IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 11, NO.

8, AUGUST 2022 1689

Exploiting User Diversity in OTFS Transmission for

Beyond 5G Wireless Systems
Armed Tusha , Saud Althunibat , Senior Member, IEEE, Mazen O. Hasna , Senior Member, IEEE,
Khalid Qaraqe , Senior Member, IEEE, and Hüseyin Arslan , Fellow, IEEE

Abstract—Orthogonal time frequency space (OTFS) modula- technologies of the wireless communication realm depend on
tion is a recent technology that offers a significant advantage orthogonal frequency division multiplexing (OFDM) and time
on link reliability due to its strong delay-Doppler resilience. In division multiplexing (TDM) which have strong frequency and
classical cellular technology, overall system performance is lim-
time selectivity resilience, respectively [2]. Unfortunately, the
ited by the fading and interference experienced by particular
user equipment (UE) over its radio resources (RR). Besides, system performance of OFDM-based and TDM-based trans-
OTFS performance depends on the channel diversity of the mitted signals is limited by interference in the case of a doubly
UEs. Therefore, this letter examines the potential of multi- dispersive wireless channel, which is inevitable to avoid in
user diversity for OTFS systems considering practical scenarios. future mobile networks. Hence, a recent two-dimensional
Specifically, we introduce a novel scheduling algorithm that pro- (2-D) signaling technique named orthogonal time frequency
poses the assignment of the available RR of OTFS frame to space (OTFS) is proposed in [3] in order to fully exploit the
the UE with the largest number of channel taps, resulting in
enhanced channel diversity. It is shown via both mathemati- communication link within a selective wireless environment.
cal analysis and simulation results that the proposed scheduling Basically, OTFS-based communication schemes convey the
technique outperforms the scheme with random access to delay- data information in delay-Doppler domain that offers a more
Doppler RR in terms of system bit error rate (BER) with comprehensive description of the environmental motion, sur-
maximum likelihood (ML) receiver. The BER results validate rounding reflectors, and number of arrival arrays, compared
the accuracy of the proposed technique for OTFS transmission to the scenarios with either time or frequency channel state
with minimum mean square error (MMSE) detector considering
various system configurations. information (CSI) [3]. Since OTFS transmission is able to
benefit from the full diversity of wireless channels, it has
Index Terms—OTFS, multi-user diversity, doubly dispersive attracted the attention of academia and industry [4]. In [5],
channel, ML, MMSE.
Raviteja et al. have exploited delay-Doppler domain repre-
sentation of the wireless channel in order to perform target
range and velocity sensing of the surrounding obstacles in the
I. I NTRODUCTION environment. Moreover, in [6], Rakib et al. have assessed the
HE EXPONENTIAL growth of new services and applica- concept of OTFS not only as a potential sensing and mod-
T tions with a wide range of requirements that need more
and more link reliability, high throughput, power efficiency,
ulation technique but also as a multiple accessing scheme,
where different users are allocated at distinct radio resources in
and spectral efficiency explicitly shows that the design of new the delay-Doppler domain. In addition, Ding et al. have com-
wireless technologies has become a necessity in beyond fifth- bined OTFS with non-orthogonal multiple accessing (NOMA),
generation (5G) wireless systems [1]. The most prominent in [7]. Ge et al. have proposed OTFS-based NOMA with
an efficient turbo receiver for uplink transmission, in [8].
Manuscript received 11 April 2022; accepted 9 May 2022. Date of publi- Moreover, authors in [9] and [10] have discussed the inte-
cation 11 May 2022; date of current version 9 August 2022. This publication gration of OTFS signals in multiple-input multiple-output
was made possible by grant number NPRP12S-0225-190152 from the Qatar technology and the design of proper pulse shaped signals,
National Research Fund, QNRF. The statements made herein are solely the
responsibility of the authors. The associate editor coordinating the review of respectively, improving the link reliability, and enhancing the
this article and approving it for publication was R. Wang. (Corresponding overall spectral efficiency of the system accordingly. On the
author: Armed Tusha.) other hand, several works in the literature have investigated the
Armed Tusha is with the Department of Electrical and Electronics
Engineering, Istanbul Medipol University, 34810 Istanbul, Turkey, and also
performance of OTFS transmission under radio frequency (RF)
with the Department of Electrical and Computer Engineering, Texas A&M imperfections, such as peak-to-average power ratio (PAPR),
University at Qatar, Doha, Qatar (e-mail: [email protected]). phase noise (PN) and in-phase and quadrature (IQ) imbalance
Saud Althunibat is with the Department of Communications Engineering, for practical wireless communication scenarios [11]–[13].
Al-Hussein Bin Talal University, Ma’an 71111, Jordan (e-mail:
[email protected]). Despite the aforementioned opportunities and applications
Mazen O. Hasna is with the Department of Electrical Engineering, Qatar of OTFS signaling, there is still limited research activities by
University, Doha, Qatar (e-mail: [email protected]). both academia and industry to take advantage of the channel
Khalid Qaraqe is with the Department of Electrical and Computer
Engineering, Texas A&M University at Qatar, Doha, Qatar (e-mail:
independence of different user equipment (UE) under practical
[email protected]). scenarios with multi-user systems [14]–[17]. Different from
Hüseyin Arslan is with the Department of Electrical and Electronics classical time/frequency-based multi-carrier schemes, in [18],
Engineering, Istanbul Medipol University, 34810 Istanbul, Turkey, and also Li et al. explicitly show that the diversity of OTFS-based
with the Department of Electrical Engineering, University of South Florida,
Tampa, FL 33620 USA (e-mail: [email protected]). wireless technologies becomes dominant with the increase of
Digital Object Identifier 10.1109/LWC.2022.3174455 the number of resolvable incoming paths. To the best of the
2162-2345 
c 2022 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: BMS College of Engineering. Downloaded on August 08,2024 at 06:20:24 UTC from IEEE Xplore. Restrictions apply.
1690 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 11, NO. 8, AUGUST 2022

