Public PhD defense

Hybrid computational methods
for 6G exposure assessment

Robin Wydaeghe

Supervisors

prof. dr. ir. Wout Joseph

prof. dr. ir. Emmeric Tanghe

At higher frequencies, absorption stays closer to the skin

ICNIRP limits: basic restrictions and reference levels

Method: ray tracing with FDTD or the surface law

Setup: FDTD on two adult and two child phantoms

Duke (34), Ella (26), Eartha (8), Thelonious (6)
550 simulations: all four up to 15 GHz, Thelonious also at 26 GHz
Plane waves at 1 W/m², 6 directions × 2 polarizations below 6 GHz
Focused case: four waves phased to meet at the phone

Chapter 2

Children’s whole-body SAR is 1.5 to 1.9 times adults’

Above 6 GHz, skin SAR increases and eye SAR decreases

Setup: 25 antennas in a factory hall at 3.5 and 28 GHz

50 × 50 × 10 m hall, 100 random machine layouts per case
25 antennas, collocated (0.5 λ apart) or distributed (10 m apart), equal total power
MRT (maximum-ratio transmission): antenna weights chosen for maximum signal at the phone
Phone 2 cm behind the right ear of the MIDA head model

Chapter 3

Method: a Huygens box couples ray tracing to FDTD

Fields add before squaring, which creates a cross term

Results: distributed antennas lower exposure 2 to 3 times

Setup: a six-minute walk through New York at 28 GHz

World Trade Center to Liberty Street, phone at the right ear
City mesh from photogrammetry, six material classes
Ray tracing for the buildings, QuaDRiGa for the fine detail, 316 snapshots
Two-stage MRT: +10 dB at the phone against a person 5 m away
Huygens box around the ear, FDTD inside

Chapter 4

Results: a hotspot under 5 mm wide persists beside the ear

Setup: measurements in an anechoic room and a stairwell

Anechoic room, 26 GHz: four sets of four antennas aimed at one point
Probe on a robot: 50 × 50 cm plane, 3.33 mm steps, 50 hours
Stairwell, 28 GHz: five planes, 31 hours
Digital twin calibrated on three planes, tested blind on a fourth, in four rotations

Chapter 6

Results: the simulated hotspot agrees with the measured one

Results: interference spacings correlate at 0.55 to 0.79

Each new body, posture or frequency needs a new simulation

Method: closed-form absorption on the skin surface

Above 6 GHz, absorption stays in the top millimeters of skin
One closed-form expression per surface triangle, no volume, no time steps
Thelonious: 23,826 triangles, 0.787 m² of skin
All triangles at once on one GPU

One skin triangle with its normal and the wave direction

Chapter 5

The surface law multiplies transmission, angle and visibility

transmission into the skin skin normal, wave direction zero where the body shadows itself

Results: self-shading reduces the mean absorption by 13.5%

Validation: the law agrees with FDTD and five studies to 100 GHz

Setup: 746,496 body maps in the factory, 8 to 100 GHz

Four phantoms, eight frequencies, 16 rooms with and without line of sight
Arrays of 16, 64 and 256 antennas, phone 6 to 22 m away
MRT: 21 phone positions on the body. ECBF: six absorption limits
2.2 hours on one GPU, about 10 ms per map

Chapter 5

Method: the same rays give the signal and the absorption matrix Q

Results: beamforming cuts absorption 6.3-fold at half signal

Setup: walking routes in ten cities at 15 GHz

Ten routes, 163 points, a 250 m city crop around each route
Rays traced backwards from the body into the surface law
Duke: 72.4 kg, 56,024 surface triangles
Whole-body SAR per transmitter density ρ_A and per watt of EIRP

Chapter 6

Results: median SAR differs 14-fold between city routes

Synthesis: the method each chapter uses at each scale

Conclusions

Far below the ICNIRP limits: at most 4% indoors, under 0.05% outdoors
Child whole-body SAR 1.5–1.9 times the adult value below 6 GHz
Distributed antennas lower exposure 2–3 times
Median SAR 14 times higher on the Mexico City route than in Milan
Surface law ÷ FDTD = 1.01 in whole-body power, 746,496 body maps in 2.2 hours
Beamforming against Q: half the signal, 6.3 times less absorbed

