Source-linked AI summary
Millimeter-Wave Human Blockage at 73 GHz with a Simple Double Knife-Edge Diffraction Model and Extension for Directional Antennas
George R. MacCartney, Sijia Deng, Shu Sun, Theodore S. Rappaport
TL;DR
The paper measures how humans block 73 GHz directional links and models the effect with double knife-edge diffraction. It finds deep attenuation and shows that antenna directivity must be included when characterizing and mitigating blockage.
Problem
Directional millimeter-wave systems are more vulnerable to human blockage because narrow beams and reduced diffraction limit angular diversity around obstacles.
Method
The paper conducts 73 GHz measurements on a 5 m indoor link and compares them with a DKED model representing the human as a rectangular screen, extended to account for antenna radiation patterns.
Results
30–40 dB or greater attenuation was observed with directional antennas, while the modified DKED METIS model showed good agreement with measurements by accounting for antenna directivity.
Takeaways & Limitations
Beam direction switching can help maintain reliable millimeter-wave connections during human blockage, and the DKED model can be incorporated into ray-tracers to simulate obstruction losses.
Abstract
from arXiv · showhide
This paper presents 73 GHz human blockage measurements for a point-to-point link with a 5 m transmitter-receiver separation distance in an indoor environment, with a human that walked at a speed of approximately 1 m/s at a perpendicular orientation to the line between the transmitter and receiver, at various distances between them. The experiment measures the shadowing effect of a moving human body when using directional antennas at the transmitter and receiver for millimeter-wave radio communications. The measurements were conducted using a 500 Megachips-per-second wideband correlator channel sounder with a 1 GHz first null-to-null RF bandwidth. Results indicate high shadowing attenuation is not just due to the human blocker but also is due to the static directional nature of the antennas used, leading to the need for phased-array antennas to switch beam directions in the presence of obstructions and blockages at millimeter-waves. A simple model for human blockage is provided based on the double knife-edge diffraction (DKED) model where humans are approximated by a rectangular screen with infinite vertical height, similar to the human blockage model given by the METIS project.
I. INTRODUCTION
Millimeter-wave systems need narrow beams to maintain SNR, but their directionality increases vulnerability to human blockage. This paper presents 73 GHz measurements over a 5 m indoor link and a DKED-based model extended for directional antennas.
- Narrow beamforming is likely needed at mmWaves because increased first-meter free-space path loss reduces received SNR.
- Directional mmWave antennas are more susceptible to blockage from humans, cars, and street furniture than omnidirectional or quasi-omnidirectional systems.
- Prior measurements at 5 and 60 GHz reported substantial human-induced attenuation, including 20–40 dB shadowing in some point-to-point links.
- Existing blockage models include DKED, ray-tracing with knife-edge diffraction, UTD, and PWL approximations, with differing agreement against measurements.
- The paper measures 73 GHz human blockage on a 5 m indoor point-to-point link using directional high-gain horn antennas and develops a modified METIS DKED comparison model.
A. Measurement Equipment and Specifications
The measurement system uses a wideband correlator channel sounder centered at 73.5 GHz and directional 20 dBi horn antennas. It records rapid successive power-delay profiles for blockage analysis.
- The sounder transmits a 500 Megachips-per-second PN sequence, producing a 1 GHz null-to-null RF bandwidth centered at 73.5 GHz.
- The TX and RX each use a 20 dBi horn antenna with 15° azimuth and elevation half-power beamwidths.
- Up to 41,000 periodic PDPs are recorded per snapshot, with a minimum capture interval of 32.752 µs.
- The TX and RX channel sounders share a cable-connected 10 MHz reference for frequency synchronization.
B. Measurement Environment and Descriptions
Measurements use a 5 m boresight-to-boresight indoor link with antennas at 1.4 m height. Nine five-second snapshots capture a human walking through the link at 500 PDPs per second.
- The TX and RX are separated by 5 m, positioned at 1.4 m height, and oriented boresight-to-boresight.
