A Survey On Low Complexity Detectors For OTFS Systems
A Survey On Low Complexity Detectors For OTFS Systems
Abstract: The newly emerging orthogonal time frequency space (OTFS) modulation can ob⁃
tain delay-Doppler diversity gain to significantly improve the system performance in high
mobility wireless communication scenarios such as vehicle-to-everything (V2X), high-speed
railway and unmanned aerial vehicles (UAV), by employing inverse symplectic finite Fouri⁃
er transform (ISFFT) and symplectic finite Fourier transform (SFFT). However, OTFS modu⁃
lation will dramatically increase system complexity, especially at the receiver side. Thus, de⁃
signing low complexity OTFS receiver is a key issue for OTFS modulation to be adopted by DOI: 10.12142/ZTECOM.202104002
new-generation wireless communication systems. In this paper, we review low complexity
https://kns.cnki.net/kcms/detail/34.1294.
OTFS detectors and provide some insights on future researches. We firstly present the OTFS
TN.20211129.1724.002.html, published
system model and basic principles, followed by an overview of OTFS detector structures, online November 30, 2021
classifications and comparative discussion. We also survey the principles of OTFS detection
algorithms. Furthermore, we discuss the design of hybrid OTFS and orthogonal frequency di⁃ Manuscript received: 2021-10-18
vision multiplexing (OFDM) detectors in single user and multi-user multi-waveform commu⁃
nication systems. Finally, we address the main challenges in designing low complexity OT⁃
FS detectors and identify some future research directions.
Citation (IEEE Format): Z. Q. Zhang, H. Liu, Q. L. Wang, et al.,“A survey on low complexity detectors for OTFS systems,”ZTE Commu⁃
nications, vol. 19, no. 4, pp. 03–15, Dec. 2021. doi: 10.12142/ZTECOM.202104002.
T
he new-generation mobile communication systems [1] support of high mobility up to 500–1 000 km/h with accept⁃
are the key enabler for the digital society in the next able quality of service (QoS)[1, 6].
ten years and are expected to satisfy the requirements However, high mobility wireless communications suffer
for high mobility applications such as vehicle-to-every⁃ from high Doppler spread, and the transmitted signals experi⁃
thing (V2X) services[2–3], high-speed railway services[4–5], as ence time-frequency doubly selective channel[7]. High Doppler
spread will result in very serious inter-carrier interference
(ICI), especially in orthogonal frequency division multiplexing
This work is supported in part by the NSFC Project under Grant No.
(OFDM) systems. Another challenge is to perform channel esti⁃
61871334, in part by the open research fund of the State Key Laboratory mation to obtain exact channel state information (CSI) of fast
of Integrated Services Networks, Xidian University under Grant No.
ISN21-15, and in part by the Fundamental Research Funds for the Central time-variant channels, even to the extent that the reported CSI
Universities, SWJTU under Grant No. 2682020CX79. FAN Pingzhi’s work is outdated. These challenges will seriously reduce the perfor⁃
is also supported by the NSFC project under Grant No. 61731017 and the
“111”project under Grant No. 111-2-14. mance of conventional OFDM systems. To tackle the challeng⁃
ZTE COMMUNICATIONS
December 2021 Vol. 19 No. 4 03
Special Topic A Survey on Low Complexity Detectors for OTFS Systems
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
es of high mobility, learning-based channel estimation, flexi⁃ study a hybrid OFDM-OTFS multi-waveform detection frame⁃
ble subcarrier spacing and length of cyclic prefix (CP), double work. Finally, we discuss some challenges for low complexity
demodulation reference signals (DMRS), i. e., front-loaded OTFS detectors, and identify some future research directions.
DMRS and additional DMRS with configurable time-domain The rest of the paper is organized as follows. A brief discus⁃
density, have been studied. However, these methods still treat sion on the OTFS system model and the principles of OTFS
high mobility as a negative factor, which results in very limit⁃ modulation are given in Section 2. In Section 3, a survey on
ed performance improvements of OFDM systems. the state-of-the-art OTFS detector structures is provided,
Recently, the orthogonal time frequency space (OTFS) mod⁃ while the research progress on OTFS detection algorithms is
ulation technology[8–9] has been proposed for high mobility given in Section 4. In Section 5, a hybrid OTFS-OFDM multi-
wireless communications, and attracted increasing attention waveform detection framework is discussed briefly, while Sec⁃
due to its excellent performance. This new two-dimensional tion 6 discusses the research challenges and identifies some
(2D) modulation transforms high mobility into a positive factor future research directions, followed with conclusions.
by introducing inverse symplectic finite Fourier transform (IS⁃
FFT)-based pre-processing before OFDM modulation and sym⁃
plectic finite Fourier transform (SFFT) -based post-processing 2 Basic Principles of OTFS Modulation
after OFDM demodulation. With ISFFT/SFFT transforms, de⁃ The OTFS system model is shown in Fig. 1, which includes
lay-Doppler (DD) domain is introduced in OTFS systems and OTFS transmitter and receiver structures. Compared with
the modulated symbols are transmitted in DD domain rather OFDM systems, OTFS systems add ISFFT-based transform
than time-frequency (TF) domain. The equivalent DD channel precoding before OFDM modulation at the transmitter side,
exhibits excellent features of separability, stability, compact⁃ while SFFT-based post-processing is employed after OFDM
ness, and possible sparsity[9], which enables OFTS systems to demodulation at the receiver side. From the perspective of sys⁃
obtain delay-Doppler diversity gain. Additionally, these excel⁃ tem structures, OTFS systems can be regarded as a type of pre⁃
lent features are also beneficial for performing channel estima⁃ coded OFDM systems and can be easily compatible with
tion under high mobility environments. OTFS modulation has OFDM systems. With the introduction of ISFFT/SFFT trans⁃
also been submitted to 3GPP as a candidate waveform for 5G form, a new domain, i. e., DD domain, is introduced. As a re⁃
systems[10–12], and is regarded as a promising waveform for sult, there are three domains in OTFS systems: DD domain,
next-generation wireless communications[13]. TF domain and time domain, while OFDM systems only have
However, since each modulated symbol is spread to the TF and time domains.
