Plugin roelens2006
1 like262 views
This paper presents a new empirical path loss model for wireless communication at 2.4 GHz above simulated human tissue. The model is valid for dipole antennas placed up to 5 cm above flat phantom tissue simulating muscle or brain, and for distances up to 40 cm. Simulation and measurement results showed excellent agreement. The study found that antenna height has a major influence on path loss, with path loss dropping quickly as height decreases. Models were developed to estimate path loss as a function of antenna height and distance for both muscle and brain tissue.
![Path loss model for wireless narrowband
communication above flat phantom
L. Roelens, S. Van den Bulcke, W. Joseph, G. Vermeeren
and L. Martens
A new empirical path loss model for wireless communication at
2.4 GHz above a flat, lossy medium, representing human tissue, is
presented. The model is valid for dipole antennas for heights up to
5 cm above the phantom and for distances up to 40 cm, and was
applied to muscle and brain simulating media. For antennas placed
close to the lossy medium, it was found that antenna height has a
major influence on path loss. The model has been validated by
measurements and simulations, which show excellent agreement.
Introduction: A wireless body area network (WBAN) is a network for
which the nodes are located in the clothes on the body or under the
skin of a person. These nodes are connected through a wireless
communication channel and form a network that typically expands
over the body of a person. According to the implementation, the nodes
consist of sensors and actuators, placed in a star or multihop topology
[1]. WBANs have many promising, new applications in medicine,
multimedia and sports, all of which make use of the unconstrained
freedom of movement a WBAN offers.
An important step in the development of a narrowband WBAN is the
estimation of path loss (PL) between two nodes on the body. This
requires detailed characterisation of the electromagnetic wave propaga-
tion near the human body. To date, only few attempts, which lack
verification by measurements and do not investigate the influence of
antenna height, have been made [2]. Therefore, we study the wave
propagation above a lossy medium including brain and muscle tissue,
using two commercial simulation tools and measurements. Our objec-
tive is to develop an empirical PL model, which takes the influence of
antenna height into account. The simulations are performed with
FEKO, a Method of Moments (MoM) program, and verified by
SEMCAD, a finite-difference time-domain (FDTD) program. Simula-
tions and measurements are performed at 2.4 GHz in the licence-free
industrial, scientific and medical (ISM) band.
Method: Model – To model the PL between two dipole antennas
above a specific lossy medium, we use the following semi-empirical
formula, expressed in dB and based on the Friis formula in free space:
Prec
Ptrans
ðd; hÞ
dB
¼ P0ðhÞjdB À nðhÞdjdB ¼ jS21jdB ð1Þ
where Prec and Ptrans are received and transmitted power, respectively,
and d is the distance between the transmitting antenna Tx and receiving
antenna Rx. P0 is path loss at a reference distance d0 (here chosen to be
40 cm) and n is the so-called PL exponent which equals 2 in free space.
Both P0 and n depend on the height h of both antennas above the
medium. The last part of (1) allows us to regard the setup as a two-port
network for which we determine the S21-parameter.
Because we want to investigate the influence of the medium on PL,
we limit ourselves to a flat, conducting, dielectric and uniform medium,
characterised by a specific relative permittivity er and conductivity s.
Furthermore, to model the PL, we use elementary half-wavelength
dipoles. The setup and parameters for determination of the PL are
shown in Fig. 1.
Fig. 1 Setup for PL measurements and simulations
Simulations and measurements – Using the MoM tool, we simulate
the PL for distances between Tx and Rx up to 40 cm and antenna
heights from 5 up to 50 mm above the lossy medium. Longer distances
and heights are impractical for a multihop WBAN and are not
considered. Both muscle (er ¼ 53.57 and s ¼ 1.81 S=m at 2.4 GHz)
and brain tissue (er ¼ 38.5 and s ¼ 1.9 S=m at 2.4 GHz) are simulated.
We make use of half-wavelength dipoles with length ‘ ¼ 0.46
l ¼ 5.75 cm and a realistic diameter t ¼ 1 mm (see Fig. 1). An applied
E-field source model, which corresponds to an applied voltage over the
source segment, is used and the length of the source segment is made
equal to the length of the other segments. Furthermore, we can model
the flat phantom by the built-in Green’s function for planar, multilayer
substrates. This greatly speeds up computation time compared to FDTD
methods, since the influence of the presence of the flat phantom is taken
into account implicitly.