- Nine separate 5-second snapshots were recorded while a human blocker walked through the point-to-point link.
- Measurements used a 500 Hz PDP interval frequency, yielding approximately 2500 PDPs per snapshot without averaging.
III. DOUBLE KNIFE-EDGE DIFFRACTION BLOCKAGE MODEL
The paper models a human blocker as a rectangular screen using double knife-edge diffraction, then extends the model to account for directional antenna radiation patterns and off-boresight paths.
- Screen approximation: The DKED model represents a human blocker as a screen with four edges or as an infinitely vertical screen with two side edges.The simplified model retains side-edge diffraction while treating screen height as infinite.
- Screen approximation: The screen remains perpendicular to the TX-RX link as it moves between the transmitter and receiver.Its dimensions represent blocker depth and height in the top-down and side projections.
- Diffraction formulation: The model computes diffraction from the screen edges using projected TX-screen and screen-RX distances, carrier wavelength, screen width, and height.The four edges are labeled w1, w2, h1, and h2, with separate top-down and side-view geometries.
- Diffraction formulation: For LOS and NLOS conditions, the model applies different edge-sign choices to determine the relevant diffraction contributions.NLOS uses a positive sign for both edges, while LOS treats the edge farthest from the link as being in the shadow zone.
- Directional-antenna extension: The original DKED loss model is accurate for omnidirectional antennas but omits the reduced angular diversity of directive antennas.When the boresight LOS path is blocked, projected paths to and from the screen do not retain the antennas’ full directive gain.
- Directional-antenna extension: The modified model weights knife-edge diffraction fields by normalized directional antenna gains at the projected off-boresight angles.This produces larger shadow-region losses than an omnidirectional model because the trajectories move away from boresight.
IV. MEASUREMENT RESULTS AND ANALYSIS
At 73 GHz, human blockage produced deep, distance-sensitive shadowing, while accounting for directional antenna patterns substantially improved DKED agreement with measurements. The simplified model omits screen height edges but can support practical blockage simulation.
- Measurement procedure: The measurement pipeline summed received-power contributions across approximately 2500 PDPs for each case while a human walked perpendicularly through the link.The experiments used an approximately 1 m/s walking speed and evaluated nine measurement cases.
- Measured blockage: Greater than 30 dB of maximum blockage attenuation occurred in the shadow region for all nine measurement cases.The approximately 4 mm wavelength produced deep oscillating fades of 30–40 dB or greater through constructive and destructive interference.
- Model comparison: The modified DKED METIS model, incorporating the 15° HPBW transmit and receive antennas, agreed well with measurements and traced the upper attenuation envelope.The original model significantly underestimated shadowing when the blocker was close to either terminal, by more than 20 dB in extreme cases.
- Model comparison: Equivalent directional antennas produced approximately symmetric shadowing durations and fade depths when the blocker was placed at equivalent distances from the transmitter or receiver.This symmetry was observed for the 0.5 m and 1 m terminal-offset pairs.
- Model scope and use: The DKED simulations ignored the screen's height edges, although the model can be extended with phase correction and non-perpendicular screen orientations.The paper presents the simple model as suitable for software implementation and ray-tracing of human blockage and similar obstructions.
V. CONCLUSION
The paper reports 73 GHz human blockage measurements over a 5 m link and compares them with simple and antenna-aware DKED models. Directional antennas produced large, long-lasting fades, while the modified model matched measurements and beam switching was identified as a practical response.
- Conclusion: 73 GHz directional-antenna measurements showed that a 0.28 m-deep human can cause 30–40 dB or more of attenuation.The measurements used perpendicular blocker paths across a 5 m transmitter–receiver link.
- Conclusion: The modified DKED METIS model showed good agreement with measurements by accounting for antenna directivity.The model represented the human as an infinitely vertical screen.
- Conclusion: Shadowing events lasted approximately 200–300 ms, and equivalent transmitter- and receiver-side blocker positions produced symmetric durations and fade depths.The conclusion identifies beamforming and phased-array beam switching as ways to seek reflections and avoid blockages.