whole TF resource grid by ISFFT operation in OTFS systems, Considering an OTFS system with an N × M DD resource
the number of equivalent DD channel dimensions is larger grid, at the OTFS transmitter side, the modulated symbols and
than that of OFDM systems, which dramatically increases the pilots are mapped to the DD resource elements. The signal car⁃
complexity of signal detection. To address this challenge, ried by the (k, l) -th DD resource element is denoted by
some efforts have been devoted to the research of low complex⁃ xDD [ k, l ] for k = 0, 1,..., N - 1, l = 0, 1,..., M - 1. Then, the
ity OTFS detector structures such as decision feedback equal⁃ symbols xDD [ k, l ] in the DD domain are converted to the sym⁃
izer (DFE)[14], iterative maximum ratio combining (MRC) detec⁃ bols xTF [ n, m ] in the TF domain using the ISFFT as
tor[15–16], non-iterative joint TF- and DD-domain detector[17], it⁃ 1
∑∑x
N - 1M - 1
xTF [ n, m ] = ISFFT ( xDD [ k, l ]) = DD
[ k, l ] e ( N M )
j2π nk - ml
MRC detector with compensation[19], learning-based detec⁃ for n = 0, 1,..., N - 1, m = 0, 1,..., M - 1. Next, the signals
tor[20–23], and separate low complexity OTFS detector[24]. Sever⁃ xTF [ n, m ] in the TF domain is converted to the symbols in the
al OTFS detection algorithms, including linear minimum dime domain signal as x ( t) = IFFT ( xTF [ n, m ]) =
mean square error (MMSE) and zero-forcing (ZF) [25–29], mes⁃ 1
∑∑xTF [ n, m ] g tx ( t - nT ) ej2πmΔf (t - nT)and is trans⁃
N - 1M - 1
sage passing (MP) [30–35] and its variants like approximate mes⁃ MN n = 0 m = 0
sage passing (AMP) [34–36], MRC[15–16], joint MP and MRC[37], mitted through the channel. At the OTFS receiver side, the re⁃
hybrid maximum a posteriori (MAP) and parallel interference
cancellation (PIC) [38], expectation propagation (EP) [39], varia⁃
ceived signal in the time domain is y ( t) = ∫∫
h (τ,υ) x ( t -
υ τ
tional Bayes (VB) [40], and iterative least squares minimum re⁃ τ)e dτdυ. After OFDM demodulation (i. e., FFT trans⁃
j2πυ ( t - τ)
sidual (LSMR)[41], have been studied. form), the symbols in the TF domain are denoted by
In this paper, a comprehensive survey on OTFS detector y TF [ n, m ]. Then, applying SFFT on y TF [ n, m ], the symbols in
structures and detection algorithms is provided. We compare the DD domain can be obtained as y DD [ k, l ] =
the advantages and disadvantages of each OTFS detector SFFT ( ( y TF [ n, m ]) = ∑∑ y TF [ n, m ] e
N - 1M - 1
-j2π ( nk )
N -
. Finally, the
ml
M
structure and detection algorithm, which can provide some in⁃ n=0m=0
sights for future research. We also provide classifications for transmitted symbols xDD [ k, l ] can be recovered from y DD [ k, l ]
OTFS detectors from different dimensions. Furthermore, we through the OTFS detector.
ZTE COMMUNICATIONS
04 December 2021 Vol. 19 No. 4
A Survey on Low Complexity Detectors for OTFS Systems Special Topic
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
Input
bits Channel Serial to Parallel Adding
Interleaving Modulation Pilot insert ISFFT IFFT DAC RF
coding parallel to serial CP
OFDM modulation
OTFS modulation
Channel
estimation
Output
bits
Channel Deinter⁃ Demodula⁃ Signal Parallel Serial to Removing
SFFT FFT ADC RF
decoding leaving tion detection to serial parallel CP
OFDM demodulation
OTFS demodulation
ADC: analog-to-digital conversion FFT: fast Fourier transform OFDM: orthogonal frequency division multiplexing SFFT: symplectic finite Fourier transform
CP: cyclic prefix IFFT: inverse fast Fourier transform OTFS: orthogonal time frequency space
DAC: digital-to-analog conversion ISFFT: inverse symplectic finite Fourier transform RF: radio frequency
(h)
Iterative joint time-, frequency- and DD-domain
signal detection
3 OTFS Signal Detector (g)
Structures DD: delay-Doppler MMSE: minimum mean square error OTFS: orthogonal time frequency space
Several works have been devot⁃ DNN: deep neural network MP: message passing TF: time frequency
ML: maximum likelihood MRC: maximum ratio combining ZF: zero forcing
ed to studying low complexity OT⁃
FS detectors. Fig. 2 illustrates ▲Figure 2. OTFS detector structures: (a) DD-domain non-iterative OTFS detector; (b) DD-domain itera⁃
several popular OTFS detector tive OTFS detector; (c) non-iterative joint TF- and DD-domain OTFS detector; (d) joint non-iterative TF-
domain and iterative DD-domain OTFS detector; (e) iterative joint time- and DD-domain OTFS detector;
structures, including DD-domain (f) iterative joint TF- and DD-main OTFS detector; (g) iterative joint time-, TF- and DD-main OTFS de⁃
non-iterative detector[25–29], DD- tector; (h) learning-enabled OTFS detector
ZTE COMMUNICATIONS
December 2021 Vol. 19 No. 4 05
Special Topic A Survey on Low Complexity Detectors for OTFS Systems
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
domain iterative detector[30–35], non-iterative joint TF- and DD- Non-iterative OTFS detector
domain detector[17], joint non-iterative TF-domain and iterative
OTFS detector
DD-domain detector[15–16], iterative joint Time- and DD-do⁃
Iterative OTFS detector
main detector[18], iterative joint TF- and DD-main detector,
(a)
and learning-enabled detector[20–23]. According to the number
of domains involved in detection processing, these OTFS de⁃ Conventional OTFS detector
tectors can be divided into two categories: the single-domain OTFS detector
OTFS detector and joint multi-domain OTFS detector. With Learning-based OTFS detector
the need of iteration, these OTFS detectors can be divided into (b)
the non-iterative OTFS detector and iterative OTFS detector.
Single domain OTFS de⁃
These OTFS detectors can also be divided into the convention⁃ tector Joint TF- and DD-domain
al OTFS detector and learning-based OTFS detector. The de⁃ OTFS detector OTFS detector
tailed classifications of OTFS detectors are shown in Fig. 3. A Joint multi-domain OTFS Joint time- and DD-do⁃
detector main OTFS detector
summary of OTFS detectors is illustrated in Table 1.