Transform (ISFFT), we convert the data symbols to time-

frequency domain. Accordingly, we apply N-point inverse Fast
Fourier Transform (IFFT) and M-point FFT operation over
the rows and columns of X u , respectively. To generate the
time domain OTFS signal, we perform Heisenberg Transform
by using M-point IFFT over time-frequency data symbols.
Consequently, time domain OTFS frame can be written as
S u = G tx F H H
M (F M X u F N ), where G tx denotes a diago-
nal matrix that has the samples of the pulse shape in time
domain, and F N denotes the DFT matrix given as
Fig. 1. Channel model for a multi-user scenario in downlink transmission.  
1 −j 2πkl
[F N ]k ,l = √ exp , k , l ∈ {0, 1, . . . , N − 1}. (2)
author’s knowledge, the influence of the multi-user diversity N N
with the focus on varying number of channel taps has not yet
The Hermitian of N-point DFT is F H N . It is worth mention-
been investigated in the literature, and the impact of diversity
ing that G tx corresponds to the identity matrix in case of
of each UE with OTFS transmission on the error performance
rectangular pulse shaping, where considering the column-wise
of the overall system is unknown. Hence, the main motivation
vectorization of S u , we can write it as
of this letter is to close this gap and to provide useful insights
for both academia and industry. In this regard, this letter pro- s u = (F H
N ⊗ G tx )x u , (3)
poses a novel scheduling algorithm for multi-user OTFS-based
wireless technology in which the BS attempts to maximize the where ⊗ denotes the Hadamard product and x u = vec(X u ).
average link reliability of the overall system via utilization of We add a cyclic prefix (CP) of L length in order to mitigate
the diverse number of channel taps of each UE in downlink the inter-symbol interference between the OTFS signals, and
transmission. thus, L is larger than the number of channel paths (Pu ). After
the transmission, the OTFS signal s u undergoes Pu −1thei time-
II. P ROPOSED S YSTEM M ODEL varying wireless channel given as hu (τ, υ) = i=0 hu δ(τ −
τui )δ(υ − υui ), where hui , τui , υui , denote the path gain, delay
In this letter, we consider a single cell with Nu active users
shift and Doppler shift of the i-th channel path, respectively,
that are served by one base station in downlink transmission, as
and δ(·) is the Dirac delta function. Moreover, the delay and
illustrated in Fig. 1. The BS selects the UE for transmission
Doppler bins for the i-th channel tap equal to τui = lui /(M Δf )
based on the corresponding CSI of each UE. Basically, in and υui = (kui + κ)/(NT ), respectively, where κ denotes
downlink systems with multi-user transmission, the ultimate the fractional Doppler. Moreover, we neglect the impact of
goal is to maximize the achievable system throughput, and fractional delay, since the typical value of the sampling time
thus, the BS in classical wireless technologies chooses the 1/(M Δf ) is sufficiently small especially in wideband wireless
UE with the best SNR for the next transmission time. Since technologies.
