PaperlayerView exact revision on arXiv

Source-linked AI summary

Millimeter Wave Communications for Future Mobile Networks

Ming Xiao, Shahid Mumtaz, Yongming Huang, Linglong Dai, Yonghui Li, Michail Matthaiou, George K. Karagiannidis, Emil Björnson, Kai Yang, Chih Lin, Amitava Ghosh

arXiv:1705.06072v1cs.IT

TL;DR

MmWave communications could provide multiple-Gbps user rates through abundant bandwidth, but mobile deployment must handle mobility, complex channels, and coordination challenges. This paper surveys recent work across channel modeling, MIMO, access, backhaul, coverage, standardization, and deployment, concluding that substantial technical challenges remain while identifying directions for future research.

  • Problem

    The central problem is exploiting mmWave’s multiple-Gbps potential in mobile networks despite moving nodes, complicated channels, and difficult coordination.

  • Method

    The paper presents a comprehensive survey of channel measurements and models, MIMO transceivers, multiple access, backhauling, coverage, connectivity, standardization, and deployment.

  • Results

    The survey summarizes recent technical progress across the major mmWave mobile-network topics and identifies directions for future research.

  • Takeaways & Limitations

    Future work must address high-mobility transmission, enhanced transmission distance, hardware impairments, and other challenges before mmWave becomes mainstream in mobile networks.

Abstract

from arXiv · show

Millimeter wave (mmWave) communications have recently attracted large research interest, since the huge available bandwidth can potentially lead to rates of multiple Gbps (gigabit per second) per user. Though mmWave can be readily used in stationary scenarios such as indoor hotspots or backhaul, it is challenging to use mmWave in mobile networks, where the transmitting/receiving nodes may be moving, channels may have a complicated structure, and the coordination among multiple nodes is difficult. To fully exploit the high potential rates of mmWave in mobile networks, lots of technical problems must be addressed. This paper presents a comprehensive survey of mmWave communications for future mobile networks (5G and beyond). We first summarize the recent channel measurement campaigns and modeling results. Then, we discuss in detail recent progresses in multiple input multiple output (MIMO) transceiver design for mmWave communications. After that, we provide an overview of the solution for multiple access and backhauling, followed by analysis of coverage and connectivity. Finally, the progresses in the standardization and deployment of mmWave for mobile networks are discussed.

I. INTRODUCTION AND BACKGROUND

Future mobile networks face rapidly growing traffic and ambitious 5G performance targets, motivating mmWave as a key high-bandwidth technology. This paper addresses the fragmented literature with a comprehensive survey spanning propagation, transceiver design, access, networking, and deployment.

  • Motivation: Future mobile networks must support growing traffic, hundreds of bit/s/Hz/km2 area spectral efficiency, and multiple-Gbps throughput per device.Smartphone traffic was predicted to reach about 50 petabytes monthly in 2021, roughly 12 times the 2016 level.
  • 5G Requirements: IMT-2020 targets peak rates above 10 Gbit/s, 1 ms over-the-air latency, 500 km/h mobility, and 10^6 connections/km2, among other KPIs.These improvements are measured relative to IMT-Advanced.
  • MmWave Motivation: MmWave is considered important for achieving 10 Gbit/s peak rates because its bands offer large amounts of bandwidth.The paper relates this capacity potential to bandwidth expansion and the channel-capacity relationship.
  • Research Scope: Using mmWave in mobile networks remains challenging because moving nodes, complicated channels, and multi-node coordination complicate operation.The paper contrasts this mobile setting with more established stationary or local-area uses.
  • Survey Contribution: The survey integrates recent work on channel measurement and modeling, MIMO, multiple access, backhauling, coverage, standardization, and deployment.Its organization follows these technical areas across the paper’s main sections.

II. KEY CHALLENGES AND TECHNICAL POTENTIALS

MmWave offers strong directional and bandwidth advantages, but mobile deployment is constrained by propagation, blockage, beam alignment, hardware, and power-related challenges. The section outlines the physical causes and engineering implications of these constraints.

