OTFS Enabled NOMA for MMTC Systems over LEO Satellite

MA Yiyan, MA Guoyu, WANG Ning, ZHONG Zhangdui, AI Bo

Abstract： As a complement of terrestrial networks， non-terrestrial networks （NTN） have ad?vantages of wide-area coverage and service continuity. The NTN is potential to play an impor?tant role in the 5G new radio （NR） and beyond. To enable the massive machine type communi ?cations （mMTC）， the low earth orbit （LEO） satellite is preferred due to its lower transmissiondelay and path loss. However， the LEO satellite may generate notable Doppler shifts to de ?grade the system performance. Recently， orthogonal time frequency space （OTFS） modulationhas been proposed. It provides the opportunity to allocate delay Doppler （DD） domain resourc ?es， which is promising for mitigating the effect of high mobility. Besides， as the LEO satelliteconstellation systems such as Starlink are thriving， the space spectrum resources have becomeincreasinglyscarce. Therefore， non-orthogonal multipleaccess（NOMA） isconsideredasacandidate technology to realize mMTC with limited spectrum resources. In this paper， the ap?plication of OTFS enabled NOMA for mMTC over the LEO satellite is investigated. The LEOsatellite based mMTC system and the OTFS-NOMA schemes are described. Subsequently， thechallenges of applying OTFS and NOMA into LEO satellite mMTC systems are discussed. Fi?nally， the potential technologies for the systems are investigated.

Keywords： mMTC; LEO satellite; OTFS; NOMA

Citation （IEEE Format）： Y. Y. Ma， G. Y. Ma， N. Wang， et al.，“OTFS enabled NOMA for mMTC systems over LEO satellite，”ZTE Com? munications， vol. 19， no. 4， pp. 63–70， Dec. 2021. doi： 10.12142/ZTECOM.202104007.

1 Introduction

1.1 Non-Terrestrial Networks

Nowadays，non-terrestrialnetworks（NTN）areplayingan importantroleinhumansociety，especiallyinnavigation，ground monitoring and communications services. It regains at? tention recently in both academia and industry， advertised by several companiesplans of launching thousands of satellites in the low earth orbit（LEO）， such as OneWeb and SpaceX. The NTN refers to the segment of network that uses air-borne or space-borne vehicles as relay nodes or base stations（BS） for transmission[ 1]. In a typical NTN， it features the following elements： the satellite-gateway that connects the NTN to the public network， satellites or unmanned aerial systems （UAS）， and user equipment.

There are kinds of available orbits for space-borne vehicles， including the high elliptical orbit （HEO） （400– 50 000 km）， geostationary earth orbit （GEO） （35 786 km）， medium earth or? bit （MEO） （7 000 – 25 000 km）， and LEO （300 – 1 500 km）. Differentorbitsexhibitdifferentcoverageandtransmission characteristics， which determine the types of services on them.The non-GEO （NGSO） satellites refer to LEO and MEO satel? lites， whoseorbitalperiodsvarybetween1.5and10h. The UAS has an altitude range of 8 – 50 km and keeps a fixed posi? tion in terms of elevation with respect to a given earth point[ 1].

1.2 MMTC System over LEO Satellite

Inthe5Gnewradio（NR）system，massivemachinetype communications （mMTC） are introduced for Internet of Things （IoT）. 5G NR considers the IoT communications with low ener? gy consumption， low data rate， burst transmissions and mas ? sive connections. In the beyond 5G （B5G） and 6G networks， IoT communications will take a larger proportion[2]. However， it is unavoidable that the required large-scale deployment of 5G or 6G BSs in the future could generate greater overhead comparedwithshort-termrevenue.Furthermore，terrestrial networks （TN） are hard to be deployed due to the geographical environment and cost in unmanned areas whereas IoT commu ? nication services are required. However， the NTN is not re? stricted by geographical environment and can still work in un ? manned and geological disaster areas. Therefore， the NTN is considered to be involved as a complement of TN to construct an air-space-ground integrated network for the sake of the su ? periority of cost， coverage and massive connections[ 1 –3]. Ref. [3] demonstrates that Release17 will study the feasibility of adaptingnarrowbandNB-IoTtosupportNTN.Additionally， potential modifications of NB-IoT at physical and higher layer aspects tosupport NTN are investigated[3]. For example， the transmission bandwidth can be reduced to improve uplink sig? nal to noise ratio （SNR） and to realize transmission under lim ? ited power of devices， based on which it is possible to support low data rate GEO satellite communication using the 23 dBm device power class. Besides， directional antenna can also be equippedbymMTCdevicestoobtaintransmissiongain. Among the candidate orbits， path loss and transmission delayof LEO communications are smaller than those of MEO andGEO， so LEO is more competitive for connection of terrestrial power-constrained mMTC devices.

