PaperlayerView exact revision on arXiv

Source-linked AI summary

Modulation and Multiple Access for 5G Networks

Yunlong Cai, Zhijin Qin, Fangyu Cui, Geoffrey Ye Li, Julie A. McCann

arXiv:1702.07673v1cs.OH

TL;DR

5G networks require modulation and multiple-access schemes that can support large-scale, heterogeneous traffic and users. This article surveys promising modulation candidates for OMA and NOMA schemes across power, code, and multiple domains. It concludes that new OMA modulations can reduce out-of-band leakage, while NOMA’s non-orthogonality can support enhanced throughput and massive connectivity.

  • Problem

    5G’s large-scale heterogeneous traffic creates design challenges that motivate new modulation and multiple-access schemes.

  • Method

    The article provides a comprehensive survey of promising modulation and MA candidates, covering OMA modulation and power-domain, code-domain, and multidomain NOMA.

  • Results

    New OMA modulations can reduce out-of-band leakage, while non-orthogonal MA is associated with enhanced throughput and massive connectivity.

  • Takeaways & Limitations

    The surveyed opportunities and challenges can inform the design of modulation and multiple access for 5G networks.

Abstract

from arXiv · show

Fifth generation (5G) wireless networks face various challenges in order to support large-scale heterogeneous traffic and users, therefore new modulation and multiple access (MA) schemes are being developed to meet the changing demands. As this research space is ever increasing, it becomes more important to analyze the various approaches, therefore in this article we present a comprehensive overview of the most promising modulation and MA schemes for 5G networks. We first introduce the different types of modulation that indicate their potential for orthogonal multiple access (OMA) schemes and compare their performance in terms of spectral efficiency, out-of-band leakage, and bit-error rate. We then pay close attention to various types of non-orthogonal multiple access (NOMA) candidates, including power-domain NOMA, code-domain NOMA, and NOMA multiplexing in multiple domains. From this exploration we can identify the opportunities and challenges that will have significant impact on the design of modulation and MA for 5G networks.

I. INTRODUCTION

5G’s massive, heterogeneous, and asynchronous traffic creates challenges that traditional OFDM and OMA cannot fully address. The article surveys new modulation and NOMA approaches intended to improve efficiency and support more users.

  • 5G must support massive connectivity, high throughput, improved spectral efficiency, and diverse user and application requirements.
  • Traditional OFDM performs well for broadband transmission but requires strong synchronization, making asynchronous narrowband mMTC transmissions vulnerable to adjacent-subband interference.
  • Filtering, pulse shaping, precoding, and delay-Doppler modulation are proposed to reduce OFDM out-of-band leakage or address high-Doppler scenarios.
  • OMA: OMA provides simple detection with minimal adjacent-block interference but supports limited users because orthogonal resource blocks are finite, restricting spectral efficiency and capacity.
  • NOMA: NOMA multiplexes users within shared time, frequency, or code resources, with power- and code-domain schemes introducing extra interference and receiver complexity.
  • The article compares modulation candidates for OMA, discusses three major NOMA types, and outlines the organization of these analyses.

II. NOVEL MODULATION FOR OMA

The paper reviews OFDM-derived and alternative modulation techniques for OMA, emphasizing lower leakage, asynchronous operation, flexible parameters, and trade-offs in complexity and time-frequency localization.

  • Traditional OFDM: OFDM is retained as a compatibility basis, but its high out-of-band leakage and guard-band requirements reduce spectral efficiency, especially with asynchronous or narrowband users.
  • Traditional OFDM: With a cyclic prefix longer than the channel delay span, OFDM converts channel distortion into per-subcarrier multiplication and enables simpler detection.
  • New OMA modulations target high spectral efficiency, loose synchronization requirements, and independently configurable subcarrier widths and symbol periods.
  • The reviewed OMA techniques include pulse shaping, subband filtering, precoding, guard-interval shortening, and delay-Doppler modulation.
  • Pulse shaping: Pulse shaping reduces out-of-band leakage but cannot simultaneously reduce time and frequency widths, so its waveforms are generally non-orthogonal and transceivers are more complex.
  • FBMC: FBMC is presented as a pulse-shaped modulation based on IDFT and DFT processing, with transmit symbols represented using pulse amplitude modulation.

