, a similar detector and Then, OTFS will be employed over T̃ frame duration accord-
rectangular waveform are employed for OTFS-IM and OFDM- ing to (24)–(26). Finally, the OTFS modulated signal will be
IM systems. Observed from Fig. 15 that the performance of vectorized and transmitted by the ith TA. At the receiver, the
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7609
where Ii is the active antenna index of the APM symbol si . Next, the DD domain X̂(i) symbol matrix of the ith subgroup
The DD domain symbol matrix is given as is denoted by
(i)
X = [v1 , . . . , vMN ] (i) (i)
⎡ ⎤ X̂ = x1 , . . . , xGi (70)
⎡ ⎤ 0, 0, ··· 0,
x1 ⎢ .. .. ⎥ and will be modulated by the OTFS technique based
⎢ x2 ⎥ ⎢ . s2 , ··· . ⎥
⎢ ⎥ ⎢ ⎥ on (24)–(26). Finally, the OTFS modulated signal of each
⎢ ⎥ ⎢ ⎥
= ⎢ ... ⎥ = ⎢ s1 , ..
. ··· 0, ⎥ . (67) group will be vectorized and transmitted by the ith TA. At
⎢ ⎥ ⎢ ⎥
⎣ xnt −1 ⎦ ⎢ .. ⎥ the receiver, the corresponding OTFS demodulation is applied
⎣ . 0, ··· sMN , ⎦ based on (27) and (28) for each RA. According to (35), the
x nt
0 0 ··· 0. received signal can be formulated as
⎡ ⎤ ⎡ ⎤⎡ (1) ⎤ ⎡ ⎤
Next, each row of the DD domain X symbol matrix xi will be Y1 H11 · · · H1nt x̂ n1
modulated by the OTFS technique based on (24)–(26). Finally, ⎢ .. ⎥ ⎢ .. .. .. ⎥⎢ .. ⎥ + ⎢ .. ⎥ (71)
⎣. ⎦ = ⎣ . . . ⎦⎣ . ⎦ ⎣ . ⎦
the OTFS modulated signal will be transmitted by the ith
TA. At the receiver, the corresponding OTFS demodulation Ynr Hnr 1 ··· Hnr nt x̂(nt ) n nr
is applied for each RA. According to (35), the received signal where Hji denotes the equivalent channel matrix defined
using equivalent channel matrix can be formulated as (i)
in (63), as well as x̂(i) = vec(X̂ ). The detectors introduced in
⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
Y1 H11 · · · H1nt x1 n1 Section IV-D can be applied for MIMO-OTFS signal recovery.
⎢ .. ⎥ ⎢ .. .. .. ⎥⎢ .. ⎥ + ⎢ .. ⎥ (68)
⎣ . ⎦=⎣ . . . ⎦⎣ . ⎦ ⎣ . ⎦ E. Performance Evaluation
Ynr Hnr 1 ··· Hnr nt xnt nnr
Fig. 16 compares the performance of the VBLAST-OTFS
where Hji has been defined in (63). systems employing ideal and rectangular waveforms to that
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7610 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7611
Fig. 20. CDF of uplink sum rate for PD-NOMA users under OTFS/OFDM
with (M, N, L, Nu ) = (8, 8, 2, 2) using the MMSE-SIC detector. The number
Fig. 19. BER performance comparison of OTFS-OMA and OFDMA systems of paths in the DD domain is P = 4 with τmax = (1/M f ), as well as
with (M, N, L, Nu , p1, p2) = (4, 4, 2, 4, 2, 2) using ML detection. The number υmax = (1/NT) for the velocity of 270 km/h.
of paths in the DD domain is P = 4 with τmax = (3/M f ), as well as
υmax = (3/NT) for the velocity of 270 km/h.
by the uth high-mobility user can be formulated as
1
N−1 M−1 nk ml
can be obtained by j2π N −M
Uu [n, m] = √ xu [k, l]e (77)
⎧ * + NM k=0 m=0
⎪
⎨ d ∈ S, if m = u/Î1 + Î2 p2
where xu [k, l] denotes the symbol of the kth row and the lth
Uu [m, n] = n = (u)Î1 + Î1 p1 (74)
⎪
⎩ column in the DD domain symbol matrix of the uth high-
0, otherwise mobility user.
Then, Heisenberg transform is exploited to convert the TF
where Î1 indicates the intervals between two users in the pres-
signal matrix Uu [n, m] to a time-domain signal. Different
ence of frequency axis, while Î2 denotes that of the time axis.
transmission powers are exploited by different users to trans-
p̂1 is the size of the interval on the frequency axis and p̂2 is
mit time-domain signals and the time-domain signal of the uth
that of the interval on the time axis.
user can be formulated as
To provide further insights, Fig. 19 compares the
performance of the OTFS modulation-based interleaved TF M−1
N−1 -
su (t) = Pu Xu (n, m) g(t − nT)ej2π m f (t−nT) (78)
MA (ITFMA) scheme and interleaved DD MA (IDDMA)
n=0 m=0
which also named as the contiguous TF MA (CTFMA) scheme
to that of OFDM access (OFDMA). It is shown in Fig. 19 where Pu denotes the transmission power adopted by the uth
that the OTFS-OMA systems (ITFMA and CTFMA) are user. Fig. 20 compares the spectral efficiency of OTFS-assisted
always capable of providing significant performance gain over PD-NOMA (OTFS-PD-NOMA) to that of OFDM-assisted PD-
OFDMA. NOMA (OFDM-PD-NOMA) under different values of the
cumulative distribution function (CDF). The parameter of
(Nu , M, N, L) = (2, 8, 8, 2), as well as MMSE-aided serial
B. OTFS-NOMA
interference cancellation (MMSE-SIC) detector are employed.
1) OTFS-Assisted DD-TF-NOMA: In the DD-TF-NOMA It is found that the spectral efficiency of OTFS-PD-NOMA and
scheme, high-mobility users occupy DD domain resources, OFDM-PD-NOMA systems is close to each other at a CDF of
meanwhile, low-mobility users occupy TF domain resource. 0.1. In particular, the spectral efficiency of OTFS-PD-NOMA
The TF domain symbol for the high-mobility user can be is increased to around 8 bits/s/Hz, while that of OFDM-based
represented as counterpart is only about to achieve 5 bits/s/Hz at a CDF
of 0.5. Therefore, the OTFS-PD-NOMA system is capable of
1
N−1 M−1 kn ml
j2π N −M
U0 [n, m] = x0 [k, l]e (75) attaining a larger spectral efficiency than OFDM-PD-NOMA.
NM Furthermore, to easy understanding, the parameters of all
k=0 l=0
the simulations are summarized at a glance in Table IV.
where n ∈ {0, . . . , N − 1} and m ∈ {0, . . . , M − 1}. The uth
low-mobility user’s TF signal is denoted as
, VII. P OTENTIAL B ENEFITS FOR I OT N ETWORK
xu [n], if m = u − 1 and u > 0 In this section, the potential advantages of OTFS for IoT
Uu [n, m] = (76)
0, otherwise. network are summarized as follows.
