Publication V Institute of Physics and Engineering in Medicine (IPEM) Reprinted by permission of Institute of Physics Publishing.

Transcription

1 Publicatin V T. Uusitupa, I. Laaks, S. Ilvnen, and K. Nikskinen. 2. SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures. Physics in Medicine and Bilgy, vlume 55, number 4, pages Institute f Physics and Engineering in Medicine (IPEM) Reprinted by permissin f Institute f Physics Publishing.

2 IOP PUBLISHING Phys. Med. Bil. 55 (2) PHYSICS IN MEDICINE AND BIOLOGY di:.88/3-955/55/4/7 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures T Uusitupa, I Laaks, S Ilvnen and K Nikskinen Department f Radi Science and Engineering, Helsinki University f Technlgy, Otakaari 5A, 25 Esp, Finland ter.uusitupa@tkk.fi Received September 29, in final frm 29 December 29 Published 28 January 2 Online at stacks.ip.rg/pmb/55/57 Abstract An extensive study n specific absrptin rate (SAR) cvering 72 simulatins and 5 vxel mdels (8 5 kg) has been perfrmed by applying the parallel finite-difference time-dmain methd. High-reslutin whle-bdy mdels have been irradiated with plane waves frm MHz t 5 GHz by applying varius incming directins and plarizatins. Detailed results f whlebdy SAR and peak g SAR are reprted, and SAR variatin in the db scale is examined. Fr an adult, the effect f incming directin n whlebdy SAR is larger in the GHz range than at arund 45 MHz, and the effect is strnger with vertical plarizatin. Fr a child (height.2 m), the effect f incming directin is similar as fr an adult, except at MHz fr hrizntal plarizatin. The effect f the phantm (8 5 kg) n whlebdy SAR is larger at arund 2 5 GHz and at vertical MHz (prximity f whle-bdy resnance fr the child) than at arund hrizntal MHz. Bdy psture has little effect n whle-bdy SAR in the GHz range, but at arund 45 MHz, ne may even expect a 2 db rise in whlebdy SAR if psture is changed frm the standing psitin. Psture affects peak g SAR much mre than whle-bdy SAR. The plarizatin f the incident electric field may have an effect f several db n whle-bdy SAR. Between 2 and 5 GHz fr adults, whle-bdy SAR is higher fr hrizntal than fr vertical plarizatin, if the incming directin is in the azimuth plane. In the GHz range, hrizntal plarizatin gives higher whle-bdy SAR, especially fr irradiatin frm the lateral directin. A cmparisn between hmgeneus and hetergeneus mdels was dne. A hmgenized mdel underestimates whle-bdy SAR, especially at 2 GHz. The basic restrictin f whle-bdy SAR, set by ICNIRP, is exceeded in the smallest mdels ( 2 kg) at the reference level f expsure, but als sme adult phantms are clse t the limit. The peak g SAR limits were never exceeded in the studied cases. The 3-955//457+2$3. 2 Institute f Physics and Engineering in Medicine Printed in the UK 57