1) FBMC:

FBMC uses filter-bank pulse shaping and OQAM to achieve real-domain orthogonality while limiting bandwidth. The section also describes GFDM and related pulse-shaped modulations as alternatives for reducing leakage and supporting flexible transmission.

  • FBMC: FBMC combines IDFT/DFT processing with synthesis and analysis polyphase filter banks, whose prototype filter performs pulse shaping.Typical prototype pulses include IOTA and PHYDYAS filters.
  • FBMC: FBMC prototype pulses are several symbol periods long and confined to a few subbands, unlike traditional OFDM pulses with long spectral tails.The time-domain length is selected according to required performance.
  • FBMC: OQAM staggers QAM-derived PAM symbols so nearby overlapped-symbol interference becomes purely imaginary and can be cancelled by a matched-filter receiver.The offset also makes adjacent block intervals half a block period.
  • GFDM: GFDM uses circularly shifted prototype filters and flexibly adjusts M frequency samples and K time samples for an application environment.Its matrix-based modulation and demodulation replace OFDM's IDFT and DFT matrices with GFDM-specific matrices.
  • Pulse-shaped modulations: Pulse-shaped modulations restrict transmit signals to narrow bandwidths, mitigating OOB leakage and supporting asynchronous operation with narrow guard bands.FBMC uses OQAM to save the guard interval and interference-cancellation cost.
  • GFDM: GFDM's circular filters avoid the long time-domain tails of linear filters, fitting sporadic transmission, and the scheme is compatible with MIMO technologies.GFDM can reduce out-of-block leakage even when orthogonality is completely given up.

C. Modulations based on Subband Filtering

Subband filtering reduces OOB leakage by filtering groups of subcarriers, with UFMC and f-OFDM offering different trade-offs in filter duration, interference handling, and flexibility. UFMC uses zero padding and short filters, whereas f-OFDM uses a cyclic prefix and can employ longer filters with stronger out-of-band attenuation.

  • Overview: UFMC and f-OFDM are representative 5G modulations based on subband filtering, a technique intended to reduce OOB leakage.Other subband-filtering variants include resource block f-OFDM.
  • UFMC: UFMC divides N subcarriers into K equal subbands, applies shifted prototype filters to the subbands, and uses OFDM modulation within each subband.Each subband contains L = N/K consecutive subcarriers.
  • UFMC: UFMC uses a zero-padding prefix to eliminate interference from short filter tails, then zero-pads received intervals to enable a 2N-point FFT.Only even subcarriers are used for signal detection after the FFT.
  • f-OFDM: f-OFDM uses a cyclic prefix and permits residual ISI, allowing longer filters than UFMC and better attenuation outside the band.Residual interference can be treated as noise, and effective channel coding can make its performance degradation negligible.
  • f-OFDM: The soft-truncated sinc filter is widely used in f-OFDM because its parameters can be adapted across applications, making the modulation flexible in frequency.f-OFDM can also use different subcarrier spacings and cyclic-prefix lengths for different users.
  • Other modulations: GI DFT-s-OFDM replaces the cyclic prefix with a known sequence that can reduce PAPR and support carrier-frequency-offset estimation.A suitable known guard interval can also avoid discontinuities between adjacent time blocks and reduce OOB leakage.

2) SP-OFDM:

The section describes SP-OFDM and related modulation approaches for reducing out-of-band leakage, then compares their spectral and BER behavior under static and high-mobility channels.

