PaperlayerView exact revision on arXiv

Source-linked AI summary

A Millimeter-Wave Channel Simulator NYUSIM with Spatial Consistency and Human Blockage

Shihao Ju, Ojas Kanhere, Yunchou Xing, Theodore S. Rappaport

arXiv:1908.09762v1cs.ITeess.SPmath.IT

TL;DR

MmWave systems need accurate channel models that account for spatial consistency, human blockage, and outdoor-to-indoor penetration loss. NYUSIM 2.0 adds trajectory-based channel updates and models for blockage and penetration. Its simulations produce correlated channel behavior, with reported NLOS blockage-loss probabilities above 10 dB and 15 dB.

  • Problem

    MmWave channel modeling must account for nearby locations having correlated channel impulse responses and for signal shadowing from human blockage and outdoor-to-indoor penetration loss.

    Source: §I · p. 1

  • Method

    NYUSIM 2.0 models trajectory-based channel evolution with multiple reflection surfaces, a four-state Markov blockage model, and a parabolic outdoor-to-indoor penetration model.

    Source: 2) Time-variant Small-scale Parameters · p. 3, B. Human Blockage Model · p. 4, §V · p. 6

  • Results

    In simulated NLOS scenarios, shadowing loss exceeded 10 dB and 15 dB with probabilities of 55% and 12%, respectively.

    Source: B. Human Blockage Model · pp. 4–5

  • Takeaways & Limitations

    NYUSIM 2.0 generates consecutive, correlated omnidirectional and directional PDPs based on a user-specified UT trajectory.

    Source: §V · p. 6

Abstract

from arXiv · show

Accurate channel modeling and simulation are indispensable for millimeter-wave wideband communication systems that employ electrically-steerable and narrow beam antenna arrays. Three important channel modeling components, spatial consistency, human blockage, and outdoor-to-indoor penetration loss, were proposed in the 3rd Generation Partnership Project Release 14 for mmWave communication system design. This paper presents NYUSIM 2.0, an improved channel simulator which can simulate spatially consistent channel realizations based on the existing drop-based channel simulator NYUSIM 1.6.1. A geometry-based approach using multiple reflection surfaces is proposed to generate spatially correlated and time-variant channel coefficients. Using results from 73 GHz pedestrian measurements for human blockage, a four-state Markov model has been implemented in NYUSIM to simulate dynamic human blockage shadowing loss. To model the excess path loss due to penetration into buildings, a parabolic model for outdoor-to-indoor penetration loss has been adopted from the 5G Channel Modeling special interest group and implemented in NYUSIM 2.0. This paper demonstrates how these new modeling capabilities reproduce realistic data when implemented in Monte Carlo fashion using NYUSIM 2.0, making it a valuable measurement-based channel simulator for fifth-generation and beyond mmWave communication system design and evaluation.

I. INTRODUCTION

The introduction motivates accurate mmWave channel modeling by pointing to its distinct channel statistics and the demands of 5G and beyond. It highlights spatial consistency, human blockage, and outdoor-to-indoor penetration loss as important modeling concerns.

I. INTRODUCTION

  • MmWave's massive available bandwidth offers capacity potential, but its directional channels have different statistics from microwave channels, making accurate modeling important for 5G and beyond.The introduction also notes applications including wireless cognition, imaging, and precise positioning.Source: §I · p. 1
  • Nearby UT locations should have highly correlated channel impulse responses, and spatial consistency can generate correlated, time-varying channel coefficients along a trajectory –.This supports continuous, realistic angular power spectra and power delay profiles based on the UT trajectory within a local area.Source: §I · p. 1
  • Human blockage matters because mmWave signals are easily blocked by people and vehicles, making blockage shadowing loss important for accurate link-budget analysis,.The introduction attributes this susceptibility to short wavelengths and directional antennas, which do not effectively diffract around human bodies or vehicles.Source: §I · p. 1
  • Outdoor-to-indoor penetration loss is prominent at mmWave frequencies, and concrete and infrared-reflecting glass can induce large losses,,.The introduction identifies accurate prediction as important for outdoor and indoor mmWave system design and deployment.Source: §I · p. 1

II. DROP-BASED NYUSIM CHANNEL SIMULATOR

The section describes NYUSIM's measurement-based channel simulator and its drop-based channel model. It outlines the path-loss and clustering approaches and notes that separate drops lack correlation across distances.

