PaperlayerView exact revision on arXiv

Source-linked AI summary

Millimeter Wave and sub-THz Indoor Radio Propagation Channel Measurements, Models, and Comparisons in an Office Environment

Yunchou Xing, Theodore S. Rappaport, Amitava Ghosh

arXiv:2103.00385v3cs.ITmath.IT

TL;DR

Indoor sub-THz propagation evidence at commercially relevant distances and its relationship to lower-frequency channels have been limited. This paper compares measurements across 28, 73, and 142 GHz in the same office, finding similar path loss exponents and decreasing multipath time dispersion at higher frequencies.

  • Problem

    Extensive indoor measurements above 100 GHz in the same environment and at commercially relevant distances were lacking, leaving the relationship between sub-THz and lower-frequency channels uncertain.

    Source: §I · p. 1

  • Method

    The paper compares indoor office propagation measurements and channel statistics at 28, 73, and 142 GHz using the same environment and locations across bands.

    Source: §I · p. 1

  • Results

    Path loss exponents are remarkably similar across 28, 73, and 142 GHz for LOS and NLOS with a 1 m free-space reference, while omnidirectional LOS mean RMS delay spread decreases from 11 ns at 28 GHz to 6 ns at 73 GHz and 3 ns at 142 GHz.

    Source: B. Path Loss Models at 28, 73, and 142… · p. 3, A. RMS Delay Spread · p. 4

  • Takeaways & Limitations

    Multipath time dispersion becomes less prominent at higher frequencies, potentially making multipath diversity harder to exploit without additive reflecting surfaces.

    Source: §I · p. 1

Abstract

from arXiv · show

This letter provides a comparison of indoor radio propagation measurements and corresponding channel statistics at 28, 73, and 140 GHz, based on extensive measurements from 2014-2020 in an indoor office environment. Side-by-side comparisons of propagation characteristics (e.g., large-scale path loss and multipath time dispersion) across a wide range of frequencies from the low millimeter wave band of 28 GHz to the sub-THz band of 140 GHz illustrate the key similarities and differences in indoor wireless channels. The measurements and models show remarkably similar path loss exponents over frequencies in both line-of-sight (LOS) and non-LOS (NLOS) scenarios, when using a one meter free space reference distance, while the multipath time dispersion becomes smaller at higher frequencies. The 3GPP indoor channel model overestimates the large-scale path loss and has unrealistic large numbers of clusters and multipath components per cluster compared to the measured channel statistics in this letter.

I. INTRODUCTION

The paper addresses limited evidence on indoor sub-THz propagation and its relationship to lower-frequency channels. It compares measurements across 28, 73, and 142 GHz in the same office, finding similar path loss exponents and less multipath time dispersion at higher frequencies.

I. INTRODUCTION

  • Indoor channels above 100 GHz lacked extensive same-environment measurements at commercially relevant distances, leaving their relationship to lower-frequency channels uncertain.Prior above-100-GHz work focused on LOS or measurements at close ranges; this paper notes the need for comparisons over many tens of meters.Source: §I · p. 1
  • Knowledge of channel variation across wide frequency ranges supports frequency-dependent models for multiband indoor wireless design.The paper identifies applications including intelligent reflecting surfaces and precise position location.Source: §I · p. 1
  • Measurements at 28, 73, and 142 GHz in the same office show remarkably similar path loss exponents for LOS and NLOS when referenced to 1 m free-space loss.The comparison uses three measurement campaigns in the NYU WIRELESS center, with coverage distances up to 40 m.Source: §I · p. 1
  • Multipath time dispersion becomes less prominent at higher frequencies, potentially making multipath diversity harder to exploit without additive reflecting surfaces.Source: §I · p. 1

II. INDOOR RADIO PROPAGATION MEASUREMENTS AND PATH LOSS MODELS AT 28, 73, AND 142 GHZ

Deployment above 100 GHz faces hardware and propagation challenges, including high phase noise, Doppler spread, limited output power, and compact-array fabrication. Higher-gain directional antennas were used at higher frequencies to compensate for receiver noise and greater first-meter path loss.