2) OTFS-Assisted Power-Domain NOMA: In the OTFS-
assisted power-domain NOMA (PD-NOMA) scheme, Nu high- A. Potential Advantages
mobility users occupy the same DD-domain resource with dif- 1) Lower PAPR: High PAPR is one of the challenging
ferent transmission power. The TF domain signal transmitted issues in the OFDM design due to the FFT operation at the
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7612 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
TABLE IV
PARAMETERS OF S IMULATIONS
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7613
in the face of high-mobility vehicular networks, even incor- in serious performance degradation [105], [106]. Owing to its
porating with the polar coding. Moreover, low-complexity robustness against phase noise [101]–[107], the OTFS tech-
tuned MMSE equalizer [77] and joint MP-maximum ration nique exhibits great potential in future mmwave and Terahertz
combining (MRC) iterative detection [95] were designed for communication.
SISO-OTFS and MIMO-OTFS in the context of the vehicular
channel, respectively. In particular, the properties of mmWave
vehicular channel are studied in the DD domain with realis- VIII. F UTURE C HALLENGING D ESIGN I SSUES
tic measurement [103]. It is found that OTFS is capable of As a promising waveform for future IoT network, the OTFS
obtaining DD diversity by prudently adjusting the parame- technique provides numerous opportunities, but it also brings
ters of M, N, as well as designing an efficient receiver that new challenges. In this section, its disadvantages are analyzed,
allows denser subcarrier spacings. Therefore, OTFS is capable as well as its design guidelines are illustrated.
of providing reliable communication for vehicular networks
by flexibly designing the system parameter as well as its
associated receiver. A. Efficient Waveform Design
2) OTFS for Underwater Acoustic Communications: The 1) Waveform Design for Low OOB: In light of the afore-
UAC channel is one of the most challenging wireless com- mentioned introduction, it is obvious that rectangular pulse
munication channels. Due to its limited bandwidth, long is one popular waveform. However, it tends to give a rise
propagation delay, and rapid time variations, it has inherent to OOB power emission, which may invoke adjacent channel
double dispersion characteristics. To mitigate the ISI and ICI interference and degrade the system performance. Moreover,
in the TF domain, a variety of redundant signal processing since 2-D convolution is employed in the DD domain, chan-
procedures have to be invoked for conventional multicarrier nel estimation and signal detection become more complex in
schemes, resulting in high implementation complexity as well the presence of fractional Doppler interference. One poten-
as a waste of resources. tial solution to eliminate this effect is to design efficient
To further relieve the effects of the double dispersion, windowing function [57], [78], [79]. Accordingly, it is nec-
OTFS modulation has been applied for UAC in [96] and [97]. essary to further explore the efficient waveform to strike an
Specifically, the performance of OTFS has been evaluated in elegant tradeoff amongst the OOB power required, the spec-
the context of different UAC channels [96]. It is found that the tral efficiency attained, as well as its detection complexity
OTFS technique using short data frames is superior to OFDM imposed.
in terms of BER, PAPR, and spectral efficiency. Moreover, 2) Waveform Coexistence Design: In order to meet the
the OFDM-based OTFS scheme has been applied for mas- diverse demands of IoT communication [113], [114], wire-
sive MIMO multiuser UAC communication system [97], which less scholars have designed specific waveform for Terahertz
predicts that it is capable of providing significant performance communication [105]–[107] as well as satellite-terrestrial com-
gain over that of conventional OFDM counterpart. munication [116], [117]. In order to be compatible with
3) OTFS for Nonterrestrial Networks: LEO communication different communication scenarios, the coexistence of wave-
network is helpful to pursuit the future objective of supporting forms has attracted a wide range of attentions [113]–[120],
the seamless coverage of IoT terminals [1]. However, the high including the coexistence of OFDM and CDMA [113], that
mobility of satellites poses challenge to the stability of LEO- of OFDM and OFDM-IM [114]–[118], that of orthogonal
based IoT communication system. To tackle this issue, OTFS- and nonorthogonal multicarrier [119], and that of OFDM and
based LEO communication network has been studied recently. FBMC.
Specifically, the performance of OTFS-based satellite to The challenging issue of waveform coexistence is the
ground communication system was first investigated in [98], effect of internumerology interference. By setting appropriate
which shows that the OTFS technique is able to outperform parameters and designing an efficient detector, the internu-
classical OFDM in terms of BER. Then, the secrecy of OTFS- merology interference can be reduced efficiently. To the best
based uplink LEO satellite communication has been inves- of the authors’ knowledge, the waveform coexistence of OTFS
tigated in [99]. Moreover, the OTFS-based NOMA system and other waveforms is blank and deserves to be further
has been conceived in [100], which is capable of support- investigated.
ing grant-free uplink access. Owing to its resilience to the 3) AI-Aided Waveform Design: Artificial intelligence tech-
Doppler effect [98]–[100], the OTFS scheme is envisioned nique has the potential to solve the complex optimization for
to be an attractive multicarrier modulation scheme for future different wireless communications research areas by learning
LEO-based IoT communication system. process using numerous data without explicitly programming,
4) OTFS for High-Frequency Band Communication: To as exemplified in [122]–[127]. As above-mentioned, to sat-
increase the system capacity of the IoT network, communi- isfy the diverse demands of IoT communication, the selection
cating over high-frequency band is the most efficient way to of the length of CP, the spacing of subcarrier as well as
achieve this goal. Since a variety of mmwave frequency bands the parameter of the waveform (i.e., window functions, filter
have been adopted in 5G communication, the Terahertz band lengths, and coefficients), becomes more complex. By intrin-
becomes more popular in future IoT network. As the frequency sically amalgamating the technique of AI to optimize the
bands increases, the distortion effect imposed by the power parameter configurations of OTFS is infancy and deserves to
amplifier and phase noise becomes non-negligible, resulting be further investigated.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7614 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
Moreover, efficient waveforms for short-packet and low- CS-based algorithms is expected to significantly reduce the
power transmissions are also deserved to be developed in the complexity of signal processing.
future. 3) Design Issues of Cell-Free MIMO: Cell-free massive
MIMO (CF-MIMO), which transmits information among a
large number of distributed access points (APs), has been
B. Multidomain-Aided OTFS Design considered as a promising technology in future IoT communi-
In order to satisfy people’s pursuit of speed and capac- cation. To combat with the delay spread of dispersive channel,
ity, wireless communication scholars have been committed to OFDM-based CF-MIMO has been conceived in [135]–[137],
explore new dimensions to transmit information. Specifically, focusing on the optimization of active users’ identifica-
the resources of time domain, frequency domain, as well as tion [135], DL-based reciprocity calibration [136], as well as
space domain are utilized to convey information for con- MP-based signal detection [137]. To the best of the authors’
ventional communication. Recently, the IM technique, which knowledge, the performance of CF-MIMO in the face of high
employs the activated indices as an additional dimension to mobility has not been investigated.
convey information, constitutes an appealing arrangement for Specifically, due to the distributed topology of CF-MIMO,
future wireless communication [128], [129]. Based on this the equivalent channel matrix of OTFS should be reana-
overview of OTFS, it is evident that the research of IM-assisted lyzed and reevaluated by exploiting the large-scale fading
OTFS is at the stage of beginning. Incorporating the mul- coefficients. Additionally, advanced low-complexity signal
tidomain concept to OTFS modulation is able to eminently processing algorithms are necessary to be designed for
enhance the attainable transmit rate and is worthy of further either centralized or decentralized implementation paradigms.
study. Moreover, for the case of high-mobility scenarios, the associ-
ated AP cluster varied dramatically with the moving speed of
the user equipment, hence, the dynamic and joint optimization
C. OTFS-Aided Massive MIMO Design of AP and OTFS are worth more intensive study.
Massive MIMO, which is capable of substantially improv- 4) Other Potential Issues: In addition, how to design
ing system capacity by equipping with a large number of sophisticated beamforming/precoding to exploit all the avail-
antennas at the BS, has been deemed as a key technology for able spatial channel is also an open issue for MIMO-
future wireless communication. By incorporating with the con- OTFS [138]. Moreover, how to eliminate the impact of RF
cept of OTFS, its associated signal processing becomes more front-end impairments (i.e., nonlinearity of the power ampli-
challenging. Its design guidelines are detailed as follows. fier, imbalance of the quadrature modulator/demodulator, and
1) Low Overhead Channel Estimation: Obtaining accurate phase noise) are also worthy of study.