the impact of channel fading reduces with the increase of the At the receiver side, after discarding the additional CP, the
number of channel taps in OTFS transmission [18], we propose received signal is given as follows
to allocate the radio resources (RR) of the OTFS frame to the
UE with the largest number of channel taps, improving the r u = H u su + w , (4)
error performance of the overall system that is formulated as
where w ∈ CMN ×1 denotes the additive white Gaussian
follow
noise (AWGN) random variable with CN (0, N0 I ). H u ∈
Δ
C MN ×MN
Pu = max{Pi } Pu −1 i l is k the channel matrix given by H u =
i=0 hu ˝ Δ , where Π = circ[0, 1, 0, . . . , 0]MN ×MN
i i
s.t. Pmin ≤ Pi ≤ Pmax , for i ∈ [0, 1, . . . , Nu − 1], (1)
is a permutation matrix based on forward cyclic shift, and
where Pu denotes the path number of the selected UE for Δ = diag[z 0 , z 1 , . . . , z MN −1 ] ∈ CMN ×MN with z =
OTFS transmission. It is worth mentioning that the number exp (j 2π/MN ). In order to get the delay-Doppler domain
of channel taps for each UE is known at the BS. In this symbols, we first perform Wigner transform as F M G rx Ru ,
letter, we consider a system model with single antennas at given that Ru = vec−1 (r u ) and G rx is a diagonal matrix
transmitter (Tx) and receiver (Rx). The system frame con- containing samples of the receiver pulse shape. Following
sists of N number of symbols and M number of subcarriers that, we perform SFFT operation to get the received signal in
with T symbol duration and Δf subcarrier spacing (SCS), the delay-Doppler domain as Y u = F H M (F M G rx R u )F N .
respectively. Therefore, the OTFS frame occupies a total band- After vectorization, the received signal is written as
width of B = M Δf with a frame duration Tf = TN ,
y u = H u,eff x u + w̄ , (5)
while T Δf = 1. Consequently, the OTFS signal conveys a
block of N x M uncoded Q-ary data symbols x [k , l ] ∈ A, where H u,eff = (F N ⊗G rx )H u (F H N ⊗G tx ) is the effective
where A = {a0 , a1 , . . . , aQ−1 } is the modulation alphabet. channel and w̄ is the noise in the delay-Doppler domain. In
X u ∈ CM ×N represents data symbols in the delay-Doppler case of using a rectangular pulse shape, i.e., G rx = I M ,
grid for the selected user u with xu [k , l ] being an element w̄ preserves the same statistical features of w. Regarding (5),
of X u . After we apply Inverse Symplectic Fast Fourier we use either maximum likelihood (ML) or minimum mean

Authorized licensed use limited to: BMS College of Engineering. Downloaded on August 08,2024 at 06:20:24 UTC from IEEE Xplore. Restrictions apply.
TUSHA et al.: EXPLOITING USER DIVERSITY IN OTFS TRANSMISSION FOR BEYOND 5G WIRELESS SYSTEMS 1691

square error (MMSE) detector to estimate the data symbols as To assess the error performance of the proposed OTFS tech-
follows nique, we utilize the union bound error performance, which is
given as
x̂ u,ML = arg min||y − H u,eff x u ||2 , (6a) 
x∈ANM
Pr{x u → x ^u } ≤ Pr{x u → x^ u |Pu }fPu (Pu )dPu , (12)
x̂ u,MMSE = HH H
u,eff (H u,eff H u,eff + N0 I )−1 y u , (6b) Pu