II. INDOOR RADIO PROPAGATION MEASUREMENTS AND PATH LOSS MODELS AT 28, 73, AND 142 GHZ

  • Above 100 GHz, high phase noise, Doppler spread, limited output power, and compact phased-array fabrication challenge network deployment.Source: §II · p. 1
  • Understanding channel characteristics above 100 GHz is described as critical to addressing deployment challenges for future wireless networks.Source: §II · p. 1
  • Higher-gain, more directional antennas at both link ends compensate for receiver noise and greater path loss in the first meter at higher frequencies.Received power does not change with signal bandwidth, while receiver noise power and noise figure do change.Source: Table I · p. 2

A. Measurement Systems and Environments

The study compares indoor hotspot measurements at 28, 73, and 142 GHz collected over six years, using the same office locations across bands. Sliding-correlation sounders and steerable horn antennas provide dual-directional power-delay profiles.

A. Measurement Systems and Environments

B. Path Loss Models at 28, 73, and 142 GHz

Across 28, 73, and 142 GHz, the 1 m-reference CI and CIF models have remarkably similar path loss exponents. The measured directional and omnidirectional results show frequency- and antenna-pattern-dependent differences, particularly in LOS and NLOS loss.

B. Path Loss Models at 28, 73, and 142 GHz

  • Directional LOS PLEs are 1.90, 1.63, and 2.05 at 28, 73, and 142 GHz; 142 GHz has 2–4 dB more loss at 10 m than 28 and 73 GHz.At 40 m, 142 GHz has 3–7 dB more average loss; the lower-frequency PLEs below 2 are attributed to waveguiding.Source: B. Path Loss Models at 28, 73, and 142… · p. 2
  • Directional NLOSBest PLEs are 2.75, 3.30, and 3.21 at 28, 73, and 142 GHz, with 73 and 142 GHz more lossy than 28 GHz.Reduced penetration and stronger reflection appear to contribute to the slight 28-to-142 GHz difference; measured scattered power is more than 20 dB below reflected signal power for most indoor materials.Source: B. Path Loss Models at 28, 73, and 142… · p. 2
  • At 142 GHz, omnidirectional PLEs are 1.74 in LOS and 2.83 in NLOS, versus directional LOS 2.05 and NLOSBest 3.21.Omnidirectional synthesis captures energy across the horizon, but lower antenna gain shortens practical link range; accurate beamforming is needed to maintain indoor NLOS links above 100 GHz.Source: B. Path Loss Models at 28, 73, and 142… · pp. 2–3
  • Omnidirectional CI PLEs (n, σ) are (1.17, 2.72 dB), (1.36, 2.30 dB), and (1.74, 3.62 dB) at 28, 73, and 142 GHz in LOS.In NLOS, the corresponding values are (2.37, 7.22 dB), (2.81, 8.71 dB), and (2.83, 6.07 dB); 28 GHz is less lossy than the higher bands.Source: B. Path Loss Models at 28, 73, and 142… · pp. 2–3
  • At f0 = 81 GHz, omnidirectional CIF fits across 28, 73, and 142 GHz yielded n_CIF = 0.29 and σ_CIF = 2.94 dB for LOS, and n_CIF = 0.11 and σ_CIF = 7.53 dB for NLOS.CI and CIF path loss models show remarkable similarity in path loss exponents across those bands.Source: B. Path Loss Models at 28, 73, and 142… · p. 3
  • Across 28, 73, and 142 GHz, CI and CIF path loss exponents are remarkably similar with a 1 m free-space reference; frequency dependence may be confined to first-meter loss.The paper suggests a single PLE may model path loss across these bands.Source: B. Path Loss Models at 28, 73, and 142… · p. 3, Figure 1 · p. 3

III. MULTIPATH STATISTICS AT 28, 73, AND 142 GHZ

Table III organizes omnidirectional InH-Office channel statistics across 28, 73, and 142 GHz alongside the 3GPP InH-Office standard.

III. MULTIPATH STATISTICS AT 28, 73, AND 142 GHZ

  • Table III covers CI and CIF large-scale path-loss model parameters across 28, 73, and 142 GHz and the 3GPP InH-Office standard.Source: Table III · p. 4
  • The comparison includes RMS delay spread for the three measured frequency bands and the 3GPP InH-Office standard.Source: Table III · p. 4
  • Table III also compares the number of clusters and multipath components per cluster across the measured bands and the standard.Source: Table III · p. 4

A. RMS Delay Spread

Measured RMS delay spread varies with frequency and propagation scenario: it declines across bands in omnidirectional LOS and NLOS channels, while directional NLOS maxima are also smaller at higher frequencies.