Joint time-, TF- and DD-do⁃
The DD-domain non-iterative OTFS detector shown in Fig. main OTFS detector
2(a) achieves signal detection in the DD domain by using non- (c)
iterative detection algorithms like MMSE/ZF, spherical detec⁃ DD: delay-Doppler OTFS: orthogonal time frequency space TF: time-frequency
tion, maximum likelihood (ML) detection, etc., where MMSE/ ▲ Figure 3. OTFS detector classifications: (a) non-iterative and itera⁃
ZF is popular and has also been adopted by 4G/5G systems tive OTFS detectors; (b) single domain and multi-domain OTFS detec⁃
due to its low complexity, while spherical detection and ML tors; (c) conventional and learning-based OTFS detectors
ZTE COMMUNICATIONS
06 December 2021 Vol. 19 No. 4
A Survey on Low Complexity Detectors for OTFS Systems Special Topic
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
detection are very complex. In general, the DD-domain non-it⁃ sequence detection even in the presence of fractional Doppler
erative signal detector adopts MMSE/ZF algorithms. Without shifts, and the computational complexity associated with the
iteration operation, the computational complexity and process⁃ domain transformation was low.
ing delay of MMSE/ZF are small, but at the cost of detection The iterative joint TF- and DD-main OTFS detector shown
performance loss. in Fig. 2(f) can be regarded as an improvement of non-iterative
The DD-domain iterative OTFS detector shown in Fig. 2(b) joint TF- and DD-main signal detector. Similar as the iterative
also achieves signal detection in the DD domain, but uses iter⁃ joint time- and DD-domain OTFS detector shown in Fig. 2(e),
ative detection algorithms like MP and its improved algo⁃ the iterative joint TF- and DD-main signal detector forms a
rithms, and the EP algorithm[39]. These algorithms iteratively large iteration loop among the TF domain and DD domain,
update information to achieve better detection performance. which is expected to obtain better performance and faster con⁃
However, the iteration operation brings some extra computa⁃ vergence by utilizing TF- and DD-domain information. Fur⁃
tional complexity. Additionally, the convergence of iterative thermore, based on the OTFS detector shown in Fig. 2(f), an it⁃
detection algorithms needs to be considered. In Ref. [39], the erative joint time-, TF- and DD-main OTFS detector with time-
iterative EP algorithm and its improvement named Approxi⁃ domain equalization is shown in Fig. 2(g).
mate EP (AEP) were studied. They exhibit better bit error rate The learning-enabled OTFS detector shown in Fig. 2(h) us⁃
(BER) performance than MMSE, MP, MRC rank and VB algo⁃ es advanced machine learning method to improve detection
rithms. performance. In Ref. [20], to reduce the complexity of conven⁃
The non-iterative joint TF- and DD-domain OTFS detector tional MP detector in OTFS systems, a damped generalized ap⁃
shown in Fig. 2(c) can be considered as an improvement of the proximate message passing (GAMP) algorithm was studied
DD-domain non-iterative OTFS detector, which utilizes both and deep learning (DL) was introduced to optimize damping
TF- and DD-domain information to improve the detection per⁃ factors. Its BER performance can outperform the classical
formance. In Ref. [17], a sliding window-assisted MMSE (SW- GAMP algorithm and MP algorithm. In Ref. [21], a two-dimen⁃
MMSE) equalization in the TF domain was studied, and a DD sional convolutional neural network (2D-CNN) based detector
equalizer like decision feedback equalizer (DFE) was intro⁃ was studied to replace the conventional OTFS detector, and an
duced. The computation complexity of this non-iterative two- MP-based data augmentation (DA) tool was employed to en⁃
stage equalizer is lower than conventional MMSE, and the large the training features of the input dataset and mitigate the
BER performance is also better than conventional MMSE. effect of the channel variations to some degree, leading to im⁃
The joint non-iterative TF-domain and iterative DD-domain provement of the robustness and learning ability of the deep
OTFS detector shown in Fig. 2(d) can be regarded as an im⁃ neural network (DNN). This 2D-CNN based detector can
provement of the DD-domain iterative OTFS detector, in achieve superior performance compared with the MP detector
which the non-iterative TF-domain equalizer provides good and similar performance as the MAP detector with a very low
initials for the iterative DD-domain OTFS detector to improve complexity. In Ref. [22], a DD-domain symbol-level DDN de⁃
its convergence performance. In Refs. [15] and [16], an itera⁃ tector was studied, which could achieve similar BER perfor⁃
tive MRC detector with initial estimates from the output of TF-
domain MMSE equalizer was studied, as shown in Fig. 4. The Received DD signals YDD,
results show that the iterative MRC detector with initial esti⁃ DD channel information
Received TF
mates can achieve better BER performance than that without signals YTF Initial DD-domain
Detector
output LLRs
TF-domain MMSE equalization, iterative MPA or MMSE. Con⁃ TF channel
TF-domain estimates MRC
detection
MMSE
sidering spatial correlation at the receiver antennas, a sample- matrix HTF equalization Detector input
LLRs
based method to estimate such correlation and the optimized
combining weights for MRC from the estimated correlation ma⁃
Deinter⁃
trix were studied in Ref. [42]. Interleaving
leaving
The iterative joint time- and DD-domain OTFS detector
shown in Fig. 2(e) forms a large iteration loop among the time
Decoder
domain and DD domain, which is expected to obtain better output
LDPC
performance and lower computational complexity by exploit⁃ LLRs
decoding Decoder
ing time domain channel sparsity and DD domain symbol con⁃ input LLRs
stellation constraints. In Ref. [18], the iterative joint time- and Infor⁃
mation
DD-domain signal detector was studied, which adopted an L- bits
MMSE estimator in the time domain and a symbol-by-symbol
DD: delay-Doppler MMSE: minimum mean square error
detection in the DD domain. The results show that this itera⁃ LDPC: low density parity check MRC: maximum ratio combining
tive joint time- and DD-domain signal detector could achieve LLR: log-likelihood ratio
almost the same error performance as the maximum-likelihood ▲Figure 4. Iterative MRC detector[15]
ZTE COMMUNICATIONS
December 2021 Vol. 19 No. 4 07
Special Topic A Survey on Low Complexity Detectors for OTFS Systems
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
mance as the full DDN detector and ML detector in static mul⁃ systems, which results in MN × MN matrix inversion. To re⁃
tipath channel with Gaussian noise, while it achieved better duce the complexity of linear signal detection in OTFS sys⁃
BER performance than the full DDN detector and ML detector tems, considering the sparsity and the block circulant nature
in static multipath channel with non-Gaussian noise. In Ref. of equivalent DD channel, some low complexity linear signal
[23], a reservoir computing (RC) -based OTFS detector was detection schemes have been studied.