where x̂ u = vec(X̂ u ) and we demodulate x̂ u ∈ CMN ×1 where Pr{x u → x^ u | Pu } denotes the PEP in case the trans-
symbols considering the modulation alphabet (A) in order to mitted symbol block x u is erroneously detected as x ^ u for
get the information bits. the desired UE with Pu channel spreading. In line with the
principle of ML receiver, we give the conditional PEP as
III. P ERFORMANCE A NALYSIS OF THE P ROPOSED follows
OTFS S CHEDULING S CHEME
^ u | h u , υ u , τ u , Pu }
Pr{x u → x
In this section, we outline the theoretical analysis in order to  
= Pr ||y u − Φυ u ,τ u (x )u h u ||2 > ||y u − Φυ u ,τ u (^
x )u h u ||2 ,
investigate the performance of the proposed multi-user OTFS
scheme. In particular, we derive the closed-form expression (13)
of the average bit error rate (ABER). To analyze the overall
error performance, we rewrite the input-output relation (5) as where, after performing some mathematical manipulations and
follows substituting (7) in (13), the upper bound of conditional PEP
is given as in [19]
y u = Φυ u ,τ u (x u )h u + w , (7) 
E{s}  2  ¯l 2
r
^u | h u , υ u , τ u , Pu } ≤ exp
Pr{x u → x − λl hu ,
where h u = [hu0 , hu1 , . . . , huPu −1 ]T ,
and each element of h u 4N0
(14)
l=1
represents an independent and identically distributed com-
plex Gaussian random variable as hui ∼ CN (μu , 1/Pu ). where E{s} is the symbol energy, λl ’s are the diagonal ele-
Φυ u ,τ u (x u ) is a concatenated matrix with size MN × Pu that ments of Γ, representing the singular values of D = Φ(^ xu)−
involves the data vector x u given as Φ(x u ). Moreover, D H D = U ΓU H and h̄ u = U h H u while
,
U U H = I . In this regard, r represents the rank of the matrix
Φυ u ,τ u (x u ) = [Ψ0 x u | Ψ1 x u | · · · | ΨPu −1 x u ], (8) Γ, and λl ∈ diag{Γ} with l ∈ [1, 2, . . . , r ]. After averag-
where Ψi = (F N ⊗ I M )˝li Δki (F H
⊗ I M ). In this let-
N
ing over (14) with the aid of the moment generating function
ter, υ u and τ u represent the vectors of Doppler values and (MGF) [20] and considering each |h̄ul |2 , we derive the upper
delay values for the user of interest, respectively, where bound of the unconditioned PEP under a given delay and
υ u = [ku0 , ku1 , . . . , kuPu −1 ] and τ u = [lu0 , lu1 , . . . , luPu −1 ]. Doppler realizations [21] as follows
 r 
 λ2 −1  γ −r
Moreover, these indices follow a discrete uniform distribution.
To calculate the pairwise error probability (PEP), we assume Pr{x u → x ^ u | υ u , τ u , Pu } ≤ l , (15)
ML estimation and ideal CSI at the receiver. Pu 4
l=1
In particular for this letter, we consider the number of chan-
nel tap of each user as a sequence Z0 , Z1 , . . . , ZNu−1 of i.i.d where γ = E{s}/N0 . Moreover, referring to [18], an approx-
random variables with support on [Pmin , Pmax ] and following imation of the upper bound for the unconditional PEP can be
a probability density function (PDF) and cumulative distribu- given as
tion function (CDF) given by fZ (z ) and FZ (z ), respectively. −r
γ
In this line, we use the random variable Pu to model the user ^ u | Pu }  Θ−r
Pr{x u → x , (16)
4Pu
performing transmission with maximum number of channel
taps, given as Pu = max Zi , where the PDF of Pu is where Θ = e H e represents the squared Euclidean distance
0≤i≤Nu −1
derived in the followings. Considering the fact that Pu < z between any pair of x
^ and x in particular, whilst e = x u −^
xu.
if and only if all elements of the sample are less than z, and Consequently, to obtain the closed form expression of the
Zi ’s are i.i.d random variables, we can deduce that unconditional PEP for the proposed OTFS-based scheduling
scheme we need to combine (11) and (16), and substitute
Pr{Pu ≤ z } = Pr{Z0 ≤ z , . . . , ZNu−1 ≤ z }, (9a) into (12). Hence, we rewrite (12) with the aid of the binomial
Nu−1
 theorem as
= Pr(Zi ≤ z ) = [FZ (z )]Nu , (9b) Nu −1
i=0 (−Pmin )Nu −1−s
Pr{x u → x
^u } ≤ ζ Υ, (17)
s!(Nu − 1 − s)!
where [FZ (z )]Nu represents the CDF of Pu . Hence, after s=0
some mathematical manipulations, the PDF of Pu is given as
where ζ = Nu !Θ−r (Pmax − Pmin )−Nu , and Υ equals
Nu −1 
fPu (z ) = Nu [FZ (z )] fZ (z ). (10) 4 r
Υ= Pus+r dPu . (18)
In case Zi is uniformly distributed on [Pmin , Pmax ], the PDF Pu γ
of Pu is given as follows
We need to integrate (18) in order to have a closed form
(z − Pmin )Nu −1 solution for the unconditional PEP in (17). Therefore, we per-
fPu (z ) = Nu . (11)
(Pmax − Pmin )Nu form some mathematical manipulations in order to simplify

