PaperlayerView exact revision on arXiv

Source-linked AI summary

Millimeter Wave Cellular Wireless Networks: Potentials and Challenges

Sundeep Rangan, Theodore S. Rappaport, Elza Erkip

arXiv:1401.2560v1cs.NI

TL;DR

The paper asks whether mmW frequencies can support reliable, longer-range NLOS cellular networks despite propagation, hardware, and coordination challenges. It surveys NYC measurements at 28 and 73 GHz and related channel and capacity analyses, finding viable urban coverage and at least order-of-magnitude capacity gains at comparable cell density. Realizing those gains requires substantial redesign across the cellular stack and heterogeneous deployment strategies.

  • Problem

    The paper addresses whether mmW cellular is feasible for longer-range NLOS scenarios despite severe shadowing, intermittent connectivity, higher Doppler spreads, and high device power demands.

  • Method

    The paper surveys NYC outdoor channel measurements at 28 and 73 GHz, statistical channel models, and capacity analyses focused on urban micro- and picocellular deployments.

  • Results

    At least an order of magnitude higher capacity than current state-of-the-art LTE is reported, with viable 100 m to 200 m NLOS coverage in NYC measurements.

  • Takeaways & Limitations

    MmW cellular can be viable for outdoor urban coverage, but adaptive beamforming, heterogeneous networks, relaying, and carrier aggregation are needed to address its design challenges.

Abstract

from arXiv · show

Millimeter wave (mmW) frequencies between 30 and 300 GHz are a new frontier for cellular communication that offers the promise of orders of magnitude greater bandwidths combined with further gains via beamforming and spatial multiplexing from multi-element antenna arrays. This paper surveys measurements and capacity studies to assess this technology with a focus on small cell deployments in urban environments. The conclusions are extremely encouraging; measurements in New York City at 28 and 73 GHz demonstrate that, even in an urban canyon environment, significant non-line-of-sight (NLOS) outdoor, street-level coverage is possible up to approximately 200 m from a potential low power micro- or picocell base station. In addition, based on statistical channel models from these measurements, it is shown that mmW systems can offer more than an order of magnitude increase in capacity over current state-of-the-art 4G cellular networks at current cell densities. Cellular systems, however, will need to be significantly redesigned to fully achieve these gains. Specifically, the requirement of highly directional and adaptive transmissions, directional isolation between links and significant possibilities of outage have strong implications on multiple access, channel structure, synchronization and receiver design. To address these challenges, the paper discusses how various technologies including adaptive beamforming, multihop relaying, heterogeneous network architectures and carrier aggregation can be leveraged in the mmW context.

I. INTRODUCTION

MmW cellular is motivated by rapidly increasing capacity demand and abundant higher-frequency spectrum, but its feasibility depends on overcoming propagation and system-design challenges. Surveyed measurements and analyses assess urban viability and the redesign needed to realize capacity gains.

  • Motivation: 40% to 70% year upon year increase in cellular traffic could require 1000 times current capacity within the next decades.The paper also notes estimates of as many as 50 billion connected devices by 2020.
  • Opportunity: 30 to 300 GHz mmW bands can provide more than 200 times the spectrum of current cellular allocations.Small wavelengths also enable large multi-element antenna arrays in compact dimensions for high-gain directional transmission.
  • Approach: The paper surveys NYC channel measurements at 28 and 73 GHz to assess mmW propagation in dense urban environments dominated by NLOS paths.NYC is treated as a challenging test case representative of likely initial deployments, with typical urban cell radii of 100 m to 200 m.
  • Findings: 100 m to 200 m signal detection with less than 1 Watt of transmit power indicates that mmW may support urban NLOS coverage without greater cell density.The reported distances are comparable to current urban UHF/microwave cell radii.
  • Findings: At least an order of magnitude higher capacity than current state-of-the-art 4G networks is reported for comparable cell density.A hypothetical 1 GHz TDD mmW system could provide a 20-fold increase in average cell throughput over a 20+20 MHz LTE system.
  • System implications: Highly directional and adaptive transmissions require major changes to cell search, broadcast signaling, random access, multiple access, and small-packet support.Directional isolation also reduces the role of interference relative to current small-cell networks.