The 3GPP[4]suggests that the connectivity of mMTC is106 devices/km2. Since the typical beam footprint size of an LEO satellite is 100 – 1 000 km[ 1] ， there could be 108– 109devices inside. Though the TN could coordinate with NTN for mMTC connection， heavy control signal overhead in the grant-based multipleaccess（MA）technologies willbegenerateddueto the massive concurrent devices. Besides， the scarce frequency resources are different to be allocated to the massive IoT de ? vices orthogonally[5]. Researchers have tried to introduce non- orthogonalmultipleaccess（NOMA）intosatelliteframe ? work[6 –8]. The related works focus on improving system spec ? trum efficiency and show the feasibility of applying NOMA in ? to NTN. Moreover， the reliable access of massive devices af? fected by power limitation and Doppler shift remains to be a challenging topic.

On the other hand， the LEO satellite rotates at a high speedand the Doppler shift effect is obvious. The Doppler character? ization of LEO satellites was investigated in Ref. [9]， as a func ? tion of the maximum elevation angular and satellite position. For instance， the normalized Doppler of the satellite with cir? cular orbit altitude1 000 km and inclination 53° is around 10-5for a terminal located at （39° N， 77° W）， which varies at the rate0. 1 ppm/sapproximately.Since thesevereDoppler shift could introduce inter carrier interference （ICI） to degrade thetransmissionperformance，variousDopplerestimation schemes have been proposed[ 10 – 14]. In general， the Doppler es ? timators are based on either geometric information or pream ? ble data. For the device capable of tracking satellites based on ephemeris， such as the ground gateway， the Doppler shift can be estimated and compensated according to the obtained Dop ? plershiftcurveanditsownpositioninformation. However， Doppler estimation at mMTC devices could raise complexity and consume extra energy[15]. Thus， the coexistence of Doppler shift at the user side needs to be considered， which could be handled at the gateway.

1.3 CombiningOTFSandNOMAforMMTCSystem over LEO Satellite

Takingtheeffectof scarcespectrumandnotableDoppler shift into account， the mMTC over LEO satellite prefers the NO ? MA technologies which are robust to multiple Doppler shift. Un ? der such vision， the classic orthogonal frequency division multi ? ple （OFDM） framework is no longer suitable for the system due to the vulnerability to Doppler shift. In Ref. [15]， the adaptabili? ty of NB-IoT technologies for NTN communications was investi ? gated， which mitigated the effect of Doppler shift by user group ? inggeographically.ThenthemaximumdifferentialDoppler shift of users in the same group is limited to a tolerable range of NB-IoT， which is 950 Hz. Subsequently， the gateway compen? sates the common Doppler shift component of users in each us ? er group. By doing so， NB-IoT is practical for IoT communica? tions over the LEO satellite. However， this scheme is not gener? al for the NOMA technologies， since the inter carrier interfer? ence （ICI） remains to be unprocessed. It will limit the system user capacity and improves the complexity of the ground gate ? way，especiallyfortheschemeswhereconcurrencyislarger than NB-IoT. Recently， orthogonal time frequency space （OT ? FS） modulation has been proposed. OTFS is a chance to utilize the Doppler diversity to realize the NOMA system robustness against multiple usersDoppler shift. In what follows， the appli? cation of OTFS enabled NOMA for mMTC over LEO satellite isinvestigated. Specifically， the system architecture is described. Adaptability research of OTFS-NOMA for mMTC over LEO sat? ellite is discussed. Moreover， the potential technologies for the system are listed.

2 System Description

Accordingtothe3GPPNTNstandard[ 1] ，therearethreeparts in the mMTC system over the LEO satellite， including ground massive devices， the satellite and the ground gateway （Fig. 1）. The link between the LEO satellite and mMTCdevic ? es refers to the service link while that between the LEO satel ? lite and ground gateway refers to feeder link. In the system， the satellite could work in a transparent or regenerative mode. In the transparent mode， the satellite performs frequency carri ? er changing， filtering and amplifying merely. For the regenera? tivepayload，thesatelliteperformssignaldigitalprocessing additionally， including demodulation/re-modulation， decoding/ re-encoding， etc. Due to the difficulty of air interface protocol application on the satellite， the transparent mode of the satel? lite is supposed in this paper. Besides， this paper concentrates on the uplink access procedure， which refers to the link from thedevicestothegroundgateway. Theinteractionbetween the LEO satellite and mMTC devices during uplink access de ? pends on the utilized MA technologies. For example， four-step random access channel （RACH） enhancements for NTN are in? vestigated in Ref. [ 1] considering the long transmission delay. After receiving mMTCdevicesdata， the LEO satellite would forward it to the ground gateway， where user identification and data recovery are performed.

As introduced before， LEO could generate a notable Dop ? plershift todegrade thesymbol performance. In the feeder link， it is assumed that the link is available all the time. Thus， the gateway could always track the satellite and compensate the Doppler shift inside. In the service link， the Doppler shift is determined by the LEO satellite altitude， the maximum and minimum elevation angles of the devices， etc. Hence， the OT? FS enabled NOMA for mMTC over the LEO satellite could be designed based on devicesgeographic locations.

