PaperlayerView exact revision on arXiv

Source-linked AI summary

A Novel Millimeter-Wave Channel Simulator and Applications for 5G Wireless Communications

Shu Sun, George R. MacCartney, Theodore S. Rappaport

arXiv:1703.08232v2cs.IT

TL;DR

NYUSIM addresses the need for realistic channel simulation to evaluate and design 5G wireless systems. It combines measurement-based statistical spatial modeling with broad frequency, bandwidth, scenario, and antenna support. The paper’s examples show that NYUSIM can produce different and more measurement-grounded MIMO and spectral-efficiency predictions than 3GPP modeling, particularly because 3GPP uses unrealistically many clusters.

  • Problem

    Wireless-system design requires channel simulators, but existing mmWave models can use cluster counts that are not supported by real-world measurements.

  • Method

    The paper develops open-source NYUSIM from extensive wideband mmWave measurements, using a statistical spatial channel model with configurable path loss, clustering, directional channels, and MIMO analysis.

  • Results

    NYUSIM recreates measured wideband PDPs, CIRs, and channel statistics, while its MIMO examples show that 3GPP can overpredict spectral efficiency; at 20 dB SNR and Ns = 4, 3GPP gives 40 bits/s/Hz versus about 27 bits/s/Hz for NYUSIM.

  • Takeaways & Limitations

    NYUSIM provides a practical alternative for analyzing MIMO channel conditions and spectral efficiency in 5G communication-system development and deployment.

Abstract

from arXiv · show

This paper presents details and applications of a novel channel simulation software named NYUSIM, which can be used to generate realistic temporal and spatial channel responses to support realistic physical- and link-layer simulations and design for fifth-generation (5G) cellular communications. NYUSIM is built upon the statistical spatial channel model for broadband millimeter-wave (mmWave) wireless communication systems developed by researchers at New York University (NYU). The simulator is applicable for a wide range of carrier frequencies (500 MHz to 100 GHz), radio frequency (RF) bandwidths (0 to 800 MHz), antenna beamwidths (7 to 360 degrees for azimuth and 7 to 45 degrees for elevation), and operating scenarios (urban microcell, urban macrocell, and rural macrocell), and also incorporates multiple-input multiple-output (MIMO) antenna arrays at the transmitter and receiver. This paper also provides examples to demonstrate how to use NYUSIM for analyzing MIMO channel conditions and spectral efficiencies, which show that NYUSIM is an alternative and more realistic channel model compared to the 3rd Generation Partnership Project (3GPP) and other channel models for mmWave bands.

I. INTRODUCTION

NYUSIM is an open-source channel simulator designed to support realistic wireless-system evaluation using measurement-based temporal and spatial channel models. It targets diverse mmWave environments and antenna configurations while improving accessibility through a platform-independent GUI.

  • Motivation: Channel simulators are important because they support communications-system evaluation and network-deployment simulation before new technologies are deployed.
  • Simulator contribution: NYUSIM was developed from extensive real-world wideband measurements at 28–73 GHz across UMi, UMa, and RMa outdoor environments.The simulator models both temporal and spatial channel impulse responses and measured signal levels.
  • Simulator contribution: The simulator supports carrier frequencies from 500 MHz to 100 GHz and RF bandwidths from 0 to 800 MHz.
  • Simulator contribution: A platform-independent GUI allows Windows and Macintosh users to run the MATLAB-written simulator without MATLAB installed.
  • Simulator contribution: NYUSIM generates temporal and spatial CIR samples from both omnidirectional and directional channel models for MIMO antenna-array design.Directional models help analyze spatial diversity and beamforming gain.

A. Path Loss Model

NYUSIM uses a close-in path-loss model referenced to 1 m, combining frequency-dependent free-space loss, distance scaling, atmospheric attenuation, and shadow fading. The model emphasizes physical interpretability and supports atmospheric effects across broad frequency and distance ranges.

  • Model formulation: NYUSIM employs a 1 m close-in free-space reference path-loss model with an atmospheric attenuation term.The model includes path-loss exponent, atmospheric attenuation, and shadow-fading components.
  • Atmospheric effects: Atmospheric attenuation aggregates the effects of dry air, water vapor, rain, and haze over 1–100 GHz.Example attenuation values use 1013.25 mbar pressure, 80% humidity, 20°C temperature, and 5 mm/hr rain.
  • Model properties: The CI model embeds frequency dependence in the free-space path-loss term, making it applicable from below 1 GHz to above 100 GHz.
  • Model properties: Compared with the ABG model, the CI model uses fewer parameters while offering physical appeal, parameter stability, and prediction performance across frequencies, distances, and scenarios.
  • Model parameters: NYUSIM Version 1.4 uses environment- and scenario-specific path-loss exponents and shadow-fading standard deviations, with user-adjustable parameters in MATLAB.

