Acoustic Transmissions for Wireless Communications and Power Supply in Biomedical Devices

Transcription

1 roeedings of th International ongress on Aoustis, IA August 1, Sydney, Australia Aousti Transmissions for Wireless ommuniations and ower Supply in Biomedial Devies Graham Wild and Steven Hinkley Shool of Engineering, Edith owan University, 7 Joondalup Drive, Joondalup WA 67, Australia AS: 43.8.Vj, 84.4.Ua, 84.3.J ABSTAT In this paper, we demonstrate the priniple of aousti transmission for ommuniations and power supply, in-vivo. The aousti transmissions are intended to be used for fixed implanted biomedial devies, suh as paemakers, but more importantly, neural implants were wired and wireless F ommuniations annot be used. The aousti transmissions an be used for both wireless ommuniations and to reharge the devie, in-vivo, using onventional piezoeletri power harvesting tehniques. urrent researh in biomedial engineering is looking at implantable devies to regulate onditions suh as arkinson s and other neuromusular onditions. Transient devies, suh as those used in the gastrointestinal trak, make use of high frequeny F, were the permittivity of the human body begins to derease. However, signifiant power is still required. This results in loal tissue heating, due to the absorption of the EM radiation. This heating has side effets that limit the exposure times for safe praties. For neural implants, were the goal is to have the produt implanted for long periods of time, without ompliations and minimal side effets, F ommuniations annot urrently be used. Aousti transmissions represent an ideal low power method of ommuniating with in-vivo biomedial devies, and for reharging them through power harvesting. In this work, we present results showing the performane of the ommuniations hannel and sample ommuniations signals, through a biologial speimen. The frequeny response, transfer funtion and transient response (at resonane) of the ommuniations hannel were measured. Due to the frequeny response of the ommuniations hannel, SK was hosen as the modulation method. Suessful ommuniation was ahieved through the ommuniations hannel. We also show the result of preliminary work on harvesting the aousti signals to provide power for reharging in-vivo Biomedial devies. INTODUTION urrent biomedial engineering researh is looking at implantable devies to regulate onditions suh as arkinson s and other neuromusular onditions [1]. Transient devies, suh as those used in the gastrointestinal trak, make use of high frequeny F, were the permittivity of the human body begins to derease []. However, signifiant power is still required for ommuniations. This results in loal tissue heating due to the absorption of the EM radiation. This heating has side effets that limit the exposure times for safe praties [3-5]. For neural implants, were the goal is to have the produt implanted for long periods of time, without ompliations and minimal side effets, F ommuniations is not urrently used. Wireless aousti ommuniations represents an ideal, low power method of ommuniating with in-vivo biomedial devies. Aousti ommuniations has previously been proposed for ommuniations in Strutural Health Monitoring (SHM) systems, were autonomous roboti agents are used for inspetion and repair [6-8]. Here, aousti signals were suessfully shown to be used for transmitting relatively high data rates (up to 1kbps) using piezoeletri transduers and aluminium panelling. The added advantage of utilising aousti ommuniations is the use of a piezoeletri reeiver. This means that urrent work into piezoeletri power harvesting [9] ould be utilised for supplying power to the in-vivo biomedial devies. The same aousti transmission used to ommuniate an in fat be used for power harvesting. In this paper, we use aousti transmissions to both ommuniate wirelessly, and supply power to in-vivo biomedial devies. The aousti transmissions hannel is made up of several layers. These inlude; a piezoeletri transduer as the transmitter, a oupling medium, the material to be ommuniated through, in this ase the first author s forearm, a seond oupling layer, and a seond piezoeletri transduer as the reeiver. The forearm was hosen due to ease, as it is relatively thin, and made up of almost parallel faets. Also, ommuniation through the bone struture of the forearm would be required for a devie loated within the hest or ranial avities. THEOY iezoeletri Transduer For a omplete understanding of piezoeletri materials and transduers, see Silk s Ultrasoni Transduers for Nonde- IA 1 1