Current MA technologies for NTN enabled communications can be divided into the grant-based and grant-free. The grant- based MA technologies include time division multiple access （TDMA）， frequency division multiple access （FDMA）， code di? vision multiple access （CDMA）， physical random access chan? nel （PRACH） in 5G-NR， etc. For example， FDMA and TDMA areappliedin the Thuraya， AceSandIridiumsatellitesys ? tems; CDMA is adopted by the Odyssey and Glonass systems. In the 3GPP 5G NR-NTN R16， PRACH formats and preamble sequences in Rel- 15 are reused for random access[ 1]. Besides，several options are provided for pre-compensating the timing and frequency offset at the device side. In general， the grantbasedMAhaveseveralroundsofinteractionsbeforedatatransmission between devices and the satellite， which are de? signed for device access and contention resolution. In addi? tion， each device occupies certain specific resources indepen ? dently. In the grant-free category， there are ALOHA， conten? tion resolution diversity slotted ALOHA （CRDSA）， etc. Devic ? es in such grant-free MA systems do not require the satellite to perform dynamic scheduling authorization， but transmit da? ta independently and obtain resources by competition. To re ? duce the control signal overhead， this paper adopts the grant- free access mode for the mMTC over LEO satellite.

3 OTFS-NOMAforLEOSatelliteMMTC System

3.1 NOMA Schemes for LEO Satellite MMTC System

Up to now， multiple NOMAschemes have been proposed for the mMTC system over the LEO satellite[6， 7， 16]. Simulation results demonstrate that the satellite communication systems can benefit from the application of non-orthogonal transmis ? sion schemes in terms of capacity and user fairness. However， the related studies rarely considered the Doppler effect and di ? rectly assumed perfect/imperfect channel satellite information （CSI） in the system. As introduced before， the coexistence of multiple Doppler shifts during the uplink transmission needs to be considered in the system due to the userspowerlimita? tion. Thus， OTFS is considered to be involved in such systems.

OTFS-NOMA was first proposed in Ref. [ 17]， which consid? ered a scenario where there was only one user with high mobil ? ity. In the mMTC system over the LEO satellite， the model in Ref. [ 17] could suffer more severe interference and the corre ? sponding performance need to be evaluated. Meanwhile， there areOTFSenabledcodeorspacedomainNOMAschemes， such as OTFS-SCMA in Ref. [ 18] and OTFS-tandem spread? ing multiple access （TSMA） in Ref. [ 19]. The applicability and improvement of differentOTFS-NOMAschemes in theLEO satellite system requires further research.

3.2 OTFS Modulation

OTFS modulation designs the transceiver in the delay Dop ? pler （DD） domain[20]. Note that the channel impulse response （CIR） of the double-selective channel is time varying in the time frequency （TF） domain. In the DD domain， the CIR re? flects thescatteringenvironment， whichcan be regardedas time-invariantduringa transmission frame.Besides，CIRin the DD domain is sparse due to the limited number of scatters. The channel equalization is simplified as a result. In the TF domain， different Doppler shifts of users occupying the same TF resource block are difficult to be equalized in the grant- free NOMA mode. Meanwhile， the effect of Doppler shifts isconsidered as interference in the TF domain， and its informa? tion entropy is not utilized. Now in the OTFS， transmission can be scheduled considering the usersDoppler shift charac? teristic， and the Doppler diversity could bring additional per? formance gain. In the DD domain， the transceiver input-output relationship depends on the DD resource granularity and ad ? opted waveform types[21]. If the waveforms ideally satisfy the bio-orthogonal property and the DD resource granularity is able to locate the CIR exactly， the output will be expressed as：

where x and y refer to the input and output in the DD domain respectively， P denotes the number of multipath， hi refers to the channel fading， τi and νi are the values of delay and Dop?pler shift of the i-th path， [ · ]N = mod （·， N）. In the TF do? main， it is assumed that the system occupies the time domain resource with NT and the frequency domain resource with MΔf， where N denotes the number of time intervals T and M denotes the number of subcarrier intervals Δf = 1/T. Then the scattered Doppler domain resource is obtained as l MΔf∈ é ? ê0， M - 1 MΔf ù ? ú， l = 0，1，...， M - 1， and the scattered delay domain resource is k NT ∈ é ? ê0， N - 1 NT ù ? ú， k = 0，1，...， N - 1[20] . Therein， the maximum Doppler and delay values are assumed to be less than Δf and T respectively. Additionally， suppose that τi and νi are dividable by 1 NT and 1 MΔf ， then kνi = νiNT and lτi = τiMΔf are obtained. Eq. （1） illustrates that based on the bio-orthogonal transceiver waveforms， y is the superposi?tion of the faded x， which is two-dimension cyclically shifted by the corresponding DD CIR. Under the non-delicate DD re? source granularity or the practical waveforms such as the rect? angular waveform， there are additional ICI and inter symbol interference （ISI） in the output.