  • A. Main Technical Challenges: MmWave mobile networks face severe pathloss, penetration loss, high power consumption, blockage, and hardware impairments.These challenges constrain the use of mmWave despite its potential for extremely high data rates.
  • A. Main Technical Challenges: In free space, received power follows the Friis relation, while measured non-free-space pathloss is commonly represented with an exponent n ranging from 2 to 6.The model uses transmit power, antenna gains, wavelength, and distance to describe received power.
  • A. Main Technical Challenges: MmWave pathloss is higher than at sub-6 GHz, but directive antennas can support urban links spanning hundreds of meters, kilometers, or demonstrated 10 km ranges.The cited ranges depend on propagation conditions and antenna directivity.
  • A. Main Technical Challenges: At 28 GHz, brick and tinted-glass penetration losses are about 28 dB and 40 dB, respectively, exceeding losses for clear glass and dry walls.The passage specifically discusses indoor non-line-of-sight conditions.
  • A. Main Technical Challenges: MIMO and beamforming provide array gain and spatial multiplexing, but fully digital mmWave arrays require many costly, power-consuming RF chains.Short wavelengths allow many antennas in compact areas, while high bandwidth makes PAs and converters expensive and power hungry.
  • A. Main Technical Challenges: Narrow beams improve directional gain but increase sensitivity to transmitter–receiver misalignment, especially under mobility.Misalignment can arise from antenna and beamforming imperfections or changing propagation geometry.
  • A. Main Technical Challenges: Hardware impairments including phase noise, nonlinear PAs, I/Q imbalance, and limited ADC resolution reduce channel capacity at high spectral efficiency.The modified Rapp model describes PA input–output nonlinearity using gain, limiting amplitude, and transition smoothness.

B. Technical Potentials

MmWave’s main technical potential is access to exceptionally wide contiguous spectrum and compact large antenna arrays. These properties can support high rates and directional transmission, although wider bandwidth does not always improve rates in noise-limited conditions.

  • B. Technical Potentials: MmWave bands can offer more than 150 GHz of potentially available bandwidth, excluding unfavorable absorption bands.At 1 b/s/Hz spectral efficiency, 150 GHz could deliver 150 Gbps.
  • B. Technical Potentials: Large bandwidth can provide high rates with relatively low spectral efficiency, simplifying implementation and reducing sensitivity to hardware impairments.The passage cautions that wider bandwidth does not always produce higher rates in the noise-limited region.
  • B. Technical Potentials: Short mmWave wavelengths allow many antennas to fit into compact arrays, enabling large-scale antenna communications.Many antenna elements also produce narrow beams.
  • B. Technical Potentials: Narrow beams can improve security against eavesdropping and jamming and increase resilience to co-user interference.These are presented as positive consequences of mmWave directionality.

III. CHANNEL MEASUREMENTS AND MODELING

This section reviews mmWave channel measurement campaigns and modeling, emphasizing measurements across candidate mobile bands and the parameters needed to characterize propagation and evaluate systems.

  • A. Millimeter Wave Measurement Campaigns: Channel measurements characterize pathloss, delay spread, shadowing, and angular spread across different environments.Pathloss and shadowing are reported for line-of-sight and non-line-of-sight cases, while delay spread describes multipath power dispersion.
  • A. Millimeter Wave Measurement Campaigns: Measurements have covered mmWave mobile communication bands including 10, 28, 38, 60, and 82 GHz.NYU WIRELESS measured at 28, 38, 60, 72, and 73 GHz; 3GPP incorporated measurement and ray-tracing results into newer modeling features.
  • B. Channel Modeling: Channel models support system-level simulation and have evolved from general mobile-network models toward mmWave-specific propagation representations.The surveyed models span historical mobile-network models and recent mmWave modeling results.
  • A. Millimeter Wave Measurement Campaigns: The survey summarizes measurement campaigns and results, but space limits prevent listing all existing campaigns and largely omit results before 2012.The limitation applies especially to older measurement literature.

2) WINNER I/II/+ Model:

The WINNER I/II/+ projects developed increasingly capable channel models for mobile networks, expanding scenario coverage, polarization modeling, frequency range, and three-dimensional propagation.

  • 1) WINNER I/II/+ Model: COST 273 models geometrically located multipath clusters to preserve spatial consistency and evaluate MIMO beamforming and multi-cell transmission more accurately.The model represents relationships between angles of arrival and departure in a two-dimensional propagation environment.
  • 1) WINNER I/II/+ Model: QuaDRiGa extends the open-source 3GPP-3D implementation with spatial consistency and multi-cell transmission features.These extensions exploit approaches from SCM-E and COST 273.

8) IMT-Advanced Model [110]:

This section describes channel-model extensions and mmWave MIMO architectures, focusing on broader propagation support and hybrid designs that balance digital flexibility against analog simplicity.

