Source-linked AI summary

Probabilistic Omnidirectional Path Loss Models for Millimeter-Wave Outdoor Communications

Mathew K. Samimi, Theodore S. Rappaport, George R. MacCartney

arXiv:1503.07612v4cs.IT

TL;DR

Dense urban mmWave path loss modeling must account for whether propagation is LOS or NLOS, using information that varies with the site. This letter builds a probabilistic omnidirectional model from New York City measurements and ray-traced LOS probabilities, finding virtually no difference between two NLOS model choices.

  • Problem

    Predicting omnidirectional path loss in dense urban mmWave channels requires models based on LOS and NLOS propagation conditions for estimating network coverage and capacity.

  • Method

    The paper combines LOS free-space path loss with NLOS close-in or floating-intercept path loss and weights them using a site-specific LOS probability versus transmitter-receiver separation.

  • Results

    The probabilistic model gives virtually no difference when NLOS path loss uses either the 1 m close-in free-space-reference model or the floating-intercept model.

  • Takeaways & Limitations

    The models can estimate signal coverage, interference, and outage as functions of distance for future mmWave systems.

Abstract

from arXiv · show

This letter presents a probabilistic omnidirectional millimeter-wave path loss model based on real-world 28 GHz and 73 GHz measurements collected in New York City. The probabilistic path loss approach uses a free space line-of-sight propagation model, and for non-line-of-sight conditions uses either a close-in free space reference distance path loss model or a floating-intercept path loss model. The probabilistic model employs a weighting function that specifies the line-of-sight probability for a given transmitter-receiver separation distance. Results show that the probabilistic path loss model offers virtually identical results whether one uses a non-line-of-sight close-in free space reference distance path loss model, with a reference distance of 1 meter, or a floating-intercept path loss model. This letter also shows that site-specific environmental information may be used to yield the probabilistic weighting function for choosing between line-of-sight and non-line-of-sight conditions.

I. INTRODUCTION

The paper develops omnidirectional mmWave path loss models from New York City measurements, incorporating the site-specific probability of LOS at each transmitter-receiver separation. The resulting hybrid model makes the mean path loss probabilistic across LOS and NLOS conditions.

  • The study models omnidirectional path loss using unique-pointing-angle 28 GHz and 73 GHz measurements collected in New York City.
  • The approach extends conventional close-in free-space-reference and floating-intercept modeling by making the estimated mean path loss probabilistic.
  • The proposed hybrid model uses LOS and NLOS data while weighting their mean path losses by a site-specific LOS probability function of transmitter-receiver separation.

II. PROBABILITY OF LINE-OF-SIGHT

LOS probability is defined by whether buildings or other obstructions block a straight path between transmitter and receiver. The study derives this probability from a 3-D site-specific urban environment model and ray-tracing tests.

  • LOS denotes an unobstructed straight propagation path with zero reflections, whereas NLOS denotes obstruction involving scattering or one or more reflections.
  • LOS and NLOS probabilities depend strongly on the physical site-specific environment containing the transmitter and receiver.
  • The method represents nearby buildings as simple 3-D geometric shapes, exports the model to XML, and discretizes it numerically in MATLAB.
  • For each transmitter, the method samples 100 evenly spaced receiver positions around a circle at the target separation distance.

R Separation (m)

Ray-tracing estimates LOS probability across four New York City transmitter sites and transmitter-receiver distances, then fits a common probability curve and site-specific parameters.

  • Fig. 2 reports ray-traced LOS probability versus transmitter-receiver separation for COL1, COL2, KAU, and KIM2, plus their mean curve.
  • The LOS probability is computed over all possible receiver locations in the modeled environments rather than only the measured receiver locations.
  • KIM1 was excluded because its location near Washington Square Park produced unusually large LOS probabilities out to 150 m.
  • Table I summarizes dBP and α values obtained by MMSE for the four transmitter locations considered.
  • The MMSE fit to the mean LOS curve yielded dBP = 27 m and α = 71 m for the shared probability model.

III. PROBABILISTIC PATH LOSS MODEL

The model combines LOS and NLOS path-loss formulations by weighting them according to the probability of LOS at each transmitter-receiver distance. It supports either a 1 m close-in NLOS model or a floating-intercept NLOS model, which produce virtually identical probabilistic curves.

  • The floating-intercept model can produce nonrealistic path-loss results when extrapolated outside the measured transmitter-receiver separation range.
  • The hybrid model weighs LOS and NLOS path losses using a distance-dependent LOS probability for omnidirectional mmWave propagation.It combines a close-in free-space reference model for LOS with either a close-in or floating-intercept model for NLOS.
  • A fixed reference distance of d0 = 1 m provides a viable alternative for modeling both LOS and NLOS conditions with little loss of accuracy.
  • The LOS and NLOS formulations use path-loss exponents, floating-intercept parameters, and a lognormal random variable to represent large-scale shadowing.The shadowing variable has standard deviation σ in dB; α and β denote the floating-intercept model's intercept and slope.
  • 28 GHz and 73 GHz measurements used 74 transmitter-receiver location combinations at each frequency, while the plotted models compare close-in and floating-intercept NLOS formulations.
  • Virtually no difference appears between probabilistic curves using the 1 m close-in NLOS model and the floating-intercept NLOS model.The probabilistic mean remains between the LOS and NLOS models and follows the LOS data where LOS probability is 100%.

IV. CONCLUSION

The probabilistic model combines LOS and NLOS omnidirectional path-loss data using a site-specific LOS probability distribution. It produces virtually identical results with either NLOS model and can estimate coverage, interference, and outage as functions of distance.

  • Virtually identical results arise when the probabilistic model uses either the NLOS 1 m close-in reference-distance or floating-intercept model.The comparison is reported for the 73 GHz New York City models in Fig. 4.
  • The model uses a probability distribution for LOS derived from a 3-D site-specific database in New York City.
  • The models may estimate signal coverage, interference, and outage as functions of distance for future millimeter-wave systems.The paper highlights sensitivity to LOS conditions for highly directional antennas and shorter wavelengths.
Loading 1503.07612v4…