PaperlayerView exact revision on arXiv

Source-linked AI summary

Architectures and Key Technical Challenges for 5G Systems Incorporating Satellites

A. Guidotti, A. Vanelli-Coralli, M. Conti, S. Andrenacci, S. Chatzinotas, N. Maturo, B. Evans, A. Awoseyila, A. Ugolini, T. Foggi, L. Gaudio, N. Alagha, S. Cioni

arXiv:1806.02088v1cs.NI

TL;DR

Satellite channel impairments challenge NR-based NTN integration for eMBB and NB-IoT services. The paper assesses architecture choices, waveforms, and PHY/MAC procedures, finding that RA, TA, and HARQ require particular attention and that solutions depend on service and architecture.

  • Problem

    Large satellite path losses, delays, and Doppler shifts complicate the realization of satellite-based NR networks for eMBB and NB-IoT.

  • Method

    The paper assesses satellite impairments across NR waveforms and PHY/MAC procedures using 3GPP NTN architecture options and eMBB and NB-IoT scenarios.

  • Results

    The main technical challenges concern Random Access, Timing Advance, and HARQ, with solutions varying by service and architecture.

  • Takeaways & Limitations

    Satellite-terrestrial and stand-alone NTN architectures can extend 5G service coverage, but NR procedures must account for satellite propagation and Doppler conditions.

Abstract

from arXiv · show

Satellite Communication systems are a promising solution to extend and complement terrestrial networks in unserved or under-served areas. This aspect is reflected by recent commercial and standardisation endeavours. In particular, 3GPP recently initiated a Study Item for New Radio-based, i.e., 5G, Non-Terrestrial Networks aimed at deploying satellite systems either as a stand-alone solution or as an integration to terrestrial networks in mobile broadband and machine-type communication scenarios. However, typical satellite channel impairments, as large path losses, delays, and Doppler shifts, pose severe challenges to the realisation of a satellite-based NR network. In this paper, based on the architecture options currently being discussed in the standardisation fora, we discuss and assess the impact of the satellite channel characteristics on the physical and Medium Access Control layers, both in terms of transmitted waveforms and procedures for enhanced Mobile BroadBand (eMBB) and NarrowBand-Internet of Things (NB-IoT) applications. The proposed analysis shows that the main technical challenges are related to the PHY/MAC procedures, in particular Random Access (RA), Timing Advance (TA), and Hybrid Automatic Repeat reQuest (HARQ) and, depending on the considered service and architecture, different solutions are proposed.

1 INTRODUCTION

5G requirements motivate integrating satellite systems into terrestrial networks to extend coverage and support diverse services. The paper frames this integration around emerging 3GPP NTN standardisation and satellite-specific challenges.

  • Motivation: 5G targets higher throughput, global connectivity, high reliability, and dense device access for diverse service and user communities.The stated peak data-rate targets are 20 Gbps downlink and 10 Gbps uplink, with 99.999% successful packet reception and 1 million-device connection density.
  • NTN Role: NTN can support 5G services in unserved and underserved areas, including remote locations, aircraft, vessels, and rural regions.The paper also associates NTN with service continuity for mission-critical communications, MTC, and IoT devices.
  • Satellite Integration: GEO satellite integration offers global large-capacity coverage but introduces large delays that challenge 5G operation.Prior work examined multicast resource allocation and TCP performance in LTE-based GEO systems.
  • Paper Scope: The paper analyses satellite impacts on NR PHY and MAC layers across eMBB and NB-IoT scenarios, following architecture options discussed in 3GPP NTN activities.The analysis covers waveform design, channel impairments, and PHY/MAC procedures.

2 SYSTEM ARCHITECTURE

The paper considers direct satellite access and relay-node access as NTN architecture options. Transparent and regenerative payload choices trade deployment complexity against propagation delay and the extent of required NR adaptations.

  • Architecture Options: NTN deployments may use stand-alone satellite systems or integrated satellite-terrestrial solutions based on different architecture options.The paper categorises the options by how user access and satellite connectivity are provided.
  • Direct Access: In direct-access architectures A1-A2, satellites connect directly to ground UEs through the NR air interface.This requires assessing satellite delays and Doppler shifts against NR waveform and PHY/MAC procedures.
  • Payload Choices: With a transparent payload, the gNB is at the gateway, whereas a regenerative payload places the gNB on the satellite.The regenerative option reduces propagation delays and can ease NR PHY/MAC modifications, but it is more complex and costly.
  • Relay-Node Access: Architectures A3-A4 use on-ground Relay Nodes for user access and satellites for backhaul connectivity.Relay Nodes connect to a Donor gNB, can terminate procedures up to Layer 3, and use NR on user-access and modified NR on backhaul links.
  • Selected Architectures: The analysis focuses on architectures A1 and A3 for eMBB and NB-IoT scenarios, respectively, using FDD framing.These represent direct access and access through Relay Nodes with transparent payload satellites.