A. RMS Delay Spread

  • Multipath components were retained using a 5 dB SNR threshold above the raw PDP's mean thermal noise floor, with peaks counted when adjacent qualifying peaks exceeded sounder time resolution.Source: A. RMS Delay Spread · p. 4
  • Directional LOS RMS delay spread has a 3–4 ns mean across the three bands, close to the probing-signal width of 2.5 ns at 28/73 GHz and 2 ns at 142 GHz.Reflections from walls, ceiling, and floor add excess delays within the original pulse width.Source: A. RMS Delay Spread · p. 4
  • Directional NLOS maximum RMS delay spread is 44 ns at 28 GHz, 31 ns at 73 GHz, and 11 ns at 142 GHz; 90% of values are below 40, 25, and 9 ns, respectively.Source: A. RMS Delay Spread · p. 4
  • 11, 6, and 3 ns are the measured mean RMS delay spreads in omnidirectional LOS at 28, 73, and 142 GHz, while 3GPP predicts 20.40 ns and 20.21 ns at 28 and 73 GHz.The measured mean falls 40% from 28 to 73 GHz and another 50% at 142 GHz; 3GPP does not specify RMS delay spread above 100 GHz.Source: A. RMS Delay Spread · p. 4
  • In omnidirectional NLOS, mean RMS delay spread is 17, 12, and 9 ns at 28, 73, and 142 GHz, respectively, exceeding LOS means by 60%, 100%, and 200%.Higher first-meter free-space and partition losses can leave weaker, long-delay components below the sounder noise floor, contributing to the lower 142-GHz RMS delay spread.Source: A. RMS Delay Spread · p. 4

B. The Numbers of Time Clusters and MPCs per Cluster

Time-cluster and multipath-component statistics depend on the chosen inter-cluster interval, and the measured omnidirectional channels become sparser at higher frequencies.

B. The Numbers of Time Clusters and MPCs per Cluster

  • The analysis uses a 6 ns minimum inter-cluster time void interval, selected based on the 2 m corridor width; larger intervals yield fewer clusters but more MPCs per cluster.Source: B. The Numbers of Time Clusters and… · p. 4
  • Mean omnidirectional time-cluster counts decline with frequency: LOS has 4.60, 2.76, and 1.90 clusters, while NLOS has 5.40, 3.20, and 2.80 at 28, 73, and 142 GHz.Cluster counts follow a Poisson distribution at all three frequencies; higher-frequency partition loss leaves weaker, longer-delay components below the noise floor.Source: B. The Numbers of Time Clusters and… · pp. 4–5
  • The number of MPCs per cluster follows a composite delta function δ(n −1) and discrete exponential distribution at all three frequencies.Source: B. The Numbers of Time Clusters and… · p. 5
  • At 28 GHz, NLOS has more mean MPCs per cluster than LOS, but this relation reverses at 73 and 142 GHz as partition loss makes higher-frequency NLOS channels sparser.Source: B. The Numbers of Time Clusters and… · p. 5

IV. CONCLUSION

Indoor path-loss exponents are remarkably similar across 28, 73, and 142 GHz in both LOS and NLOS, while multipath time dispersion and distribution means decrease at higher frequencies.

IV. CONCLUSION

  • 28, 73, and 142 GHz show remarkably similar path-loss exponents in both LOS and NLOS when referenced to the first-meter free-space distance.The reported similarity excludes first-meter path loss, where energy spreads into the far field.Source: §IV · p. 5
  • THz channels resemble today’s mmWave propagation channels except for first-meter path loss as energy spreads into the far field.Source: §IV · p. 5
  • Multipath time dispersion is much smaller at higher frequencies, and distribution means for cluster count, RMS delay spread, and subpaths per cluster decrease with frequency.Source: §IV · p. 5
  • The results differ from existing 3GPP InH prediction models and support frequency-dependent indoor office channel models spanning mmWave and THz bands.Source: §IV · p. 5
Loading 2103.00385v3…