Authorized licensed use limited to: BMS College of Engineering. Downloaded on August 08,2024 at 06:20:24 UTC from IEEE Xplore. Restrictions apply.
1692 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 11, NO. 8, AUGUST 2022

Fig. 2. Theoretical vs. simulation results of the proposed scheme.

the integral part of (17) as follows

 γ s+1 
Υ= κ (s+r ) d κ, (19)
4 κ

where κ = 4P u
γ . After combining (19) with (17), the upper
bound of the unconditional PEP reduces to
u −1 
γ s+1 (−Pmin )Nu −1−s κ s+r +1 κx
N
Pr{x u → x
^u } ≤ ζ , (20)
4 s!(Nu − 1 − s)! s + r + 1 κn
s=0

where κn = 4Pγmin and κx = 4Pγmax . Then, the ABEP of the

proposed system model can be evaluated by
B −1
Pr{e} = ^u }ε (x u , x
Pr{x u → x ^u ), (21) Fig. 3. BER performances for the proposed multi-user OTFS scheme
NM log2 Q considering various practical scenarios.
x x
^

where ε(x u , x^ u ) represents the bit-based Hamming distance

scheme enhances significantly the total error performance of
among particular x u and x ^ u pair, NM log2 Q is the total num-
OTFS systems as illustrated via the solid-line with circle
ber of bits transmitted per OTFS transmission, and B = Q NM .
marker. Moreover, the error performance of the system get
worse while the UE with minimum channel taps performs
IV. N UMERICAL R ESULTS AND D ISCUSSION transmission via the solid-line with square marker. Thus, Fig. 2
In this section, we assess the performance of the proposed explicitly illustrates the wide range on the performance dif-
multi-user OTFS scheme via Monte-Carlo based simulations ference between the two possible user scheduling scenarios
considering various practical scenarios. In this letter, we with extreme channel tap configurations. In line with this, the
assume the delay index and Doppler index are uniformly gen- dashed-line with circle marker validates the theoretical results,
erated from the set {0, . . . , M − 1} and {0, . . . , N − 1}, which matches with the simulation results for the proposed
respectively. To have a better understanding of the derived scheduling scheme. We numerically average the conditional
mathematical equations and to assess their validity, we first error performance subjected to all possible delay and Doppler
simulate the BER performance of the proposed OTFS-based realizations in order to obtain the multi-user diversity gain.
multiple accessing scheme with binary phase shift keying The proposed scheme enhances the reliability of OTFS by at
(BPSK) modulation, considering ML receiver. It is worth men- least 10dB compared with the random scheduling scenario at
tioning that since the complexity of ML detector increases 10−4 BER. It is worth noting that the performance of OTFS
exponentially with the parameters of OTFS frame. Hence, we transmission with low values of N, M and Pu does not guar-
select small values of N and M to assess the multi-user diver- antee the full diversity within all possible realizations of Γ,
sity coming from the time/frequency dispersion of the wireless resulting in gap between simulation results and theory due
channel of each UE in OTFS-based systems. to the union bound technique (16). Different from Fig. 2,
Fig. 2 shows the BER performance of three different Fig. 3 and Fig. 3(b) illustrate the error performance of the
multi-user OTFS schemes with N = 2, M = 4 for BPSK trans- proposed scheduling scheme with practical values of (N, M)
mission. The solid-line with diamond marker represents the and MMSE detector. Even in the case of small OTFS param-
performance of multi-user OTFS scenario in case the RR are eters (M = 2, N = 4), our proposed scheme offers at least
randomly assigned to a single UE, whilst the number of chan- 2dB more gain compared to the random scenario. It is impor-
nel taps Pi for each UE is uniformly distributed in the range tant to notice that, since the diversity of OTFS transmission
of [Pmin , . . . , Pmax ]. In other words, this result shows the with small parameters is significantly reduced, ML and MMSE
performance of OTFS in case no scheduling mechanism exists detectors offer similar system error performance. Fig. 3 shows
to exploit the multi-user diversity. Alternatively, an intelligent the performance of an OTFS system with practical parameters
assignment of the RR with the aid of our proposed scheduling (M = 16, N = 7) where the proposed scheme offers at least