3 SCENARIOS

The eMBB scenario uses relay-based transparent-satellite backhaul, while the NB-IoT scenario uses direct access and excludes GEO because of delay and path-loss constraints.

  • eMBB Scenario: The eMBB scenario provides broadband connectivity through architecture A3, with transparent satellite backhaul to on-ground NR cells formed by Relay Nodes.The RN-UE and RN-gNB links use NR, and the RN operates as a normal gNB from the UEs’ perspective.
  • NB-IoT Scenario: The NB-IoT scenario addresses massive machine-type communications through direct access with a transparent satellite payload in architecture A1.IoT terminals’ stringent battery-life requirements make propagation delay especially important.
  • NB-IoT Scenario: GEO is considered infeasible for the NB-IoT scenario because of its extremely large delays and path-loss values.The paper therefore focuses the subsequent NB-IoT analysis on non-GEO satellite conditions.

4 SATELLITE CHANNEL IMPAIRMENTS

The paper models propagation delay and Doppler shift as key satellite impairments affecting NR PHY/MAC feasibility. Their effects depend on link geometry, orbital conditions, and procedure timing.

  • Impairments: Satellite channel assessment focuses on large Doppler shifts and propagation delays that can affect NR PHY and MAC layers.These impairments are evaluated for the previously defined eMBB and NB-IoT scenarios.
  • Delay: Propagation analysis distinguishes one-way delay from RTT because procedures may terminate at the gNB or require interaction with the NGC.RTT is approximated as twice the propagation delay when satellite signal-processing time is negligible.
  • Delay: Worst-case path distances and associated delays are computed using non-aligned transmitter-receiver elevation angles and scenario-specific satellite altitudes.The analysis assumes 5° gateway elevation and 10° UE or RN elevation for the listed distances and delays.
  • Doppler Shift: Doppler shift depends on relative motion, carrier frequency, and satellite geometry, producing substantially larger shifts at Ka-band than below 6 GHz.At 4 GHz and 500 km/h, the paper gives a maximum Doppler shift of 1.9 kHz; it then considers 20 GHz Ka-band conditions.
  • Doppler Shift: The Doppler model uses satellite orbital velocity and elevation angle to represent frequency shift over the satellite link.The cited formulation incorporates orbital-motion and Earth-geometry parameters.

5 TECHNICAL CHALLENGES ANALYSIS

The section assesses NR physical-layer waveforms and timing procedures under satellite-specific impairments, focusing on eMBB over GEO backhaul links. It identifies waveform nonlinear-distortion sensitivity and timing-advance constraints as key design concerns.

  • PHY: NR eMBB uses CP-OFDM as its baseline waveform, with DFT-s-OFDM additionally supported for single-stream, link-budget-limited uplink cases.NR provides flexible subcarrier spacing, FFT size, subframe duration, CP length, and windowing.
  • Candidate Waveforms: Candidate alternatives include f-OFDM, W-OFDM, FBMC, UFMC, and GFDM, trading lower OOBE and timing-offset tolerance against higher PAPR or complexity.f-OFDM is identified as a promising candidate for cellular NR networks.
  • Candidate Waveforms: In a GEO satellite nonlinear channel, F-OFDM is highly sensitive to nonlinear distortion because its filtering increases PAPR.The analysis uses an OFDM design with FFT size N = 1024 and 600 used subcarriers.
  • Candidate Waveforms: At IBO = 20 dB, F-OFDM degrades to legacy OFDM performance in Figures 3c–3d.This result motivates future investigation of nonlinear compensation techniques.
  • Timing Advance: NR timing advance must accommodate differential delays between fixed relay nodes and remain valid longer than the satellite round-trip time.Otherwise, uplink and downlink frame synchronization can fail and produce disruptive interference.
  • Timing Advance: With maximum subcarrier spacing, the maximum timing advance is TT A = 0.0209 ms, corresponding to a maximum distance of 3.135 km.The timing-advance range depends on subcarrier spacing and determines the supported differential distance between relay nodes.

Random Access

NR Random Access retains its terrestrial algorithmic structure, but satellite delay affects contention-based procedures involving the core network. Contention-free access through the relay node requires no satellite-channel modification in the considered eMBB architecture.