II. MILLIMETER WAVE CELLULAR NETWORKS

MmW cellular extends established mmW applications toward longer-range NLOS cellular use, enabled by wide bandwidth and compact antenna arrays. Its main obstacles include shadowing, rapidly changing channels, multiuser coordination, and processing power consumption.

  • Definition and context: 1 to 10 mm wavelengths correspond to approximately 30 to 300 GHz, the frequency range used for mmW signals in this paper.Existing mmW applications include satellite communications, cellular backhaul, and short-range 60 GHz WLAN and PAN links.
  • Motivation: The feasibility of longer-range NLOS mmW cellular remains debated because signals suffer severe shadowing, intermittent connectivity, and higher Doppler spreads.This contrasts with many existing mmW systems designed for short-range or point-to-point LOS settings.
  • Enabling technologies: Compact arrays and advances in CMOS RF and digital processing support reconsideration of mmW cellular viability.Multiple arrays can also be distributed across a device to provide path diversity against human blockage.
  • Challenges: 40 to 80 dB attenuation from brick and 20 to 35 dB loss from the human body illustrate the severity of mmW shadowing.Reflective people and outdoor materials can nevertheless act as important scatterers, while rain and humidity fades are described as minimal for cellular systems.
  • Challenges: Over 3 kHz Doppler spread at 60 km/h and 60 GHz implies channel changes on the order of hundreds of µs.High shadowing can make obstacles produce dramatic swings in channel quality and intermittent connectivity.
  • Challenges: Greater spatial reuse and spectral efficiency require new mechanisms to coordinate simultaneous transmissions on multiple mmW links.Existing point-to-point and limited-user mmW applications do not provide the same coordination requirements as cellular systems.
  • Challenges: Greater than 250 mW is estimated for 100 Ms/s, 12-bit A/D conversion across 16 antennas, creating a substantial mobile-device power burden.A/D power scales linearly with sampling rate and exponentially with bits per sample.

C. Deployment Models

The paper focuses on small outdoor mmW cells in challenging urban environments, where NLOS coverage and deployment economics shape the network architecture. Because mmW cannot provide uniform coverage alone, deployments are expected to combine heterogeneous cell types, conventional overlays, relays, and shared infrastructure.

  • Deployment Models: 100 m to 200 m urban micro- and picocellular radii define the paper’s deployment focus.Urban coverage is more challenging than campus or stadium settings because NLOS propagation occurs more frequently.
  • Deployment Models: Street furniture such as lampposts or building sides can host picocells for dense street coverage with minimal shadowing.The placement is intended to provide direct coverage onto urban streets.
  • Deployment Models: MmW networks are inherently heterogeneous because propagation limitations prevent uniform, robust, high-capacity coverage across deployments.The paper anticipates coexistence between cellular and local-area networks over time.
  • Deployment Models: Backhaul and spectrum may be shared more extensively among operators and third parties to support indoor coverage and efficient spectrum use.The figure frames this sharing alongside coordination with conventional microwave cells.
  • Deployment Models: Limited mmW range requires coexistence with a conventional UHF/microwave cellular overlay for universal coverage.This creates heterogeneity beyond cell size, including different radio technologies and coverage roles.
  • Deployment Models: Relays or repeaters can reduce the impracticality and expense of running fiber to every small cell.They can build on existing mmW backhaul technology and exploit mmW degrees of freedom.
  • Deployment Models: Microcells and picocells may need to coexist because microcells offer larger range while picocells provide shorter-range LOS coverage.The paper contrasts this with more diffuse NLOS coverage from microcells.
  • Deployment Models: Indoor penetration challenges may favor third-party ownership of indoor mmW cells, creating multi-operator heterogeneous networks.Such architectures introduce design issues in cell selection and networking.