  • 9) METIS Model [111]: The described channel-model extensions combine deterministic map-based and stochastic components, with mmWave parameters and measurements available for 50–70 GHz bands.The stochastic model supports 3D shadowing maps, power-angular-spectrum sampling, and frequency-dependent pathloss.
  • 9) METIS Model [111]: The statistical model targets 6–100 GHz link- and system-level simulations across many scenarios.It adds improved spatial accuracy, spherical-wave subpaths, frequency consistency, time variation, blockage, clustering, scattering, and ground or floor reflection.
  • A. MIMO Architectures: Fully digital mmWave MIMO requires many energy-intensive RF chains, creating high hardware cost and energy consumption with large antenna arrays and bandwidth.Alternative architectures are therefore needed for emerging mobile networks.
  • A. MIMO Architectures: Fully analog architecture uses one RF chain and phase-based analog processing, reducing hardware cost and energy consumption but limiting adaptation and single-stream transmission.Its restricted signal control can cause performance loss, particularly for mobile users, and prevents multiplexing gain from multiple streams.
  • A. MIMO Architectures: Hybrid analog-digital architecture divides large-scale analog processing from dimension-reduced digital processing using fewer RF chains.Its suitability follows from the typically low-rank mmWave channel and smaller optimal number of data streams.
  • A. MIMO Architectures: Hybrid analog circuits use fully connected phase-shifter networks, sub-connected phase-shifter networks, or lens antenna arrays, each imposing different hardware constraints.Sub-connected networks reduce array gain and directivity relative to fully connected networks, while lens arrays combine signal emission and phase-shifting.
  • A. MIMO Architectures: Finite-resolution phase shifters introduce phase noise and can degrade hybrid precoding and combining performance.The survey identifies signal-processing design under finite phase-shifter resolution as a future research topic.

B. Channel Estimation with Hybrid Architecture

Hybrid mmWave architectures make channel estimation difficult because few RF chains prevent simultaneous sampling across all antennas and mobile channels require timely CSI. Proposed solutions reduce estimation dimension through beam training or exploit channel sparsity for low-overhead recovery.

  • Channel estimation challenges: Hybrid architectures cannot simultaneously sample all receive antennas, making conventional channel estimation impose unaffordable pilot overhead.The number of RF chains is much smaller than the antenna count.
  • Beam-training-based estimation: Dimension-reduced estimation first obtains analog precoders and combiners through beam training, then estimates the smaller effective channel matrix.The effective matrix has dimension N_D × N_D, with N_D much smaller than N_T and N_R.
  • Beam-training-based estimation: IEEE 802.11ad-style training begins with wide beam pairs and uses received-SNR feedback to identify a strong beam pair.Multi-resolution or multi-beam extensions can reduce the search burden, but local-search methods may lack robustness.
  • Sparsity-based estimation: The sparse-channel formulation uses pilot precoders and combiners to construct measurements from which the channel representation is recovered.With multiple RF chains, orthogonal pilots can simplify receiver-side estimation.
  • Sparsity-based estimation: Sparse recovery directly estimates the complete channel by representing vec(H) with a sparse path-gain vector over quantized AoA/AoD dictionaries.Adaptive angle searches, low-coherence sensing designs, OMP, LASSO, and structural sparsity can improve accuracy or reduce pilot overhead.

C. Channel Tracking

Mobile mmWave channels vary rapidly because of mobility-induced Doppler effects and short coherence times, while hybrid hardware limits estimation speed. Channel tracking therefore uses retained candidate beams or motion-informed sparse support prediction, with real-environment validation still required.

  • Tracking challenge: Mobility causes large Doppler effects and short coherence times, so mmWave channels can change quickly even with short wideband symbol durations.Hybrid implementations also lack sufficient time to perform frequent full channel estimation.
  • Beam-based tracking: Candidate-beam tracking retains several high-SNR beam pairs and tests them as the channel changes, rerunning complete training only when all candidates fail.This low-complexity approach is used in IEEE 802.11ad WLAN but is mainly suited to single-stream transmission.
  • Motion-aided tracking: Motion-aided tracking predicts future LoS channel support from temporal AoA/AoD variation and previous channel estimates, reducing pilot overhead.The approach performs well for LoS paths but is difficult to apply to NLoS paths caused by complicated scattering.
  • Validation: Beam-tracking efficiency must be tested in real environments to determine supported mobility speeds and reliable channel characteristics.The paper identifies this as necessary for realizing mmWave mobile networks.