2 3-7 August 1, Sydney, Australia roeedings of th International ongress on Aoustis, IA 1 strutive Testing [1]. A brief overview is inluded here for ompleteness. In linear elasti solids, the strain (S) and stress (T) are related by the elasti stiffness (). In the same material, the eletri displaement (D) is related to the eletri field (E) by the permittivity (ε r ) of the material. These equations are referred to as the onstitutive equations In a piezoeletri linear elasti material, the onstitutive equations are oupled. Hene, a hange in stress or strain orresponds to a hange in the harge distribution within the material. The onstitutive equations for a piezoeletri material are [9], T = S+ he D= ε r E+ hs, where h is the piezoeletri oupling oeffiient. ommuniations For the benefit of inreased bandwidth, only digital ommuniations methods were used for the aousti ommuniations. Due to the thikness of the ommuniations medium and omposite struture (bone, musle et), only hase Shift Keying (SK) was onsidered. In SK, the digital information is enoded onto the arrier wave via a phase modulation. The state of eah bit of information is determined aording to the state of the preeding bit. If the phase of the arrier wave does not hange, then the logi level stays the same. If the phase of the arrier wave hanges by 18 degrees, then the logi level hanges, from zero to one, or from one to zero. Deoding SK uses some simple mathematis to retrieve the phase information. The SK signal; where (1) f( = A os( π f )), () t for data= φ ( = (3) 18 for data= 1. is multiplied by a synhronous sine and osine, giving, and g( = A h( = A A = [sin os(πf ) sin(πf (( 4πf )) + sin( )]. os(πf ) os(πf A = [os ( φ) + os ( 4πf ) ( )], These two omponents are alled the in-phase (I) and quadrature (Q) omponents, respetively. Both I and Q ontain high and low frequeny omponents, where the low frequeny omponent is the sine or osine of the time dependent phase. Using a low pass filter, the high frequeny omponents are removed, leaving only the phase omponents, and (4) (5) A g( = sin( ). (6) A h( = os( ), (7) Then by taking the artan of I on Q, the time dependent phase information is reovered, g( y( = artan h( sin( ) = artan os( ) = artan =. ( tan( ) ) The filter used is a raised osine filter [1]. Figure 1 shows the steps of the SK enoding and deoding proess used in the ommuniations hannel experiments. Figure 1 a) shows the digital information to be transmitted. This information is then enoded onto the square wave arrier, shown in figure 1 b), as the 18 degree phase hange. This signal is then reeived, as shown in Figure 1 ). The phase information is then reovered, as shown in Figure 1 d), and by sampling at suitable points a phase greater than is equated to a 1, and a phase less than is equated to a. Figure 1. The SK ommuniations enoding and deoding; a) The initial digital information, b) the phase enoded signal, ) the reeived signal, and d) the reovered phase information ower Harvesting For the power harvesting, the piezoeletri reeiver is modelled as a urrent soure, i p, in parallel with with a apaitor, p. The soure urrent an be written as [9], (8) i ( = I sin( ω, (9) where I p is the peak urrent, also refered to as the short iruit urrent, and ω is the angular frequeny of the A signal. The open iruit voltage, V O, an then be defined in terms of the short iruit urrent and the reatane of the apaitor (X ) [11], that is, V O I = I X =, (1) ω To harvest power, the piezoeletri element needs to be onneted aross a load. In the ase of the A analysis, this is simply a load resistane. There is a 9 degree phase shift between the urrent flowing through the load resistor () and the urrent flowing through the apaitor. The total power an be expressed as the geometri sum of the power stored in the apaitor, and the power dissipated through the resistor. That is, IA 1