As for the OTFS receiver， the data detection and channel estimation scheme should be able to reverse the two-dimen? sional cyclic shift in Eq. （2）. In terms of data detection， a be? spoke optimal maximum a posteriori （MAP） detector was pro? posed in Ref. [21]. Such kind of detector is designed based on the sparsity of the DD domain channel matrix H ∈NM × NM， and is with excessive complexity. Therefore， researchers have focused on complexity reduction of the MAP detector[22]. A variational Bayes （VB） OTFS detector was proposed in Ref. [23]. In such a detector， the distributions of OTFS symbols are constructed adaptively according to corresponding interfer? ence patterns to make the OTFS detection converge rapidly. Besides， Ref. [24] proposed a detector named cross-domain it? erative detection （CDID）. In this detector， a linear minimum mean squared error （L-MMSE） estimator is adopted for equalization in the time domain and low-complexity symbol-by-sym? bol detection is utilized in the DD domain. Both VB and CDID

based detectors are shown to be with close-to-optimal perfor? mance， where the computational complexity is reduced[22]. There are other potent schemes such as the combination of hy ? brid MAP and PIC detection proposed in Ref. [25].

In terms of channel estimation， OTFS modulation outper? forms the schemes designed in the TF domain， due to the DD domain channel sparsity and quasi-stationarity[22]. When the channel sparsity is damaged （e.g.， there are fractional Doppler shifts）， the requirement of larger guard space would result in heavy training overhead. The solution to enhancing the chan? nel sparsity has been proposed by TF domain window design ? ing[26]. Specifically， a Dolph-Chebyshev （DC） window is ap? plied in the transceiver of OTFS to suppress the channel spreading. It is demonstrated that better channel estimation accuracy is obtained by the DC windowing compared with the conventional rectangular window. Moreover， coded OTFS can be used to improve the system performance[27].

In the mMTC system over the LEO satellite， the satellite- land channel is different from that in terrestrial communica? tions. In terms of large-scale fading， there are obvious atmo? spheric absorption， rain attenuation， cloud attenuation and scintillation in the satellite-land channel， which should be considered in the link budget. In terms of the small-scale fad? ing， the LOS probability of the satellite-land channel is higher in the urban， suburban and rural scenarios[1]. Additionally， the additive white Gaussian noise （AWGN） channel can be as? sumed in the open area， such as the devices on boats or air? crafts. Therefore， the DD domain CIR is sparser than that in the terrestrial communication， and the DD domain input/out? put relationship under the ideal pulses is written as：

It should be noted that k ν and lτ are mainly related to the de? vicesgeographical position. Thus， the unique Doppler shift and delay values of difficult users can be regarded as a novel orthogonal space. It provides opportunity to schedule the sys ? tem transmission from the perspective of devicesgeography to mitigate multi-user interference （MUI）. As a result， the pre- compensation of Doppler shift at devices could be omitted.

3.3 Combining NOMA with OTFS for MMTC over LEO Satellite

To embed NOMA into OTFS framework for mMTC over the LEO satellite， a novel transceiver could be designed. In terms of the transmitter， the main concern is to design the resource allocation scheme adapting to the DD domain channel charac ? teristics. In terms of the receiver， equalization technologies need to be considered for the grant-free transmission. Further? more， strategies to scale up the system capacity， such as mas? sive multiple input multiple output （MIMO） and access point（AP） assistant communications， could be involved in. For the code domain grant-free OTFS-NOMA schemes， data spreading and interleaving are able to reverse two-dimension cyclic shift of the DD domain， like those shown in OTFS-TSMA[19].

TSMA is a code domain NOMA scheme， which innovates on transceiver design and data settings. At the TSMA transmit? ter， usersdata are segmented and encoded to generate redun ? dant segments. Subsequently， each user is allocated with a unique tandem spreading codeword， considering the indexes of the spreading sequences utilized by each segment. Then the segments are tandemly spread according to the codeword. Therein， the tandem spreading codebook C is designed accord? ing to the maximum distance separable （MDS） code， which maximizes the user capacity under the limitation of the num ? ber of colliding segments. At the TSMA receiver， the superpo? sition of active usersdata is separated by orthogonal correla? tion detection on each segment according to C. Then the ac? tive users set and corresponding data are obtained. Subse? quently， the colliding segments inside are deleted and the re ? dundant segments are used for segment decoding. Therefore， TSMA scales up the user capacity by tandem spreading， en? sures transmission reliability by orthogonal spreading se ? quences and sacrifices data rate.