II. DROP-BASED NYUSIM CHANNEL SIMULATOR

  • NYUSIM produces omnidirectional and directional CIRs, PDPs, and 3-D angular power spectra, using a 3-D spatial statistical channel model developed from extensive 28–140 GHz field measurements.It supports carrier frequencies from 500 MHz to 100 GHz, RF bandwidths up to 800 MHz, and ULA and URA antenna arrays.Source: §II · p. 1
  • Each drop generates a static, independent CIR at one T-R separation distance, so even nearby distances have independent channel coefficients.Each drop uses i.i.d. large- and small-scale parameters.Source: §II · p. 2
  • NYUSIM uses a close-in path-loss model with a 1 m reference distance, plus an attenuation term for atmospheric conditions,,.The model also includes shadow fading and total atmospheric absorption; its free-space reference loss is evaluated at 1 m and carrier frequency f [14],,.Source: §II · p. 2
  • The time-cluster spatial-lobe approach generates MPC temporal and spatial properties separately, unlike joint time-angle distributions in the current 3GPP model.Time clusters group MPCs close in time; spatial lobes group MPCs from similar directions, potentially across hundreds of nanoseconds.Source: §II · p. 2

III. NEW FEATURES IMPLMENTED IN NYUSIM 2.0

Spatial consistency is described as realistic channel evolution along a UT trajectory, using correlated large-scale parameters and time-varying small-scale parameters. The section illustrates the spatial shadow-fading map and a sample UT path.

A. Spatial Consistency Procedure

  • Spatial consistency generates spatially correlated SF and LOS/NLOS conditions alongside time-varying MPC angles, power, delay, and phase along a UT trajectory.A channel segment is 10–15 m long, and channels within it are treated as highly correlated and updated across snapshots.Source: A. Spatial Consistency Procedure · p. 2
  • Figure 1 presents a spatially correlated SF map with BS and UT locations for a UMi LOS scenario; SF is distributed as N(0,4) dB, with 100 m T-R separation.Source: Figure 1 · p. 2
  • The UT track is a clockwise partial hexagon with a 40 m total distance and 10 m sides.The figure caption identifies this as the UT's movement path.Source: Figure 2 · p. 2

1) Spatially Correlated Large-scale Parameters:

NYUSIM creates spatially correlated large-scale parameters by filtering an independent shadow-fading map, and applies a 2-D exponential filter to generate correlated LOS/NLOS conditions. Along a UT track, correlated SF yields smoother path-loss evolution than independent values.

1) Spatially Correlated Large-scale Parameters:

  • SF values are initialized independently on a 1 m grid and passed through a 2-D exponential filter to form a correlated map.The filter length is set to 8 times the SF correlation distance; correlation is negligible beyond 4 times that distance.Source: 1) Spatially Correlated Large-scale… · pp. 2–3
  • Path loss varies rapidly by about 16 dB with independent SF values, but varies smoothly with spatially correlated SF along the UT trajectory.The SF map spans -10 dB to 10 dB continuously, with a 10 m correlation distance in the UMi LOS scenario.Source: 1) Spatially Correlated Large-scale… · p. 3
  • A 2-D exponential filter also generates spatially correlated LOS/NLOS conditions.The SF map is illustrated over a 200 m x 200 m area, with a UT moving 40 m along a partial hexagon track.Source: 1) Spatially Correlated Large-scale… · p. 3

2) Time-variant Small-scale Parameters:

NYUSIM updates small-scale channel parameters along the UT trajectory using geometry. For NLOS paths, multiple reflection surfaces convert the UT to a virtual LOS case, allowing LOS update equations to be reused.

2) Time-variant Small-scale Parameters:

  • LOS angles are updated geometrically from BS and UT locations, UT speed, and direction.The four MPC angles are AOD, ZOD, AOA, and ZOA.Source: 2) Time-variant Small-scale Parameters · p. 3
  • For NLOS components, multiple reflection surfaces transform the UT location into a virtual LOS location, so the components can be updated using the LOS equations.Fig. 3 depicts UT movement and AOD change in the azimuth plane for an NLOS multiple-reflection case.Source: 2) Time-variant Small-scale Parameters · p. 3, Figure 3 · p. 3
  • For NLOS paths with M reflections, angle relations are derived iteratively, with odd and even reflection counts treated as separate cases.The binary variable B represents the cases: B = -1 for an even number of reflection surfaces and B = 1 for an odd number.Source: 2) Time-variant Small-scale Parameters · p. 3
  • Each MPC’s delay and phase are updated from path-length change using the law of cosines.For LOS, path length is the T-R separation; for NLOS, it is obtained from the initialized absolute delay multiplied by the speed of light.Source: 2) Time-variant Small-scale Parameters · p. 4
  • MPC powers are redistributed using the initialization procedure, with cluster and intracluster powers decaying exponentially with their respective excess delays.Total received power is updated based on time-variant large-scale path loss.Source: 2) Time-variant Small-scale Parameters · p. 4

3) Smooth transitions:

A smooth transition procedure connects channel segments whose initial coefficients are generated independently.

3) Smooth transitions:

  • Initial channel coefficients are generated independently across channel segments, so a smooth transition procedure connects the segments through cluster birth and death.The power of an old cluster ramps down as a new cluster ramps up.Source: 3) Smooth transitions · p. 4
  • When adjacent segments have different numbers of time clusters, individual cluster powers can ramp up or down.Source: 3) Smooth transitions · p. 4
  • Cluster birth and death occur for only one time cluster in a channel snapshot.Source: 3) Smooth transitions · p. 4

B. Human Blockage Model

NYUSIM models human blockage as time-varying shadowing using a four-state Markov process, with blockage statistics adjusted for antenna directionality. Simulations show stronger losses for narrower directional antennas and close agreement between simulated and measured channel-outage results.