Authorized licensed use limited to: BMS College of Engineering. Downloaded on August 08,2024 at 06:20:24 UTC from IEEE Xplore. Restrictions apply.
TUSHA et al.: EXPLOITING USER DIVERSITY IN OTFS TRANSMISSION FOR BEYOND 5G WIRELESS SYSTEMS 1693

number of channel taps in order to improve the exploitation

of the multi-user channel diversity. Moreover, we have derived
the closed-form BER upper-bound expression for the proposed
scheduling scheme. Accordingly, both the simulation and the
theoretical results show that the proposed technique offers a
significant improvement in error performance compared with
random RR assignment scheme. It is worth mentioning that the
generalization of this letter by taking into account fading char-
acteristics, various UE velocities, and inter-user interference in
uplink are left as a future work.

R EFERENCES
Fig. 4. The proposed scheme under imperfect CSI and fractional Doppler. [1] A. Yazar, S. D. Tusha, and H. Arslan, “6G vision: An ultra-flexible radio
access technology perspective,” ITU J. Future Evolv. Technol., vol. 1,
no. 1, pp. 1–18, Jan. 2020.
5dB better BER performance compared with the random sce- [2] A. B. Kihero et al., “Wireless channel and interference,” in Design
nario for BPSK modulation. Moreover, in Fig. 3(b), similar and Analysis of Wireless Communication Signals: A Laboratory-Based
effect occurs on the error performance for OTFS system with Approach. Newark, NJ, USA: Wiley, 2021.
[3] G. D. Surabhi, R. M. Augustine, and A. Chockalingam, “On the diver-
QPSK modulation. sity of uncoded OTFS modulation in doubly-dispersive channels,” IEEE
Fig. 4 assesses the performance of the proposed Trans. Wireless Commun., vol. 18, no. 6, pp. 3049–3063, Jun. 2019.
OTFS scheduling scheme under fractional Doppler (|κ| ∈ [4] P. Raviteja, Y. Hong, E. Viterbo, and E. Biglieri, “Effective diversity
of OTFS modulation,” IEEE Wireless Commun. Lett., vol. 9, no. 2,
[0, . . . , 0.1]) and imperfect CSI, considering channel errors pp. 249–253, Feb. 2020.
as ĥui = hui + Δhui , where Δhui is modeled as an [5] P. Raviteja, K. T. Phan, Y. Hong, and E. Viterbo, “Orthogonal time
−γdB
CN (0, 10 10 ) and ĥui = hui if = 0. The error performance frequency space (OTFS) modulation based radar system,” in Proc. IEEE
Radar Conf., Dec. 2019, pp. 1–6.
of the proposed technique is robust against CSI error, leading [6] S. Rakib et al., “Multiple access in wireless telecommunications system
to a shift for all the results in accordance with the level of error for high-mobility applications,” U.S. Patent 9 722 741, Aug. 2017.
(Δhi ). It is worth noting that the effect of channel impairments [7] Z. Ding, R. Schober, P. Fan, and H. V. Poor, “OTFS-NOMA: An efficient
approach for exploiting heterogeneous user mobility profiles,” IEEE
increases with the diversity of the OTFS system. For this rea- Trans. Commun., vol. 67, no. 11, pp. 7950–7965, Nov. 2019.
son, we assess the performance of the system under imperfect [8] Y. Ge, Q. Deng, P. C. Ching, and Z. Ding, “OTFS signaling for
CSI in case of extreme = 0.1 value and practical = 0.05 uplink NOMA of heterogeneous mobility users,” IEEE Trans. Commun.,
vol. 69, no. 5, pp. 3147–3161, May 2021.
value, shown via dotted line and dot-dashed line, respectively. [9] A. R. Reyhani, A. Farhang, M. Ji, R. R. Chen, and B. Farhang-
The minimum error channel impact occurs in case of min- Boroujeny, “Analysis of discrete-time MIMO OFDM-based orthogonal
imum channel spread, since the UE diversity is minimized. time frequency space modulation,” in Proc. IEEE Int. Conf. Commun.,