B. Wideband Temporal/Spatial Clustering Algorithm

NYUSIM’s statistical spatial channel model represents broadband channels through time clusters and spatial lobes derived from mmWave field observations. Its smaller, measurement-based cluster counts contrast with the larger 3GPP configuration that can distort simulated channel rank and spectral-efficiency predictions.

  • B. Wideband Temporal/Spatial Clustering Algorithm: NYUSIM’s SSCM models omnidirectional CIRs and joint AOD/AOA power spectra using time clusters and spatial lobes.Time clusters contain closely timed MPCs, while spatial lobes represent dominant arrival or departure directions.
  • B. Wideband Temporal/Spatial Clustering Algorithm: 3GPP TR 38.900 uses up to 12 LOS and 19 NLOS clusters in the UMi street-canyon scenario, which the paper describes as unsupported by mmWave measurements.
  • B. Wideband Temporal/Spatial Clustering Algorithm: NYUSIM uses 1–6 time clusters and a mean of about 2 spatial lobes, upper-bounded by 5, based on field observations.
  • B. Wideband Temporal/Spatial Clustering Algorithm: The larger 3GPP cluster count produces higher simulated channel rank and unrealistic eigen-channel distributions, yielding inaccurate mmWave spectral-efficiency predictions.

III. GRAPHICAL USER INTERFACE AND SIMULATOR BASICS

NYUSIM provides a GUI for Monte Carlo generation of channel impulse responses at user-specified transmitter–receiver distances. Users control the number of samples and separation-distance range for simulations.

  • III. GRAPHICAL USER INTERFACE AND SIMULATOR BASICS: NYUSIM’s GUI runs Monte Carlo simulations that generate CIR samples at specified T-R separation distances.
  • III. GRAPHICAL USER INTERFACE AND SIMULATOR BASICS: Users specify both the number of CIR samples and the range of T-R separation distances.

A. Input Parameters

NYUSIM exposes 28 input parameters organized into channel parameters and antenna properties for configuring propagation and transmitter/receiver arrays.

  • Input organization: 28 input parameters are grouped into Channel Parameters and Antenna Properties.Channel Parameters contains 16 propagation inputs, while Antenna Properties contains 12 transmitter and receiver antenna-array inputs.

1) Output Figure Files:

NYUSIM generates per-run channel visualizations and aggregates path-loss results across continuous runs, including omnidirectional and directional analyses.

  • Per-run and aggregate outputs: Five figures are generated for each simulation run, with an additional path-loss scatter plot after N continuous runs using identical inputs.The first run’s five figures and the aggregate figure are displayed for visual inspection.
  • Path-loss analysis: The path-loss scatter plot compares omnidirectional and directional values across the distance range from N continuous simulation runs.It also reports fitted path-loss exponent and shadow-fading standard deviation using the MMSE method.
  • Directional processing: Directional path loss is computed by searching TX/RX pointing angles in azimuth and elevation HPBW increments after generating the omnidirectional PDP.The directional path loss combines transmit power and antenna gains with directional received power; the example uses 10.9° azimuth and 8.6° elevation HPBWs at both ends.
  • Validation: The simulated path-loss exponent and shadow-fading standard deviation agree well with MMSE-fitted values.

2) Output Data Files:

NYUSIM exports per-simulation and aggregate text or MATLAB files containing angular spectra, power-delay profiles, channel parameters, and antenna-array measurements.

  • Per-run files: Each simulation run produces five paired .txt and .mat file sets for AOD spectra, AOA spectra, omnidirectional PDPs, directional PDPs, and small-scale PDPs.The filenames identify the simulation run with n and the spatial lobe with x where applicable.
  • Aggregate files: After N continuous runs, NYUSIM produces paired BasicParameters, OmniPDPInfo, and DirPDPInfo files.These aggregate files summarize inputs and omnidirectional or directional PDP information across runs.
  • Aggregate contents: BasicParameters files contain all input values, while OmniPDPInfo files record separation distance, received power, path loss, and RMS delay spread for each omnidirectional PDP.
  • Angular spectra: AOD lobe files contain path delay, path power, path phase, AOD, and ZOD for each resolvable multipath component.Corresponding AOA lobe files contain similar parameters for the 3D AOA power-spectrum output.
  • Power-delay profiles: OmniPDP files contain propagation time delay and received power, while SmallScalePDP files additionally include RX antenna separation in wavelengths.The values are stored in nanoseconds and dBm, with small-scale outputs spanning different RX antenna elements.