As shown in Fig. 2， novel data interleaving and corresponding data recovery strategies are proposed for OTFS-TSMA. The chip-level data interleaving/de-interleaving procedures are demonstrated in Fig. 3. It shows that based on the inter? leaving/de-interleaving strategies， the two-dimension cyclic shift of DD domain resources is transformed into cyclic shifts of Doppler domain elements， segments， systems and chips. Additionally， OTFS-TSMA adopts the cyclic orthogonal spreading sequences， namely discrete Fourier transform （DFT） sequences， whose orthogonality is maintained by the interleav ? ing/de-interleaving strategies. At the OTFS-TSMA receiver， the segment-level cyclic shift is recovered during user identifi ? cation， the symbol-level cyclic shift is recovered after de- spreading， phase rotation caused by the chip level cyclic shift is recovered before channel equalization， and Doppler diversi? ty brought by the Doppler elements level cyclic shift is uti ? lized during data combination. Therefore， two-dimension cy? clic convolution of the DD domain is de-solved from the per? spective of data design. Besides， OTFS-TSMA cleverly com ? bines the multiple access characteristics of TSMA and the ro ? bustness of OTFS to dual selective channels. It could be in? volved into OTFS-NOMA enabled mMTC over the LEO satel? lite and inspire other OTFS-NOMA schemes of the code do? main. Additionally， to make full use of DD domain resources， the schemes need to consider usersgeometry positions.Therefore， users with certain longitude and latitude values are scheduled to transmit during certain timeslots according to their DD domain CIR characteristics.

For the power domain OTFS-NOMA， MUI brought by the DD domain channel could be mitigated by resource， rate and power allocation policies. In Ref. [ 17]， users with high mobil? ity are served in the DD domain and those with low mobility are serviced in the TF domain. Inspired by Ref.[ 17]， both the TF and DD domains could be utilized for mMTC over the

LEO satellite. For example， two users with the same Doppler shift and delay values （symmetrical with the sub satellite tra? jectory） could be serviced in two domains respectively. Be ? sides， the users could be scheduled according to the channel conditions， whereas adaptive-rate transmission or fixed-rate transmission could be considered depending on the NOMA usersability. Furthermore， power allocation schemes could bedesignedtofacilitatetheimplementationofsuccessful success interference cancellation （SIC）， according to the us ? ersQoS requirements and channel conditions. Note that the power domain OTFS-NOMA schemes rarely consider the up ? link channel estimation， which always assume perfect CSI at transceiver. They perform better in system design in terms of communicationcapacityandtransmissionfairness. Onthe fast-varyingsatellite-groundchannel，theperformanceof power domain OTFS-NOMA under imperfect CSI remains to be studied. Furthermore， in terms of the grant-free OTFS-NO ? MA design for mMTCsystems over the LEOsatellite ， data colliding caused by non-cooperative transmission would ef? fectuseridentificationanddatarecovery. Aclassicgrant- free OTFS-NOMA scheme utilizes the sparsity of devices[28] ， which is the idea from compressive sensing based multi-user detection（CSMUD）.Therein，eachuserisprovidedwitha preamble sequence known to the receiver， and the sequences areusedforuseridentificationandchannelequalization based on compressed sensing.Nextly， the data could be di? vided by SIC. There is also another way to design the grant- free OTFS-NOMA in the time domain using the idea of ALO ? HA. Overall， the performance degradation brought by non-co? operative transmission in the grant-free OTFS-NOMA mode canbecompensatedbyadditionaldesigninspace，code， power， time or frequency domains.

3.4 ChallengesofCombiningOTFSandNOMAfor MMTC over LEO Satellite

Firstly， the effect of a notable Doppler shift and heteroge ? neous delay is challenging for the system design. In the OTFS modulation， the DD domain resource plane requires that the subcarrier interval should be larger than the maximum Dop ? pler shift， and the time interval should be larger than the maxi ? mumdelay.Therefore，therequiredtimeandfrequencyre ? sources are dozens of times that of similar services on ground communications，duetothenormalizedDopplershiftat24 ppm and the propagation delay at 25.77 ms. It seems difficultto realize the system design with the rare space spectrum at sub-6G which is preferred for mMTC. Therefore， the Doppler and delay processing schemes need to be designed for OTFS- NOMA in the grant-free mode. For example， the system could concentrate on the differential values of the Doppler shift and delay， rather than the absolute values. In addition， the flexible configuration and mobility of the satellite system will bring dif? ferent delay and Doppler characteristics. It also puts forward requirements for the flexibility of Doppler and delay process ? ing schemes.

Secondly， the performance of OTFS-NOMA under rectangu? lar pulses and fractional Doppler shift needs to be investigat? ed. The DD domain input/output relationship in this scenario is written as：

Eqs.（3）– （5）illustratethatICIandISIareintroduced. Therefore，thesystemcapacityandtransmissionreliability performance of OTFS-NOMA remain to be investigated. Be ? sides， the OTFS-NOMA transmitter can be re-designed consid? ering the geometry position of signal x and the LEO satellite， which determines k νand lτ ， where the unknown quantity is on? ly h. The novel resource allocation scheme can be proposed correspondingly.

