PaperlayerView exact revision on arXiv

Source-linked AI summary

Filtered-OFDM - Enabler for Flexible Waveform in The 5th Generation Cellular Networks

Xi Zhang, Ming Jia, Lei Chen, Jianglei Ma, Jing Qiu

arXiv:1508.07387v1cs.IT

TL;DR

The paper addresses the difficulty of using unified-numerology OFDM for 5G services with diverse requirements and channel characteristics. It presents filtered-OFDM, which splits bandwidth into filtered subbands with independently tailored waveforms and numerologies. In a four-service scenario, f-OFDM provides up to 46% throughput gains over conventional OFDM.

  • Problem

    Unified OFDM numerology cannot simultaneously suit diverse service requirements and associated channel characteristics, while 5G also requires lower OOBE and greater misalignment tolerance.

  • Method

    The paper develops f-OFDM by splitting assigned bandwidth into subbands, tailoring waveform numerology per subband, and applying subband-based filtering.

  • Results

    Up to 46% throughput gains over conventional OFDM are reported for four distinct services in a specific simulation scenario.

  • Takeaways & Limitations

    f-OFDM supports flexible waveform configurations across subbands while reducing OOBE, guard-band consumption, and synchronization requirements.

Abstract

from arXiv · show

The underlying waveform has always been a shaping factor for each generation of the cellular networks, such as orthogonal frequency division multiplexing (OFDM) for the 4th generation cellular networks (4G). To meet the diversified and pronounced expectations upon the upcoming 5G cellular networks, here we present an enabler for flexible waveform configuration, named as filtered-OFDM (f-OFDM). With the conventional OFDM, a unified numerology is applied across the bandwidth provided, balancing among the channel characteristics and the service requirements, and the spectrum efficiency is limited by the compromise we made. In contrast, with f-OFDM, the assigned bandwidth is split up into several subbands, and different types of services are accommodated in different subbands with the most suitable waveform and numerology, leading to an improved spectrum utilization. After outlining the general framework of f-OFDM, several important design aspects are also discussed, including filter design and guard tone arrangement. In addition, an extensive comparison among the existing 5G waveform candidates is also included to illustrate the advantages of f-OFDM. Our simulations indicate that, in a specific scenario with four distinct types of services, f-OFDM provides up to 46% of throughput gains over the conventional OFDM scheme.

I. INTRODUCTION

5G services and channels require more flexible waveform support than unified-numerology OFDM provides. The paper introduces f-OFDM, which splits bandwidth into filtered subbands to tailor waveforms and numerologies.

  • 5G waveforms should tailor services to channel needs, reduce out-of-band emission, and tolerate time-frequency misalignment.
  • Unified OFDM numerology makes it difficult to serve diverse requirements such as ultra-low-latency V2V and low-power IoT simultaneously.4G LTE numerology was selected mainly for mobile broadband, while different services may require different waveform choices.
  • f-OFDM splits the assigned bandwidth into subbands and filters them so independent OFDM systems or other waveforms can coexist.Subband filtering relaxes global synchronization requirements and supports inter-subband asynchronous transmission.
  • f-OFDM uses filters to reduce guard-band consumption and optimized numerology within each subband to match service needs.
  • The paper presents f-OFDM as promising among 5G waveform candidates in overall performance, complexity, cost, and evolution smoothness.

II. GENERAL FRAMEWORK

f-OFDM creates flexible waveform operation by dividing bandwidth into subbands, tailoring numerology within each, and filtering subband signals.

  • f-OFDM divides a wider assigned bandwidth into several subbands for independently tailored waveform configurations.Each subband can use conventional OFDM or potentially another waveform suited to its service and channel.
  • Each subband can select suitable subcarrier spacing, cyclic-prefix length, and transmission time interval for its service and channel characteristics.
  • Subband filtering suppresses inter-subband interference while intentionally breaking consecutive-symbol time-domain orthogonality to lower out-of-band emission.The paper states that this incurs negligible performance loss in other aspects.

A. Transceiver Structure

The f-OFDM transceiver contains independently configured waveforms in separate subbands, with guard tones accommodating filter transitions and asynchronous transmission.