IV. APPLICATIONS OF NYUSIM

NYUSIM outputs support channel-response simulation, MIMO channel analysis, and BER simulation for wireless-system studies.

  • Applications: NYUSIM output figures and data files can be used to simulate mmWave channel impulse responses, investigate MIMO channel performance, and perform BER simulation.

A. MIMO Channel Condition Number

NYUSIM can generate OFDM MIMO channel coefficients from resolvable multipath parameters and evaluate channel condition numbers across sub-carriers. In one simulation, the 3×3 channel condition numbers averaged about 18 dB higher than those of the 2×2 channel, possibly because the matrix was rank deficient.

  • Condition-number interpretation: The condition number characterizes MIMO channel quality as the ratio of the largest to smallest singular values.Values near 0 dB indicate equal singular values and full rank, while values above 20 dB indicate a minimum singular value close to zero.
  • Channel-matrix construction: NYUSIM generates each OFDM sub-carrier’s Nr×Nt channel matrix from resolvable multipath components and their amplitudes, phases, delays, and angular parameters.The required parameters are extracted from DirPDPInfo.txt or DirPDPInfo.mat for the channel-coefficient calculation.
  • 2×2 simulation: 81 sub-carriers were simulated with a 10 MHz frequency interval across an 800 MHz bandwidth, revealing frequency-selective fading across antenna-element combinations.The four coefficient magnitudes varied with both sub-carrier frequency and transmit–receive antenna pairing.
  • 3×3 comparison: 1601 sub-carriers were simulated with a 500 kHz interval, then antenna counts and per-row elements were changed from 2×2 to 3×3.The remaining simulation input parameters were held constant while comparing the two antenna configurations.
  • 3×3 comparison: 18 dB: the 3×3 MIMO channel’s individual-sub-carrier condition numbers were higher on average than those of the 2×2 channel.The paper suggests that the relatively large 3×3 condition number may stem from matrix rank deficiency.

B. Spectral Efficiency Comparison Between 3GPP and NYUSIM Channel Models

The paper compares 3GPP and NYUSIM spectral-efficiency predictions for a 28 GHz, 256×16 mmWave MIMO system using hybrid beamforming. NYUSIM predicts substantially lower spectral efficiency at four data streams, which the paper attributes to its measurement-based dominant-cluster structure.

  • System setup: The comparison uses a 28 GHz single-cell, single-user UMi MIMO system with 256 base-station and 16 user antenna elements.The base station and user use uniform rectangular arrays, while SNR is fixed for spectral-efficiency evaluation.
  • Simulation procedure: 200 random channel realizations were generated for each model with transmitter–receiver distances from 10 to 435 m.The comparison evaluates the 3GPP TR 38.900 Release 14 and NYUSIM channel models under the same system study.
  • Spectral-efficiency results: At 20 dB SNR and Ns = 4, 3GPP achieves 40 bits/s/Hz versus about 27 bits/s/Hz for NYUSIM, a difference of about 13 bits/s/Hz.For Ns = 1, the 3GPP result is only slightly smaller than the NYUSIM result.
  • Spectral-efficiency results: NYUSIM produces one or two strong dominant clusters and weaker non-dominant clusters, whereas 3GPP distributes directional energy less realistically.The paper links this structural difference to the models’ divergent spectral-efficiency predictions.
  • Interpretation: The paper concludes that 3GPP is optimistic about mmWave diversity and achievable spectral efficiency, while NYUSIM provides more realistic predictions based on extensive measurements.The comparison is motivated by the 3GPP model’s larger cluster and ray counts, which the paper describes as unsupported by measurements.

V. CONCLUSION

The paper presents NYUSIM as an open-source simulator built from extensive broadband mmWave measurements and supporting diverse channel, antenna, and environmental configurations. Its examples analyze MIMO channel conditions and compare 3GPP with NYUSIM, positioning the simulator for 5G development and deployment.

  • Conclusion: NYUSIM recreates wideband power-delay profiles, channel impulse responses, and channel statistics across varied frequencies, bandwidths, antenna beamwidths, scenarios, and atmospheric conditions.The simulator includes a GUI and its simulated results match measured data.
  • Conclusion: More than 7000 downloads by corporations and universities worldwide are reported for NYUSIM.
  • Conclusion: Application examples examine channel conditions and compare MIMO channel performance between 3GPP and NYUSIM models.The simulator can also support other analyses beyond the examples presented.
  • Conclusion: NYUSIM is presented as useful for 5G communication system development and deployment.
Loading 1703.08232v2…