  • Procedure: NR Random Access can be contention-based or contention-free, and both procedures begin with a UE-transmitted random-access preamble on gNB-indicated resources.Contention-free access comprises only the first two steps of the contention-based procedure.
  • Contention-Free RA: Contention-free RA in each on-ground cell terminates at the corresponding relay node and therefore does not require modification for the satellite channel.The relay node terminates protocols through Layer 3 in this scenario.
  • Contention-Based RA: Contention-based RA requires relay-node communication with the core network through the satellite, making propagation delay relevant in steps 3 and 4.The procedure is otherwise algorithmically inherited from LTE and NR.
  • Contention-Based RA: The contention-resolution timer can be set up to 15 ms, significantly below GEO round-trip time, so it may need extension for GEO systems.The Random Access Response window and contention timer also matter during relay-node startup.

HARQ

NR HARQ timing is tightly specified for acknowledgements, creating substantial scaling demands when satellite propagation delay is much larger than in terrestrial systems. The considered GEO eMBB case therefore requires more parallel processes or larger buffers and control fields.

  • HARQ Operation: NR HARQ uses multiple parallel Stop-and-Wait processes with asynchronous retransmissions and adaptive transmission attributes.Incremental Redundancy and Chase Combining can combine retransmitted transport blocks.
  • HARQ Timing: HARQ acknowledgement timing is tightly specified, with a default four-subframe delay from transport-block reception to ACK/NACK transmission.Higher layers may indicate an offset k that accounts for propagation delay.
  • HARQ Timing: The HARQ processing-time budget includes ACK/NACK timing and the delay from uplink-grant reception to uplink-data transmission.The paper expresses the required process count as NHARQ = THARQ/TTI.
  • Satellite Impact: The large process count increases UE soft-buffer requirements and requires DCI fields of at least 10 bits to identify HARQ processes.The paper discusses larger buffers or alternative ACK signalling as possible responses.

5.2 NarrowBand-IoT

The NB-IoT analysis examines deployment, waveform, timing, and access constraints for satellite integration, focusing on non-GEO systems because GEO delay and path loss are unsuitable for the assumed IoT requirements. Key concerns include latency-sensitive timers, Doppler and phase impairments, battery life, and link budget.

  • Requirements: NB-IoT targets ultra-low-complexity devices, improved coverage, massive low-throughput connectivity, long battery life, and low exception-report latency.Reported targets include 164 dB MCL, at least 52547 devices per sector, and 10-year battery life.
  • Applications: NB-IoT supports smart metering, smart cities and buildings, environmental and agricultural monitoring, and animal or people tracking.These applications motivate low-throughput, wide-coverage device support.
  • Satellite Scenario: The satellite NB-IoT study focuses on LEO systems because latency and link-budget constraints make GEO unsuitable for the assumed IoT scenario.The analysis treats link closure as an assumption, for example through suitable satellite antenna gains.
  • Latency: Existing LTE-derived RA and RRC timers, including the RAR and contention-resolution windows, may be incompatible with satellite RTT delays.The RAR time-window size and contention-resolution window size are identified as specific timing constraints.
  • Doppler and Phase: Narrow and closely spaced NB-IoT subcarriers make Doppler and phase impairments, especially for LEO or MEO satellites, potential causes of unsuccessful transmission.Differential Doppler and CFO among users are highlighted for Scenario A.
  • Battery and Link Budget: Longer satellite RTT can lengthen device wake-up periods, while higher transmit power may be needed to close the link, threatening the target battery lifetime.Power constraints must be considered on both forward and return feeder and user links.
  • Deployment: NB-IoT can operate in-band, guard-band, or stand-alone, with stand-alone deployment allowing all base-station transmit power to serve NB-IoT.In-band deployment uses one 180 kHz LTE resource block.

Frequency error constraints for NB-IoT

NB-IoT satellite links must handle carrier-frequency errors and Doppler within tight synchronization and subcarrier-spacing constraints. Downlink differential Doppler is largely manageable at each UE, whereas uplink differential Doppler requires mitigation to preserve SC-FDMA structure.