CSI plays a vital role in future IoT network [130]. The current
channel estimation has mainly three categories: 1) embed- D. OTFS-Aided Multiple Access Design
ded pilot-aided method [27], [28], [35], [52]; compressed
MA technique is one of the key technologies to support
sensing-based approach [26], [48], [49]; as well as Bayesian
massive connectivity in future IoT network. To further satisfy
theory-based algorithm [34]. Specifically, in the embedded
the demands of MA in IoT network, where a huge chunk of
pilot-aided scheme, pilot, GI, and data symbols are efficiently
terminals are moving at a high speed, OTFS-based MA tech-
arranged in the DD plane to mitigate the interference between
niques have been extensively studied in [83]–[91]. By reusing
the pilot and data symbols at a cost of a certain waste of
the resources of both the DD and the TF domains, it is capa-
resources. By exploiting the sparsity of channel, the CS-
ble of efficiently utilizing the spectrum resources. However,
based approach is capable of acquiring CSI at a reduced
the superposition of signals of the DD and TF domains makes
complexity, but it is not suitable for strongly correlated chan-
signal recovery more challenging. Additionally, the current
nels. The Bayesian theory-based algorithm performs better in
OTFS-based MA paradigms mainly rely on the assumption of
terms of pilot overhead, pilot power consumption, and anti-
biorthogonal as well as the integer delay Doppler. Therefore,
noise interference, but it imposes a slightly higher complexity.
the performance of the OTFS-MA scheme using real wave-
Accordingly, low complexity channel estimation algorithm
forms and fractional delay/Doppler is also required further
as well as the advanced training optimization techniques for
analysis. Moreover, for the sake of exploiting the spatial diver-
massive MIMO are worth more intensive study.
sity and multiplexing gain, the performance of MIMO-aided
2) Low Complexity Signal Detector: Although a wide
OTFS-MA is worth more intensive study.
range of related contributions has concentrated on mas-
sive MIMO signal detection [131], [132], they are mainly
designed for flat-fading channel, so that these detectors may E. OTFS-Aided RIS Design
not be directly applicable to massive MIMO-OTFS. On Reconfigurable intelligent surface (RIS) is proposed as a
the other hand, the increased dimension of the equivalent low hardware cost and energy consumption technique for
channel matrix further increases the complexity of signal future IoT network [139]–[143]. As a passive relay, RIS
detection. Fortunately, its equivalent channel matrix exhibits is capable of providing beneficial beamforming gain with-
inherent sparsity, which can be fully exploited by CS algo- out self-interference cancellation. For the sake of improving
rithms [133], [134]. Consequently, making full use of the the performance of RIS in the context of broadband and
characteristics of the current large-dimensional detector and full-duplex communication scenarios, OFDM-assisted RIS has
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7615
been extensively investigated in [139]–[142], which are not [11] W. Cui, D. Qu, T. Jiang, and F. Behrouz, “Coded auxiliary pilots
suitable for the time-variant wireless channel. for channel estimation in FBMC-OQAM systems,” IEEE Trans. Veh.
Technol., vol. 65, no. 5, pp. 2936–2946, May 2016.
Naturally, as a benefit of the OTFS technique, OTFS- [12] W. Liu, D. Chen, K. Luo, and D. Qu, “FDM-structured preamble
assisted RIS is capable of substantially improving user’s optimization for channel estimation in MIMO-OQAM/FBMC
achieve rate in the time-variant wireless channel. Moreover, systems,” IEEE Trans. Wireless Commun., vol. 17, no. 12,
pp. 8433–8443, Dec. 2018.
it also has the potentiality to eliminate interference with a [13] A. A. Zaidi et al., “Evaluation of waveforms for mobile radio communi-
reduced energy consumption in the context of full-duplex cations above 6 GHz,” in Proc. IEEE Globecom workshop, Washington,
communication. Nevertheless, because the communication DC, USA, 2016, pp. 1-6.
between RIS and its AP/user involves additional information, [14] P. Yang, Y. Xiao, M. Xiao, and S. Li, “6G wireless communica-
tions: Vision and potential techniques,” IEEE Netw., vol. 33, no. 4,
RIS generates more channel estimation overhead. Additionally, pp. 70–75, Jul./Aug. 2019.
since passive RIS components fail to carry out signal process- [15] F. Hlawatsch and G. Matz, Wireless Communications Over Rapidly
ing, the number of components may be extremely large in Time-Varying Channels. Cambridge, MA, USA: Elsevier Press, 2011.
[16] F.-M. Han and X.-D. Zhang, “Wireless multicarrier digital transmission
practice, which impedes the design and deployment of RIS. On via Weyl-Heisenberg frames over time–frequency dispersive channels,”
the other hand, due to the expensive cost of ON/OFF switch- IEEE Trans. Wireless Commun., vol. 57, no. 6, pp. 1721–1733, Jun.
ing of large-scale RIS element, individual amplitude control 2009.
[17] W. Guo , W. Zhang, P. Mu, F. Gao, and H. Lin, “High-mobility wide-
is required for each unit, so that the phase shift design in the band massive MIMO communications: Doppler compensation, analysis
scenes of high speed becomes challenging. It is noted that the and scaling laws,” IEEE Trans. Wireless Commun., vol. 18, no. 6,
researches of RIS in the context of high-speed mobile scenar- pp. 3177–3191, Jun. 2019.
ios is infancy. Hence, further research could be conducted on [18] R. Hadani et al., “Orthogonal time frequency space modulation,” in
Proc. IEEE Wireless Commun. Netw. Conf., San Francisco, CA, USA,
the switching frequency and pilot design of OTFS-RIS. 2017, pp. 1–6.
[19] R. Hadani and A. Monk, “OTFS: A new generation of modulation
addressing the challenges of 5G,” 2018, arXiv:1802.02623.
IX. C ONCLUSION [20] M. K. Ramachandran, G. D. Surabhi, and A. Chockalingam, “OTFS:
A new modulation scheme for high-mobility use cases,” J. Indian Inst.
