An Ultra Wideband Communication Channel Model for ... - IEEE Xplore

3 downloads 0 Views 510KB Size Report
Abstract—Capsule endoscopy is an increasingly popular alter- native to a tube-based endoscope used in diagnosing gastroin- testinal diseases. It enables the ...
An Ultra Wideband Communication Channel Model for Capsule Endoscopy Stig Støa1,2,3, Raul Chavez-Santiago1,2,4, and Ilangko Balasingham1,2,4 1 The Interventional Centre, Oslo University Hospital, Oslo, Norway 2 Institute of Clinical Medicine, University of Oslo, Oslo, Norway 3 Novelda AS, Oslo, Norway 4 Dept. Electronics & Telecommunications, Norwegian University of Science & Technology, Trondheim, Norway @rr-research.no

Abstract—Capsule endoscopy is an increasingly popular alternative to a tube-based endoscope used in diagnosing gastrointestinal diseases. It enables the inspection of areas that are not easily accessible using traditional endoscopy and reduces patient discomfort. In addition to transferring high-capacity demanding image data, the capsule’s wireless interface must provide a wireless link that enables real-time positioning and tracking of the capsule. Ultra wideband (UWB) interfaces have great potential for the communication links of this application due to their inherent low power consumption, high transmission rates, accurate localization properties and simple electronics. However, accurate knowledge of the propagation channel is essential for efficient design of such UWB wireless communication systems. This paper presents a channel model for the propagation of a UWB pulse in the digestive tract in the 3.4-4.8 GHz frequency band. For the development of this model, numerical electromagnetic (EM) simulations were conducted using a voxel anatomical model that includes the dielectric properties of human tissues; using this EM simulator the channel responses of many in-body probes were computed. Based on the analysis of the obtained data we provide the mathematical expressions to calculate the average path loss and its distribution at several receiver locations surrounding the abdomen. Our proposed model gives designers an important tool that approximates well the digestive tract’s in-body channel properties, thereby eliminating the need for time consuming and complex numerical simulations. Keywords - channel model; capsule endoscopy; in-body communication; ultra wideband; wave propagation

I. I NTRODUCTION Capsule endoscopy is a clinical procedure that uses a tiny wireless camera in the form of a pill-sized capsule to produce images or video of the digestive tract. The image data is then transmitted to a recorder that is worn around the waist through a wireless interface. This procedure enables the visual inspection of areas like the small intestines, which are not easily accessible using traditional endoscopy. Traditional endoscopy involves passing a flexible tube containing a small video camera into the throat or through the rectum, which causes the patient much more discomfort. Several challenges are associated with the wireless interface of capsule endoscopy. The human body is not a good conductor for radio waves, This work is partly sponsored by the MELODY Project, which is funded by the Research Council of Norway under the contract number 187857/S10.

and because of the complex geometry of the digestive tract it is difficult to localize the capsule endoscope as it travels inside the body. In addition, relatively high data rates are required for image transmission, and there is a restriction on size and power consumption of the capsule. In February 2002, the FCC authorized the unlicensed use of low power ultra wideband (UWB) applications in the frequency range of 3.1-10.6 GHz [1]. As a result, parts of this band are now available in large parts of the world. UWB technology has great potential for the wireless interface of capsule endoscopy, due to its simple electronics, high transmission speed, and inherent low power consumption [2]. It also offers more accuracy in terms of localizing the capsule. Accurate knowledge of the propagation channel is necessary for efficient design of UWB wireless communication systems. Considerable effort has been devoted to the characterization of the propagation of UWB radio signals from on-body sensors. However, significantly fewer results are reported for the case of UWB devices operating inside the body. A number of path loss models for narrowband (NB) implantable sensors [3]–[7] are available in the literature, which regrettably are not suitable for modeling UWB propagation conditions. The IEEE 802.15.6 standardization group [8] has issued several propagation models for medical and non-medical devices, both in-body and on-body, for wireless body area networks (WBANs). Nevertheless, the in-body channel models therein characterize NB applications [9], [10], whereas the UWB models only characterize on-body communication channels. Recently, the lack of a UWB channel model for implant sensor communications was overcome with the investigations reported in [11], [12]. In that work, the characterization of a UWB link in the 3.4-4.8 GHz frequency band is presented. The channel model was developed assuming 20 arbitrary locations of a transmitting implant device inside the human chest between 6-18 mm depth. Subsequently, our research led to a more general UWB in-body channel model for the chest [13], which describes the propagation channel for depths between 5-120 mm, and for the abdominal region [14] for depths 10-150 mm in the 1-6 GHz frequency band. However, the latter model is applicable only for implantable