3 58 T Uusitupa et al present ICNIRP guidelines shuld be revised by lwering the reference levels, especially at arund 2 5 GHz. (Sme figures in this article are in clur nly in the electrnic versin). Intrductin Mdern sciety invlves many applicatins and services based n electrmagnetic (EM) fields, and it has been fund reasnable t cntrl the EM-field expsure f the general public and wrkers in rder t avid pssible adverse health effects. Thus, e.g. ICNIRP (998) has defined basic restrictins (limits) and reference levels n EM-field expsure. In the radi frequency and micrwave range, the mst relevant dsimetric quantity is the specific absrptin rate (SAR) fr which limits, based n established health effects, have been defined. The purpse f reference levels is t prvide a practical way t assess expsure: if measured r mdeled field strength r pwer density is belw the crrespnding reference level, SAR is belw the basic restrictin. SAR can be cmputed by the finite-difference time-dmain (FDTD) methd with realistic human bdy mdels. SAR depends n expsure cnditins such as frequency, field plarizatin and the wave incming directin, and n the bdy size and anatmy. T btain data fr simple SAR frmulas r statistical mdels f SAR variatin ( uncertainty ), it is necessary t study SAR via a large number f full-wave field cmputatins. Furthermre, it is imprtant t examine under which cnditins the SAR values stay belw the limits set by e.g. ICNIRP (998). Cnil et al (28) studied SAR variability in adults and children, frm 2 MHz t 2.4 GHz, fr a plane wave irradiating an islated bdy frm the frnt with the electric field vertically plarized. In child mdels, btained via mrphing frm adults, whle-bdy SAR typically exceeded the basic restrictin at the reference level f expsure at arund 2 GHz. Similar results were btained in Kühn et al (29) where SAR was studied under plane wave expsure, between 5 and 245 MHz, with fur mdels f different size, and als incming directin and plarizatin were varied. In Kühn s paper the child mdels had been btained frm MRI (magnetic resnance imaging) data. Dimbylw and Blch (27) adapted University f Flrida child phantms fr SAR calculatin between 5 MHz and 4 GHz (frntally incident plane waves, vertical plarizatin). In their paper, t, basic restrictin f whle-bdy SAR was exceeded by smaller children at the reference level f expsure. Hwever, it can be remarked that Hirata et al (28) shwed by a cmputatinal thermal mdel that the same whle-bdy SAR causes less temperature elevatin in a small child than in an adult. The effect f bdy psture n SAR was studied by Findlay and Dimbylw (25) by applying fur pstures in the range MHz. Results were presented fr frntally incident plane waves under islated and grunded cnditins. Seated vxel mdels f an adult and scaled versins representing children were studied by Findlay and Dimbylw (26) fr vertically and hrizntally plarized frntally incident plane wave, under islated and grunded cnditins, between MHz and 3 GHz. A nvel psture transfrmatin methd was reprted in Nagaka and Watanabe (28) where sitting mdels were expsed t frntally incident plane waves, fr vertical plarizatin, between 3 MHz and 3 GHz. In this wrk, 72 free-space plane-wave simulatins have been perfrmed in the frequency range frm MHz t 5 GHz fr tw rthgnal plarizatins and fr varying incming directin. This frequency range cvers many applicatins, e.g. GSM (Glbal System

4 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures 59 φ=36 (back) φ=33 φ= θ=5, φ= z z z φ=27 y θ=2, φ= x φ k E v E h θ mdels lk in the +x directin y RIGHT LEFT HAND HAND x φ=2 φ= (frntal) φ=24 azimuth plane (θ= ) θ=6, φ= x θ=3, φ= elevatin plane y Figure. Left: in the simulatins, the prpagatin directin f an incming plane wave (k) is specified using the angles θ and φ. Vertical plarizatin means that the electric field vectr (E v ) lies in the shaded plane. Hrizntal plarizatin means that the electric field vectr (E h ) is nrmal t this shaded plane. E h des nt have a u z cmpnent. Vectrs E v, E h and k are rthgnal. All the mdels, in every simulated case, are lking twards the +x directin. Right: incming directins illustrated. fr Mbile Cmmunicatins) and varius wireless bradband services: 45 MHz can be allcated fr e.g. t wireless Internet in sparsely-ppulated areas, GSM is widely used at MHz, UMTS (Universal Mbile Telecmmunicatins System) at 2 MHz, and WiMAX (Wrldwide Interperability fr Micrwave Access) at 35 MHz. 5 bdy mdels f different type and weight (8 5 kg) have been irradiated, and als bdy psture has been varied. It is wrth emphasizing that in the gigahertz range high mdel reslutin (abut mm) has t be used in FDTD, which requires huge memry and cmputing resurces, especially fr adult phantms and pstures ccupying larger cmputatin vlume (e.g. arms stretched ut). Due t an exceptinally large number f mdels and incming directins, and high frequencies, this wrk invlves a very large number f high-reslutin adult phantm simulatins, which has necessitated the usage f parallel cmputatin. Als, the effect f psture n SAR has been btained in a wide frequency range up t 5 GHz. All the key results f the FDTD-SAR simulatins, whle-bdy averaged SAR (SAR WB ) and peak g SAR in the limbs and in the head/trunk regin have been reprted and presented in the figures, i.e. nt nly SAR WB. 2. Methds 2.. Expsure setup and FDTD-based SAR cmputatin The bdy mdels were expsed t free-space plane waves fr tw rthgnal plarizatins at, 45,, 2, 35 and 5 MHz. The SAR simulatins were perfrmed by applying the FDTD methd. Excitatin signals were tapered sine waves at the pint frequencies specified abve. In each plarizatin frequency case, the incming directin and the human mdel were varied. Figure shws the essential symbls f plane-wave expsure. In-huse parallel FDTD and SAR cdes were run in an HP supercluster in CSC (Finnish IT Center fr Science). The FDTD cde uses the effective cnvlutinal perfectly matched layer (CPML) absrbing bundary cnditin. The SAR cdes (SAR averaging) have been implemented accrding t IEEE Std C95.3 (22), Annex E, and Caputa et al (999). The methds and cdes were validated befre the simulatins by an interlabratry cmparisn. The cell size in the unifrm FDTD grid was always equal t the phantm reslutin which is shwn in table fr each phantm and fr the lw and high reslutin cases (crrespnding This was dne with the prject partners f the MMF-GSMA dsimetry prgram, phase II, which has been backgrund wrk fr the develpment f expsure assessment standard IEC