III. CELLULAR CHANNEL MEASUREMENTS

The paper surveys prior outdoor mmW propagation work and presents NYC measurements at 28 and 73 GHz for realistic urban microcellular assessment.

  • Outdoor micro- and picocellular propagation is less understood than indoor mmW channels, while earlier studies often relied on analytic or ray-tracing models.Those models commonly assumed LOS or a few strong specular reflections, assumptions the paper evaluates with measurements.
  • The study measured 28 and 73 GHz channels in New York City to assess street-level propagation in dense urban environments.Transmitters were placed on rooftops, with measurements taken at street locations up to 500 m away.
  • The 28 and 73 GHz bands were selected as likely initial operating frequencies for mmW cellular systems.The paper emphasizes their relevance to prospective early deployments within the mmW range.
  • The measurements used directional, rotatable horn antennas to characterize both bulk path loss and spatial channel structure.The antennas had 10-degree beamwidths, 24.5 dBi gain, and 30 dBm RF output at both transmitter and receiver.

C. Large-Scale Path Loss Model

NYC-derived models show higher omnidirectional mmW path loss than conventional low-frequency models, but fixed-size directional antennas can compensate for that increase.

  • The NYC channel models convert directional measurements into unity-gain omnidirectional models by summing received angular powers and removing 49 dB of antenna gain.The resulting statistical models support the paper’s capacity and design analysis.
  • The floating-intercept fit cautions against assigning physical meaning to its estimated intercept or slope parameters.A close-in free-space reference model is less sensitive to data perturbations and has only slightly greater fitting error.
  • At d = 100 m, free-space propagation predicts approximately 30 dB less path loss than the experimentally derived omnidirectional NYC model.This makes free-space-based capacity predictions potentially optimistic.
  • Beamforming can entirely compensate the modeled mmW path-loss increase for fixed antenna area, with possible improvement when applied at both ends.The paper notes that spatial multiplexing may provide additional gains.

D. Angular and Delay Spread Characteristics

NYC measurements reveal multiple angular and delay-separated path clusters, providing spatial diversity but imposing receiver-design implications.

  • Several distinct path clusters with significant angular and delay separation were observed in the channel-sounding measurements.The measurements used 10-degree rotatable horn antennas and 400 MHz baseband bandwidth for high-resolution temporal and angular characterization.
  • Three angular clusters were visible at a typical 28 GHz location, while several clusters also appeared in the power-delay profile.Each cluster had relatively narrow angular and delay spread.
  • Approximately two clusters per location are modeled on average, and many locations potentially support two or three spatial degrees of freedom.The paper attributes this potential to multiple clusters and within-cluster angular spread.
  • Angular spread occurs mainly in azimuth at both transmitter and receiver, with some receiver elevation spread potentially caused by ground reflections.Within-cluster rms beamspread varies substantially and is modeled exponentially.
  • Power is not concentrated overwhelmingly in the strongest cluster; substantial power often appears in the second or third cluster.This distribution indicates possible spatial multiplexing gains between a base station and mobile.

E. Outage Probability

Outage is a central urban mmW risk because signals depend on line-of-sight or strong reflections, with measured reliability declining beyond roughly 175 m.

  • Outdoor mmW reception relies on LOS links or strong reflections and scattering because signals cannot penetrate many building walls.Shadowing can therefore cause outage when suitable reflective or scattering paths are absent.
  • All 30 Manhattan measurement locations within 175 m had detectable signals, whereas most locations beyond 175 m experienced outage.The results were obtained from measurements extending to locations as far as 500 m from the transmitter.
  • The measured outage behavior is highly environmentally dependent and may worsen with local obstacles, handheld receivers, or indoor operation.The paper explicitly cautions against broad generalization from these measurements.
  • The outage analysis used experimentally derived NYC channel models and standard random-placement small-cell network simulations.The network evaluation considered BS and UE distributions similar to the 3GPP UMi model.
  • The simulations used a conservative single-stream beamforming model that excluded single-user and multi-user spatial multiplexing gains.Coordinated beamforming and MIMO could provide additional gains, particularly near the cell.

