A Model-Based Approach for Realizing a Safe Wireless Biotelemetry System Kerron Duncan Advisor: Dr. Ralph Etienne-Cummings Johns Hopkins University Department of Electrical and Computer Engineering PhD candidate, John Hopkins University
A Model-Based Approach for
Realizing a Safe Wireless
Biotelemetry System
Kerron Duncan
Advisor: Dr. Ralph Etienne-Cummings
Johns Hopkins University
Department of Electrical and Computer Engineering
PhD candidate, John Hopkins University
OverviewA Model-Based Framework for Trade Space Exploration
• Development of a wireless biotelemetry communications system model thatenables early architecture trades and decision making for safety, performanceand power efficiency.
• This integrated system model is constructed using a model-based systemsengineering (MBSE) framework which enables the digital integration ofrequirements, architecture and disparate performance models to determinecandidate solutions for an implanted communication system.
• In the conceptual and early design phases of a biotelemetry system one mustconsider the constraints of implanted wireless devices in humans and animals,as well as those imposed on short range, high data rate wirelesscommunications.
• The integrated model greatly improves the ability of the designer to quicklyanalyze thousands of system permutations while considering systemrequirements.
2
What Is Biotelemetry?
• Measuring, recording and monitoring of real time physiological
parameters such as ECG, EOG, EMG and neural spikes, over
some distance via radio-frequency signals
• Why is it important ?
– Health Care, Neuroscience & Brain Machine Interfaces
• Why implant ?
– Reduce infections
– Higher fidelity3
Rx
Medical Care Advances
Family
Doctors
Router/Receiver
Implanted
Transmitter
Cloud
Un-tethered
Neuroscience
Experiments(eliminates wires, etc.)
Source: Guevremont et al, Neurophysiology 2007
Wireless Biotelemetry Standards
• Wireless Medical Telemetry Services (WMTS)
– On-Body Devices: Transmits at periodic intervals, short duration, low power consumption
– 608-614 MHz, 1395-1400 MHz & 1427-1432 MHz
• MedRadio ( was Medical Implant Communications Service (MICS) )
– In-Body Devices: Propagation characteristics conducive to transmission of RF through the body
– Little to no interference with other RF in band, compatible internationally
– Bandwidth: 401-406 MHz
• Industrial, Scientific & Medical (ISM)
– In-Body Devices: Used in conjunction with MedRadio devices, band differs internationally slightly
– Bandwidth : 433 MHz, 2.4-2.48 GHz, 5.725 GHz
• Ultra-wideband (UWB)
– In-Body and On-body devices
– Higher frequencies allow smaller antenna designs, higher data rates, but more losses
– Bandwidth : 3.1 – 10.6 GHz.
4
Wireless Standard Comparison Leads us to UWB
• UWB allows
– Wide bandwidth leads to higher data rates
– Low EIRP requirements which leads to low power consumption and dissipation
– Higher operational frequency leads to smaller possible antenna designs for ease of implantation
– Small penetration depth leads to short range comm. or implanted a few mm from skin surface but acceptable for most applications
5
Wireless Standard Bandwidth EIRP Wavelength (m)
Penetration
Depth (mm)MedRadio 5 MHz 25 uW 0.743 52
ISM 80 MHz 4 W 0.123 22.4
UWB 7500 MHz 556 uW 0.044 6
EIRP:
Equivalent Isotropic
Radiated Power
What is Ultra-wideband (UWB)?
6
(FCC Spectral Mask)
Source: Wang, EMC 2007
BW
of interest
Gaussian derivatives are used-41.3 dBm/MHz
or 556 uW
5th derivative Gaussian fits UWB spectral mask
7
UWB allows full occupation or sub-bands
Multiband UWB
Uses Sub-bands
IR-UWB signal
shaped for BW
Source: Gharpurey, UWB Circuits, Transceivers and Systems
The “Hostile” Implanted Environment Has Been Studied
81 2 3 4 5 6 7 8 9 10
x 109
0
10
20
30
40
50
60
70
Frequency (Hz)
Rela
tive P
erm
ittivity
Relative Permittivity vs. Frequency for Biological Materials
wet skin
dry skin
fat
grey matter
white matter
blood
cancellous bone
cortical bone
bone marrow
muscle
High Permittivity
(dielectric “constant”)
Conductivity
402 MHz Data
UWB Range Data
• We must understand the electrical characteristics
of biological material as they vary over frequency
• Dispersive
• High εr (muscle ~50, air is 1 )
• Conductive (blood ~ 1.35, sea water is 4.8)
Source: Jaehoon, MTT 2004Source: Gabriel, Phys Med Biol. 1996
Finite Element Analysis Tools Used To Model Implanted
Environ.