In this article, we present a comprehensive overview of a Sci., vol. 100, no. 2, pp. 315–336, 2020.
wide range of contributions associated with OTFS, which con- [21] F. Wiffen, L. Sayer, M. Z. Bocus, A. Doufexi, and A. Nix, “Comparison
stitutes a compelling paradigm for future IoT network. This of OTFS and OFDM in ray launched sub-6 GHz and mmWave line-
of-sight mobility channels,” in Proc. IEEE 29th Annu. Int. Symp. Pers.
constructive scheme modulates information in the DD domain Indoor Mobile Radio Commun., Bologna, Italy, 2018, pp. 73–79.
instead of the conventional TF domain and is capable of pro- [22] A. Nimr, M. Chafii, M. Matthe, and G. Fettweis, “Extended GFDM
viding full diversity by exploiting a time-domain equalizer. framework: OTFS and GFDM comparison,” in Proc. IEEE Global
Commun. Conf., Abu Dhabi, United Arab Emirates, 2018, pp. 1–6.
In particular, by means of adjusting the system parameter of [23] V. S. Bhat, S. G. Dayanand, and A. Chockalingam, “Performance anal-
OTFS, it is also capable of offering a flexible waveform design, ysis of OTFS modulation with receive antenna selection,” IEEE Trans.
which becomes attractive for IoT network. Veh. Technol., vol. 70, no. 4, pp. 3382–3395, Apr. 2021.
[24] G. D. Surabhi, R. M. Augustine, and A. Chockalingam, “On the diver-
sity of uncoded OTFS modulation in doubly-dispersive channels,” IEEE
Trans. Wireless Commun., vol. 18, no. 6, pp. 3049–3063, Jun. 2019.
R EFERENCES [25] R. Marsalek, J. Blumenstein, A. Prokes, and T. Gotthans, “Orthogonal
[1] J. Lin, W. Yu, N. Zhang, X. Yang, H. Zhang, and W. Zhao, “A sur- time frequency space modulation: Pilot power allocation and nonlin-
vey on Internet of Things: Architecture, enabling technologies, security ear power amplifiers,” in Proc. IEEE Int. Symp. Signal Process. Inf.
and privacy, and applications,” IEEE Internet Things J., vol. 4, no. 5, Technol., Ajman, United Arab Emirates, 2019, pp. 1–4.
pp. 1125–1142, Oct. 2017. [26] M. Zhang, F. Wang, X. Yuan, and L. Chen, “2D structured turbo com-
[2] “The Internet of Things: How the next evolution of the Internet is pressed sensing for channel estimation in OTFS systems,” in Proc.
changing everything,” San Jose, CA, USA, Cisco, White Paper, 2011. IEEE Int. Conf. Commun. Syst., Chengdu, China, 2018, pp. 45–49.
[3] J. A. Stankovic, “Research directions for the Internet of Things,” IEEE [27] P. Raviteja, K. T. Phan, Y. Hong, and E. Viterbo, “Embedded delay-
Internet Things J., vol. 1, no. 1, pp. 3–9, Feb. 2014. doppler channel estimation for orthogonal time frequency space mod-
[4] S. Chen, J. Hu, Y. Shi, and L. Zhao, “LTE-V: A TD-LTE-based V2X ulation,” in Proc. IEEE 88th Veh. Technol. Conf., Chicago, IL, USA,
solution for future vehicular network,” IEEE Internet Things J., vol. 3, 2018, pp. 1–5.
no. 6, pp. 997–1005, Dec. 2016. [28] P. Raviteja, K. T. Phan, and Y. Hong, “Embedded pilot-aided channel
[5] B. Farhang-Boroujeny and H. Moradi, “OFDM inspired waveforms for estimation for OTFS in delayĺCdoppler channels,” IEEE Trans. Veh.
5G,” IEEE Commun. Surveys Tuts., vol. 18, no. 4, pp. 2474–2492, 4th Technol., vol. 68, no. 5, pp. 4906–4917, May 2019.
Quart., 2016. [29] K. R. Murali and A. Chockalingam, “On OTFS modulation for high-
[6] A. Sahin, I. Guvenc, and H. Arslan, “A survey on multicarrier com- doppler fading Channels,” in Proc. IEEE Inf. Theory Appl. Workshop,
munications: Prototype filters, lattice structures, and implementation San Diego, CA, USA, 2018, pp. 1–10.
aspects,” IEEE Commun. Surveys Tuts., vol. 16, no. 3, pp. 1312–1338, [30] P. Raviteja, K. T. Phan, Q. Jin, Y. Hong, and E. Viterbo, “Low-
3rd Quart., 2014. complexity iterative detection for orthogonal time frequency space
[7] T. Jiang and Y. Wu, “An overview: Peak-to-average power ratio reduc- modulation,” in Proc. IEEE Wireless Commun. Netw. Conf., Barcelona,
tion techniques for OFDM signals,” IEEE Trans. Broadcast., vol. 54, Spain, 2018, pp. 1–6.
no. 2, pp. 257–268, Jun. 2008. [31] P. Raviteja, K. T. Phan, Y. Hong, and E. Viterbo, “Interference
[8] B. Farhang-Boroujeny, “OFDM versus filter bank multicarrier,” IEEE cancellation and iterative detection for orthogonal time frequency
Signal Process. Mag., vol. 28, no. 3, pp. 92–112, May 2011. space modulation,” IEEE Trans. Wireless Commun., vol. 17, no. 10,
[9] D. Chen, Y. Tian, D. Qu, and T. jiang, “OQAM-OFDM for wire- pp. 6501–6515, Oct. 2018.
less communications in future Internet of Things: A survey on key [32] G. D. Surabhi and A. Chockalingam, “Low-complexity linear equal-
technologies and challenges,” IEEE Internet Things J., vol. 5, no. 5, ization for OTFS modulation,” IEEE Commun. Lett., vol. 24, no. 2,
pp. 3788–3809, Oct. 2018. pp. 330–334, Feb. 2020.
[10] R. Nissel, S. Schwarz, and M. Rupp, “Filter bank multicarrier modu- [33] A. Pfadler, P. Jung, and S. Stanczak, “Mobility modes for pulse-shaped
lation schemes for future mobile communications,” IEEE J. Sel. Areas OTFS with linear equalizer,” in Proc. IEEE Global Commun. Conf.,
Commun., vol. 35, no. 8, pp. 1768–1782, Aug. 2017. Taipei, Taiwan, 2020, pp. 1–6.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7616 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
[34] L. Zhao, W. Gao, and W. Guo, “Sparse bayesian learning of delay- [56] F. Long, K. Niu, C. Dong, and J. Lin, “Low complexity iterative
doppler channel for OTFS system,” IEEE Commun. Lett. vol. 24, LMMSE-PIC equalizer for OTFS,” in Proc. IEEE Int. Conf. Commun.,
no. 12, pp. 2766–2769, Dec. 2020. Shanghai, China, 2019, pp. 1–6.
[35] V. K. Singh, M. F. Flanagan, and B. Cardiff, “Maximum likeli- [57] P. Raviteja, Y. Hong, E. Viterbo, and E. Biglieri, “Practical pulse-
hood channel path detection and MMSE channel estimation in OTFS shaping waveforms for reduced-cyclic-prefix OTFS,” IEEE Trans. Veh.
systems,” in Proc. IEEE 92nd Veh. Technol. Conf., Victoria, BC, Technol., vol. 68, no. 1, pp. 957–961, Jan. 2019.
Canada, 2020, pp. 1–5. [58] P. Raviteja, E. Viterbo, and Y. Hong, “OTFS performance on static
[36] X. Xu, M. Zhao, M. Lei, and M. Zhao, “A damped GAMP detection multipath channels,” IEEE Wireless Commun. Lett., vol. 8, no. 3,
algorithm for OTFS system based on deep learning,” in Proc. 92nd pp. 745–748, Jun. 2019.