studied, in which one-shot online learning was sufficiently In Ref. [27], the eigenvalues of G MMSE was computed from
flexible to cope with channel variations among different OTFS the eigenvalues of DD channel matrix H, which can signifi⁃
frames and explicit CSI was not required. cantly reduce the complexity. This MMSE with low complexity
is summarized as follows:
1) Compute the eigenvalues of each block of H, by comput⁃
4 OTFS Detection Algorithms ing DFTs of the first row of each circulant block;
OTFS detection algorithms include linear MMSE/ZF, MP 2) Compute the eigenvalues of H;
and its improvements, MRC, MAP, EP, and VB algorithms. A 3) Compute the eigenvalues of G MMSE, by using the eigenval⁃
summary of these detection algorithms including their compu⁃ ues of H;
tational complexity and BER performance is presented in Ta⁃ 4) Compute G MMSE y.
ble 2. This idea was also adopted by Ref. [28] to study the detec⁃
tion in MIMO-OTFS systems. Unlike the SISO-OTFS channel,
4.1 Linear MMSE/ZF Detection Algorithm the eigenvalue matrix D in MIMO-OTFS channel is not diago⁃
Linear signal detection mainly includes MMSE and ZF, nal, however, the inverse of the DA constructed by the matrix
while MMSE has been adopted by 4G/5G OFDM systems, due D can be performed block-wise by two steps: matrix partition⁃
to its low complexity. The detection matrices of classical ing and backtracking[28].
MMSE and ZF in OTFS systems are G MMSE = ( H H H + The computational complexity of MMSE is mainly caused
σ 2 Ι ) -1 H H and G ZF = ( H H H ) -1 H H, respectively. However, by large matrix inversion. Considering the sparsity of equiva⁃
when these classical MMSE and ZF detection algorithms are lent DD channel matrix and quasi-banded structure of matri⁃
used for OTFS systems directly, they suffer from very high ces in MMSE detection, a lower-upper (LU) factorization-
complexity Ο( M 3 N 3 ). This is because the number of dimen⁃ based low complexity MMSE detection algorithm was studied
sions of equivalent DD channel matrix is MN × MN in OTFS for OTFS systems with reduced CP[25] and full CP[29], in which
high complexity channel inversion is replaced by low complex⁃
▼Table 2. Summary of computational complexity and performance ity LU factorization operation. Further, the final estimate sym⁃
Detection Algorithm Computational Com⁃ Perfor⁃ bols step can be performed by Fast Fourier Transform (FFT).
Reference
Algorithm Characteristic plexity mance Its detailed procedure is illustrated in Fig. 5.
Ref. [39]
Classical
Non-iterative Ο( M 3 N 3 ) There are some other low complexity MMSE/ZF detection
MMSE
algorithms. For example, the one-tap MMSE detection algo⁃
Low complexi⁃
Ref. [27] Non-iterative Ο( MN log ( MN ) ) rithm studied in Ref. [26] achieved low complexity detection
ty MMSE
in pulse-shaped OTFS systems over doubly-dispersive chan⁃
lower-upper
Ref. [25] factorization Non-iterative Ο( MN log ( N ) ) nels, which only estimated the channel main diagonal and the
-based MMSE UAMP>EP> self-interference power instead of interference cancellation
Refs. [31]
MP Iterative Ο(2 IMNS )
Q
AEP and considered the power of the channel estimation error and
>MRC-rake
and [32]
>VB
self-interference as additional tuning variance parameters.
Ref. [33] MF-MP-PC Iterative Ο( IMN (2Q/2 + S ) )
>MP
Ref. [34] GAMP Iterative Ο(2Q IMNS ) >Classical 4.2 MRC Detection Algorithm
Ο( IMN log ( MN ) ) MMSE
The MRC detection algorithm extracts received multipath
Ref. [35] UAMP Iterative ≥low com⁃
+Ο(2Q IMN ) plexity components of the transmitted symbols in the delay-Doppler
Ref. [36] ICMP Iterative Ο(2Q IMNGS ) MMSE grid and combines them by using MRC to improve the signal-
Refs. [15] Ο( IMN ( L + log ( N ) ) ) to-noise ratio (SNR) of the combined signal. The detailed
MRC-rake Iterative
and [16] +Ο( MN ( L + log ( M ) ) ) steps of MRC algorithm are shown as follows[15–16]:
EP Iterative Ο( IMN (2Q + S ) ) 1) Construct circulant matrix having element K m,l, accord⁃
Ref. [39]
AEP Iterative Ο( IMN (2Q + S ) ) ing to the channel Doppler spread vector at each delay tap;
Ref. [40] VB Iterative Ο(2Q IMNS ) 2) Construct matrix R by using a circulant matrix, where
AEP: approximate expectation propagation MMSE: minimum mean square error R m = ∑l ∈ { l } K mH + l,l K m + l,l , m = 0, 1,..., M - l max;
EP: expectation propagation MP: message passing
i
GAMP: generalized approximate message MRC: maximum ratio combining 3) Construct the equations for the symbol vector estimates
passing PC: probability clipping b lm, by using the estimates of symbol vectors from previous iter⁃
ICMP: iterative combining message passing UAMP: unitary approximate message passing
MF: matched filtering VB: variational Bayes ation;
ZTE COMMUNICATIONS
08 December 2021 Vol. 19 No. 4
A Survey on Low Complexity Detectors for OTFS Systems Special Topic
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
ZTE COMMUNICATIONS
December 2021 Vol. 19 No. 4 09
Special Topic A Survey on Low Complexity Detectors for OTFS Systems
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
PMF of alphabet symbols and then passes it back to the con⁃ tion of the a posteriori distribution p(d|y). The detailed proce⁃
nected observation nodes; 3) The joint convergence indicator dures are shown as follows:
of all beamforming branches is calculated in the MRC fashion 1) Formulate the approximation q*(d) as an optimal problem
after each iteration. Finally, when the convergence is satisfing, by minimizing the Kullback-Leibler divergence;
the soft output of each transmitted symbol is computed, fol⁃ 2) Construct the approximation q(d) by mean filed approxi⁃
lowed with hard decision. mation as q ( d ) = ∏k,l q k,l ( d k,l ). Note that in this form, all vari⁃
ables are mutually independent;
4.4 MAP Detection Algorithm
3) Transform p(d|y) into a pairwise form;
The MAP detection algorithm uses all received signals to es⁃
4) Obtain the variational function in the optimization prob⁃
timate all transmitted symbols, which can be formulated as x̂ =
lem in Step 1, by substituting q(d) and p(d|y) into the optimiza⁃
arg max Pr ( x|y, H ). Obviously, its complexity increases with
x ∈ Α NM × 1 tion problem;
exponent in NM[32]. To reduce the complexity, a near-optimal 5) Find a stationary point of the variational function, by iter⁃
symbol-wise MAP detection algorithm was studied in Ref. atively updating each local function q k,l ( d k,l );
[38], and its detection rule is expressed as x ( k, l) = 6) Approximate a posteriori distributions for all the data
arg max Pr ( x ( k, l)|y, H ). symbols iteratively, resulting in the approximate marginals
x ( k,l) ∈ Α
q* k,l ( d k,l );
4.5 EP Detection Algorithm 7) Estimate the transmitted symbols by finding the maxi⁃
The OTFS system can be represented by a sparsely-connect⁃ mum of marginal distribution q* k,l ( d k,l ).