B. SINR and Rate Distributions

The evaluated mmW configurations provide high SINR and rate distributions, with cell capacities often exceeding 1 Gbps and substantial cell-edge gains. Higher-frequency operation can be offset by using more antenna elements within a similar physical area.

  • Cell capacities are often greater than 1 Gbps, while the lowest 5% cell-edge rates exceed 10 Mbps.These rates are presented as potentially satisfying many Beyond 4G requirements.
  • For the same number of antenna elements, 73 GHz rates are approximately half those at 28 GHz.
  • A 28 GHz 4x4 λ/2 array occupies about the same area as a 73 GHz 8x8 λ/2 array: roughly 1.5× 1.5 cm2.The paper notes that this area could fit in a handheld mobile device.
  • 73 GHz 8x8 rate and SNR distributions are very close to those of 28 GHz 4x4 arrays when UE antenna size is approximately constant.
  • Only about 10% of mobiles fall below 0 dB SINR, a lower fraction than in typical cellular deployments.The comparison uses the Fig. 10 SINR distributions against traditional UHF or microwave cellular studies.
  • The 5% cell-edge rates provide a 9 to 10 fold increase, but edge users remain power-limited and cannot fully exploit the added spectrum.The paper identifies repeaters or relays as needed for more uniform performance in these NLOS scenarios.

D. Interference vs. Thermal Noise

The evaluated mmW small-cell links are generally not interference-dominated: directional isolation makes thermal noise comparable to or larger than interference for many mobiles. Without spatial multiplexing, links remain bandwidth-limited despite high SINR, while outage threatens uniform edge performance.

  • Interference is not dominant; for most mobiles, thermal noise is comparable to or larger than interference, particularly downlink.This conclusion comes from the 28 GHz uplink and downlink interference-to-noise simulations.
  • Without spatial multiplexing, mmW links are bandwidth-limited rather than power-limited despite non-dominant interference.The paper characterizes this as a distinct operating point with relatively high-SINR, directionally isolated links.
  • Spatial multiplexing has not yet been evaluated; it would lower SINR per stream for higher-SINR mobiles while leaving the INR distribution largely unchanged.The paper expects links to remain limited by thermal noise because total transmit power stays constant.
  • Under a threshold outage model, signals beyond T = 175m are treated as undetectable and links are considered in outage.The threshold is based on the paper’s measured channel data.
  • A soft-outage model assigns approximately 20% outage probability even at 80 m, reducing 5% uplink and downlink cell-edge rates by 50%.Average cell capacity is not significantly reduced under this more conservative model.

F. Other Studies

Earlier analyses using diverse scenarios and methods broadly corroborate high mmW capacity estimates based on experimentally derived channel models. The paper then turns to system redesign challenges created by directional communication.

  • F. Other Studies: A 500 MHz system was estimated to provide approximately 300 Mbps per-cell throughput in an earlier study.That estimate corresponds to somewhat lower spectral efficiency than the measurements-based results discussed here.
  • F. Other Studies: Ray tracing of a campus network found a median total capacity of 32 Gbps across five sites with four cells per site.With 2 GHz bandwidth, this corresponds to approximately 0.8 bps/Hz/cell; the study used QPSK modulation.
  • F. Other Studies: A stochastic-geometry study predicted almost 5.4 bps/Hz, nearly twice the paper’s estimated spectral efficiency, assuming Shannon-limit operation without a maximum spectral efficiency.
  • F. Other Studies: Across varied scenarios and analysis methods, mmW systems can offer orders-of-magnitude increases in capacity and cell-edge rate over state-of-the-art current-band systems.The paper presents this as the broad message from comparisons with prior work.
  • F. Other Studies: Achieving the full mmW gains may require redesigning cellular systems around highly directional transmissions.The paper identifies synchronization, broadcast signaling, cell search, and random access as affected functions.
  • F. Other Studies: Intermittent communication requires transmitters to maintain spatial channel-state information so mobile or base-station transmissions can begin quickly.This issue arises in DRX and DTX modes used for low-power always-on connectivity.