3 3-7 August 1, Sydney, Australia roeedings of th International ongress on Aoustis, IA 1 T = = + I + I X. (11) Sine the iruit is an A urrent divider, the short iruit urrent an be expressed as, Testing the ommuniations involved looking at a number of different quantities. These inluded, the transfer funtion, the frequeny response, the transient response, and, the data rate. First, the transfer funtion of the ommuniations hannel was measured. The funtion generator was set to give a ontinuous sine wave at the resonant frequeny of the ZT transduers, 1MHz. The amplitude was then varied from 1 volt to 1 volts. Values were reorded at 1 volt inrements. This proess was repeated several time to give an average and statistial unertainty. I = I + I. (1) The peak power will then our when the urrent flow through the apaitor and the resistor is equal. That is, the load resistane is equal to the apaitor s reatane, 1 =. (13) ω The resistor urrent at peak power is then, I I =. (14) The voltage at peak power is then, I V = I X =. (15) ω We an also express the voltage out as a funtion of the resistane. From (1) we see that, V = I = I I, (16) The apaitor urrent is also a funtion of the voltage, so with a little algebra we see [1], I V =. (17) 1+ ( ω ) p The power as a funtion of the resistane an then be expressed as, METHOD V I = = 1+ ( ω ). (18) p Aousti Transmissions hannel Setup The experimental setup of the aousti-transmissions hannel is shown in Figure. The ZT transduers used were unbaked, and oupled to the forearm using aousti oupling gel. The piezoeletri transduers used were Steiner and Martins SMQA ZTs. They had a thikness of.1mm, orresponding to a resonant frequeny of 1MHz, and a radius of 1mm. Aousti ommuniations Figure. The setup of the aousti-transmissions hannel Next, the frequeny responses of the ommuniations hannel were determined. The funtion generator was set to give a ontinuous sine wave at maximum voltage, 1 Volts peak. The frequeny was then varied from 1 kilohertz to Megahertz. Values were reorded every 1 kilohertz. Finally, the transient response of the ommuniations hannel was investigated, using a low rate sine wave burst at 1MHz with 1 yles. The trailing signal is also examined to determine if it will have any adverse effets on the performane of the ommuniations hannel. The ommuniations signals were generating on an Agilent 331A arbitrary waveform generator. ASK (speifially OOK) signals were generated using the burst funtion of the generator. A 1MHz sine wave arrier was used with a data rate of 4kbps. The SK signals were generated in the Waveform Editor software for the waveform generator. The signals were then flashed to the devie via the omputer interfae. The waveform generated onsisted of a sine wave arrier, with a data rate of 1/1 the arrier frequeny (the software does not generate time so the frequeny is set and varied on the generator, and hene a ratio is used). So for the arrier wave frequeny of 1MHz, the data rate was 1kbps. All of the ommuniations signals were reorded on the digital osillosope, and downloaded to a. The demodulation of the signals was then implemented in Matlab TM [13]. ower Harvesting For the preliminary aousti power harvesting, the A performane was analysed. In the A iruit experiments, first the apaitane of the piezoeletri element was measured using a apaitane meter. After alulating the reatane at the resonant frequeny, the output of the piezoeletri reeiver was applied to a variety of suitable load resistors. The voltage drop aross the load resistor was measured using a 1MΩ Digital Storage Osillosope (DSO). To ompare the experimental results to the theoretial analysis, the A iruit was also simulated in Spie. The value of I p was obtained using (1), with the measured values of p and V O. A parametri analysis was performed, varying the value of the load resistane in a frequeny domain analysis. The load value IA 1 3