978-1-4244-8132-3/10/$26.00 ©2010 IEEE

Fig. 1. Section of the human body model used in the simulation scenario shown at 0◦ (left), and 90◦ (center) angle. The figure on the right illustrates the receiver belt.

sensors and actuators such as pressure detectors in the bladder and implanted automatic-insulin-release devices in the pancreas. These devices operate at a fixed location in the abdomen. Therefore, a UWB in-body channel model for endoscopy capsule is necessary for the design of the communication interfaces of this specific application. This paper presents such model. In the case of capsule endoscopy, the transceiver moves through the digestive tract, constantly changing its location and depth inside the body. Hence, in this paper we propose a model for the propagation in the abdomen that allows computing the average path loss and its distribution at different receiver positions surrounding the body specified by angle and height from a reference location. The proposed channel model takes the movement of the capsule endoscope into account and covers the frequency band of 3.4-4.8 GHz. This is an attractive band for UWB applications since it does not overlap with the many NB systems operating in the ISM bands. Also, when considering the band from 3.4 GHz to 10.6 GHz, the electromagnetic waves are less affected by attenuation, caused by the tissues, at lower frequencies. The remainder of the paper is organized as follows: Section II presents the in-body simulation scenario used for the numerical simulations. Section III presents the computational characterization of the channel and Section IV describes the UWB channel model. Finally, our conclusions are drawn in Section V. II. N UMERICAL S IMULATION The time-domain finite integration technique (FIT) was used to solve the Maxwell’s equations for our numerical simulations. An anatomical model based on the Visible Human Project of the National Library of Medicine (NLM) [15] was embedded in the FIT electromagnetic (EM) simulator. A voxel representation of the human body with a resolution of 2 mm was used. The dielectric properties of the human tissues, permittivity and conductivity, were provided by Gabriel [16]

(a)

(b)

Fig. 2. Simulation scenario where the antennas (a) surround the body with 30◦ angular spacing (b). The localization of the intestines can be seen (dark gray) in the cross section of the human model which is located 30 mm above the navel.

based on the Cole-Cole model, which describes the frequency dependent characteristics of the tissue materials. However, incorporating the highly complex material properties in the numerical simulation causes extremely long computational time. Previous investigations [17], [18] provide a simplified representation of the complex permittivity of the different human tissues, which was used in our simulations. A grid of field probes is placed within the model where the intestines of the human model are located as shown in Fig. 1. The spacing between the probes are 10 mm in the horizontal axis and 20 mm along the vertical one. The field probes are ideal frequency independent isotropic antennas with a specified polarization that is co-polar with the incident field polarization. Only vertical polarization of the electric field with respect to the standing body is considered. The probe arrays do not have any coupling among them. Perfectly matched layer (PML) absorbing boundary condition (ABC) is used for the simulations, thus the body environment reflections are disregarded. The arms and hands of the model are excluded

978-1-4244-8132-3/10/$26.00 ©2010 IEEE

III. C OMPUTATIONAL R ESULTS A UWB Gaussian pulse with 3.4-4.8 GHz bandwidth shaped by a Hamming window is transmitted by the antenna depicted in Fig. 2(a). The antenna provides a 90 Ω impedance matching with return loss < 10 dB for frequencies between 2-6 GHz and the excitation power is normalized to 0 dBW. The dipole has a 3 dB, 6 dB and 10 dB angular width of 60◦ , 90◦ and 120◦, respectively, in the vertical plane and at the center frequency. Because of this, the calculated path loss will have some inaccuracy at the probes on the higher and lower planes. In the horizontal plane the radiation pattern is approximately uniform. Because of the isotropic materials of the human tissues, the radio channel behaves in the same way for transmissions either from the outer to the inner body or in the opposite direction.

0 −5 −10

PDP [dB]

−15 −20 o

B. Path Loss versus Angle and Height The field probes were positioned within 13 horizontal layers (planes) spaced 20 mm apart along the vertical axis. By grouping the probes belonging to each layer we could determine the height of each probe, which is independent of the distance to the antenna located on the surface of the body. The average path loss for the receiver probes was calculated for each of the height layers, and then for each of the angles around the body. Figure 5(a) shows the average path loss for different angles at a given height. From the figure we

−120

−30

o

−60

−35

0

−40

60

o o o

120

−45 −50

0

1

2

3

4

5

Time [ns]

Fig. 3. probes.