  • 2) SP-OFDM:: SP-OFDM uses DFT spreading, spectral precoding, and iterative detection to project signals into a lower-dimensional subspace with lower leakage than traditional OFDM.The rank-deficient precoder can suppress OOB leakage at the cost of only a few reduced dimensions.
  • 2) SP-OFDM:: SP-OFDM avoids filter-tail ISI, can notch selected frequencies in fragmented bands, and can combine precoding with filtering for further improvement.These advantages distinguish it from filtering-based modulation schemes.
  • 2) SP-OFDM:: OTFS maps symbols into the delay-Doppler domain and uses a two-dimensional symplectic Fourier transform before time-frequency-domain transmission.Pulse shaping and subband filtering can also be combined with OTFS to reduce leakage.
  • 2) SP-OFDM:: FBMC and f-OFDM achieve the lowest leakage, while UFMC, GFDM, GI DFT-s-OFDM, and SP-OFDM also outperform traditional OFDM in OOB leakage.The comparison uses PSDs under a 15.36 MHz bandwidth and 1024-point DFT setting.
  • 2) SP-OFDM:: FBMC, UFMC, GFDM, and SP-OFDM have similar BER performance, whereas f-OFDM is slightly worse because its extra ISI cannot be completely canceled.The degradation for f-OFDM is especially apparent at high SNR.
  • 2) SP-OFDM:: In high mobility, OTFS maintains good performance through its channel-estimation method, and its performance can exceed the zero-Doppler case because of Doppler diversity.The delay-Doppler channel representation also provides a stable model for time-varying fading and supports localized pilot-based MIMO CSI estimation.

F. Open Issues

The open issues concern balancing leakage, time-frequency dispersion, PAPR, filtering, and application-specific requirements, while NOMA offers resource sharing and connectivity benefits.

  • F. Open Issues: In narrowband IoT subbands, interference from a short CP can degrade performance, requiring additional processing such as filtering, SIC, or RISIC.The relevant detection treatment depends on the application setting.
  • F. Open Issues: Fixed-length subband filters may not suit different users and application scenarios, which motivates more adaptable filtering designs.The text identifies filter length and leakage requirements as scenario-dependent.
  • F. Open Issues: Time and frequency dispersions cannot both be reduced, so prototype-filter design must balance them according to application scenarios.The trade-off follows the Heisenberg-Gabor uncertainty principle.
  • F. Open Issues: FBMC and UFMC retain large PAPR, while conventional PAPR-reduction methods introduce distortion that can degrade performance.Extending PAPR reduction appropriately to new modulations remains an open design problem.
  • F. Open Issues: NOMA improves spectral efficiency by allowing multiple users to share one time, frequency, or code resource block.This resource sharing is presented as a key NOMA feature.
  • F. Open Issues: NOMA can support massive connectivity because multiple users can occupy one resource block, including billions of low-rate smart devices.This capability is particularly relevant to IoT scenarios.
  • F. Open Issues: NOMA relaxes uplink channel-feedback requirements and can enable grant-free transmission without scheduling requests, reducing latency drastically.Perfect uplink CSI is not required at the base station.

A. Power-Domain NOMA

Power-domain NOMA multiplexes users on the same resource block using different power levels and SIC, improving connectivity and throughput but introducing interference, complexity, and scalability limits.

  • Basic principle: Power-domain NOMA distinguishes multiple users sharing one resource block through different power levels.Downlink transmission superimposes users’ signals under a total power constraint.
  • Power allocation and detection: Downlink power allocation gives more power to users with poorer CSI to support fairness and reasonable received signal power.Users with better CSI receive less power, while SIC decodes the higher-power signal first and subtracts it before detecting subsequent users.
  • Power allocation and detection: SIC decodes the weaker-channel, higher-power user first, but its detection error propagates to later users and requires sufficient allocated power.The first detected user also experiences the largest inter-user interference.
  • Performance and limitations: NOMA achieves lower outage probability than OMA, while requiring more complex transmitters and receivers to mitigate interference.The cited comparison considers two users served by the same base station under NOMA.
  • Performance and limitations: Power-domain NOMA generally works best with two or a few users per resource block; increasing multiplexed users intensifies MAI and degrades performance.NOMA can support more connectivity and higher throughput with limited resources, but this benefit is constrained by multiuser interference.