Thirdly，theretransmission，beamformingandhandover strategies remain to be studied. Due to propagation fading， the retransmissionstrategy formassivedevicesneedstobede ? signed.Itisconstrainedbytheshorttransmissionwindow， limiteddevicespowerandoverallthroughput.Inaddition， since the massive terrestrial devices need to be served by the earth-fixedbeams， beamformingof theLEOsatelliteshould be investigated. It could expand the system user capacity in the meantime. Besides， owing to the high mobility of the LEO satellite， a handover proposal for reliable communications is required.

3.5 Potential Technologies for OTFS Enabled NOMA for MMTC over LEO Satellite

Inadditiontothetechnologies forOTFSmodulationand NOMA， some additional technologies are still needed by the mMTC over LEO satellite to achieve reliable access and com ? munications.

Since the cost of cyclic pilot （CP） could be huge under the largedelayand theDopplershiftisnotable， frequencyand time advance at devices can be adopted. Therein， the propaga? tion delay and Doppler shift could be formulated by the loca? tion information and the LEO ephemeris. If the devices do not supportsuchkindofcalculation，certainvalueofDoppler shiftandtimeadvancecouldbeinitializedinthedevices. Thereafter， the LEO satellite concentrates on the differential Doppler shift and delay， which is much smaller than the abso ? lute value.

To tackle with the differential Doppler shift and delay， user grouping based on geography is a potential solution. It is able to limit the differential values of the usersDoppler shift and delayinacertaininterval.Subsequently，theinterference could be mitigated and the resources required by theOTFS could be reduced. Besides， resource allocation and data map ? ping schemes of OTFS could be designed correspondingly. In order to realize grouping and seamless convergence of the mas ? sive devices in the earth-fixed beams， beamforming strategies are required. They can also be combined with spectrum shar? ing technologies， such as the cognitive radio， to increase the spectrum resources efficiency.

In the wide area application of mMTC over the LEO satel ? lite， handover strategies need to be considered to enable the constantservice.Firstly，thespotbeamhandover，which switches the connection to a difficult spot-beam， could be re? searched to achieve the maximum system throughput. Second? ly， fast inter-satellite handover could be investigated to con ? nect the worldwide mMTC devices， under the high mobility of the LEO satellite. Therein， the design of LEO satellite cancel? lations is required.

In addition to mMTC over the LEO satellite， the air network consistingofthehigh-altitudeplatforms（HAPs）andun? manned aerial vehicles（UAVs） could complement the NTN. In detail， the HAPs are with lower mobility and smaller com ? munication coverage properties. Therefore， heterogeneous ar? chitecturecanbeadoptedaccordingtotheservicerequire ? ment. Additionally， the mobility management and routing algo ? rithms need to be considered.

4 Conclusions

In this paper， we investigate the OTFS enabled NOMA for mMTCsystemsovertheLEOsatellite.Thearchitectureand scenarios of the LEO satellite enabled mMTC systems are de ? scribed. The MA schemes in the current NTN are also listed. Under the contradiction between the massive access and scarcespectrum， the NOMA technology is introduced in the system. Moreover， in the grant-free access system， OTFS is introduced to tackle with the notable Doppler shift by Doppler diversity， rather thanpre-compensation. Thedesignsandchallengesof applying OTFS and NOMA into mMTC systems over the LEO satellite are then analyzed. In the end， the potential technolo? gies and further studies of the system are suggested.

References

[ 1] 3GPP. Solutions for NR to support non-terrestrial networks （NTN） （Release 16）： TR 38.821 v 16.0.0 [S]. 2019

[2] SHEN X M S）， CHENG N， ZHOU H B， et al. Air-space-ground integrated net? work technology： exploration and prospects [J]. Chinese journal on Internet of Things， 2020， 4（3）： 3 – 19. DOI： 10. 11959/j.issn.2096-3750.2020.00142

[3] LIBERGO，L?WENMARKSE，EULERS，etal.NarrowbandInternetof Things for non-terrestrial networks [J]. IEEE communications standards maga? zine， 2020， 4（4）： 49 –55. DOI： 10. 1109/MCOMSTD.001.2000004

[4] YAN X J， AN K， LIANG T， et al. The application of power-domain non-orthogo? nal multiple access in satellite communication networks [J]. IEEE access， 2019， 7： 63531 –63539. DOI： 10. 1109/ACCESS.2019.2917060

[5] CHAE S H， JEONG C， LEE K. Cooperative communication for cognitive satel ? lite networks [J]. IEEE transactions on communications， 2018， 66（ 11）： 5140 – 5154. DOI： 10. 1109/TCOMM.2018.2850813

[6] PEREZ-NEIRA A I， CAUS M， VAZQUEZ M A. Non-orthogonal transmission techniques for multibeam satellite systems [J]. IEEE communications magazine， 2019， 57（12）： 58 –63. DOI： 10. 1109/MCOM.001. 1900249

