Sep 2008 doc IEEE 802 15 08 0714

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Project: IEEE P 802. 15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: Channel modeling for medical implanted communication systems by numerical simulation Date Submitted: [xx Sep, 2008] Source: Jaehwan Kim[ETRI], Hyung. Soo Lee[ETRI], Jeong Ki Pack[CNU], Tae Hong Kim[CNU] Contact: Jae Whan Kim, ETRI, Korea Voice: : +82 -42 -860 -5338, E-mail: kimj@etri. re. kr Re: [n/a] Abstract: Provide needs of channel modeling for medical implanted communication system Purpose: To provide basic channel characteristics for the manufacture of medical implantable communication system Notice: This document has been prepared to assist the IEEE P 802. 15. It is offered as a basis for discussion and is not binding on the contributing individual's or organization's. The material in this document is subject to change in form and content after further study. The contributor's reserves the right to add, amend or withdraw material contained herein. Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and maybe made publicly available by P 802. 15. Submission Slide 1

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Contents • Channel models for BAN • Methods for channel modeling • Channel modeling 1 • Channel modeling 2 • Channel modeling 3 • Conclusions Submission Slide 2

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Channel models for WBAN Scenario Submission Description Frequency Band Channel Model S 1 Implant to Implant 402 -405 MHz CM 1 S 2 Implant to Body Surface 402 -405 MHz CM 2 S 3 Implant to External 402 -405 MHz CM 2 S 4 Body Surface to Body Surface (LOS) TBD (f 1, … fn) CM 3 S 5 Body Surface to Body Surface (NLOS) TBD (f 1, … fn) CM 3 S 6 Body Surface to External (LOS) TBD (f 1, … fn) CM 4 S 7 Body Surface to External (NLOS) TBD (f 1, … fn) CM 4 Slide 3

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Methods for channel modeling • Channel modeling - Scenario 1(CM 1), Scenario 2(CM 2 ) - FDTD method was used for channel modeling using Remcom XFDTD 6. 5 - Frequency: 403. 5 MHz • Human body model - Korean male phantom model (voxel size : 3 mm) • TX antenna - Hertzian dipole - Channel models must not be affected by the transmitting antenna pattern. So the proper compensation for directive gain as well as polarization are needed. Submission Slide 4

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 • Simulation scenario − Transmitter location : 17 positions 16 15 1 2 3 1114 4 127 13 586 10 9 Ø Near surface implants : 9 Ø Deep tissue implants : 8 Location Device Deep tissue Capsule endoscope, Capsule for drug delivery(1 -4, 69) Glucose-Insulin(10 -11), Insulin pump(12 -13) Pacemaker(14) Near surface Deep brain stimulator, Parkinson’s disease, Cortical stimulator, Visual neuro-stimulator, Audio-neuro stimulator(15 -16) Central-nerve stimulator(11) Healthcare shoes (17) 17 Submission − Receiver location : 138 points Ø Implants(in-body) : 78 Ø Body surface : 60 Slide 5

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Channel modeling 1 • Path loss model - Path loss model used in IEEE P 802. 15 -08 -0033 -05 -0006 - PL(d)=PL(d 0)+10 nlog 10(d/d 0)+S [d. B] Ø d 0 : reference distance, 50 mm Ø n : path loss exponent Ø S : random scatter around the regression line, N(0, σs) • Channel modeling - Grouping of the transmitter location Ø Deep tissue implant Ø Near surface implant - Grouping of the receiver location Ø In-body (implant) Ø Body surface : from the skin to 2 cm away from the skin Submission Slide 6

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(CM 1) • Deep tissue implant to another implant • Near surface implant to another implant (the maximum path length is about 180 cm) ( n=4. 33, PL(d 0)=47. 70, σs=5. 77 ) Submission Slide 7 (n=3. 65, PL(d 0)=46. 52, σs=8. 84)

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(CM 2) • Deep tissue implant to body surface (the maximum path length is about 180 cm) • Near surface implant to body surface (the maximum path length is about 180 cm) (n=5. 22, PL(d 0)=37. 28, σs=5. 76) (n=3. 53, PL(d 0)=45. 37, σs=9. 46) Submission Slide 8

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(CM 1) • Deep tissue implant to another implant (fitted up to 50 cm) • Near surface implant to another implant (fitted up to 50 cm) (n=6. 17, PL(d 0)=34. 89, σs=5. 44) Submission (n=5. 34, PL(d 0)=35. 48, σs=8. 42) Slide 9