9
Source:www.cst.com
• Industry standards are
• CST Microwave
• ANSYS HFSS
• Model antenna and implanted environment in 3-D
• Accurately models dispersive material
• Plot far/near field fields, S-parameters
• Post process SAR and Temp Rise
• Parameterized and Optimize Structures
• Requires minimal to significant computing power
• Expensive
CST Microwave Voxel
Human Model
Used in our research
E-field
SAR
Temperature
Device
Recent SAR Study IR-UWB Implanted In Head Model
10Source: Yuce, et al. MTT, 2013
CST Microwave Head Model w/ Radiating Fields
SAR From Voxel Model of Adult Human Head, SAR < 2 W/kg limit
Power Limits Increased main lobe direction 180˚
SAR Study on MICS & ISM for Retinal Prosthesis
11Source: Permana, et al. IET Microwaves, Antennas and Propagation Journal, 2013
CST Microwave Head Model
Head Model Tissue Stack-up
Antenna Design 2.4 GHz
Units in mm
SAR (W/kg) vs Input Power
Input Power
Calculated SAR
2.4 GHz has worst SAR in this study
2.4 GHz
Med Radio
SAR Margin
12
Biotelemetry
System
Modeling
System Block Diagram
13
PostProcessing
SignalAcquisition
DataModulation
Detection & De-
Modulation
Environment
Data Amplifier AmplifierRetrieved
Data
Transmitter Receiver
Acquires
Processes
Modulates
Amplifies
Transmits
Receives
Amplifies
Detects
Demodulates
Completes
Tissue
System Link Budget
14
Signal
Strength
(dBm)
Communications Path
Receiver
Sensitivity
Received
Signal
Rx
System Margin
Modulated
DataDe-Modulated
Data
2
4
dGGPP wb
rxtxtxrx
Signal link budget calculates
the power, gain and noise
figure at each step in the
system for transmit and
receive.
In Watts
Receive
Power
Transmit
PowerTransmit
GainReceive
Gain DistanceFree Space
Path Loss
Wide-Band
WavelengthModified Friis equation for
wideband signals computes the
received power at some
distance (d) between a
transmitter and receiver. Uses
the geometric mean of the
upper and lower frequencies.
Figures of Merit (FOM) include:• Equivalent Radiated Power (ERP)
• Sensitivity
• Signal-to-Noise Ratio (SNR)
• Specific Absorption Rate (SAR)
SAR
SNR
NF
ERP
System Performance Model – 1st Principles ApproachExpanded Friis Equation Including Statistical Variation
15
2
4
dGGPP wb
rxtxtxrx
Empirical Approach To Path Loss Modeling is Used
• Generally accepted path loss model (log-distance)
– Average receive signal power decreases logarithmically w/ distance
– Probability distribution function incorporates path loss measurements
16
Xd
dndPLdBPL
o
o
log10)()(
Reference Path Loss at distance, do
Path Loss Exponent
Distance-dependent mean
CDF of scattering within human chest fits zero-mean Gaussian
Path loss measurements within human chest fits log-distance eq.
Shadowing Effect R.V. (in dB)
Source: Khaleghi, et al. IET Journal, 2010
System Performance Model – 1st Principles ApproachExpanded Friis Equation Including Statistical Variation
17
2
4
dGGPP wb
rxtxtxrx
Xd
dn
d
cGGPP
o
dBrxdBtxdBtxdBrxdBin
10
0
10 log104
log20
System Safety Model – 1st Principles ApproachSpecific Absorption Rate
18
2E
dV
dW
dt
d
dm
dW
dt
dSAR
Product of the tissue conductivity (S/m) and electric field (V/m)
squared divided by the tissue density (kg/m3)
Average SAR (in W/kg) (or whole-body average SAR) is defined as the ratio of the total absorbed power in the exposed body to its mass where the local SAR refers to the absorbed energy value per unit volume or mass, which can be arbitrarily small
System Performance Model – 1st Principles ApproachReceiver Sensitivity & Data Modulation
19
SNRNFkTBdBmS )(min
BW
R
N
ESNR b
o
b
BWTkkTB
Sensitivity is the minimum detectable
signal of the receiver based on a given
bit error rate (BER)
Signal-to-Noise Ratio (SNR) of the receiver is a
function of the system modulation scheme and BER.
The noise power of the receiver is a function of
temperature (T) and Bandwidth (BW).