Power delay profile for different receiver angles averaged over all

0.7 0.65 0.6 0.55 0.5 0.45 0.4 0.35

A. Power Delay Profile and Delay Spread A multipath channel is characterized by its time dispersive properties such as excess delay spread and root-meansquare (RMS) delay spread, which are useful in assessing the potential for intersymbol interference (ISI) in high data rate transmissions. The intensity of a signal received through a multipath channel as a function of time delay, τ , is known as the power delay profile (PDP). Figure 3 shows the PDP of received UWB signals at different propagation angles. Figure 4 shows the RMS delay spread for different receiving angles. It is important to notice that the very low RMS delay spread enables very high data rates.

o

−180

−25

Average RMS Delay Spread [ns]

from the EM simulation. The human model was exposed to a UWB pulse, which excites a center feed elliptic dipole antenna like the one illustrated in Fig. 2(a). A cross section of the simulation scenario within the anatomical model of an adult male is illustrated in Fig. 2(b). Separate simulations were done for each antenna location such that only one antenna was present in the simulator at a given time. In order to calculate the average path loss and its distribution as it is observed from receivers surrounding the body, 12 antennas with uniform angular spacing are positioned on the same horizontal plane surrounding the body. This plane is placed 30 mm above the navel and forms the reference height. An imaginary line drawn from the center of the body to the navel forms a reference angle of 0 degrees. The position of the antennas is 10 mm from the skin at the different angles shown in Fig. 2(b).

−150

Fig. 4. probes.

−100

−50 0 Angle [degrees]

50

100

150

RMS delay spread at different receiver angles averaged over all

see that depending on the height, the optimal position of the receiver antenna changes. This indicates that for certain heights, a larger part of the digestive tract is closer to one of the receiver antennas and is not symmetric within the body. Figure 5(b) shows the second dimension, namely average path loss, of a given angle for different heights. This illustrates how the path loss increases as the probe moves further from the cross section of the receiver belt. As the height between the transmitter and receiver increases, the angle of the propagation has a varying influence on the received power. IV. C HANNEL M ODELING The path loss for different angles around the body was calculated by averaging the attenuation observed at the receiver probes. Since the probes are placed within the digestive tract with varying depth from the skin, considerable path loss variations around the average was observed. However, one

978-1-4244-8132-3/10/$26.00 ©2010 IEEE

h=−120 h=−100 h=−80 h=−60 h=−40 h=−20 h=0 h=20 h=40 h=60 h=80 h=100 h=120

−60

Path Loss [dB]

−70 −80 −90 −100 −110 −120 −130 −140 −180

−120 −60 0 60 120 Angle relative to reference point

(a) −50

θ=−180 θ=−150 θ=−120 θ= −90 θ= −60 θ= −30 θ= 0 θ= 30 θ= 60 θ= 90 θ= 120 θ= 150

−60

Path Loss [dB]

−70 −80 −90 −100 −110 −120 −130 −140

−120 −80 −40 0 40 80 120 Height relative to reference point

(b) Fig. 5.

The average path loss versus angle (a), and versus height (b).

can expect large path loss variations with standard deviation of more than 9 dB even at a fixed depth [14], often referred to as shadowing, which is caused by the dielectric property variations of different tissues that surround each probe. We found that the path loss for each angle was approximately lognormal distributed. A model of the average path loss at different angles can be written as a Fourier sine and cosine series and the path loss distribution can be added as a random variable with lognormal distribution with parameters μ and σ for mean and standard deviation, respectively. Note that the Fourier series is used because it provides the best fitting with a reasonable amount of coefficients. The Fourier series of order I is expressed in generic form as I = a0 + [ i=1 ai × cos( iπθ (1) L[dB] (θ) p ) +bi ×

sin( iπθ p )

] + N (μ(θ), σ(θ))

where p is a scaling constant, θ is the angle from the reference line in degrees, a and b are the vectors containing path loss fitting coefficients related to angle, and N is a normal distributed random variable with mean value μ and standard deviation σ. A good fit to the average path loss from -150◦ to +180◦ is achieved with a second order Fourier series with the coefficients p=207, {ai }=[91.0,-27.2,1.61] and {bi }=[-0.371,-3.75]. Figure 6 shows the path loss at individual