2) Multiple Antenna based NOMA:

Multiple-antenna NOMA uses spatial processing alongside power-domain multiplexing, offering performance gains while creating beam-design, user-ordering, and optimization challenges.

  • Multiple-antenna design: Multiple antennas add a spatial degree of freedom to NOMA, enabling beam-based designs that can improve performance.Users may share a beam, while different clusters use carefully designed beams to suppress inter-cluster interference.
  • Multiple-antenna design: User ordering is a central challenge in multiple-antenna NOMA because channels are represented by vectors or matrices rather than scalars.Designs serve one or multiple users with a single beamforming vector, or assign different users different beams in one resource block.
  • Reported performance: Multiple-antenna NOMA schemes have shown significant performance improvement over conventional OMA schemes.Reported approaches include power optimization and beamforming-based designs.
  • Reported performance: MIMO-based NOMA can achieve better outage performance than MIMO-based OMA even for users experiencing strong co-channel interference.A precoding and detection framework with fixed power allocation was proposed to address uncertainties from random beamforming.
  • Power allocation: Power allocation must account for CSI ordering and fairness, while ordered power constraints are non-convex and make optimization difficult.Separate formulations are needed for perfect and imperfect CSI, and further research on optimal ordering is expected.
  • Cooperative NOMA: Cooperative NOMA uses better-CSI users or coordinated base stations to relay or jointly support poorer-CSI and cell-edge users.Relay-assisted schemes use DF or AF relaying, while CoMP coordinates multiple base stations for cell-edge transmission.

5) Spectral and Energy Efficiency in NOMA:

NOMA’s extra power-domain degree of freedom supports spectral efficiency and massive connectivity, but spectral and energy efficiency cannot generally be optimized simultaneously.

  • Efficiency objectives: Spectral efficiency and energy efficiency are important NOMA metrics, and resource-allocation methods have been studied to maximize energy efficiency under rate constraints.Reported approaches include sub-optimal allocation and formulations with minimum required data rates.
  • Efficiency objectives: NOMA is considered capable of boosting spectral efficiency in 5G networks through an additional power-domain degree of freedom.This property is especially suited to IoT networks requiring massive connectivity with low sensor-node power consumption.
  • Efficiency tradeoff: Spectral efficiency and energy efficiency cannot be achieved simultaneously in NOMA networks, motivating further work on their tradeoff.The review identifies this tradeoff as an area for future research.
  • Code-domain NOMA: Code-domain NOMA assigns different codes to users sharing a time-frequency resource block, gaining spreading and shaping benefits at the cost of extra signal bandwidth.Examples include LDS-CDMA, LDS-OFDM, and SCMA.
  • Code-domain NOMA: Sparse signatures enable low-complexity, near-optimal message-passing detection for LDS-CDMA, significantly improving performance.LDS-OFDM maps signatures onto OFDM subcarriers, supporting wideband channels and flexible resource allocation.

3) SCMA:

SCMA combines sparse spreading with multidimensional codebooks and message-passing detection, enabling overloaded resource blocks and reduced receiver complexity, while broader NOMA designs multiplex across multiple domains.