Veh. Technol. Conf., Victoria, BC, Canada, 2020, pp. 1–5. [59] E. Biglieri, P. Raviteja, and Y. Hong, “Error performance of orthogonal
[37] M. K. Ramachandran and A. Chockalingam, “MIMO-OTFS in high- time frequency space (OTFS) modulation,” in Proc. IEEE Int. Conf.
Doppler fading channels: Signal detection and channel estimation,” in Commun. Workshops, Shanghai, China, 2019, pp. 1–6.
Proc. IEEE Global Commun. Conf., Abu Dhabi, United Arab Emirates, [60] S. G. Neelam and P. R. Sahu, “Error performance of OTFS in the pres-
2018, pp. 206–212. ence of IQI and PA nonlinearity,” in Proc. IEEE Nat. Conf. Commun.,
[38] R. M. Augustine, G. D. Surabhi, and A. Chockalingam, “Space-time Kharagpur, India, 2020, pp. 1–6.
coded OTFS modulation in high-doppler channels,” in Proc. IEEE 89th [61] H. Zhang, X. Huang, and J. A. Zhang, “Comparison of OTFS diversity
Veh. Technol. Conf., Kuala Lumpur, Malaysia, 2019, pp. 1–6. performance over slow and fast fading channels,” in Proc. IEEE/CIC
[39] G. D. Surabhi and A. Chockalingam, “Low-complexity linear equaliza- Int. Conf. ICCC, Changchun, China, 2019, pp. 828–833.
tion for 2×2 MIMO-OTFS signals,” in Proc. IEEE 21st Int. Workshop [62] T. Thaj and E. Viterbo, “OTFS modem SDR implementation and exper-
SPAWC, Atlanta, GA, USA, 2020, pp. 1–5. imental study of receiver impairment effects,” in Proc. IEEE Int. Conf.
[40] A. Farhang, A. RezazadehReyhani, L. E. Doyle, and B. Farhang- Commun. Workshops, Shanghai, China, 2019, pp. 1–6.
Boroujeny, “Low complexity modem structure for OFDM-based [63] P. Raviteja, Y. Hong, E. Viterbo, and E. Biglieri, “Effective diversity
orthogonal time frequency space modulation,” IEEE Wireless Commun. of OTFS modulation,” IEEE Wireless Commun. Lett., vol. 9, no. 2,
Lett., vol. 7, no. 3, pp. 344–347, Jun. 2018. pp. 249–253, Feb. 2020.
[41] A. RezazadehReyhani, A. Farhang, M. Ji, R. R. Chen, and B. Farhang- [64] V. Rangamgari, S. Tiwari, S. S. Das, and S. C. Mondal, “OTFS:
Boroujeny, “Analysis of discrete-time MIMO OFDM-based orthogonal Interleaved OFDM with block CP,” in Proc. IEEE Nat. Conf. Commun.,
time frequency space modulation,” in Proc. IEEE Int. Conf. Commun., Kharagpur, India, 2020, pp. 1–6.
Kansas City, MO, USA, 2018, pp. 1–6. [65] X. Wu, S. Ma, and X. Yang, “Tensor-based low-complexity channel
[42] W. Anwar, A. Kumar, N. Franchi, and G. Fettweis, “Performance anal- estimation for mmWave massive MIMO-OTFS systems,” J. Commun.
ysis using physical layer abstraction modeling for 5G and beyond Inf. Netw., vol. 5, no. 3, pp. 324–334, Sep. 2020.
waveforms,” in Proc. IEEE Global Commun. Conf., Waikoloa, HI, [66] H. Zhao, Z. Kang, and H. Wang, “A novel channel estimation scheme
USA, 2019, pp. 1–6. for OTFS,” in Proc. IEEE 20th Int. Conf. Commun. Technol., Nanning,
[43] W. Shin, K. Kim, and Y. Ko, “Performance comparison of equalization China, 2020, pp. 12–16.
schemes for OTFS over time-varying multipath channels,” in Proc. [67] S. Tiwari, S. S. Das, and V. Rangamgari, “Low complexity
IEEE Int. Conf. Inf. Commun. Technol. Convergence, Jeju-do, South LMMSE receiver for OTFS,” IEEE Commun. Lett., vol. 23, no. 12,
Korea, 2020, pp. 1433–1438. pp. 2205–2209, Dec. 2019.
[44] B. Park, H. M. Kwon, and H. Ryu, “SFFT-based OTFS communication [68] W. Yuan, Z. Wei, J. Yuan, and D. W. K. Ng, “A simple variational bayes
system robust to high doppler and long delay channel,” in Proc. IEEE detector for orthogonal time frequency space (OTFS) Modulation,”
Int. Conf. Inf. Commun. Technol. Convergence, Jeju-do, South Korea, IEEE Trans. Veh. Technol., vol. 69, no. 7, pp. 7976–7980, Jul. 2020.
2020, pp. 850–853. [69] L. Jing, H. Wang, C. He, Y. Zhang, and H. Yin, “Two dimensional adap-
[45] Z. Wang, X. Chen, and X. Ning, “BER analysis of integrated WFRFT- tive multichannel decision feedback equalization for OTFS system,”
OTFS waveform framework over static multipath channels,” IEEE IEEE Commun. Lett., vol. 25, no. 3, pp. 840–844, Mar. 2021.
Commun. Lett., vol. 25, no. 3, pp. 754–758, Mar. 2021. [70] Y. Ge, Q. Deng, P. C. Ching, and Z. Ding, “Receiver design for OTFS
[46] B. Xu, Z. Xia, R. Liu, Y. Zhang, J. Hu, and W. Xie, “Research on with a fractionally spaced sampling approach,” IEEE Trans. Wireless
OTFS modulation applied in LTE-based 5G terrestrial broadcast,” in Commun., vol. 20, no. 7, pp. 4072–4086, Jul. 2021.
Proc. IEEE Int. Wireless Commun. Mobile Comput., Limassol, Cyprus, [71] H. Zhang and T. Zhang, “A low-complexity message passing detec-
2020, pp. 514–519. tor for OTFS modulation with probability clipping,” IEEE Wireless
[47] R. Bomfin, M. Chafii, A. Nimr, and G. Fettweis, “A robust base- Commun. Lett., vol. 10, no. 6, pp. 1271–1275, Jun. 2021.
band transceiver design for doubly-dispersive channels,” IEEE Trans. [72] B. C. Pandey, S. K. Mohammed, P. Raviteja, Y. Hong, and E. Viterbo,
Wireless Commun., vol. 20, no. 8, pp. 4781–4796, Aug. 2021. “Low complexity precoding and detection in multi-user massive
[48] W. Shen, L. Dai, S. Han, I. Chih-Lin, and R. W. Heath, “Channel esti- MIMO OTFS downlink,” IEEE Trans. Veh. Technol., vol. 70, no. 5,
mation for orthogonal time frequency space (OTFS) massive MIMO,” pp. 4389–4405, May 2021.
in Proc. IEEE Int. Conf. Commun., Shanghai, China, 2019, pp. 1–6. [73] S. S. Das, S. Tiwari, V. Rangamgari, and S. C. Mondal, “Performance
[49] W. Shen, L. Dai, J. An, P. Fan, and R. W. Heath, “Channel estimation of iterative successive interference cancellation receiver for LDPC
for orthogonal time frequency space (OTFS) massive MIMO,” IEEE coded OTFS,” in Proc. IEEE Int. Conf. Adv. Netw. Telecommun. Syst.,
Trans. Signal Process., vol. 67, no. 16, pp. 4204–4217, Aug. 2019. New Delhi, India, 2020, pp. 1–6.