B. Multiple Access and Front-End / Baseband Considerations

mmW front-end constraints make multiple access a central design issue: analog beamforming may force TDMA, sacrificing frequency granularity and uplink efficiency. Relaying and redesigned peer-to-peer coordination can help address coverage and backhaul challenges.

  • TDD is attractive for mmW because small cells need bandwidth flexibility, beamforming support, and low cost.
  • FDMA can improve uplink power, small-packet efficiency, and UE power consumption, but analog beamsteering makes same-slot frequency access difficult.
  • Analog phased-array front-ends may prevent spatial multiplexing and same-slot multi-user transmission, forcing one-user-at-a-time TDMA.Large antenna arrays and wide bandwidths make per-element high-resolution ADCs impractical from a power or cost perspective.
  • An order of magnitude improvement in uplink rate is shown for FDMA with multi-user transmission versus baseline TDMA under TDD.The comparison uses beamforming with the 28 GHz isotropic channel model.
  • Relaying can extend mmW coverage through clutter and shadowing, support indoor-outdoor coverage, and provide wireless backhaul.The paper suggests that fully exploiting relaying may require peer-to-peer topologies and synchronous coordination across relays, base stations, and mobiles.

D. An End to Interference?

Directional mmW links can reduce interference, shifting emphasis away from network-wide interference mitigation toward point-to-point technologies. Channel sparsity and small angular spreads may also make channel tracking more manageable despite high carrier frequencies.

  • With appropriate beamforming, mmW links become directionally isolated and interference is greatly reduced, limiting gains from conventional interference-mitigation techniques.
  • Point-to-point technologies may therefore contribute more to mmW capacity gains than network-wide interference coordination.
  • Despite potentially high Doppler from carrier frequency and mobility, measurements show signals arriving in a few clusters with small angular spreads.
  • A channel parametrization using cluster count, Doppler, arrival angles, and time-varying cluster gains can support more accurate tracking.Measurements typically found 1 to 5 clusters, whose gains and arrival angles vary slowly.
  • The nonlinear channel parametrization is analogous to finite-rate-of-innovation and compressed-sensing models, suggesting value for wideband large-array systems.

F. Heterogeneous Networking Issues

mmW networks require heterogeneous support because shadowing can make individual links unreliable and coverage cannot be provided standalone. Faster path switching, carrier aggregation, and shared-spectrum coordination are central design concerns.

  • Standalone mmW deployment cannot provide uniform, reliable coverage, so fallback to conventional UHF or microwave cellular systems is necessary.
  • Shadowing can make links to individual cells inherently unreliable and rapidly changing, requiring faster cell selection and path switching than current systems.
  • Carrier aggregation could provide macro-diversity by connecting mobiles to multiple base stations, but requires network support for path switching and scheduling.
  • A single operator may not fully utilize mmW spectrum, while multi-operator sharing would require sophisticated direction-aware inter-cell interference coordination.
  • Measurements and capacity analysis indicate 100 m to 200 m NLOS viability and at least an order-of-magnitude capacity over current state-of-the-art LTE for outdoor coverage.
  • Realizing mmW potential requires redesigning cell search, synchronization, random access, intermittent communication, multiple access, and channelization around directional transmissions and analog front-ends.
  • Carrier aggregation and multihop relaying may become prominent in mmW, while interference mitigation may have less impact because of directional isolation.
Loading 1401.2560v1…