  • CFO and subcarrier constraints: NB-IoT synchronization requires each UE to search an in-band NB-PRB carrier with CFO up to ±7.5 kHz.This requirement applies during synchronization procedures.
  • CFO and subcarrier constraints: The minimum NB-IoT subcarrier spacing is 3.75 kHz, which constrains tolerable Doppler.
  • Doppler decomposition: Doppler for user i is decomposed into a common component shared within a footprint and a differential component determined by relative user positions.The decomposition is fdi = fdcommon + ∆fdi.
  • Link-specific impact: On the downlink, each UE compensates its own Doppler and CFO, so differential Doppler has negligible impact on individual subcarriers across the 180 kHz bandwidth.
  • Link-specific impact: On the uplink, differential Doppler among independently transmitting SC-FDMA users must be mitigated when it exceeds the 950 Hz LTE Doppler constraint to prevent subcarrier information from overlapping.

Analysis on frequency errors for NB-IoT waveforms

The analysis models CFO and Doppler separately in a LEO satellite scenario, then examines how user-specific frequency differences affect the uplink waveform. It concludes that common Doppler can be compensated centrally, while differential Doppler requires user-side pre-compensation.

  • Scenario and assumptions: The study considers a transparent-payload LEO satellite with variable zenith heights from 600 to 1500 km and a 90° elevation case identified as worst for differential Doppler.
  • Scenario and assumptions: Satellite orbital speed is computed from Earth and orbit parameters, using the gravitational constant and Earth mass defined in the analysis.The velocity formulation uses satellite height and Earth radius together with G and M_E.
  • Scenario and assumptions: Because propagation speed greatly exceeds source and receiver velocities, Doppler is assumed symmetric in the uplink and downlink.
  • CFO analysis: CFO and Doppler are investigated separately, with the user index k denoting the k-th user and downlink CFO compensation based on a frequency-advance method.
  • CFO analysis: Differential frequency offsets among nUEs degrade the received SC-FDMA signal and must be compensated at each nUE through frequency advance based on the downlink broadcast signal.
  • Doppler analysis: Users at different footprint positions experience different Doppler values simultaneously, creating differential Doppler and correlation among uplink users.
  • Doppler analysis: Doppler curves are presented for reference UEs in the NB-IoT scenario, including a case with satellite height hSAT = 1500 km.
  • Doppler analysis: The uplink received signal is modeled as the superposition of user signals, with each Doppler term split into a reference-user component and a differential component.The common reference component can be compensated at the eNB, while each nUE must pre-compensate its differential Doppler.

Differential Doppler assessment

The analysis examines Doppler, Timing Advance, Random Access, and HARQ procedures for satellite NB-IoT, identifying constraints and solutions for differential impairments and propagation delay.

  • Differential Doppler: Higher satellite orbits reduce differential Doppler, while the common Doppler can be compensated using a GNSS receiver.The maximum differential Doppler should remain below the LTE constraint of 950 Hz.
  • Differential Doppler: Position tracking can reduce Doppler sufficiently to facilitate NB-IoT over LEO, but GNSS operation can consume UE battery life.Frequency advance can reduce forward-link frequency offsets on the reverse link, although adaptations may be needed for differential Doppler.
  • Timing Advance: NB-IoT Timing Advance compensates uplink misalignment up to a maximum of 0.67 ms, matching legacy LTE support.The allowed TA must cover the maximum differential travel time among UEs; a propagation-delay offset is an alternative.
  • Random Access: Satellite NB-IoT Random Access can retain the LTE/NR message flow because extended response windows and contention timers accommodate channel delay.The contention resolution timer can be extended up to 10.24 s without modifying the standard procedure.
  • HARQ: NB-IoT uses one HARQ process by default, with up to two parallel processes from Release 14 when supported.Retransmissions within a bundle are non-adaptive and the whole bundle is retransmitted after a NACK.

6 CONCLUSIONS

The paper assesses satellite impairments across NR waveforms and PHY/MAC procedures for eMBB and machine-type services. It finds that HARQ is strongly affected by large delays, whereas architecture-dependent solutions mitigate other challenges.

  • 6 CONCLUSIONS: The study evaluates satellite path loss, delay, and Doppler impacts on NR waveforms and PHY/MAC procedures for eMBB and machine-type services.It considers architecture and deployment options discussed in 3GPP NTN standardisation.
  • 6 CONCLUSIONS: The f-OFDM waveform offers bandwidth benefits but shows greater sensitivity to TWTA non-linearities and larger PAPR than CP-OFDM.The comparison is conducted in terms of out-of-band emissions under TWTA non-linearities.
  • 6 CONCLUSIONS: In the eMBB architecture using on-ground Relay Nodes, Random Access and Timing Advance pose no peculiar issues, while HARQ is deeply impacted by large delays.The delay may require many more parallel HARQ processes, affecting UE soft-buffer sizes.
Loading 1806.02088v1…