TABLE I PATH LOSS DISTRIBUTION PARAMETERS , N (μ, σ), FOR DIFFERENT ANGLES , θ CONSIDERING ALL PROBES θ -180 -150 -120 -90

N (μ, σ) 11.36 , 10.37 26.43 , 15.46 31.36 , 19.03 25.97 , 16.89

θ -60 -30 0 30

N (μ, σ) 23.06 , 15.98 20.99 , 14.63 22.12 , 18.57 23.79 , 22.23

θ 60 90 120 150

N (μ, σ) 24.55 , 23.93 18.56 , 20.24 17.09 , 17.84 16.99 , 13.71

probes (circles), the average path loss (solid line), and the fitted expression given by (1) (dashed line). The path loss distribution parameters for different angles are given in Table I. In order to account for the height between the transmitter and receiver, the series can be extended to a Fourier double series to include angle and height of order I and J respectively. In the generic form L[dB] (θ, H) = c0 +

I  J 

ci,j × sin(

i=1 j=1

× sin(

iπθ π + ) pθ 4

(2)

jπH π + ) pH 4

where θ is the angle, H is the height, and ci,j are the Fourier double series fitting coefficients given by Table II. By using the method of nonlinear least squares and the Trust Region algorithm, we found that the average path loss for different heights and angles can be approximated by a double series of order 4. Figure 7 shows the surface plot generated by the fourth order Fourier double series (2) when using the coefficients of Table II. The hight axis (-120 mm to 120 mm) are scaled with scaling parameter pH =870, and the angles (-180◦ to 150◦) with scaling parameter pθ =590. For simplicity, we suggest that the expression for the path loss distribution, N (μ(θ), σ(θ)), given by (1) and the corresponding coefficients of Table I, be used to model the distribution for both (1) and (2).

180 160 140 Path Loss [dB]

−50

120 100 80 60 Average PL Fitted PL Probe PL

40 20

−150

−100

−50

0 50 Angle [degrees]

100

150

200

Fig. 6. Path loss calculated for individual probes (circles), average path loss (solid line) and fitted path loss (dashed line).

978-1-4244-8132-3/10/$26.00 ©2010 IEEE

−70 −40 −80

Path Loss [dB]

−60

−90

−80 −100

−100

−120

−110

−140 100

−120 50

100 0

0 −50

−100

−130

−100 Angle [degrees]

Height [mm]

Fig. 7. Surface plot of the Fourier double series fitted to the average path loss for different angles and heights. Circles show the original data. TABLE II C OEFFICIENT VALUES FOR F OURIER DOUBLE SERIES c(θ, H) c(1,H) c(2,H) c(3,H) c(4,H) c(0,0)

c(θ,1) -1986.1652 1876.8755 -562.7475 281.9745 55.2556

c(θ,2) 1398.9216 102.9257 -1837.829 225.708

c(θ,3) 89.4395 -2333.7691 3439.0036 -773.7729

c(θ,4) -288.0136 1310.1936 -1610.6079 430.382

V. C ONCLUSION A computational analysis and path loss model for the UWB in-body channel in the digestive tract for 3.4-4.8 GHz was presented. The model characterizes the propagation loss experienced by a radio link between a capsule endoscope traveling inside the digestive tract and a receiver located at different angles around the torso. The model takes into consideration the mobility of the capsule endoscope by averaging the path loss observed at different heights in the abdomen; this enables the design of a single belt receiver that minimizes the strain on the patient. An important contribution of this work is the description of the path loss variations at different points around the waist. This can be used to improve signal reception by combining two or more receivers. Furthermore, our results might give important insight into novel applications in the areas of localization and tracking of capsule endoscopes and diversity combining. It will also ease the design and performance evaluation tasks of UWB interfaces for implanted sensors and actuators. In future studies, we will expand our model to characterize multipath characteristics with respect to angle.