D. Hybrid Precoding and Combining

Hybrid precoding and combining must optimize achievable rate under architecture-specific analog hardware constraints rather than the unconstrained structure of fully digital MIMO. The resulting optimization is non-trivial because feasible precoders and combiners depend on the selected hybrid network.

  • Optimization formulation: Hybrid precoding and combining differ from fully digital MIMO because analog and digital components must satisfy special hardware constraints.The hybrid precoder and combiner are composed of analog and digital stages.
  • Optimization formulation: The design objective is to maximize achievable rate over sets of feasible analog precoders and combiners, accounting for noise and interference covariance.The feasible sets differ across the N1–N3 architectures.

n WHHFFHHHW ,

The survey reviews architecture-specific hybrid precoding and combining methods, including decomposed optimization, beam selection, and interference-aware designs. Multi-user extensions are possible for some architectures, while the N2 multi-user problem remains open.

  • N1 architecture: For N1, spatially sparse OMP-based precoding decomposes the problem and can achieve near-optimal performance by exploiting mmWave channel sparsity.Related methods approximate subproblems as convex optimizations.
  • N2 architecture: For N2, proposed methods optimize rate, spectral efficiency, or energy efficiency by decomposing the subarray problem and alternating among subproblems.The sub-connected structure motivates this decomposition.
  • N3 architecture: For N3, analog beam selection is central because digital processing can be obtained after selecting DFT beams.Magnitude-maximization and interference-aware selection are representative approaches.
  • Multi-user extensions: N3 schemes can extend to multi-user multi-stream transmission, while N1 can use two-stage precoding to combine desired-signal maximization with interference cancellation.The N2 multi-user hybrid precoding problem remains open.

E. Low-Resolution ADC Based Architecture

Low-resolution ADC architectures reduce mmWave MIMO hardware complexity, energy consumption, and cost, but quantization limits high-SNR capacity and leaves broadband signal processing unresolved.

  • Architecture: Low-resolution ADCs replace high-resolution converters to reduce RF complexity in mmWave MIMO systems.The architecture reduces converter requirements rather than primarily reducing the number of RF chains.
  • Benefits: The architecture reduces energy consumption and hardware cost while simplifying circuit modules such as automatic gain control.Its ADCs require fewer comparator components.
  • Performance and challenges: Capacity-approaching performance is achievable in low and medium SNR regions, but severe quantization limits capacity at high SNR.Quantization nonlinearity also creates new signal-processing challenges, illustrated by 1-bit sampling’s discrete outputs.
  • Open issues: Open issues include designing signal processing for broadband mmWave MIMO channels.The paper identifies this as requiring future investigation.
  • Related developments: Subsequent work investigates mmWave MIMO signal processing, including antenna selection, structured-sparsity channel estimation, and training-sequence design.These examples are presented among schemes studied in the cited special issue.

A. Multiple-access Technologies

mmWave multiple access combines spatial, power-domain, and random-access approaches, while directional propagation offers multiplexing opportunities but creates scheduling, receiver-complexity, and beam-discovery challenges.

  • SDMA: SDMA multiplexes users spatially and can increase sum rate with the number of users when sufficient RF chains and favorable propagation are available.Highly directional mmWave beams can separate users from different directions, although hybrid-system channel estimation requires further work.
  • SDMA: When users outnumber antennas, SDMA requires grouping users and serving each group orthogonally or semi-orthogonally across time, frequency, or code.Scheduling algorithms have been proposed to enhance throughput.
  • NOMA: NOMA superimposes coded and modulated user signals on shared time-frequency resources and uses successive interference cancellation at receivers.It can enhance spectral efficiency and connectivity density, but increases receiver complexity relative to OMA.
  • NOMA: Simulation results report that NOMA achieves better channel capacity than OMA in uplink and downlink mmWave systems.The cited studies also examine NOMA in mmWave massive-MIMO systems.
  • Random access: Random access is challenging because initial access and handover cannot fully exploit beamforming without information about the best transmit-receive beam pair.Design issues include initial access, handover, link configuration, and scheduling.
  • Random access: Directional cell discovery scans angular space in time-varying random directions, contrasting with exhaustive 360-degree scanning.These approaches are discussed for mmWave discovery procedures.
  • Backhauling: Ultradense-network deployment uses densely placed small-cell base stations to support Gbps user experience and seamless coverage, increasing the need for high-bandwidth backhaul.Existing lower-frequency backhaul spectrum is described as limited for 5G demand.
  • Backhauling: In-band backhauling shares the same mmWave spectrum between access and backhaul, offering hardware and frequency-reuse benefits while requiring interference control.Beamforming and signal processing are cited as control mechanisms.