ed factor graph where each variable node (VN) is connected to
factor nodes D. The main idea of EP algorithm[39] is to use a
Gaussian distribution through distribution projection to ap⁃ 5 Hybrid OFDM-OTFS Multi-Waveform
proximate the sophisticated posterior distribution in the mes⁃ Detector Structure
sage updating steps, which leads to the complicated belief To satisfy the requirements for various scenarios and appli⁃
computation being replaced by means and variances computa⁃ cations, mobile communication systems have evolved from sin⁃
tion. The detailed steps of EP algorithm are represented as fol⁃ gle waveform to multi-waveform systems. For example, in 4G
lows[39]: systems, high-spectrum efficiency CP-OFDM is adopted by
1) Compute the joint distribution p ( x DD , y DD ); the downlink, while the uplink adopts single-carrier frequency
2) Compute the likelihood function p ( y b |x DD ); division multiple access (SC-FDMA) with a low peak to aver⁃
3) Compute the means and variances passed from FNs and age power ratio (PAPR). In 5G systems, the downlink adopts
VNs as u if b - > x a and v if b - > x a; CP-OFDM, while CP-OFDM and discrete Fourier transform-
4) Compute the means and variances passed from VNs and spread orthogonal frequency-division multiplexing (DFT-s-
FNs as u ix a - > f b and v ix a - > f b; OFDM) with low-PAPR are adopted by the uplink. In general,
when UE is in the cell center, UE can still obtain the expected
5) Compute the a posteriori LLR of each coded bit as c qa;
QoS with low transmit power, thus UE can adopt CP-OFDM
6) Stop criteria by stopping iteration, when some conditions
waveform with higher spectrum efficiency. When UE is at the
are satisfied.
cell edge, UE should increase transmit power to obtain the ex⁃
Note that the main computational complexity of the EP algo⁃
rithm depends on the number of non-zero elements D of chan⁃ pected QoS, which requires UE to adopt DFT-s-OFDM wave⁃
nel matrix. In case of rich scattering scenarios and fractional form with low PAPR. Since OFTS exhibits excellent perfor⁃
Doppler shift, D is relatively large. To further reduce the com⁃ mance in high mobility environments, if OTFS is accepted by
putational complexity, small channel coefficients can be ap⁃ future mobile communication systems (FMCS), its downlink
proximated to a fixed value (e. g., the median value of these waveform will be CP-OFDM or OTFS, while its uplink wave⁃
small elements) during the message passing from FNs to VNs, form will be CP-OFDM, DFT-s-OFDM or OTFS. To determine
which is named channel coeffificients-aware approximate EP each user’s uplink (UL)/downlink (DL) waveform, the base
(AEP) algorithm[39]. station shall call UL and DL waveform decision algorithms
with the input of user mobility speed and user type. Further,
4.6 VB Detection Algorithm the base station also needs to dynamically switch user’s down⁃
The optimal MAP detection algorithm suffers from very link and uplink waveform type if some conditions are trig⁃
high complexity, which increases exponentially with the size gered.
of data symbol vector. To reduce the complexity of MAP algo⁃ Fig. 6 shows the hybrid OFDM-OTFS multi-waveform detec⁃
rithm, a variational Bayes algorithm was studied in Ref. [40]. tor structure, in which Fig. 6(a) is for single user OTFS sys⁃
The main idea of VB algorithm is to find a distribution q(d) tems and Fig. 6(b) is for multi-user OTFS systems. The base
from a tractable distribution family as an optimized approxima⁃ station first determines the DL and UL waveform types accord⁃
ZTE COMMUNICATIONS
10 December 2021 Vol. 19 No. 4
A Survey on Low Complexity Detectors for OTFS Systems Special Topic
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
ZTE COMMUNICATIONS
December 2021 Vol. 19 No. 4 11
Special Topic A Survey on Low Complexity Detectors for OTFS Systems
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
Start
Start
▲Figure 9. Multiple access schemes for downlink hybrid OFDM-OTFS
systems in multi-user scenario: (a) hybrid orthogonal frequency divi⁃
Obtain user mobility speed sion multiple access (OFDMA) and orthogonal time frequency space
and user type (CCU or CEU) multiple access (OTFSMA) in both DD and TF domains; (b) hybrid
OFDMA and OTFSMA with overlap in the TF domain
Y N
User type is CEU? achieve dramatic BLER performance gain in a low-speed sce⁃
nario while some BLER performance gain can be obtained in
Output uplink
waveform
User mobility N a high-speed scenario. Therefore, it is suggested that low-com⁃
speed ≥ threshold?
DFT-s-OFDM plexity OFDM modulation is used to low-speed users, while
Y high-speed users adopt OTFS modulation. Since different us⁃
Output uplink Output uplink ers with different speeds coexist in the base station, hybrid
waveform waveform
OTFS CP-OFDM OTFS and OFDM systems should be considered.