  • Different OFDM systems can occupy different subbands with distinct subcarrier spacing, cyclic-prefix length, and transmission-time-interval duration.
  • Nonoverlapping subbands use a small number of guard tones to accommodate inter-subband interference and asynchronous transmission.The required guard-tone count depends on the filters’ transition regions.
  • f-OFDM filtering can also be combined with DFT-spread-OFDM for uplink transmission.
  • Compared with uniform 4G LTE allocation, f-OFDM enables more flexible time-frequency arrangements, including shortened TTIs and enlarged subcarrier spacing for V2V.

C. Evolution Path

The proposed evolution path moves from 4G OFDM toward f-OFDM incrementally, initially reusing existing guard-band resources and later allocating additional bandwidth to flexible 5G operation.

  • C. Evolution Path: The initial evolution stage reuses the 10% 4G guard band for f-OFDM data transmission without changing legacy 4G mobile devices.
  • C. Evolution Path: Over time, remaining and newly available spectrum can be allocated to 5G for more flexible and efficient spectrum utilization.
  • C. Evolution Path: The proposed path is intended to provide both backward and forward compatibility.

A. Filter Design and Implementation

f-OFDM filter design balances time-domain containment, frequency-domain transition sharpness, and implementation complexity. The paper considers soft-truncated sinc and equiripple filters, while noting that further improvement remains open.

  • A. Filter Design and Implementation: Filter design trades off time-domain energy spread, frequency-domain transition sharpness, and implementation complexity.Containing time-domain energy restrains inter-symbol interference, while sharp transitions help alleviate inter-subband interference.
  • A. Filter Design and Implementation: Two filter families are investigated: soft-truncated sinc filters and equiripple filters.The investigated filters use order 1024 and a 720-KHz passband.
  • A. Filter Design and Implementation: Soft-truncated sinc filters use Hann or root-raised-cosine windows to shorten sinc impulse responses and limit inter-symbol interference.Frequency taps below -30 dB may be excluded, potentially reducing fixed-point frequency-domain implementation complexity.
  • A. Filter Design and Implementation: Equiripple filters minimize maximum frequency-response error and provide sharper transition regions than soft-truncated filters.Extremely narrow transitions make their impulse responses longer, creating a time-domain tradeoff.
  • A. Filter Design and Implementation: Further improvement and generalization of soft-truncated sinc filters require additional investigation.

B. Guard Tone Arrangement

f-OFDM uses guard tones between subbands to accommodate inter-subband interference, with their quantity determined by filter transition regions. Proper filter design can minimize this overhead.

  • B. Guard Tone Arrangement: Guard tones are placed between subbands as multiples of the 15 KHz standard subcarrier spacing.The system is expected to support a finite set of subcarrier spacings, including integer or fractional copies of 15 KHz.
  • B. Guard Tone Arrangement: The required number of guard tones depends on the filters’ transition regions and can be minimized through proper filter design.Minimizing guard tones is intended to maximize spectrum utilization.

C. Treatment with Filter Tails

f-OFDM filtering creates forward and backward time-domain tails, so extending the cyclic prefix across all tails is generally inefficient. A special extended-CP receiver treatment is reserved for extremely narrow subbands.

  • C. Treatment with Filter Tails: f-OFDM filters create both forward and backward time-domain tails in addition to multipath delay spread.
  • C. Treatment with Filter Tails: Extending the cyclic prefix over the full filter tails would create high overhead and is not generally recommended.For medium-to-large subbands, the filter mainlobe is reasonably narrow.
  • C. Treatment with Filter Tails: For extremely narrow 180 KHz passbands, the transmitter can include the filter mainlobe in the cyclic prefix while the receiver shifts its window forward by half the mainlobe.This suppresses forward and backward filtering-induced inter-symbol interference.
  • C. Treatment with Filter Tails: Extended cyclic-prefix and receiver-window processing is needed only for extremely narrow subbands, which rarely occur in practical systems.