For verification purposes, simulations were performed at an antenna
height of 1 cm using the FDTD method. The flat phantom is now
modelled by a rectangular box with a height of 75 mm. The FDTD
cell size varies between 0.25 and 8.75 mm to ensure a maximum cell
dimension around l=10 ¼ 0.2 mm in the medium and well below
l=10 ¼ 12.5 mm in free space at 2.4 GHz. Because of these small cell
sizes, the fields in a maximum grid size of 11 million cells have to be
calculated for 10 periods of the excitation to obtain accurate results. This
leads to long computation times compared to the MoM simulations.
To verify the simulations, we performed measurements with a
network analyser (Rohde Schwarz ZVR). Two half-wavelength
dipoles with length of 5.75 cm and diameter of 1 mm are placed
close to a rectangular box phantom recommended by the CENELEC
standard EN50383 [3] with dimensions 80 Â 50 Â 20 cm. The phantom
shell has a thickness of 1 cm (Æ1 mm) and is filled with brain
simulating liquid (er ¼ 38.5 and s ¼ 1.9 S=m at 2.4 GHz) and was
also modelled in the MoM and FDTD simulations.
The measurements are performed in a non-anechoic environment,
resulting in undesired reflections. Therefore, we performed a de-
embedding step [4]. At each position of the antennas, we measured
S21( f ) from 300 kHz up to 4 GHz. This frequency range of
3.9997 GHz is necessary to distinguish the direct and reflected waves
in the time domain. Next, we took the inverse FFT to derive S21(t).
Reflections in S21(t) are then mitigated by a tenth-order Butterworth
digital bandpass filter, characterised by a flat passband and the absence
of sidelobes. Finally, we took the FFT to obtain S21,filtered( f ), corrected
for reflections.
Results: Validation – Fig. 2 shows simulation and measurement
results for Tx and Rx both 1 cm above the phantom and distances
up to 40 cm. We obtained very good agreement between MoM and
FDTD and an average deviation of 1.8 dB between measurements and
simulations. Contributing to this deviation is the difference of 1.4 dB
found between the measured (2.8 dB) and simulated (4.2 dB) gain of
the system consisting of both antennas in free space. Also, the
effective thickness of the phantom shell and positional errors cause
measurement inaccuracies. The small deviations between MoM and
FDTD allow us to perform the further investigation in this Letter with
the faster MoM tool.
Fig. 2 Influence of distance between Tx and Rx on jS21j (brain tissue)
Influence of height – We performed a series of simulations with the
MoM tool for varying antenna heights above the phantom, from 5 up to
50 mm, and distances up to 40 cm. The results for jS21jdB are shown in
Fig. 3 for brain simulating tissue. It is clear that jS21jdB strongly depends
on antenna height and drops quickly as the height decreases. The PL
model of (1) is now fitted to the data obtained from these simulations.
ELECTRONICS LETTERS 5th January 2006 Vol. 42 No. 1](https://image.slidesharecdn.com/plugin-roelens2006-140328203415-phpapp01/85/Plugin-roelens2006-1-320.jpg)
Plugin roelens2006
- 1. Path loss model for wireless narrowband communication above flat phantom L. Roelens, S. Van den Bulcke, W. Joseph, G. Vermeeren and L. Martens A new empirical path loss model for wireless communication at 2.4 GHz above a flat, lossy medium, representing human tissue, is presented. The model is valid for dipole antennas for heights up to 5 cm above the phantom and for distances up to 40 cm, and was applied to muscle and brain simulating media. For antennas placed close to the lossy medium, it was found that antenna height has a major influence on path loss. The model has been validated by measurements and simulations, which show excellent agreement. Introduction: A wireless body area network (WBAN) is a network for which the nodes are located in the clothes on the body or under the skin of a person. These nodes are connected through a wireless communication channel and form a network that typically expands over the body of a person. According to the implementation, the nodes consist of sensors and actuators, placed in a star or multihop topology [1]. WBANs have many promising, new applications in medicine, multimedia and sports, all of which make use of the unconstrained freedom of movement a WBAN offers. An important step in the development of a narrowband WBAN is the estimation of path loss (PL) between two nodes on the body. This requires detailed characterisation of the electromagnetic wave propaga- tion near the human body. To date, only few attempts, which lack verification by measurements and do not investigate the influence of antenna height, have been made [2]. Therefore, we study the wave propagation above a lossy medium including brain and muscle tissue, using two commercial simulation tools and measurements. Our objec- tive is to develop an empirical PL model, which takes the influence of antenna height into account. The simulations are performed with FEKO, a Method of Moments (MoM) program, and verified by SEMCAD, a finite-difference time-domain (FDTD) program. Simula- tions and measurements are performed at 2.4 GHz in the licence-free industrial, scientific and medical (ISM) band. Method: Model – To model the PL between two dipole antennas above a specific lossy medium, we use the following semi-empirical formula, expressed in dB and based on the Friis formula in free space: Prec Ptrans ðd; hÞ dB ¼ P0ðhÞjdB À nðhÞdjdB ¼ jS21jdB ð1Þ where Prec and Ptrans are received and transmitted power, respectively, and d is the distance between the transmitting antenna Tx and receiving antenna Rx. P0 is path loss at a reference distance d0 (here chosen to be 40 cm) and n is the so-called PL exponent which equals 2 in free space. Both P0 and n depend on the height h of both antennas above the medium. The last part of (1) allows us to regard the setup as a two-port network for which we determine the S21-parameter. Because we want to investigate the influence of the medium on PL, we limit ourselves to a flat, conducting, dielectric and uniform medium, characterised by a specific relative permittivity er and conductivity s. Furthermore, to model the PL, we use elementary half-wavelength dipoles. The setup and parameters for determination of the PL are shown in Fig. 1. Fig. 1 Setup for PL measurements and simulations Simulations and measurements – Using the MoM tool, we simulate the PL for distances between Tx and Rx up to 40 cm and antenna heights from 5 up to 50 mm above the lossy medium. Longer distances and heights are impractical for a multihop WBAN and are not considered. Both muscle (er ¼ 53.57 and s ¼ 1.81 S=m at 2.4 GHz) and brain tissue (er ¼ 38.5 and s ¼ 1.9 S=m at 2.4 GHz) are simulated. We make use of half-wavelength dipoles with length ‘ ¼ 0.46 l ¼ 5.75 cm and a realistic diameter t ¼ 1 mm (see Fig. 1). An applied E-field source model, which corresponds to an applied voltage over the source segment, is used and the length of the source segment is made equal to the length of the other segments. Furthermore, we can model the flat phantom by the built-in Green’s function for planar, multilayer substrates. This greatly speeds up computation time compared to FDTD methods, since the influence of the presence of the flat phantom is taken into account implicitly. For verification purposes, simulations were performed at an antenna height of 1 cm using the FDTD method. The flat phantom is now modelled by a rectangular box with a height of 75 mm. The FDTD cell size varies between 0.25 and 8.75 mm to ensure a maximum cell dimension around l=10 ¼ 0.2 mm in the medium and well below l=10 ¼ 12.5 mm in free space at 2.4 GHz. Because of these small cell sizes, the fields in a maximum grid size of 11 million cells have to be calculated for 10 periods of the excitation to obtain accurate results. This leads to long computation times compared to the MoM simulations. To verify the simulations, we performed measurements with a network analyser (Rohde Schwarz ZVR). Two half-wavelength dipoles with length of 5.75 cm and diameter of 1 mm are placed close to a rectangular box phantom recommended by the CENELEC standard EN50383 [3] with dimensions 80 Â 50 Â 20 cm. The phantom shell has a thickness of 1 cm (Æ1 mm) and is filled with brain simulating liquid (er ¼ 38.5 and s ¼ 1.9 S=m at 2.4 GHz) and was also modelled in the MoM and FDTD simulations. The measurements are performed in a non-anechoic environment, resulting in undesired reflections. Therefore, we performed a de- embedding step [4]. At each position of the antennas, we measured S21( f ) from 300 kHz up to 4 GHz. This frequency range of 3.9997 GHz is necessary to distinguish the direct and reflected waves in the time domain. Next, we took the inverse FFT to derive