[50] Y. Liu, S. Zhang, F. Gao, J. Ma, and X. Wang, “Uplink-aided high [74] T. Thaj and E. Viterbo, “Low complexity iterative rake detector for
mobility downlink channel estimation over massive MIMO-OTFS orthogonal time frequency space modulation,” in Proc. IEEE Wireless
system,” IEEE J. Sel. Areas Commun., vol. 38, no. 9, pp. 1994–2009, Commun. Netw. Conf., Seoul, South Korea, 2020, pp. 1–6.
Sep. 2020. [75] T. Thaj and E. Viterbo, “Low complexity iterative rake decision feed-
[51] L. Zhao, W. Guo, Y. Liu, J. Yang, and W. Wang, “Pilot optimization back equalizer for zero-padded OTFS systems,” IEEE Trans. Veh.
for OFDM-based OTFS systems over doubly selective channels,” Proc. Technol., vol. 69, no. 12, pp. 15606–15622, Dec. 2020.
IEEE Global Commun. Conf., Taipei, Taiwan, 2020, pp. 1–6. [76] S. Tiwari and S. S. Das, “Circularly pulse-shaped orthogonal time
[52] S. S. Das, V. Rangamgari, S. Tiwari, and S. C. Mondal, “Time domain frequency space modulation,” IEEE Electron. Lett., vol. 56, no. 3,
channel estimation and equalization of CP-OTFS under multiple frac- pp. 157–160, Feb. 2020.
tional dopplers and residual synchronization errors,” IEEE Access, [77] A. Pfadler, P. Jung, and S. Stanczak, “Pulse-shaped OTFS for V2X
vol. 9, pp. 10561–10576, 2021. short-frame communication with tuned one-tap equalization,” in Proc.
[53] L. Li, Y. Liang, P. Fan, and Y. Guan, “Low complexity detection algo- IEEE 24th Int. ITG WSA, Hamburg, Germany, 2020, pp. 1–6.
rithms for OTFS under rapidly time-varying channel,” in Proc. IEEE [78] M. N. Hossain, Y.e Sugiura, T. Shimamura, and H. Ryu, “Waveform
89th Veh. Technol. Conf., Kuala Lumpur, Malaysia, 2019, pp. 1–5. design of low complexity WR-OTFS system for the OOB power reduc-
[54] H. Qu, G. Liu, L. Zhang, S. Wen, and M. A. Imran, “Low-complexity tion,” in Proc. IEEE Wireless Commun. Netw. Conf., Seoul, South
symbol detection and interference cancellation for OTFS system,” IEEE Korea, 2020, pp. 1–8.
Trans. Commun., vol. 69, no. 3, pp. 1524–1537, Mar. 2021. [79] Z. Wei, W. Yuan, S. Li, J. Yuan, and D. W. K. Ng, “Transmitter
[55] C. Jin, Z. Bie, X. Lin, W. Xu, and H. Gao, “A simple two-stage and receiver window designs for orthogonal time–frequency space
equalizer for OTFS with rectangular windows,” IEEE Commun. Lett., modulation,” IEEE Trans. Commun., vol. 69, no. 4, pp. 2207–2223,
vol. 25, no. 4, pp. 1158–1162, Apr. 2021. Apr. 2021.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
XIAO et al.: OVERVIEW OF OTFS FOR INTERNET OF THINGS: CONCEPTS, BENEFITS, AND CHALLENGES 7617
[80] R. Bomfin, A. Nimr, M. Chafii, and G. Fettweis, “A robust and low- [104] R. Marsalek, J. Blumenstein, D. Schützenhöfer, and M. Pospisil,
complexity walsh-hadamard modulation for doubly-dispersive chan- “OTFS modulation and influence of wideband RF impairments mea-
nels,” IEEE Commun. Lett., vol. 25, no. 3, pp. 897–901, Mar. 2021. sured on a 60 GHz testbed,” in Proc. IEEE 21st Int. Workshop SPAWC,
[81] Y. Liang, L. Li, P. Fan, and Y. Guan, “Doppler resilient orthogonal Atlanta, GA, USA, 2020, pp. 1–5.
time–frequency space (OTFS) systems based on index modulation,” in [105] Z. Chen et al., “A survey on terahertz communications,” China
Proc. IEEE 91st Veh. Technol. Conf., Antwerp, Belgium, 2020, pp. 1–5. Commun., vol. 16, no. 2, pp. 1–35, Feb. 2019.
[82] H. Zhao, D. He, Z. Kang, and H. Wang, “Orthogonal time frequency [106] O.Tervo, T. Levaneny, K. Pajukoski, J. Hulkkonen, P. Wainio, and
space (OTFS) with dual-mode index modulation,” IEEE Wireless M. Valkama, “5G new radio evolution towards Sub-THz communi-
Commun. Lett., vol. 10, no. 5, pp. 991–995, May 2021. cations,” in Proc. 2nd 6G Wireless Summit, Levi, Finland, 2020, pp.
[83] V. Khammammetti and S. K. Mohammed, “OTFS-based multiple- 1–6.
access in high doppler and delay spread wireless channels,” IEEE [107] H. Ji, Y. Kim, T. Kim, and J. Oh, “Compressed OTFS transmission
Wireless Commun. Lett., vol. 8, no. 2, pp. 528–531, Apr. 2019. over Millimeter-wave and Terahertz spectrum,” in Proc. ICC workshop,
[84] R. M. Augustine and A. Chockalingam, “Interleaved time–frequency Montreal, QC, Canada, Jun. 2021, pp. 1–6.
multiple access using OTFS modulation,” in Proc. IEEE 90th Veh. [108] G. D. Surabhi, R. M. Augustine, and A. Chockalingam, “Peak-to-
Technol. Conf., Honolulu, HI, USA, 2019, pp. 1–5. average power ratio of OTFS Modulation,” IEEE Commun. Lett.,
[85] Z. Ding, R. Schober, P. Fan, and H. Vincent Poor, “OTFS-NOMA: An vol. 23, no. 6, pp. 999–1002, Jun. 2019.
efficient approach for exploiting heterogenous user mobility profiles,” [109] C. Naveen and V. Sudha, “Peak-to-average power ratio reduction in
IEEE Trans. Commun., vol. 67, no. 11, pp. 7950–7965, Nov. 2019. OTFS modulation using companding technique,” in Proc. IEEE Int.
[86] Z. Ding, “Robust beamforming design for OTFS-NOMA,” IEEE Open Conf. Devices Circuits Syst., Coimbatore, India, 2020, pp. 140–143.