4 3-7 August 1, Sydney, Australia roeedings of th International ongress on Aoustis, IA 1 was swepted from 1Ω to the value of the DSO, 1MΩ at 1 points per deade. Figure 3 shows the iruit diagram for the power harvesting simulations. For the experiments, this simply required the load resistane to be plaed between the reeived line and ground of Figure SIgnal Stength (mv) Figure 3. The Spie simulation iruit, with the A urrent soure and the soure apaitor as the piezoeletri reeiver in parallel with the Load and DSO ESULTS Frqueny (Hz) Figure 5. The frequeny response of the aousti ommuniations hannel Transfer Funtion Figure 4 shows the transfer funtion of the aoustiommuniations hannel at 1MHz. The relationship between the input signal strength and the output signal strength is linear, with a oeffiient of 1, whih orresponds to a 1% effiieny. The noise in the urve is due to small movements in the transmission medium. elative Amplitude (A.U.) y = x = Figure 6. The transient response of the aousti ommuniations hannel to an input tone burst Output Signal (mv) Input Signal (V) Figure 4. The transfer funtion of the aousti ommuniations hannel at resonane Frequeny esponse The frequeny response of the aousti-ommuniations hannel is show in figure 5. As expeted, a strong peak in the frequeny spetrum ours at the resonant frequeny of the piezoeletri transduers, that is, 1MHz. A seondary peak is notieable at 1kHz. Transient esponse Figure 6 show the transient response of the aoustiommuniations hannel for 1 yles at 1MHz. The ompat burst has a minor tail effet, elongating in time. One of the main reasons for this is due to the omposite nature of the ommuniations hannel. The various materials whih the body is made up of all have different aousti veloities. The result of this is that the various paths travelled by the ultrasound in the medium will result in signifiant temporal dispersion, and then interferene. Figure 7 show the transient response with enough yles to ahieve steady-state. The rise time is then given by approximately 5 yles, at 1MHz, giving 5µs. elative Amplitude (A.U.) E+.E-5 4.E-5 6.E-5-1 Figure 7. The transient response of the aousti ommuniations hannel showing the rise time Aousti ommuniation The results of the aousti ommuniations test are shown in Figures 8 to 11. Figure 8 shows the transmitted ASK (OOK) signal. inging is notieable as the signal is swithed off. Figure 9 shows the reeived ASK (OOK) signal. A low pass filter at above the data rate, but below the arrier frequeny, will reover the envelope, and the use of a omparator with a suitable ompare level will enable the digital information to be reovered. Figure 1 shows the reeived SK signal, whih ontains the data stream [ ]. The deoded SK signal is then shown in Figure 11. The original digital information an be reovered by seleting a digital 1 as a phase less than degrees, and a digital as a phase greater than degrees. 4 IA 1

5 3-7 August 1, Sydney, Australia roeedings of th International ongress on Aoustis, IA 1 Note that the transmitted SK signal is not shown, as no information is visible on the time sale of the entire signal. 8 4 Figure 13 shows the load urrent as a funtion of the output voltage (IV urve), and Figure 14 shows the power delivered to the load as a funtion of the output voltage (V urves), for the experimental, theoretial and simulated results. The V urve shows a measured peak power of 1mW, while theory and simulation give peak power values of 1.11mW and 1.15mW, respetively. Amplitude (V) Figure 8. The transmitted ASK (OOK) aousti ommuniations signal.1.5 Figure 11. The deoded SK aousti ommuniations signal [ ] Amplitude (V) Figure 9. The reeived ASK (OOK) aousti ommuniations signal.45.4 Figure 1. Voltage as a funtion of load resistane for the power harvesting.35 Amplitude (V) Figure 1. The reeived SK aousti ommuniations signal ower Harvesting The apaitane of the piezoeletri reeiver was measured to be 1.86nF. At the resonant frequeny of 1.35MHz, this gives a reatane of 141Ω. With an open iruit voltage of 57mV, (1) gives a short iruit urrent of 4mA. These values where then used in the Spie simulation of the A iruit. Figure 1 shows the omparison between the applied load and the voltage drop aross it, for both the experimental values and the simulated results. As expeted, as the load resistane dereases in size, the output voltage also dereases. Figure 13. Load urrent as a funtion of the voltage, IV urve DISUSSION As expeted, the transfer funtion is linear. Some randomness is notieable in the signal, hene the unertainty. It is worth noting that a similar unertainty would be expeted on all other results. The experiments were preformed with the arm as immobile as possible. A signifiant variation was notied when the arm/hand was allowed to artiulate. The peak value varied from around 14mV to 8mV, a fator of. This flutuation may be an issue, in partiular if ASK is used as IA 1 5