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(CM 2) • Deep tissue implant to body surface (fitted up to 50 cm) (n=6. 06, PL(d 0)=31. 95, σs=5. 75) Submission • Near surface implant to body surface (fitted up to 50 cm) (n=4. 34, PL(d 0)=40. 17, σs=10. 09) Slide 10

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Summary of the channel modeling 1 • When we fit the path loss model for the whole receiver locations (the maximum path length is about 180 cm), the modeling parameters are slightly different from the results of IEEE 802. 15 -08 -0519 -00 -0006. • However, the parameter values are well within the statistical error bound (one σs value). <Implant to Implant CM 1 (Scenario S 1)> Implant to Implant PL(d 0) n σs Deep tissue 47. 70 4. 33 5. 77 Near surface 46. 52 3. 65 8. 84 <Implant to Body surface CM 2 (Scenario S 2)> Implant to body surface PL(d 0) n σs Deep tissue 37. 28 5. 22 5. 76 Near surface 45. 37 3. 53 9. 46 Submission Slide 11

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 • When fitted up to 50 cm as shown in IEEE 802. 1508 -0519 -00 -0006, we obtained similar results as in IEEE P 802. 15 -08 -0033 -05 -0006. <Implant to Implant CM 1 (Scenario S 1)> Implant to Implant PL(d 0) n σs Deep tissue 34. 89 6. 17 5. 44 Near surface 35. 48 5. 34 8. 42 <Implant to Body surface CM 2 (Scenario S 2)> Implant to body surface PL(d 0) n σs Deep tissue 31. 95 6. 06 5. 75 Near surface 40. 17 4. 34 10. 09 Submission Slide 12

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Channel modeling 2 • Path loss model - Same as the model used in IEEE P 802. 15 -08 -0033 -05 -0006 • Channel modeling - We tried channel modeling for different scenarios (different grouping of TX’s or RX’s). - Modeling 2 A Ø Path loss was modeled for total implant locations (i. e. , the whole TX’s are grouped together for fitting) - Modeling 2 B Ø Path loss was modeled for the total implant locations with the RX points grouped differently. Ø Receiver groups: head, trunk, lower parts of the body, arms - Modeling 2 C Ø Path loss was modeled for the type of the transmitting implanted devices (Implant to Implant) Ø Type of devices : capsule endoscope, glucose-insulin, pacemaker, deep brain stimulator, healthcare shoes Submission Slide 13

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(2 A) • Total implant(deep tissue and near surface) to implant • Total implant(deep tissue and near surface) to body surface (n=3. 83, PL(d 0)=48. 46, σs=8. 53) Submission Slide 14 (n=4. 01, PL(d 0)=43. 97, σs=8. 89)

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Summary of the channel modeling 2 A • Modeling parameters fitted for all transmitting implanted devices PL(d 0) n σs Implant to Implant 48. 46 3. 83 8. 53 Implant to Body surface 43. 57 4. 01 8. 89 Submission Slide 15

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(2 B) • Implant to head (n=4. 03, PL(d 0)=47. 11, σs=8. 65 ) Submission • Implant to lower parts of the body (n=3. 75, PL(d 0)=47. 89, σs=9. 78) Slide 16

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(2 B) • Implant to trunk (n=4. 26, PL(d 0)=44. 99, σs=8. 21 ) Submission • Implant to arms (n=3. 81, PL(d 0)=48. 02, σs=7. 97) Slide 17

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Summary of the channel modeling 2 B • Modeling parameters for different RX locations ( Implant to Implant) Head Lower parts of the body Trunk Arms PL(d 0) 47. 11 47. 89 44. 99 48. 02 n 4. 03 3. 75 4. 26 3. 81 σs 8. 65 9. 78 8. 21 7. 97 Submission Slide 18

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(2 C) • Capsule endoscope (gullet) • Glucose Insulin (right hand) (n=4. 48, PL(d 0)=45. 57, σs=6. 42 (n=3. 23, PL(d 0)=43. 52, σs=6. 46) ) Submission Slide 19

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(2 C) • Pacemaker (heart) (n=4. 29, PL(d 0)=46. 75, σs=6. 55 ) Submission • Deep brain stimulator (right throat) (n=3. 60, PL(d 0)=47. 96, σs=10. 07) Slide 20