System Summary
20
System Model Architecture
21
System & Safety
Requirements
Integrated System Performance & Safety Model
Receiver (Ant, Amp)
Gtx
SAR
Ptx
n
σ
PLd0
Grx
NF
kTBSNR
Database of Key
Component
Characteristics (Model
or Measured)
Implanted Antenna
(Transmitter)
Environment Effects
Data Modulation
Database of Key
System Variables
& Parameters
comparisonTrade Space Visualization
& Analysis
Frequency
Bandwidth
Pin
Data Rate
SAR
Received Power
Sensitivity
ERP
Minimum Comm. Distance
Specific Absorption Rate
Equivalent Radiated Power
System Performance & Safety Roll-Up for Visualization & Analysis
Decision Makers
• Apply Model-Based Systems
Engineering (MBSE)
descriptive modeling
concepts using the systems
modeling language (sysML)
• Link descriptive and
analytical models using
IBM’s Rhapsody, PTC’s
Integrity Modeler and
ModelCenter®
22
Application of Model-Based Systems Engineering
Use Cases
Reqm’ts
Block
Definition
Diagrams
Parametric
Diagrams
System ModelingImplanted Transmit Antenna Model
23
1Determine System
& Safety
Requirements
Design & Optimization
of Implanted Antenna
using FEA
Sweep Frequency
Sweep Input Power
Add Tissue Stack-Up & Frequency Dependent
Characteristics
S11
f
2
3
Minimize Return Loss
S11(dB)
Across Band
Duncan, ISCAS 2017
System ModelingImplanted Transmit Antenna Model
24
1Determine System
& Safety
Requirements
Design & Optimization
of Implanted Antenna
using FEA
Gtx(f)
f
ηrad
Pin
S11
f
SAR(f)
fSAR(Pin)
Pin
Antenna Performance
and Safety as a function
of frequency (f) and input
power (Pin)
Ptx(Pin)
Pin
2
3
4a
4b
Minimize Return Loss
S11(dB)
Across Band
Multivariate Performance and Safety Results from FEA Simulations
Sweep Frequency
Sweep Input Power
Add Tissue Stack-Up & Frequency Dependent
Characteristics
25
Gtx(f)
f
ηrad
Pin
SAR(f)
fSAR(Pin)
Pin
Ptx(Pin)
Pin
System ModelingImplanted Transmit Antenna Model
Antenna Performance
and Safety as a function
of frequency (f) and input
power (Pin)
MATLAB model contains all the data and
interdependencies between the various inputs
and outputs of the transmit antenna model
Antenna
Performance
& Safety Database
),(
,,
frequencyPinf
SARGtxERP
*Could also include f(location, implant distance)
*
System ModelingEnvironment Model
26
Determine System
& Safety
Requirements
Non Line-of-Sight
Line-of-Sight
Sweep Frequency
Variability of the Environment
Path Loss as a
function of
frequency (f) and
environment
(NLOS, LOS)
)max,min,,/( distffNLOSLOSf
LossPath
LOS
NLOS
Assuming 10.6 – 3.1 GHz Bandwidth
Assuming 10.6 – 3.1 GHz Bandwidth
Assuming LOS Environment
3.6 – 3.1 GHz
10.6 – 3.1 GHz
27
Empirical Model
from Published
Works
Measured /
Modeled Database
Integrated Performance & Safety Model Modular and Extensible
Database of Key
Component
Characteristics
(Model or
Measured)
System & Safety
Requirements (ERP,
SAR, Distance)
System Parameters
(Frequency,
Temperature, …)
Analytical Models
Modelcenter® Integration Environment Used to Link Models
28
Model links and interdependencies
System Performance & Safety Calculation
Integrated Performance & Safety Model Parameters Linked w/ Interdependency
System & Safety
Requirements (ERP,
SAR, Distance)
System Parameters
(Frequency,
Temperature, …)
Modelcenter® Integration Environment Used to Link Models
29
Model links and interdependencies
System Performance & Safety Calculation
Integrated Performance & Safety Model Trade Space Exploration Enabled
Parallel Coordinates
J. Johansson and C. Forsell, "Evaluation of Parallel Coordinates: Overview, Categorization and Guidelines for Future Research," in IEEE Transactions on Visualization and Computer Graphics, vol. 22, no. 1, pp. 579-588, Jan. 31 2016.
Design of Experiments
Scatter Plots
T. H. Huang, M. L. Huang and K. Zhang, "An Interactive Scatter Plot Metrics Visualization for Decision Trend Analysis," 2012 11th International Conference on Machine Learning and Applications, Boca Raton, FL, 2012, pp. 258-264.