ZTE COMMUNICATIONS
12 December 2021 Vol. 19 No. 4
A Survey on Low Complexity Detectors for OTFS Systems Special Topic
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
BLER
BLER
10 -1
10-3
10 15 20 25 10 15 20 25 6.2.1 Advanced Low Complexity
SNR/dB SNR/dB OTFS Detectors and Detec⁃
(a) (b)
tion Algorithms
BLER: block error rate SNR: signal-to-noise ratio
The computational complexity
▲Figure 10. BLER performance: (a) two-user orthogonal time frequency space (OTFS) systems without
inter-user interference (IUI); (b) two-user hybrid OTFS-orthogonal frequency division multiplexing
of current OTFS detection algo⁃
(OFDM) systems rithms has been reduced to
Ο( MN log ( N ) ), but it is still
rich scattering or a large number of paths such as MIMO-OT⁃ much higher than the acceptable complexity in practical sys⁃
FS systems, the block circulant and sparsity will not be satisf⁃ tems. Therefore, in the future, the first important work is to
ing. Furthermore, integer Doppler shifts are assumed in many study advanced low complexity OTFS detector structures and
works, while the assumption of fractional Doppler shifts is detection algorithms. As for OTFS detector structures, even
more reasonable in practical OTFS systems. However, fraction⁃ though some single domain and joint multi-domain OTFS de⁃
al Doppler shifts will increase computational complexity and tectors have been studied, there are still some novel OTFS de⁃
result in more serious inter-Doppler interference. Therefore, tector structures to be studied such as joint channel estimation
the research on low complexity OTFS detection algorithm is a and detection. As for OTFS detection algorithms, some non-it⁃
great challenge. erative and iterative detection algorithms have been studied,
but their computation complexities are still very high, up to
6.1.2 Decoupling Between MIMO-OTFS Detector and Precoder Ο( MN log ( N ) ). The properties of DD channel matrix together
Current research on OTFS receivers mainly focuses on with some simplified and approximated matrix operations
SISO-OTFS systems, while just a few works study MIMO-OT⁃ should be further exploited to develop novel OTFS detection
FS systems. However, when extending SISO-OTFS detection algorithms. Furthermore, iterative detection algorithms can
algorithms to MIMO-OTFS systems, it will face some new achieve better performance, but they are needed to further an⁃
problems. For example, when the MRC detector is used to MI⁃ alyze the convergence by employing some tools such as extrin⁃
MO-OTFS systems, it needs to obtain the precoding matrix. sic information transfer (EXIT) chart and design efficient itera⁃
However, when a non-codebook-based precoding scheme is tion stopping schemes to reduce the number of iterations.
adopted, it is difficult for the OTFS receiver to obtain the pre⁃
coding matrix. That is, in order to match the detection algo⁃ 6.2.2 Learning-Based OTFS Detectors
rithms, some detectors require special design at the MIMO- Several OTFS detection algorithms have been studied, but
OTFS transmitter side. This strong coupling design between they reduce the computational complexity by exploiting the
the MIMO-OTFS receiver and transmitter reduces the flexibili⁃ block circulant and sparsity, as well as simplified and approxi⁃
ty of MIMO-OTFS system design and processing. Therefore, mated mathematical methods. It will become more difficult to
the research on MIMO-OTFS detectors and detection algo⁃ find more low complexity OTFS detection algorithms. A learn⁃
rithms, which is decoupled with the MIMO-OTFS transmitter, ing-based method provides a new way for the OTFS detector,
is another challenge. which considers OTFS detection processing as a black box
and performs OTFS detection by deploying online learning
6.1.3 Multi-Waveform Hybrid OTFS Detector model trained offline. Currently, there are few research works
OTFS modulation can obtain delay-Doppler diversity gain, on learning-based OTFS detector, and thus many efforts are
and thus it can achieve better performance than conventional needed to study learning models and performance verification.
OFDM systems in high mobility scenarios. However, OTFS Therefore, the learning-based OTFS detector is a future re⁃
modulation cannot obtain obvious performance gain in low mo⁃ search direction.
bility scenarios. When users experience different scenarios, it
would be better to switch waveform to obtain better perfor⁃ 6.2.3 Unified Multi-Waveform Detector Design
mance. The coexistence of multiple waveforms, such as If OTFS modulation is adopted, the downlink and uplink of
OFDM and OTFS, requires that the receiver supports multi- future mobile communication systems will be multi-waveform.
ZTE COMMUNICATIONS
December 2021 Vol. 19 No. 4 13
Special Topic A Survey on Low Complexity Detectors for OTFS Systems
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
However, there are few works to study the receiver design to es and solutions [J]. China communications, 2016, 13 (Supplement 2): 1 – 13.
DOI: 10.1109/CC.2016.7833456
support coexistence of multiple waveform. To support multi-
[8] HADANI R, RAKIB S, TSATSANIS M, et al. Orthogonal time frequency space mod⁃
waveform systems, the receiver should support signal detec⁃ ulation [C]//IEEE Wireless Communications and Networking Conference (WCNC).
tion of each waveform. A simple way is to deploy multiple sig⁃ San Francisco, USA: IEEE, 2017: 1–6. DOI: 10.1109/WCNC.2017.7925924
nal detection modules and switch among these detection mod⁃ [9] WEI Z Q, YUAN W J, LI S Y, et al. Orthogonal time-frequency space modula⁃
tion: a promising next-generation waveform [J]. IEEE wireless communications,
ules according to configured waveform type. However, this
2021, 28(4): 136–144. DOI: 10.1109/MWC.001.2000408
way is inefficient and will increase the processing complexity. [10] 3GPP TSG RA WG1. Overview of OTFS waveform for next generation RAT:
A better way is to design a unified multi-waveform receiver, Meeting #84-bis R1162929 [R]. Busan, South Korea: 3GPP, 2016
which can flexibly and efficiently support signal detection of [11] 3GPP TSG RA WG1. OTFS modulation waveform and reference signals for
new RAT: Meeting #84-bis R1162930 [R]. Busan, South Korea: 3GPP, 2016
different waveform. Therefore, unified multi-waveform receiv⁃
[12] 3GPP TSG RA WG1. Performance results for OTFS modulation: Meeting #85
er design is another future research direction. R1165620 [R]. Nanjing, China: 3GPP, 2016
[13] MobileChina. White paper on 2030+ technology trends [R]. 2019
[14] JING L Y, WANG H, HE C B, et al. Two dimensional adaptive multichannel
7 Conclusions decision feedback equalization for OTFS system [J]. IEEE communications let⁃
ters, 2021, 25(3): 840–844. DOI: 10.1109/LCOMM.2020.3039982
In this paper, the research works on low complexity OTFS [15] THAJ T, VITERBO E. Low complexity iterative rake detector for orthogonal
detectors have been surveyed comprehensively. Firstly, we time frequency space modulation [C]//IEEE Wireless Communications and
present the OTFS system model and basic principles. And Networking Conference (WCNC). Seoul, Korea (South): IEEE, 2020: 1 – 6.