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path Loss vs. Distance Scatter Plot(2 C) • Healthcare shoes (sole) (n=3. 83, PL(d 0)=36. 72, σs=5. 54 ) Submission Slide 21

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Summary of the channel modeling 2 C • Modeling parameters for different transmitting implanted devices (Implant to Implant) Capsule endoscope (gullet) Glucose. Insulin (right hand) Pacemaker (heart) Deep brain stimulator (right throat) Healthcare shoes (sole) PL(d 0) 45. 57 43. 52 46. 75 47. 96 36. 72 n 4. 48 3. 23 4. 29 3. 60 3. 83 σs 6. 42 6. 46 6. 55 10. 07 5. 54 Submission Slide 22

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Summaries of the channel modeling 2 • We have modeled WBAN channels for three different modeling scenarios(2 A, 2 B, 2 C). • The results of the modeling scenario 2 A are similar to those of the channel modeling 1. Thus, it seems that the modeling scenario in IEEE 802. 1508 -0519 -00 -0006 could be simplified. • The results of the modeling scenario 2 B show that there are no large difference in the model parameters for different body parts. So, the modeling scenario in IEEE 802. 15 -08 -0519 -000006 seems to be good. Submission Slide 23

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 • When we classify the transmitting implants in more detail(modeling scenario 2 C), the parameter values are similar to the results of the modeling scenario 2 A. • Thus, the detailed classification of the transmitting implants does not seem to be necessary. Submission Slide 24

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Channel modeling 3 • Path loss model - PL(d)=PL(d 0)+10 nlog 10(d/d 0)+ a*d +S [d. B] Ø d 0 : reference distance, 50 mm Ø a : coefficient for absorption loss Ø n : path loss exponent, N(0, σs) Ø S : random scatter around the regression line - The absorption loss of biological tissues is very large. Because the absorption loss is a linear term in a log scale, we tried a modified path loss model by adding the first order term to WBAN channel more accurately. • Channel modeling - Same as in the channel modeling 1 Submission Slide 25

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path loss vs. Distance Scatter Plot(CM 1) • Deep tissue implant to another implant • Near surface implant to another implant (n=5. 87, PL(d 0)=36. 80, a=-19. 70, σs=8. 22) (n=7. 39, PL(d 0)=33. 58, a=-28. 00, σs=5. 07) Submission Slide 26

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Path loss vs. Distance Scatter Plot(CM 2) • Deep tissue implant to body surface • Near surface implant to body surface (n=4. 89, pl=39. 83, a=-13. 44, σs=9. 30) (n=7. 38, PL(d 0)=29. 37, a=-24. 96, σs=5. 45) Submission Slide 27

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Summary of the channel modeling 3 • The comparison of the results of the channel modeling 3 shows that the modified pass loss model seems to be better, in terms of the modeling accuracy (i. e. standard deviation σs). <Implant to Implant CM 1 (Scenario S 1)> Implant to Implant PL(d 0) n a σs Deep tissue 33. 58 7. 39 -28. 00 5. 07 Near surface 36. 80 5. 87 -19. 70 8. 22 <Implant to Body surface CM 2 (Scenario S 2)> Submission Implant to body surface PL(d 0) n a σs Deep tissue 29. 37 7. 38 -24. 96 5. 45 Near surface 39. 83 4. 89 -13. 44 9. 30 Slide 28

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 Conclusions • We have modeled WBAN channel by two path loss models. • We also tried different scenarios for channel modeling. • Extensive simulation to characterize the MICS path loss has been performed and statistical path loss models at 403. 5 MHz are derived. • The models are based on 9 near surface implants and 8 deep tissue implants for the Korean male phantom model. Submission Slide 29

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 • The modeling results show that Ø Classification of the implants to two groups(deep tissue and near surface) might not be necessary. Ø The modified path loss model seems to work better. Ø The modeling parameters could be different depending on the path length for fitting, the location of TX’s or RX’s, but they are well within the statistical error bound (one σs value). Submission Slide 30

Sep 2008 doc. : IEEE 802. 15 -08 -0714 -00 -0006 References • Doc : 15 -08 -0519 -00 -0006 A statistical path loss model for MICS • Doc : 15 -08 -0033 -05 -0006 Channel model for body area network (BAN) Submission Slide 31