[7] CHU J H， CHEN X M， ZHONG C J， et al. Robust design for NOMA-based mul? tibeamLEOsatellite Internet of Things[J]. IEEEInternet of Things journal， 2021， 8（3）： 1959 – 1970. DOI： 10. 1109/JIOT.2020.3015995

[8] LIU X， ZHAI X B， LU W D， et al. QoS-guarantee resource allocation for multibeam satellite industrial Internet of Things with NOMA [J]. IEEE transactions on industri? al informatics， 2021， 17（3）： 2052 –2061. DOI： 10.1109/TII.2019.2951728

[9] ALI I， AL-DHAHIR N， HERSHEY J E. Doppler characterization for LEO satel? lites [J]. IEEE transactions on communications，1998， 46（3）： 309 – 313. DOI： 10. 1109/26.662636

[ 10] YOU M H， LEE S P， HAN Y. Adaptive compensation method using the predic ? tion algorithm for the doppler frequency shift in the LEO mobile satellite com ? municationsystem[J].ETRI journal，2000，22（4）：32– 39.DOI：10.4218/ etrij.00.0100.0404

[ 11] LIN J N， HOU Z W， ZHOU Y Q， et al. Map estimation based on Doppler char? acterization in broadband and mobile LEO satellite communications [C]//83rd Vehicular Technology Conference （VTC Spring）. Nanjing， China： IEEE， 2016： 1 –5. DOI： 10. 1109/VTCSpring.2016.7504336

[ 12] LIU Y J， ZHU X， LIM E G， et al. High-robustness and low-complexity joint es? timationof TOAsandCFOsformultiuserSIMOOFDMsystems[J].IEEE transactions on vehicular technology， 2018， 67（8）： 7739 –7743. DOI： 10. 1109/TVT.2018.2821152

[ 13] TIAN D， ZHAO Y， TONG J F， et al. Frequency offset estimation for 5G based LEO satellite communication systems [C]//IEEE/CIC International Conference on Communications in China （ICCC）. Changchun， China： IEEE， 2019： 647 – 652. DOI： 10. 1109/ICCChina.2019.8855824

[ 14] PAN M G， HU J L， YUAN J H， et al. An efficient blind Doppler shift estima? tion and compensation method for LEO satellite communications [C]//20th In? ternational Conference on Communication Technology（ICCT）. Nanning， Chi? na： IEEE， 2020： 643 –648. DOI： 10. 1109/ICCT50939.2020.9295821

[15] KODHELIO，ANDRENACCIS，MATURON，etal.AnuplinkUEgroup- basedschedulingtechniquefor5GmMTCsystemsoverLEOsatellite[J].IEEE access， 2019， 7： 67413–67427. DOI： 10.1109/ACCESS.2019.2918581

[16] ZHANG Z J， LI Y， HUANG C W， et al. User activity detection and channel estimation for grant-free random access in LEO satellite-enabled Internet ofThings [J]. IEEE Internet of Things journal， 2020， 7（9）： 8811 – 8825. DOI：10.1109/JIOT.2020.2997336

[17] DING Z G， SCHOBER R， FAN P Z， et al. OTFS-NOMA： an efficient approach forexploiting heterogenous user mobility profiles [J]. IEEE transactions on communications， 2019， 67（11）： 7950–7965. DOI： 10.1109/TCOMM.2019.2932934

[18] DEKA K， THOMAS A， SHARMA S. OTFS-SCMA： A code-domain NOMA approach for orthogonal time frequency space modulation [J]. IEEE transactions oncommunications， 2021， 69（8）： 5043–5058. DOI： 10.1109/TCOMM.2021.3075237

[19] MA Y Y， MA G Y， WANG N， et al. OTFS-TSMA for massive Internet ofThings in high-speed railway [J]. IEEE transactions on wireless communications， Early access， 2021. DOI： 10.1109/TWC.2021.3098033

[20] HADANI R， RAKIB S， TSATSANIS M， et al. Orthogonal time frequency spacemodulation [C]//Wireless Communications and Networking Conference （WCNC）.San Francisco， USA： IEEE， 2017： 1–6. DOI： 10.1109/WCNC.2017.7925924

[21] RAVITEJA P， PHAN K T， HONG Y， et al. Interference cancellation and iterative detection for orthogonal time frequency space modulation [J]. IEEE transactions on wireless communications， 2018， 17（10）： 6501 – 6515. DOI：10.1109/TWC.2018.2860011

[22] WEI Z， YUAN W， LI S， et al. Orthogonal time-frequency space modulation： apromising next generation waveform [J]. IEEE wireless communications. 2021，28（4）： 136–144. DOI： 10.1109/MWC.001.2000408

[23] YUAN W， WEI Z， YUAN J， et al. A simple variational bayes detector for orthogonal time frequency space （OTFS） modulation [J]. IEEE transactions on vehicular technology， 2020， 69（7）： 7976–7980. DOI： 10.1109/TVT.2020.2991443