Various Approaches Exist for Multivariable Visualization & Analysis
30
Parametric Diagrams Developed of Each Module in Rhapsody
31
Parametric Diagrams Developed of Each Module in RhapsodyReceiver Noise Power Example (kTB) Linked to Model Center
Integrated Performance & Safety Model Descriptive Integrated to Analytical Models
32
• Varied
– Maximum frequency from 3.6 GHz to 10.6 GHz (500 – 7500 MHz)
– Input power to transmit antenna from 1e-5 to 1e-3 mW
– Transmit to Receive distance from 1 to 10 meters
• Analyzed SAR, ERP, Prx, Sensitivity
• Ran 6000 design of experiments (DOE) – 1 ½ hours
• Used Latin hypercube sampling (LHS)
– A statistical method for generating a near-random sample of parameter values
from a multidimensional distribution. -Wikipedia
33
Integrated Performance & Safety Model Trade Space Exploration Setup
34
Receiver Sensitivity (dBm) increases (receiver is less
sensitive) as bandwidth increases (500 MHz to 7500 MHz)
Integrated Performance & Safety Model Trade Space Analysis - Sensitivity
Smin = -62 dBm
Bandwidth (MHz)
Smin = -73.8 dBm
Se
ns
itiv
ity (
dB
m)
Sensitivity vs. Bandwidth
35
Integrated Performance & Safety Model Trade Space Analysis – Received Power
Received Power vs Distance
Distance (meters)
Prx trend
Sensitivity vs.
Bandwidth
10 m1 m
Re
ce
ive
d P
ow
er
(dB
m)
36
Integrated Performance & Safety Model Trade Space Analysis – Received Power (Apply Constraints)
Received Power vs Distance
Smin = -73.8 dBm
@ 500 MHz
Prx trend
Model shows
minimum transmit
distance < 10 meters
10 m1 m 5 m
Re
ce
ive
d P
ow
er
(dB
m)
Distance (meters)
37
Integrated Performance & Safety Model Trade Space Analysis – Received Power (Apply Constraints)
Received Power vs Distance
Re
ce
ive
d P
ow
er
(dB
m)
Prx vs Distance trend
Model shows
minimum transmit
distance < 5 meters
1 m
Smin = -62 dBm
& 7500 MHz
5 m 10 m
Distance (meters)
38
Integrated Performance & Safety Model Trade Space Analysis – ERP
ERP is visualized to
show its dependency
on bandwidth and
input power
ERP vs Input Power
Input Power (mW)
ER
P (
dB
m)
39
Integrated Performance & Safety Model Trade Space Analysis – Specific Absorption Rate
SAR vs Input Power
Input Power (mW)
SA
R (
W/k
g)
SAR is visualized to
show its dependency
on bandwidth and
input power
40
Integrated Performance & Safety Model Trade Space Analysis – SAR (Apply Constraints)
SAR vs Input Power
Input Power (mW)
SA
R (
W/k
g)
Prx > Smin= -73.82 dBm @ 500 MHz
Constraints on the received
power affects SAR
41
Integrated Performance & Safety Model Trade Space Analysis – SAR (Apply Constraints)
SAR vs Input Power
Input Power (mW)
SA
R (
W/k
g)
Prx > Smin= -62 dBm @ 7500 MHz
Constraints on the received
power affects SAR
42
Integrated Performance & Safety Model Trade Space Analysis – SAR (Apply Constraints)
SAR vs Input Power
Input Power (mW)
SA
R (
W/k
g)
Prx > Smin= -62 dBm @ 7500 MHz
SAR < 1.6 W/kg
Apply the SAR requirement
of < 1.6 W/kg
43
Integrated Performance & Safety Model Trade Space Analysis – SAR (Apply Constraints)
SAR vs Input Power
Input Power (mW)
SA
R (
W/k
g)
Prx > Smin= -62 dBm @ 7500 MHz
SAR < 1.6 W/kg
Pin < 2.38 mW
Determine the max input
power to meet SAR
requirement
Apply the SAR requirement
of < 1.6 W/kg
Integrated Performance & Safety Model Trade Space Visualization & Analysis
Parallel Coordinates
DOE Cube
Carpet Plots
Scatter Matrix
• A decision support tool has been developed to support a broad trade space
of implanted wireless communication system permutations using MBSE
• Allows the user to simulate numerous alternatives early in the system
lifecycle
• Leverages system requirements and constraints to prune and explore the
results to determine the best solutions for the system
• Enables the use of models or data to inform the user such as to consider
what-if analysis
45
Conclusion
Summary & Future WorkComparison of Published Implant Positions vs Performance
Implanted Antenna
(Transmitter)
• Leverage published UWB implant
data and locations
• Add double-hope system trades
• Add thermal effects on modeled
performance
• Add power efficiency calculation for
implanted system-on-chip
• Add implant size as a constraint
Receiver (Ant, Amp)
)()( bTTBASARTkt
TpC
Brain
Neck
Body Cavity
Arm
47
Thank You!
Questions?