  • SCMA design: SCMA can distinguish colliding users through additional resource blocks and multidimensional constellation structure.The cited example has six users sharing four resource blocks.
  • SCMA design: SCMA uses sparse multidimensional constellations to reduce receiver complexity and improve spectral efficiency.A four-point QAM constellation, for example, can be projected into a three-point constellation in a resource-block subspace.
  • SCMA detection: SCMA maintains excellent detection performance when resource blocks are overloaded because of sparse spreading and large multidimensional constellation distance.MPA detection approaches near-optimal performance with lower complexity than ML and BCJR, though device-side complexity remains relatively high.
  • Multi-domain multiplexing: Combining SIC across clusters with MPA within clusters can significantly reduce receiver complexity.Different clusters may be separated by power and space, while users sharing a signature matrix are detected using MPA.
  • Multi-domain multiplexing: NOMA multiplexing can span power, code, and spatial domains through PDMA, BOMA, and LPMA to support massive connectivity.Multiple-antenna NOMA already combines power and spatial multiplexing.
  • Multi-domain multiplexing: PDMA uses sparse signature patterns whose resource-block occupancy can vary, increasing system capacity through overloading.Users may additionally be multiplexed by power or space, with MPA, SIC, or MPA-SIC used for detection.

2) BOMA:

BOMA multiplexes users through constellation building blocks, supporting simple implementation and increased multiuser capacity. Compared with other NOMA schemes, its simplicity trades off against required user pairing and reduced flexibility.

  • BOMA principle: The good-CSI user can detect the building block’s constellation points and decode its own bits.
  • Performance and complexity: BOMA significantly increases multiuser-system capacity while requiring no complex power allocation or SIC receiver.
  • BOMA principle: BOMA embeds a user with good CSI’s data within a coarse constellation assigned to a user with poor CSI.The building block is treated as interference by the poor-CSI user, while its small size relative to the coarse constellation spacing limits detection degradation.
  • Performance and complexity: BOMA has a simple, 4G-compatible structure requiring only minor software changes and supporting massive MIMO and high-frequency bands.
  • Comparison with other NOMA schemes: Power-domain NOMA is simple and compatible with MIMO and cooperative networks, but clustering and pairing increase system complexity.
  • Comparison with other NOMA schemes: BOMA requires user pairing, reducing flexibility, whereas LPMA multiplexes users across multilevel power and code domains without complex user clustering.
  • Comparison with other NOMA schemes: LPMA’s multilevel coding requires corresponding non-binary or nested binary channel coding.
  • Comparison with other NOMA schemes: SCMA is feasible for bad channel conditions or concentrated user locations because of shaping gain and near-optimal MPA detection.

E. Future Work

Future work centers on jointly designing modulation and multiple-access schemes, improving NOMA detection and clustering, and extending analysis to high-frequency bands. These directions address interference, noise, propagation, and hardware-impairment challenges while targeting enhanced 5G connectivity and spectral efficiency.

  • NOMA scalability: Further research is needed to improve NOMA schemes and support more users under limited resource blocks.The paper identifies this as a direction beyond existing work.
  • NOMA detection and clustering: User clustering strongly affects MPA-SIC detection, with asynchronous users performing better when similar delays share a cluster.Large delay variation within a cluster increases inter-user interference and may disrupt sparsity.
  • NOMA detection and clustering: Multi-branch processing can improve clustering performance by evaluating branches in parallel and selecting the best result instead of relying on one clustering approach.The passage describes each cluster as a branch and reports improved performance relative to single clustering.
  • Joint modulation and NOMA design: Joint modulation–NOMA design remains an open direction, particularly because f-OFDM can introduce ISI and ICI that degrade SCMA detection.RISIC-based interference cancellation would require SCMA multiuser detection within CP-reconstruction iterations.
  • High-frequency bands: Designing modulation and multiple access above 40 GHz is gaining interest because mmWave and THz bands may reduce spectrum scarcity.Poor propagation, noise, transmit-power constraints, CFO, and phase noise create substantial system-design challenges.
  • Paper-level conclusions: The survey concludes that new OMA modulations can reduce out-of-band leakage, while NOMA can improve throughput, massive connectivity, and spectral efficiency.Its coverage spans promising modulation and MA candidates for 5G networks.
Loading 1702.07673v1…