Even though the proposed scheme suffers the most from chan- Oct. 2018, pp. 1–6.
[10] P. Raviteja, Y. Hong, E. Viterbo, and E. Biglieri, “Practical pulse-shaping
nel errors, it provides the best BER performance. Moreover, waveforms for reduced-cyclic-prefix OTFS,” IEEE Trans. Veh. Technol.,
the superiority of this scheme becomes more and more promi- vol. 68, no. 1, pp. 957–961, Jan. 2019.
nent considering practical CSI error values (i.e., = 0.05). [11] G. D. Surabhi, M. K. Ramachandran, and A. Chockalingam, “OTFS
modulation with phase noise in mmWave communications,” in Proc.
Likewise, the proposed scheduling mechanism outperforms IEEE Veh. Technol. Conf., Jun. 2019, pp. 1–5.
the random RR assignment in the presence of the fractional [12] A. Tusha, S. Dogan-Tusha, F. Yilmaz, S. Althunibat, K. Qaraqe, and
Doppler, as illustrated via the starred lines. Besides, the H. Arslan, “Physical effect of in-phase and quadrature imbalance in
receiver complexity of the OTFS-based transmission depends delay-Doppler domain,” in Proc. IEEE 89th Veh. Technol. Conf., Sep.
2021, pp. 1–6.
on the dimensions of H u,eff matrix for the MMSE recep- [13] A. Tusha, S. Dogan-Tusha, F. Yilmaz, S. Althunibat, K. Qaraqe, and H.
tion. Since the size of H u,eff ∈ CMN ×MN is irrelevant Arslan, “Performance analysis of OTFS under in-phase and quadrature
of the number of channel taps Pu of the desired UE, the imbalance at transmitter,” IEEE Trans. Veh. Technol., vol. 70, no. 11,
pp. 11761–11771, Nov. 2021.
proposed multi-user OTFS scheme does not bring extra com- [14] D. Tse, “Multiuser diversity in wireless networks,” in Proc. Wireless
plexity. Specifically, the number of complex multiplications Commun. Seminar, 2001, pp. 1–60.
(CM) needed by the conventional MMSE receiver (6b) is [15] R. M. Augustine and A. Chockalingam, “Interleaved time-frequency
O(M 3 N 3 ). In addition, CSI determines the user fairness for multiple access using OTFS modulation,” in Proc. IEEE Veh. Technol.
Conf., 2019, pp. 1–5.
the real systems. However, in this letter, we consider that [16] V. Khammammetti and S. K. Mohammed, “OTFS-based multiple-access
the channel characteristics are similar for each UE since the in high doppler and delay spread wireless channels,” IEEE Wireless
impact of channel fading reduces with the increase of the Commun. Lett., vol. 8, no. 2, pp. 528–531, Apr. 2019.
[17] S. M. Alamouti, “A simple transmit diversity technique for wire-
number of channel taps in OTFS systems. less communications,” IEEE J. Sel. Areas Commun., vol. 16, no. 8,
pp. 1451–1458, Oct. 1998.
[18] S. Li, W. Yuan, J. Yuan, Z. Wei, B. Bai, and D. W. K. Ng, “Performance
V. C ONCLUSION analysis of coded OTFS systems over high-mobility channels,” IEEE
Trans. Wireless Commun., vol. 20, no. 9, pp. 6033–6048, Sep. 2021.
In this letter, we have investigated the performance of OTFS [19] B. Vucetic and J. Yuan, Space-Time Coding. Hoboken, NJ, USA: Wiley,
transmission from the perspective of exploiting the multi-user 2003.
diversity effect in downlink transmission. Specifically, we have [20] M. K. Simon and M.-S. Alouini, Digital Communication Over Fading
Channels, vol. 95. Hoboken, NJ, USA: Wiley, 2005.
proposed a novel scheduling algorithm with the assignment [21] H. Jafarkhani, Space-Time Coding: Theory and Practice. Cambridge,
of the radio resources to the UE associated with the largest U.K.: Cambridge Univ. Press, 2005.

Authorized licensed use limited to: BMS College of Engineering. Downloaded on August 08,2024 at 06:20:24 UTC from IEEE Xplore. Restrictions apply.