DOI: 10.1109/WCNC45663.2020.9120526
then, we focus on low complexity OTFS detector structures,
[16] THAJ T, VITERBO E. Low complexity iterative rake decision feedback equal⁃
and give the categories and discussions of all surveyed OTFS izer for zero-padded OTFS systems [J]. IEEE transactions on vehicular technol⁃
detector structures. According to different classification regu⁃ ogy, 2020, 69(12): 15606–15622. DOI: 10.1109/TVT.2020.3044276
lations, two classifications of OTFS detectors are the single-do⁃ [17] JIN C X, BIE Z S, LIN X H, et al. A simple two-stage equalizer for OTFS with
rectangular windows [J]. IEEE communications letters, 2021, 25(4): 1158 –
main OTFS detector and joint multi-domain OTFS detector;
1162. DOI: 10.1109/LCOMM.2020.3043841
the non-iterative OTFS detector and iterative OTFS detector. [18] LI S Y, YUAN W J, WEI Z Q, et al. Cross domain iterative detection for or⁃
As for their performance, the joint multi-domain OTFS detec⁃ thogonal time frequency space modulation [J]. IEEE transactions on wireless
tor is superior to the single-domain one, while the iterative OT⁃ communications, early access. DOI: 10.1109/TWC.2021.3110125
[19] SHAN Y R, WANG F G. Low-complexity and low-overhead receiver for OTFS
FS detector is better than the non-iterative one. We also pro⁃
via large-scale antenna array [J]. IEEE transactions on vehicular technology,
vide an overview on the principles of popular OTFS detection 2021, 70(6): 5703–5718. DOI: 10.1109/TVT.2021.3072667
algorithms, and discuss them in terms of complexity and per⁃ [20] XU X K, ZHAO M M, LEI M, et al. A damped GAMP detection algorithm for
formance. Furthermore, considering the coexistence of multi⁃ OTFS system based on deep learning [C]//IEEE 92nd Vehicular Technology
Conference (VTC2020-Fall). Victoria, Canada: IEEE, 2020: 1 – 5. DOI:
ple waveforms such as OTFS and OFDM, we discuss the de⁃
10.1109/VTC2020-Fall49728.2020.9348493
sign for hybrid multi-waveform detectors in single user and [21] ENKU Y K, BAI B M, WAN F, et al. Two-dimensional convolutional neural
multi-user OTFS systems, and waveform switching procedures network based signal detection for OTFS systems [J]. IEEE wireless communi⁃
and algorithms. Finally, we present main challenges for low cations letters, 10(11): 2514–2518. DOI: 10.1109/LWC.2021.3106039
[22] NAIKOTI A, CHOCKALINGAM A. Low-complexity delay-doppler symbol
complexity OTFS detectors and identify some future research
DNN for OTFS signal detection [C]//IEEE 93rd Vehicular Technology Confer⁃
directions. ence (VTC2021-Spring). Helsinki, Finland: IEEE, 2021: 1–6. DOI: 10.1109/
VTC2021-Spring51267.2021.9448630
[23] ZHOU Z, LIU L J, XU J R, et al. Learning to equalize OTFS [EB/OL]. (2021-
07-17)[2021-08-31]. https://arxiv.org/abs/2107.08236
[24] PANDEY B C, MOHAMMED S K, RAVITEJA P, et al. Low complexity pre⁃
coding and detection in multi-user massive MIMO OTFS downlink [J]. IEEE
transactions on vehicular technology, 2021, 70(5): 4389–4405. DOI: 10.1109/
References TVT.2021.3061694
[1] ITU-R. Framework and overall objectives of the future development of IMT for [25] TIWARI S, DAS S S, RANGAMGARI V. Low complexity LMMSE Receiver
2020 and beyond: ITU-R M.2083-0 [R]. 2015 for OTFS [J]. IEEE communications letters, 2019, 23(12): 2205– 2209. DOI:
[2] 3GPP. Study on NR vehicle-to-everything (V2X) (Release 16): 3GPP TR 38.885 10.1109/LCOMM.2019.2945564
[R]. 2019 [26] PFADLER A, JUNG P, SZOLLMANN T, et al. Pulse-shaped OTFS over doubly-
[3] CHEN S Z, HU J L, SHI Y, et al. Vehicle-to-everything (v2x) services supported dispersive channels: one-tap vs. full LMMSE equalizers [C]//IEEE International
by LTE-based systems and 5G [J]. IEEE communications standards magazine, Conference on Communications Workshops (ICC Workshops). Montreal, Cana⁃
2017, 1(2): 70–76. DOI: 10.1109/MCOMSTD.2017.1700015 da: IEEE, 2021: 1–6. DOI: 10.1109/ICCWorkshops50388.2021.9473535
[4] 3GPP. Mobile communication system for railways (Release 17): 3GPP TS [27] SURABHI G D, CHOCKALINGAM A. Low-complexity linear equalization for
22.289 [S]. 2019 OTFS modulation [J]. IEEE communications letters, 2020, 24(2): 330 – 334.
[5] AI B, GUAN K, RUPP M, et al. Future railway services-oriented mobile commu⁃ DOI: 10.1109/LCOMM.2019.2956709
nications network [J]. IEEE communications magazine, 2015, 53(10): 78 – 85. [28] SINGH P, MISHRA H B, BUDHIRAJA R. Low-complexity linear MIMO-OT⁃
DOI: 10.1109/MCOM.2015.7295467 FS receivers [C]//IEEE International Conference on Communications Work⁃
[6] ZHANG Z Q, XIAO Y, MA Z, et al. 6G wireless networks: vision, requirements, shops (ICC Workshops). Montreal, Canada: IEEE, 2021: 1–6. DOI: 10.1109/
architecture, and key technologies [J]. IEEE vehicular technology magazine, ICCWorkshops50388.2021.9473839
2019, 14(3): 28–41. DOI: 10.1109/MVT.2019.2921208 [29] ZOU T T, XU W J, GAO H, et al. Low-complexity linear equalization for OTFS
[7] FAN P Z, ZHAO J, I C L. 5G high mobility wireless communications: challeng⁃ systems with rectangular waveforms [C]//IEEE International Conference on
ZTE COMMUNICATIONS
14 December 2021 Vol. 19 No. 4
A Survey on Low Complexity Detectors for OTFS Systems Special Topic
ZHANG Zhengquan, LIU Heng, WANG Qianli, FAN Pingzhi
Communications Workshops (ICC Workshops). Montreal, Canada: IEEE, MIMO-OTFS [J]. IEEE wireless communications letters, 2021, early access.