Outlook

Near field: phones and wearables on the body
Outdoor measurement campaign for the city model
Many bodies in one beamformer, dosimetry inside the network loop, about 10 ms per body
Child phantoms for 7.125–8.4 GHz, the band under study for WRC-27

Thank you

Robin Wydaeghe

https://phd.rwydaeghe.be/talk/

Hybrid computational methods
for 6G exposure assessment

At higher frequencies, absorption stays closer to the skin

ICNIRP limits: basic restrictions and reference levels

Method: ray tracing with FDTD or the surface law

Setup: FDTD on two adult and two child phantoms

Children’s whole-body SAR is 1.5 to 1.9 times adults’

Above 6 GHz, skin SAR increases and eye SAR decreases

Setup: 25 antennas in a factory hall at 3.5 and 28 GHz

Method: a Huygens box couples ray tracing to FDTD

Fields add before squaring, which creates a cross term

Results: distributed antennas lower exposure 2 to 3 times

Setup: a six-minute walk through New York at 28 GHz

Results: a hotspot under 5 mm wide persists beside the ear

Setup: measurements in an anechoic room and a stairwell

Results: the simulated hotspot agrees with the measured one

Results: interference spacings correlate at 0.55 to 0.79

Each new body, posture or frequency needs a new simulation

Method: closed-form absorption on the skin surface

The surface law multiplies transmission, angle and visibility

Results: self-shading reduces the mean absorption by 13.5%

Validation: the law agrees with FDTD and five studies to 100 GHz

Setup: 746,496 body maps in the factory, 8 to 100 GHz

Method: the same rays give the signal and the absorption matrix Q

Results: beamforming cuts absorption 6.3-fold at half signal

Setup: walking routes in ten cities at 15 GHz

Results: median SAR differs 14-fold between city routes

Synthesis: the method each chapter uses at each scale

Conclusions

Outlook

Thank you

Focused APD against the restriction, 7 to 15 GHz

Exposure per radiated watt and per incident watt

Distributed antennas even out exposure along the walk

Which restriction binds first

Surface law error budget at 28 GHz

Absorption maps at 28 GHz under MRT and ECBF

The held-out stairwell plane

Hybrid computational methodsfor 6G exposure assessment

At higher frequencies, absorption stays closer to the skin

ICNIRP limits: basic restrictions and reference levels

Method: ray tracing with FDTD or the surface law

Setup: FDTD on two adult and two child phantoms

Children’s whole-body SAR is 1.5 to 1.9 times adults’

Above 6 GHz, skin SAR increases and eye SAR decreases

Setup: 25 antennas in a factory hall at 3.5 and 28 GHz

Method: a Huygens box couples ray tracing to FDTD

Fields add before squaring, which creates a cross term

Results: distributed antennas lower exposure 2 to 3 times

Setup: a six-minute walk through New York at 28 GHz

Results: a hotspot under 5 mm wide persists beside the ear

Setup: measurements in an anechoic room and a stairwell

Results: the simulated hotspot agrees with the measured one

Results: interference spacings correlate at 0.55 to 0.79

Each new body, posture or frequency needs a new simulation

Method: closed-form absorption on the skin surface

The surface law multiplies transmission, angle and visibility

Results: self-shading reduces the mean absorption by 13.5%

Validation: the law agrees with FDTD and five studies to 100 GHz

Setup: 746,496 body maps in the factory, 8 to 100 GHz

Method: the same rays give the signal and the absorption matrix Q

Results: beamforming cuts absorption 6.3-fold at half signal

Setup: walking routes in ten cities at 15 GHz

Results: median SAR differs 14-fold between city routes

Synthesis: the method each chapter uses at each scale

Conclusions

Outlook

Thank you

Focused APD against the restriction, 7 to 15 GHz

Exposure per radiated watt and per incident watt

Distributed antennas even out exposure along the walk

Which restriction binds first

Surface law error budget at 28 GHz

Absorption maps at 28 GHz under MRT and ECBF

The held-out stairwell plane

Hybrid computational methods
for 6G exposure assessment