IV. COMPARISON AMONG 5G WAVEFORM CANDIDATES

The paper compares f-OFDM with GFDM, FBMC, UFMC, and OFDM as 5G waveform candidates pursuing reduced out-of-band emission or relaxed synchronization. It emphasizes f-OFDM’s quasi-orthogonal subbands, short filters, and compatibility with multi-antenna transmission.

  • IV. COMPARISON AMONG 5G WAVEFORM CANDIDATES: The compared waveform candidates share motivations including reduced out-of-band emission and relaxed synchronization, but apply filtering differently.
  • IV. COMPARISON AMONG 5G WAVEFORM CANDIDATES: GFDM uses non-orthogonal, closely spaced subcarriers and therefore requires high-order filtering, tail biting, and interference-cancellation processing.f-OFDM retains quasi-orthogonal subcarriers within each subband and uses comparatively short filters without complicated pre- or post-processing.
  • IV. COMPARISON AMONG 5G WAVEFORM CANDIDATES: OFDM, GFDM, UFMC, f-OFDM, and FBMC are compared through typical power spectral densities under the same overhead consumption.
  • IV. COMPARISON AMONG 5G WAVEFORM CANDIDATES: FBMC typically uses filters longer than three symbol durations, making them resource-consuming compared with f-OFDM filters.The paper also identifies difficulty combining FBMC with multi-antenna transmission, whereas f-OFDM requires no special processing for that combination.

V. SIMULATION RESULTS

The simulations verify f-OFDM’s out-of-band emission and block error rate under LTE-based configurations, then preliminarily examine throughput performance.

  • The simulation program evaluates f-OFDM’s out-of-band emission, block error rate, and throughput along the proposed evolution path.The first part follows 4G LTE for detailed OOBE and BLER verification; the second preliminarily investigates throughput.
  • The 20-MHz LTE bandwidth is divided into three subbands, with 1-MHz guard-band subbands surrounding an 18-MHz loaded subband.The setup loads the outer subbands with 4 resource blocks each and the central subband with 100 resource blocks.
  • The simulation setup is documented as Table II, with PSD comparisons including power-amplifier distortions.
  • Uniform OFDM numerology is applied across the three subbands, while different guard-tone counts and a half-symbol delay model inter-subband interference.The setup uses uniform subcarrier spacing, normal CP, no special filter-tail treatment, and asynchronous transmission conditions.
  • The reported BLER comparisons assess f-OFDM in the first subband against single-OFDM operation without interference under equal and unequal subband transmit powers.The unequal-power case raises the second subband’s transmit power by 10 dB.

B. Throughput

The throughput study compares f-OFDM and OFDM across four service types using service- and channel-adapted numerology. In the stated toy scenario, f-OFDM achieves up to 46% total throughput gain.

  • Up to 46% total throughput gain is achieved by f-OFDM over OFDM in the stated four-service toy scenario.The comparison uses pedestrian, urban, highway, and vehicle-to-vehicle services.
  • The OFDM baseline uses worst-case numerology, extended CP, and a 10% guard band across the supported services.
  • f-OFDM minimizes guard tones and evenly divides bandwidth into four subbands, optimizing numerology for each channel and service requirement.The adaptations include CP length and subcarrier spacing choices suited to propagation and frequency-selectivity conditions.
  • Throughput gains arise from both guard-tone savings and adaptations such as reduced CP length and reduced subcarrier spacing.

VI. CONCLUSIONS AND FUTURE WORK

The paper presents filtered-OFDM as an enabler for flexible 5G waveforms and compares it with other 5G waveform candidates. Simulations are encouraging, while prototyping and field testing are in progress.

  • Filtered-OFDM is presented as an enabler for flexible waveform configuration designed to meet 5G cellular-network expectations.
  • The paper outlines f-OFDM’s framework and methodology and compares it with existing 5G waveform candidates to illustrate its advantages.
  • The authors state that encouraging simulation results make f-OFDM appear to be the most promising 5G waveform candidate.This assessment is explicitly qualified as the authors’ view.
  • Prototyping and field testing of f-OFDM are in progress.
Loading 1508.07387v1…