6 3-7 August 1, Sydney, Australia roeedings of th International ongress on Aoustis, IA 1 the enoding method. It is for this reason that SK would be a far more robust enoding method. Figure 14. ower delivered to the load as a funtion of the voltage, V urve The result of the transient response suggests that relatively high data rates may be ahievable, speifially if a transduer with a higher resonant frequeny was to be utilised. The use of a high ommuniations rate would redue the effet of flutuations due to motion of the ommuniations medium. The ASK ommuniations signal show the relatively high data rate that may be ahievable; that is, a data rate of 4kbps with a 1MHz arrier frequeny. The result of the SK ommuniations signal, figure 1, also shows that a relatively high data rate is possible. The phase transitions are not as quik as those show in previous work [6], when ommuniating through an aluminium panel, but the data rate is relatively high for the intended appliation. The preliminary results for the power harvesting are promising. The value of 1mW was signifiant ompared to values expeted. However, in the attempt to implement an A to D onverter, the very high frequeny appears to be limiting the ability to suessfully retify the output of the transduer. This is mainly due to the high juntion apaitane of the retifier diodes. In the onversion from A to D, the apaitane is an important onsideration to ahieve peak power output [9]. To resolve this issue, we intend to aquire transduers with a lower resonant frequeny, in the kilohertz range, and quantify the performane of the power onversion as a funtion of frequeny. EFEENES 1 V. K. Varadan, The role of nanotehnology and nano and miro-eletronis in monitoring and ontrol of ardiovasular diseases and neurologial disorders, ro. SIE 658 (7). Kim, T. Lehmanna and S. Nooshabadib, An ultrawideband transeiver for biotelemetry systems, ro. SIE 6798 (7) 3. Gabriel, S. Gabriely and E. orthout, The dieletri properties of biologial tissues: I. Literature survey, hys. Med. Biol. 41, (1996) 4 S. Gabriely,.W. Lau and. Gabriel, The dieletri properties of biologial tissues: II. Measurements in the frequeny range 1 Hz to GHz, hys. Med. Biol. 41, (1996) 5 S. Gabriely,.W. Lau and. Gabriel, The dieletri properties of biologial tissues: III. arametri models for the dieletri spetrum of tissues, hys. Med. Biol. 41, (1996) 6 G. Wild, Design and evaluation of an eletro-aousti ommuniations hannel for use by autonomous agents in the strutural health monitoring of ageless aerospae vehiles, Honours Thesis (Edith owan University, 5) 7 G. Wild and S. Hinkley, Eletro-aousti and aoustoopti ommuniations for roboti agents in smart strutures, ro. SIE 6414 (6) 8 G. Wild and S. Hinkley, Wireless aousti ommuniations for autonomous agents in strutural health monitoring sensor networks, ro. SIE 6798 (7) 9 G.K. Ottman, et al., Adaptive piezoeletri energy harvesting iruit for wireless remote power supply IEEE Trans. ower Eletron. 17(5), () 1 M.G. Silk, Ultrasoni Transduers for Nondestrutive Testing (Adam Hilger Ltd, Bristol, 1984). 11 M. Guan and W.H. Liao, omparative analysis of piezoeletri power harvesting iruits for rehargeable batteries in roeedings of the 5 IEEE International onferene on Information Aquisition (IEEE, 5) pp J. roakis and M. Salehi, ommuniation Systems Engineering (rentie Hall, New Jersey, 1994) 13 Mathworks In. 14 V.S. Mallela, V. Ilankumaran and N.S. ao, Trends in ardia paemaker batteries Indian aing Eletrophysiol. J. 4(4), 1 1 (4) However, with the suessful implementation of an A to D onverter, the measured power levels ould easily be utilised for the in-vivo reharging of a devie suh as a paamaker [14]. ONLUSION In onlusion, we have suessfully used aousti transmissions to both ommuniate and harvest power through a biologial medium, in-vivo. The aousti ommuniations show great promise for utilisation in pratial ommuniations with in-vivo biomedial devies; speifially in those appliations where loal heating effets of wireless F transmission is prohibitive. Even if the data rate was lowered in order to redue the bit error rate, and inrease reliability, a signifiant data rate ould be utilised, more than is neessary for stati biomedial devies. The power that an be delivered through the hannel is also promising, espeially for low power devies, suh as paemakers and neural implants. 6 IA 1