B. Human Blockage Model

  • The four-state Markov model represents blockage as unshadowed, decay, shadowed, and rising stages, with transition rates and mean attenuation dependent on antenna HPBW.The underlying measurements used 7°, 15°, and 60° antennas; they were peer-to-peer measurements, which the authors consider reasonable to extend to BS–UT settings.Source: B. Human Blockage Model · p. 4
  • For omnidirectional channels, spatial-lobe width replaces antenna HPBW in calculating blockage, and MPCs within one lobe are assumed to experience approximately equal loss.The lobe-width assumption is related to a measured 10.5° LOS azimuth spread and a typical width below 63° with 99.7% probability; directional channels instead group MPCs within antenna HPBW.Source: B. Human Blockage Model · p. 4
  • NYUSIM superimposes 1–5 independent Markov traces per spatial lobe to form the total loss trace, then samples a time instance for the shadowing loss.Omnidirectional and directional blockage procedures are simulated independently because their transition rates and average mean attenuation differ.Source: B. Human Blockage Model · p. 4
  • Simulated outage probability and 5% SNR levels are close to measurements, supporting realistic recreation of directional channels with human blockage.Table I compares simulated and measured user-SNR CDFs at 73 GHz, with and without blockage, using 1000 Monte Carlo simulations.Source: Table I · p. 5, B. Human Blockage Model · p. 5
  • In NLOS, shadowing loss exceeds 10 dB and 15 dB for 55% and 12% of users, respectively; 31% of users with 7° HPBW antennas exceed 15 dB.The omnidirectional comparison covers UMi LOS and NLOS; the directional comparison uses UMi NLOS channels.Source: B. Human Blockage Model · p. 5
  • The NLOS omnidirectional shadowing loss is slightly larger than LOS loss because NLOS spatial-lobe angular spread is smaller.The comparison used 1000 UT locations over T–R distances from 100 m to 500 m.Source: B. Human Blockage Model · p. 5

C. O2I Penetration Loss Model

NYUSIM implements a parabolic model for outdoor-to-indoor building penetration loss, with low- and high-loss forms selected according to the external building surface.

C. O2I Penetration Loss Model

IV. NYUSIM 2.0

NYUSIM 2.0 provides spatial-consistency and drop-based operating modes, alongside human-blockage and outdoor-to-indoor penetration capabilities.

IV. NYUSIM 2.0

  • NYUSIM 2.0 can generate continuous time-varying CIRs and simulate power loss from random human blockage and external building penetration.Source: §IV · p. 5
  • When spatial consistency is enabled, the simulator generates successive, correlated CIRs along the UT trajectory.Source: §IV · p. 5
  • When spatial consistency is disabled, the drop-based mode generates independent CIRs for different distances; human blockage operates in both modes.Source: §IV · p. 5

A. Input parameters

NYUSIM 2.0 adds user-editable input groups for spatial consistency, human blockage, and outdoor-to-indoor penetration, as well as two spatial-consistency track types.

A. Input parameters

  • The interface adds 10 spatial-consistency, five human-blockage, and two O2I-penetration inputs to the existing channel and antenna parameters.The existing groups contain 16 Channel Parameters and 12 Antenna Properties.Source: A. Input parameters · p. 5
  • Spatial-consistency modeling offers linear and hexagon user tracks.Source: A. Input parameters · p. 5
  • The human-blockage defaults fit transition rates and mean attenuation to and adjust with RX antenna azimuth HPBW; users can instead edit these values.The editable option allows user preferences or other Markov-model parameters.Source: A. Input parameters · p. 5

B. Sample Output Figures and Output Data

NYUSIM 2.0 adds spatial consistency, human blockage, and O2I penetration loss modeling. Its outputs include trajectory-based channel visualizations and stored time-varying parameters.

B. Sample Output Figures and Output Data

  • Spatial-consistency outputs include correlated shadow-fading and LOS/NLOS maps, consecutive omnidirectional and directional PDPs, 3D angular power spectra, and stored trajectory-dependent parameters.Directional PDPs apply the antenna pattern in the strongest received MPC direction to the omnidirectional PDP.Source: B. Sample Output Figures and Output Data · p. 5
  • The GUI presents new dialogue boxes for spatial consistency, human blockage, and O2I penetration loss.Source: Figure 5 · p. 6
  • Each consecutive omnidirectional PDP output stores and displays absolute time delay for its channel snapshot.Source: Figure 6 · p. 6

V. CONCLUSION

  • NYUSIM 2.0 combines multiple reflection surfaces for spatial consistency, a four-state Markov model for human blockage, and a parabolic model for O2I penetration loss.It generates consecutive, correlated omnidirectional and directional PDPs along a user-specified trajectory.Source: §V · p. 6
Loading 1908.09762v1…