J. Commun. Soc., vol. 1, pp. 33–40, 2020. [110] S. Gao and J. Zheng, “Peak-to-average power ratio reduction in pilot-
[87] O. K. Rasheed, G. D. Surabhi, and A. Chockalingam, “Sparse embedded OTFS modulation through iterative clipping and filtering,”
delay-doppler channel estimation in rapidly time-varying channels for IEEE Commun. Lett., vol. 24, no. 9, pp. 2055–2059, Sep. 2020.
multiuser OTFS on the uplink,” in Proc. IEEE 91st Veh. Technol. Conf., [111] T. Jiang, W. Xiang, P. C. Richardson, J. Guo, and G. Zhu, “PAPR
Antwerp, Belgium, 2020, pp. 1–5. reduction of OFDM signals using partial transmit sequences with low
[88] A. K. Sinha, S. K. Mohammed, P. Raviteja, Y. Hong, and computational complexity,” IEEE Trans. Broadcast., vol. 53, no. 3,
E. Viterbo, “OTFS based random access preamble transmission for pp. 712–724, Sep. 2007.
high mobility scenarios,” IEEE Trans. Veh. Technol., vol. 69, no. 12, [112] T. Jiang, C. Li, and C. Ni, “Effect of PAPR reduction on spectrum and
pp. 15078–15094, Dec. 2020. energy efficiencies in OFDM systems with class-A HPA over AWGN
[89] W. Shao, S. Zhang, C. Zhong, X. Lei, and P. Fan, “Angle-delay-doppler channel,” IEEE Trans. Broadcast., vol. 59, no. 3, pp. 513–519, Sep.
domain NOMA over massive MIMO-OTFS networks,” in Proc. IEEE 2013.
Int. Conf. Commun. China, Chongqing, China, 2020, pp. 74–79. [113] A. Maatouk, E. Caliskan, M. Koca, M. Assaad, G. Gui, and H. Sari,
[90] M. Li, S. Zhang, P. Fan, and O. A. Dobre, “Multiple access for massive “Frequency-domain NOMA with two sets of orthogonal signal wave-
MIMO-OTFS networks over angle-delay-doppler domain,” Proc. IEEE forms,” IEEE Commun. Lett., vol. 22, no. 5, pp. 906–909, May 2018.
Global Commun. Conf., Taipei, Taiwan, 2020, pp. 1–6. [114] S. Dogan, A. Tusha, and H. Arslan, “NOMA with index modulation for
[91] M. Li, S. Zhang, F. Gao, P. Fan, and O. A. Dobre, “A new path division uplink URLLC through grant-free access,” IEEE J. Sel. Topics Signal
multiple access for the massive MIMO-OTFS networks,” IEEE J. Sel. Process., vol. 13, no. 6, pp. 1249–1257, Oct. 2019.
Areas Commun., vol. 39, no. 4, pp. 903–918, Aug. 2021.
[115] G. Beradinelli, “Generalized DFT-s-OFDM waveforms without cyclic
[92] A. Chatterjee, V. Rangamgari, S. Tiwari, and S. S. Das, “Nonorthogonal
prefix,” IEEE Access, vol. 6, pp. 4677–4689, Dec. 2017.
multiple access with orthogonal time–frequency space signal transmis-
[116] Y. Yang, L. Zhu, and X. Mao, “A GFDM-CDMA scheme for hybrid
sion,” IEEE Syst. J. vol. 15, no. 1, pp. 383–394, Mar. 2021.
satellite-terrestrial communication system,” in Proc. Int. Symp. Netw.
[93] T. Blazek, H. Groll, S. Pratschner, and E. Zochmann, “Vehicular chan-
Comput. Commun., Istanbul, Turkey, 2019, pp. 1–6.
nel characterization in orthogonal time–frequency space,” in Proc.
[117] D. Chen, W. Wang, and T. Jiang, “New multicarrier modulation for
IEEE Int. Conf. Commun. Workshops, Shanghai, China, 2019, pp. 1–5.
satellite-ground transmission in space information networks,” IEEE
[94] W. Anwar, A. Krause, A. Kumar, N. Franchi, and G. P. Fettweis,
Netw., vol. 34, no. 1, pp. 101–107, Jan./Feb. 2020.
“Performance analysis of various waveforms and coding schemes in
V2X communication scenarios,” in Proc. IEEE Wireless Commun. [118] A. Tusha, S. Dogan, and H. Arslan, “A hybrid downlink NOMA with
Netw. Conf., Seoul, South Korea, 2020, pp. 1–8. OFDM and OFDM-IM for beyond 5G wireless networks,” IEEE Signal
[95] J. Cheng, C. Jia, H. Gao, W. Xu, and Z. Bie, “OTFS based receiver Process. Lett., vol. 27, pp. 491–495, 2020.
scheme with multi-antennas in high-mobility V2X systems,” in Proc. [119] X. Liu, T. Xu, and I. Darwazeh, “Coexistence of orthogonal and non-
IEEE Int. Conf. Commun. Workshops, Dublin, Ireland, 2020, pp. 1–6. orthogonal multicarrier signals in beyond 5G scenarios,” in Proc. 2nd
[96] M. J. Bocus, A. Doufexi, and D. Agrafiotis, “Performance of OFDM- 6G Wireless Summit, levi, Finland, 2020, pp. 1–5.
based massive MIMO OTFS systems for underwater acoustic commu- [120] Z. E. Ankarali, B. Peköz, and H. Arslan, “Flexible radio access beyond
nication,” IET Commun., vol. 14, no. 4, pp. 588–593, Jan. 2020. 5G: A future projection on waveform, numerology, and frame design
[97] X. Feng et al., “Underwater acoustic communications based on OTFS,” principles,” IEEE Access, vol. 5, pp. 18295–18309, 2017.
in Proc. 15th IEEE Int. Conf. Signal Process., Beijing, China, 2020, [121] S. N. Jenkins, B. A. Kenney, A. J. Majidy, H. Moradiy, and B. Farhang-
pp. 439–444. Boroujeny, “CP-DSSS: A novel waveform for multiple access in IoT,”
[98] T. Li, R. He, B. Ai, M. Yang, Z. Zhong, and H. Zhang, “OTFS mod- in Proc. 2nd 6G Wireless Summit, levi, Finland, 2020, pp. 1–5.
ulation performance in a satellite-to-ground channel at sub-6-GHz and [122] Z. Zhang et al., “6G wireless networks: Vision, requirements, architec-
millimeter-wave bands with high mobility,” Front. Inform. Tech. El., ture, and key technologies,” IEEE Veh. Technol. Mag., vol. 14, no. 3,
vol. 22, no. 4, pp. 517–526, 2021. pp. 28–41, Sep. 2019.