[24] LI S， YUAN W， WEI Z， et al. Cross domain iterative detection for orthogonaltime frequency space modulation [J]. IEEE transactions on wireless communications， 2021，early access. DOI： 10.1109/TWC.2021.3110125

[25] LI S， YUAN W， WEI Z， 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

[26] WEI Z， YUAN W， LI S， et al. Transmitter and receiver window designs for orthogonal time-frequency space modulation [J]. IEEE transactions on communications， 2021， 69（4）： 2207–2223. DOI： 10.1109/TCOMM.2021.3051386

[27] LI S， YUAN J， YUAN W， et al. Performance analysis of coded OTFS systemsover high-mobility channels [J]. IEEE transactions on wireless communications， 2021， 20（9）： 6033–6048. DOI： 10.1109/TWC.2021.3071493

[28] WANG F G， MA G Y. Massive machine type communications： multiple accessschemes [M]. Heidelberg， Germany： Springer， 2019

Biographies

MA Yiyan received the B.S. degree in applied physics from Beijing Jiaotong University， China in 2019， and is currently working toward the Ph.D. degree at the State Key Laboratory of Rail Traffic Control and Safety， Beijing Jiaotong University. His current research interests include the field of Internet of Things and massive machine type communications.

MA Guoyu （magy@bjtu.edu.cn） received the B.S. and Ph.D. degrees in electri?cal engineering from Beijing Jiaotong University， China in 2012 and 2019， respectively. Currently he is an associate professor at the State Key Laboratory ofRailTrafficControlandSafety，BeijingJiaotongUniversity.Hiscurrentre? search interests include machine-type communications and random access.

WANGNingreceivedtheB. E.degreeincommunicationengineering from Tianjin University， China in 2004， the M.A.Sc. degree in electrical engineering from The University of British Columbia， Canada in 2010， and the Ph.D. degree in electrical engineering from the University of Victoria ， Canada in 2013. He wasontheFinalistof theGovernorGeneralsGoldMedalforOutstanding Graduating Doctoral Student with the University of Victoria in 2013. From 2004 to 2008， he was with the China Information Technology Design and Consulting Institute as a mobile communication system engineer， specializing in planning and design of commercial mobile communication networks， network traffic analysis， and radio network optimization. He was a postdoctoral research fellow with the Department of Electrical and Computer Engineering， The University of British Columbia， from 2013 to 2015. Since 2015， he has been with the School of Information Engineering， Zhengzhou University，China， where he is currently an associate professor. He also holds adjunct appointments with the Department of Electrical and Computer Engineering， McMaster University， Canada， and the Department of Electrical and Computer Engineering， University of Victoria. He has served on the technical program committees of international conferences ， including the IEEE GLOBECOM， IEEE ICC， IEEE WCNC， and CyberC. His research interests include resource allocation and security designs of future cellular networks， channel modeling for wireless communications， statistical signal processing， and cooperative wireless communications.

ZHONG Zhangdui is currently a professor and advisor of Ph. D. candidates withBeijing Jiaotong University，China. He is also a chief scientist with the State Key Laboratory of Rail Traffic Control and Safety， Beijing JiaotongUniversity. He is a director of the Innovative Research Team of Ministry of Education and a chief scientist of Ministry of Railways in China. He is an executive council member of Radio Association of China and a deputy director of Radio Association of Beijing. He has authored/coauthored seven books， five invention pat ents， and more than 200 scientific research papers in his research area. His in terests include wireless communications for railways， control theory and techniques for railways， and GSM-R system. His research results have been widely used in railway engineering， including the Qinghai-Xizang railway， Datong-Qin huangdao heavy haul railway and many high-speed railway lines in China. Prof ZHONG was the recipient of a Maoyisheng Scientific Award of China ，Zhantian youRailwayHonoraryAwardofChina，andTopTenScience/Technology Achievements Award of Chinese Universities. He is a fellow of IEEE.

AI Bo （boai@bjtu.edu.cn） graduated from Tsinghua University， China， with the honor of Excellent Postdoctoral Research Fellow in 2007. He received the mas? ters and Ph. D. degrees from Xidian University， China in 2002 and 2004， re? spectively. He is currently working as a full professor and a Ph. D. advisor with Beijing Jiaotong University， China， where he is also the deputy director of the State Key Laboratory of Rail Traffic Control and Safety and the International Joint ResearchCenter.He has authored or coauthored eight books and pub ? lished over 300 academic research papers in his research area. He holds 26 in? vention patents. He has been the research team leader for 26 national projects and has won some important scientific research prizes. Five of his papers have been the ESI highly cited papers. He has been notified by the Council of Cana? dian Academies （CCA） that， based on Scopus database， he has been listed as one of the Top 1% authors in his field all over the world. His research interests includetheresearchandapplicationsof channelmeasurementandchannel modeling and dedicated mobile communications for rail traffic systems. He is afellow of IEEE and IET.