C. Coverage and Connectivity

mmWave coverage and connectivity are constrained by blockage and limited line-of-sight regions, motivating relaying, cooperation, large arrays, multi-connectivity, and heterogeneous deployments.

  • Blockage modeling: A stochastic blockage model represents base stations and users as homogeneous Poisson point processes and buildings as random rectangles.The model addresses blockage of moving users by walls and other objects.
  • Blockage modeling: The blockage parameters β and p require experimental fitting in practical mmWave systems.This is an explicit modeling assumption for applying the blockage formulation.
  • Coverage: The average visible LoS region is 2πe^-p/β^2, and the average number of LoS-connected base stations is 2πµe^-p/β^2.The paper links acceptable coverage to denser deployment or shorter links through intermediate relays.
  • Connectivity: Multi-hop relaying greatly improves connectivity over single-hop transmission, with near-optimal performance when the relaying route window is about the obstacle-building size.Coverage probability is defined as the probability of receiving a signal above threshold SNR T.
  • Coverage enhancement: Cooperation among randomly located base stations can effectively increase downlink coverage probability.The result comes from a stochastic-geometry analysis of mmWave networks.
  • Coverage enhancement: Coverage probability increases as a non-decreasing concave function of antenna-array size, and large arrays are required for satisfactory mmWave coverage.The result is obtained using a tractable stochastic-geometry framework with arbitrary interference distributions and antenna patterns.
  • Connectivity: Multi-connectivity improves session-level mmWave operation by addressing urban propagation, human-induced blockage, and session continuity.Even simpler multi-connectivity schemes produce notable improvements in ultradense urban deployments.
  • Deployment architecture: A heterogeneous architecture pairs microwave macro cells for discovery, signaling, and control with mmWave small cells for high-rate data.This division is proposed to provide robust coverage and connectivity.

A. 3GPP’s New Radio at mmWave Band

3GPP’s mmWave New Radio vision combines high-capacity mmWave operation with massive MIMO, while standardization and deployment remain constrained by unresolved beam-management, measurement, and mobility challenges.

  • Vision and use cases: 3GPP targets mmWave for eMBB services requiring extraordinary data rates, especially in small cells and dense urban scenarios, while defining channel models above 6 GHz up to 100 GHz.Other envisioned uses include last-mile fiber replacement through backhaul and wireless fronthaul, but latency and reliability effects remain unclear.
  • mmWave and massive MIMO: mmWave and massive MIMO are identified as core 5G PHY technologies with complementary roles across coverage, mobility, and high-capacity hotspot scenarios.3GPP limits RF chains to 32 at the base station and 8 at the user equipment, with up to 1024 and 64 antenna elements respectively at 70 GHz.
  • PHY layer design: Beam management remains an open PHY-design area spanning beam sweeping, beam selection, tracking, recovery, reference signals, and control channels.Beam sweeping affects initial access, tracking under changing channels, and connection maintenance after blockage or link failure.
  • Prototypes and deployment plan: Hardware platforms and proof-of-concept prototypes are needed to test mmWave across frequency bands, use cases, and their corresponding KPIs before mass deployment.The measurement campaigns cover basic throughput, non-line-of-sight transmission, coverage, and beam tracking.
  • Prototypes and deployment plan: Current measurements report at most two spatial streams and below 20 bps/Hz spectral efficiency, with coverage and tracking varying substantially by setting and mobility.Reported coverage reaches indoor ≤100 m, urban outdoor ≤350 m, and outdoor-to-indoor ≤20 m; single-user tracking is demonstrated up to 40 km/h, while multi-user tracking remains lacking.
  • Conclusions: The survey identifies high-mobility operation, longer transmission distance, hardware impairments, energy efficiency, and cost-efficient fully digital RF implementation as important future challenges.It states that mmWave mobile-network mass deployment would likely occur after 2020, depending on standardization progress and resolution of these challenges.
Loading 1705.06072v1…