[99] J. Hu, J. Shi, S. Ma, and Z. Li “Secrecy analysis for orthogonal [123] H. Viswanathan and P. E. Mogensen, “Communications in the 6G era,”
time frequency space scheme based uplink LEO satellite communi- IEEE Access, vol. 8, pp. 57063–57074, Mar. 2020.
cation,” IEEE Wireless Commun. Lett., vol. 10, no. 8, pp. 1623–1627, [124] F. Tang, Y. Kawamoto, N. Kato, and J. Liu, “Future intelligent and
Aug. 2021. secure vehicular network toward 6G: Machine-learning approaches,”
[100] X. Zhou and Z. Gao, “Joint active user detection and channel estimation Proc. IEEE, vol. 108, no. 2, pp. 292–307, Feb. 2020.
for grant-free NOMA-OTFS in LEO constellation Internet-of-Things,” [125] K. Letaief, W. Chen, Y. Shi, J. Zhang, and Y. Zhang, “The roadmap to
2021, arXiv:2108.01520. 6G: AI empowered wireless networks,” IEEE Commun. Mag., vol. 57,
[101] R. Hadani et al., “Orthogonal time frequency space (OTFS) modulation no. 8, pp. 84–90, Aug. 2019.
for millimeter-wave communications systems,” in Proc. IEEE MTT-S [126] N. Kato, B. Mao, F. Tang, Y. Kawamoto, and J. Liu, “Ten challenges in
Int. Microw. Symp., Honololu, HI, USA, 2017, pp. 681–683. advancing machine learning Technologies toward 6G,” IEEE Wireless
[102] G. D. Surabhi, M. K. Ramachandran, and A. Chockalingam, “OTFS Commun., vol. 27, no. 3, pp. 96–103, Jun. 2020.
modulation with phase noise in mmWave communications,” in Proc. [127] A. Yazar and H. Arslan, “A waveform parameter assignment framework
IEEE 89th Veh. Technol. Conf., Kuala Lumpur, Malaysia, 2019, for 6G with the role of machine learning,” IEEE Open J. Veh. Technol.,
pp. 1–5. vol. 1, pp. 156–172, May 2020.
[103] T. Blazek and D. Radovic, “Performance evaluation of OTFS over [128] T. Mao, Q. Wang, Z. Wang, and S. Chen, “Novel index modulation
measured V2V channels at 60 GHz,” in Proc. IEEE MTT-S Int. Conf. techniques: A survey,” IEEE Commun. Surveys Tuts., vol. 21, no. 1,
Microw. Intell. Mobility, Linz, Austria, 2020, pp. 1–4. pp. 315–348, 1st Quart., 2019.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.
7618 IEEE INTERNET OF THINGS JOURNAL, VOL. 9, NO. 10, MAY 15, 2022
[129] P. Yang, M. Di Renzo, Y. Xiao, S. Q. Li, and L. Hanzo, “Design guide- [139] Y. Yang, B. Zheng, S. Zhang, and R. Zhang, “Intelligent
lines for spatial modulation,” IEEE Commun. Surveys Tuts., vol. 17, reflecting surface meets OFDM: Protocol design and rate
no. 1, pp. 6–26, 1st Quart., 2015. maximization,” IEEE Trans. Commun., vol. 68, no. 7, pp. 4522–4535,
[130] P. Liu and T. Jiang, “Channel estimation performance analysis of mas- Jul. 2020.
sive MIMO IoT systems with ricean fading,” IEEE Internet Things J., [140] C. Pradhan, A. Li, L. Song, J. Li, B. Vucetic, and Y. Li, “Reconfigurable
vol. 8, no. 7, pp. 6114–6126, Apr. 2021. intelligent surface (RIS)-enhanced two-way OFDM communica-
[131] S. Yang and L. Hanzo, “Fifty years of MIMO detection: The road tions,” IEEE Trans. Veh. Technol., vol. 69, no. 12, pp. 16270–16275,
to large-scale MIMOs,” IEEE Commun. Surveys Tuts., vol. 17, no. 4, Dec. 2020.
pp. 1941–1988, 4th Quart., 2015. [141] B. Zheng and R. Zhang, “Intelligent reflecting surface-enhanced
[132] M. Albreem, M. Juntti, and S. Shahabuddin, “Massive MIMO detection OFDM: Channel estimation and reflection optimization,” IEEE Wireless
techniques: A survey,” IEEE Commun. Surveys Tuts., vol. 21, no. 4, Commun. Lett., vol. 9, no. 4, pp. 518–522, Apr. 2020.
pp. 3109–3132, 4th Quart., 2019. [142] S. Lin, B. Zheng, G. C. Alexandropoulos, M. Wen, F. Chen, and
[133] L. Xiao et al., “Efficient compressive sensing detectors for generalized S. sMumtaz, “Adaptive transmission for reconfigurable intelligent
spatial modulation systems,” IEEE Trans. Veh. Technol., vol. 66, no. 2, surface-assisted OFDM wireless communications,” IEEE J. Sel. Areas
pp. 1284–1298, Feb. 2017. Commun., vol. 38, no. 11, pp. 2653–2665, Nov. 2020.
[134] L. Xiao, P. Xiao, Z. Liu, W. Yu, H. Hass, and L. Hanzo, “A compres- [143] B. Zheng, C. You, and R. Zhang, “Intelligent reflecting surface
sive sensing assisted massive SM-VBLAST system: Error probability assisted multi-user OFDMA: Channel estimation and training design,”
and capacity analysis,” IEEE Trans. Wireless Commun., vol. 19, no. 3, IEEE Trans. Wireless Commun., vol. 19, no. 12, pp. 8315–8329,
pp. 1990–2005, Mar. 2020. Dec. 2020.
[135] J. Gao, Y. Wu, Y. Wang, W. Zhang, and W. Fan, “Uplink transmis- [144] H. Li, G. Ru, Y. Kim, and H. Liu, “OFDMA capacity analysis in MIMO
sion design for crowded correlated cell-free massive MIMO-OFDM channels,” IEEE Trans. Inf. Theory, vol. 56, no. 9, pp. 4438–4446, Sep.
systems,” Sci. China Inf. Sci., vol. 64, 2021, Art. no. 182309, 2010.
doi: 10.1007/s11432–020-3103–3. [145] S. Rakib and R. Hadani, “Multiple access in wireless telecommuni-
[136] N. Athreya, V. Raj, and S. Kalyani, “Beyond 5G: Leveraging cell cations system for high-mobility applications,” U.S. Patent 9 722 741
free TDD massive MIMO using cascaded deep learning,” IEEE B1, Aug. 2017.
Wireless Communi. Lett., vol. 9, no. 9, pp. 1533–1537, Sep. 2020, [146] V. Khammammetti and S. K. Mohammed, “OTFS based multiple-
doi: 10.1109/LWC.2020.2996745. access in high Doppler and delay spread wireless channels,”
[137] M. Ke, Z. Gao, Y. Wu, X. Gao, and K.-K. Wong, “Massive IEEE Wireless Commun. Lett., vol. 8, no. 2, pp. 528–531,
access in cell-free massive MIMO-based internet of Things: Apr. 2019.
Cloud computing and edge computing Pparadigms,” IEEE J. [147] G. D. Surabhi, R. M. Augustine, and A. Chockalingam, “Multiple
Sel. Areas Commun., vol. 39, no. 3, pp. 756–772, Mar. 2021, access in the delay-Doppler domain using OTFS modulation,” 2019,
doi: 10.1109/JSAC.2020.3018807. arXiv1902.03415.
[138] Y. Chen, D. Chen, Y. Tian, and T. Jiang, “Spatial lobes division-based [148] Y. Ge, Q. Deng, P. Ching, and Z. Ding, “OTFS signaling
low complexity hybrid precoding and diversity combining for mmWave for uplink NOMA of heterogeneous mobility users,” IEEE
IoT systems,” IEEE Internet Things J., vol. 6, no. 2, pp. 3228–3239, Trans. Commun., vol. 69, no. 5, pp. 3147–3161, May 2021,
Apr. 2019. doi: 10.1109/TCOMM.2021.3059456.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY DELHI. Downloaded on November 02,2022 at 04:21:10 UTC from IEEE Xplore. Restrictions apply.