2021: 1–6. DOI: 10.1109/ICCWorkshops50388.2021.9473771 DOI: 10.1109/LWC.2021.3125986
[30] KOLLENGODE RAMACHANDRAN M, CHOCKALINGAM A. MIMO-OTFS in
high-doppler fading channels: signal detection and channel estimation [C]//IEEE
Global Communications Conference (GLOBECOM). Abu Dhabi, United Arab
Emirates: IEEE, 2018: 206–212. DOI: 10.1109/GLOCOM.2018.8647394
[31] RAVITEJA P, PHAN K T, JIN Q Y, et al. Low-complexity iterative detection for
orthogonal time frequency space modulation [C]//IEEE Wireless Communica⁃ Biographies
tions and Networking Conference (WCNC). Barcelona, Spain: IEEE, 2018: 1–
6. DOI: 10.1109/WCNC.2018.8377159 ZHANG Zhengquan (zhangzqswjtu@163. com) received the Ph. D. degree in
[32] RAVITEJA P, PHAN K T, HONG Y, et al. Interference cancellation and itera⁃ information and communication engineering from Southwest Jiaotong Universi⁃
tive detection for orthogonal time frequency space modulation [J]. IEEE trans⁃ ty, China in 2019. From 2008 to 2013, he was with ZTE Corporation as a com⁃
actions on wireless communications, 2018, 17(10): 6501 – 6515. DOI: munication engineer. From 2016 to 2017, he was a guest Ph.D. student with the
10.1109/TWC.2018.2860011 KTH Royal Institute of Technology, Sweden. Since 2019, he has been with the
Department of Communication Engineering, School of Information Science and
[33] ZHANG H J, ZHANG T T. A low-complexity message passing detector for OT⁃
Technology, Southwest Jiaotong University. His research interests include B5G
FS modulation with probability clipping [J]. IEEE wireless communications let⁃
and 6G wireless communication technologies. He is a member of IEEE.
ters, 2021, 10(6): 1271–1275. DOI: 10.1109/LWC.2021.3063904
[34] XIANG L P, LIU Y S, YANG L L, et al. Gaussian approximate message pass⁃
ing detection of orthogonal time frequency space modulation [J]. IEEE transac⁃ LIU Heng received the Ph.D. degree in information and communication engi⁃
tions on vehicular technology, 2021, 70(10): 10999 – 11004. DOI: 10.1109/ neering from Southwest Jiaotong University, China in 2013. Since 2013, he has
TVT.2021.3102673 been with the Department of Communication Engineering, School of Informa⁃
[35] YUAN Z D, LIU F, YUAN W J, et al. Iterative detection for orthogonal time tion Science and Technology, Southwest Jiaotong University. His research inter⁃
frequency space modulation with unitary approximate message passing [J]. ests include next-generation wireless communications, rail transit wireless com⁃
IEEE transactions on wireless communications, early access. DOI: 10.1109/ munications, machine learning and intelligent wireless communications.
TWC.2021.3097173
[36] GE Y, DENG Q W, CHING P C, et al. Receiver design for OTFS with a frac⁃
tionally spaced sampling approach [J]. IEEE transactions on wireless commu⁃ WANG Qianli received the B.S. and M.S. degrees in electronic and informa⁃
nications, 2021, 20(7): 4072–4086. DOI: 10.1109/TWC.2021.3055585 tion engineering and the Ph.D. degree from the University of Electronic Science
[37] CHENG J Q, JIA C L, GAO H, et al. OTFS based receiver scheme with multi-an⁃ and Technology of China in 2013, 2016 and 2021, respectively. Since 2021, he
tennas in high-mobility V2X systems [C]//IEEE International Conference on Com⁃ has been with the Department of Communication Engineering, School of Infor⁃
munications Workshops (ICC Workshops). Dublin, Ireland: IEEE, 2020: 1 – 6. mation Science and Technology, Southwest Jiaotong University, China. His re⁃
DOI: 10.1109/ICCWorkshops49005.2020.9145313 search interests include estimation theory, array signal processing, radar sensor
network, and compressed sensing.
[38] LI S Y, YUAN W J, WEI Z Q, et al. Hybrid MAP and PIC detection for OTFS
modulation [J]. IEEE transactions on vehicular technology, 2021, 70(7): 7193–
7198. DOI: 10.1109/TVT.2021.3083181
FAN Pingzhi is currently a distinguished professor of Southwest Jiaotong Uni⁃
[39] LI H, DONG Y Y, GONG C H, et al. Low complexity receiver via expectation
versity, China, and a visiting professor of Leeds University, UK (1997–). He is
propagation for OTFS modulation [J]. IEEE communications letters, 2021, 25 a recipient of the UK ORS Award (1992), the National Science Fund for Distin⁃
(10): 3180–3184. DOI: 10.1109/LCOMM.2021.3101827 guished Young Scholars (1998, NSFC), IEEE VT Society Jack Neubauer Memo⁃
[40] YUAN W J, WEI Z Q, YUAN J H, et al. A simple variational Bayes detector rial Award (2018), IEEE SP Society SPL Best Paper Award (2018), IEEE WC⁃
for orthogonal time frequency space (OTFS) modulation [J]. IEEE transactions SP 10-Year Anniversary Excellent Paper Award (2009–2019), and IEEE/CIC
on vehicular technology, 2020, 69(7): 7976 – 7980. DOI: 10.1109/ ICCC Best Paper Award (2020). He served as a chief scientist of the National
TVT.2020.2991443 “973”Plan Project between January 2012 and December 2016. He is an IEEE
[41] QU H Y, LIU G H, ZHANG L, et al. Low-complexity symbol detection and in⁃ VTS Distinguished Speaker (2019–2022), a fellow of IEEE, IET, CIE and CIC.
terference cancellation for OTFS system [J]. IEEE transactions on communica⁃ His research interests include high mobility wireless communications, massive
tions, 2021, 69(3): 1524–1537. DOI: 10.1109/TCOMM.2020.3043007 random-access techniques, and signal design and coding.
[42] THAJ T, VITERBO E. Low-complexity linear diversity-combining detector for
ZTE COMMUNICATIONS
December 2021 Vol. 19 No. 4 15