[3] W. G. Scanlon, N. E. Evans, and Z. M. McCreesh, “RF performance of a 418-MHz radio telemeter packaged for human vaginal placement”, IEEE Trans. Biomed. Eng., vol. 44, no. 5, pp. 427-430, May 1997. [4] W. G. Scanlon, J. B. Burns, and N. E. Evans, “Radio wave propagation from a tissue-implanted source at 418 MHz and 916.5 MHz”, IEEE Trans. Biomed. Eng., vol. 47, no.4, pp. 527-534, April 2000. [5] A. J. Johansson, “Performance of a radio link between a base station and a medical implant utilising the MICS standard”, in Proc. 26th Annu. Intl. Conf. IEEE IEMBS, San Francisco, CA, pp. 2113-2116, September 2004. [6] A. Alomainy, Y. Hao, Y. Yuan, and Y. Liu, “Modelling and characterisation of radio propagation from wireless implants at different frequencies”, in Proc. 9th European Conf. on Wireless Technol., pp. 119122, September 2006. [7] A. Alomainy, and Y. Hao, “Modeling and characterization of biotelemetric radio channel from ingested implants considering organ contents”, IEEE Trans. Antennas and Propag., vol. 57, no. 4, pp. 999-1005, April 2009. [8] Channel Model for Body Area Network (BAN), IEEE P802.15-08-078009-0006, April 27, 2009. [9] T Aoyagi, K. Takizawa, T. Kobayashi, J. Takada, and R. Kohno,“Development of a WBAN channel model for capsule endoscopy”, Antennas and Propagation Society International Symposium, 2009 APSURSI’09. IEEE, Charleston, SC, June 1-5, 2009. [10] K. Takizawa, T Aoyagi, K. Hamaguchi, and R. Kohno, “Performance evaluation of Wireless Communications through Capsule Endoscope”, 31st Annual International Conference of the IEEE EMBS Minneapolis, Minnesota, USA, September 2-6, 2009. [11] Q. Wang, K. Masami, and J. Wang, “Channel modeling and BER performance for wearable and implant UWB body area links on chest”, in Proc. IEEE Intl. Conf. on Ultra-Wideband (ICUWB 2009),Vancouver, Canada, September 9-11, 2009, pp. 316-320. [12] J. Wang, and Q. Wang, “Channel modeling and BER performance of an implant UWB body area link”, in Proc. 2nd Intl. Symp. on Applied Sciences in Biomedical and Commun. Technol. (ISABEL ’09), Bratislava, Slovak Republic, November 24-27, 2009. [13] A. Khaleghi, R. Chávez-Santiago, X. Liang, I. Balasingham, V. C. M. Leung, and T. A. Ramstad, “On ultra wideband channel modeling for inbody communications”, In Proc. IEEE Int. Symp. on Wireless Pervasive Computing (ISWPC), Modena, Italy, May 5-7, 2010. [14] S. Støa, R. Chavez-Santiago, and I. Balasingham, “An Ultra Wideband Communication Channel Model for the Abdominal Region”, In Proc. IEEE Globecom 2010 Workshop on Advanced Sensor Integration Technology (ASIT 2010), Miami, Florida, USA, December, 2010, to appear. [15] M. J. Ackerman, “Viewpoint: The visible human project”, J. Biocommun., vol. 18, no. 2, p.14,1991. [16] C. Gabriel and S. Gabriel, “Compilation of the dielectric properties of body tissues at RF and microwave frequensies”, Brooks Air Force, AL/OE-TR-1996-0037, San Antonio, TX, 1996. [17] A. Khaleghi, and I. Balasingham, “On non-line-of-sight on-body ultra wideband (1-6 GHz) channel characterization using different antenna polarizations”, IET Microwaves, Antennas and Propagation, vol. 3, no. 7, pp. 1019-1027, 2009. [18] A. Khaleghi, and I. Balasingham, “Improving in-body ultra wideband communication using near-field coupling of the implanted antenna”, Microwave and Optical Technology Letters, vol. 51, no. 3, pp. 585589,March 2009. [19] A. F. Molisch, “Ultra-wide-band propagation channels”, Proc. IEEE, vol.97, no2, pp.353-371, February 2009.

R EFERENCES [1] FCC Order, ET Docket No. 98-153 (FCC 02-48), February 14, 2002. [2] B. Allen, T. Brown, K. Schwieger, E. Zimmermann, W. Malik, D. Edwards, L. Ouvry, and I. Oppermann, “Ultra wideband: Applications, technology and future perspectives”, in Proc. Intl. Workshop on Convergent Technologies (IWCT), 2005.

978-1-4244-8132-3/10/$26.00 ©2010 IEEE