S21(t). Reflections in S21(t) are then mitigated by a tenth-order Butterworth digital bandpass filter, characterised by a flat passband and the absence of sidelobes. Finally, we took the FFT to obtain S21,filtered( f ), corrected for reflections. Results: Validation – Fig. 2 shows simulation and measurement results for Tx and Rx both 1 cm above the phantom and distances up to 40 cm. We obtained very good agreement between MoM and FDTD and an average deviation of 1.8 dB between measurements and simulations. Contributing to this deviation is the difference of 1.4 dB found between the measured (2.8 dB) and simulated (4.2 dB) gain of the system consisting of both antennas in free space. Also, the effective thickness of the phantom shell and positional errors cause measurement inaccuracies. The small deviations between MoM and FDTD allow us to perform the further investigation in this Letter with the faster MoM tool. Fig. 2 Influence of distance between Tx and Rx on jS21j (brain tissue) Influence of height – We performed a series of simulations with the MoM tool for varying antenna heights above the phantom, from 5 up to 50 mm, and distances up to 40 cm. The results for jS21jdB are shown in Fig. 3 for brain simulating tissue. It is clear that jS21jdB strongly depends on antenna height and drops quickly as the height decreases. The PL model of (1) is now fitted to the data obtained from these simulations. ELECTRONICS LETTERS 5th January 2006 Vol. 42 No. 1
- 2. The results of these fits are shown in Fig. 4. The following PL model for wave propagation above brain tissue is obtained: nbrainðhÞ ¼ À25:0h þ 4:0 P0;brainðhÞjdB ¼ 7:7 lnðhÞ À 11:9; h 0:15l ¼ 388:7h À 49:4; h 0:15l ð2Þ where ln is the natural logarithm. For wave propagation above muscle simulating tissue we obtain: nmuscleðhÞ ¼ À25:4h þ 4:0 P0;muscleðhÞjdB ¼ 7:7lnðhÞ À 12:1; h 0:15l ¼ 404:1h À 49:9; h 0:15l ð3Þ In (2) and (3), we use a linear approximation for the PL exponent n. For P0 we use a logarithmic fit for heights below the breakpoint value 0.15 l and a linear fit for heights above this breakpoint (see Fig. 4). Excellent agreement is obtained, with a maximal and average deviation of only 0.24 and 0.08 dB, respectively. From (2) and (3) it can be seen that brain and muscle tissue only result in small differences and that antenna height is the determining factor. Using these models, we are able to make an accurate estimation of the PL above a flat phantom. Fig. 3 Influence of antenna height on PL for several distances (brain tissue) 4.0 3.5 3.0 2.5 PLexponent,n antenna height ,h l 0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36 0.40 –28 –30 –34 –38 –42 –46 –50 –54 P0,dB Fig. 4 Results of fit for n(h) and P0(h)jdB against antenna height (brain tissue) m FEKO simulations for n –Á–Á linear fit for n FEKO simultions for P0 - - - logarithmic fit for P0 . . . Á linear fit for P0 Conclusions: We have presented a new, accurate model for the path loss near a flat and homogeneous phantom for both brain and muscle tissue at 2.4 GHz. The influence of antenna height was characterised and found to be very important. The model has been validated using simulations and measurements for which excellent agreement is reported. # IEE 2006 24 August 2005 Electronics Letters online no: 20063062 doi: 10.1049/el:20063062 L. Roelens, S. Van den Bulcke, W. Joseph, G. Vermeeren and L. Martens (Department of Information Technology, Ghent Univer- sity, Gaston Crommenlaan 8 Box 201, B-9050 Ghent, Belgium) E-mail: laurens.roelens@intec.ugent.be References 1 Latre´, B., et al.: ‘Networking and propagation issues in body area networks’. 11th Symp. on Communications and Vehicular Technology in the Benelux 2004, SCVT 2004, Ghent, Belgium, November 2004 2 Ryckaert, J., et al.: ‘Channel model for wireless communication around human body’, Electron. Lett., 2004, 40, (9), pp. 543–544 3 CENELEC EN50383: ‘Basic standard for the calculation and measurement of electromagnetic field strength and SAR related to human exposure from radio base stations and fixed terminal stations for wireless telecommunication systems (110 MHz–40 GHz)’, September 2002 4 Joseph, W., Verloock, L., and Martens, L.: ‘Accurate low-cost measurement technique for occupational exposure assessment of base station antennas’, Electron. Lett., 2003, 39, (12), pp. 886–887 ELECTRONICS LETTERS 5th January 2006 Vol. 42 No. 1