5 6 T Uusitupa et al Table. Applied mdels. The table shws the mdel name, weight, reslutins (lw/high), type f each mdel and the number f incming directins fr each mdel. Lw- and high-reslutin mdel weights may slightly differ. The type f the mdel is hmgeneus (tissue-equivalent liquid phantm) r hetergeneus. A single incming directin means frntal expsure. 2 directins refer t frntal and back expsure. Mdel name Weight (kg) Reslutin (mm) Type Incming directins Nrman /. Hetergeneus 2 Japan Male /. Hetergeneus 2 Japan Female /. Hetergeneus 2 VHP Male 5.38/ /. Hetergeneus 2 Psture sit /. Hmgeneus Psture examine /. Hmgeneus Psture arms up /. Hmgeneus Psture standing /. Hmgeneus Psture sit n grund /.25 Hmgeneus Psture arms ut /.25 Hmgeneus VF Male Hmgeneus 68.54/ /. Hmgeneus VF Male Heterg /. Hetergeneus VF By Hmgeneus 8.28/ /. Hmgeneus VF By Heterg 9.4/ /. Hetergeneus VF Female Heterg /. Hetergeneus 2 cell sizes are 2 mm and mm fr mst f the phantms). In general, lwer reslutin was used at 2 MHz and higher reslutin at 35 5 MHz (see sectin 2.2). The CPML layer thickness was seven cells, and the CPML-phantm distance was seven r eight cells, which has been fund sufficient earlier (Laaks et al 27) Human bdy mdels The FDTD-SAR cmputatins were dne by applying 5 different human bdy mdels: the set includes seven hetergeneus mdels, tw hmgenized versins and a hmgeneus phantm in six pstures. The mdels are listed in table with their weights and reslutins and are illustrated in figures 2 and 3. Actually, there are tw versins f each mdel: a lw-reslutin mdel was applied at, 45, and 2 MHz and a high-reslutin mdel was applied at 35 and 5 MHz (but the high-reslutin versin f Visible Human Prject (VHP) Male was used als at 2 MHz). Fr VHP Male and Virtual Family (VF), there were lw- and high-reslutin versins riginally available. Fr Nrman, Japan Male, Japan Female and the six psture mdels, the riginal phantms were available at abut 2 mm reslutin. The high reslutin mdels were btained by dividing each cell int eight cells. The vxel data f Nrman were btained frm the Health Prtectin Agency/UK (Dimbylw 997). Japanese Male and Female were btained frm the Bimedical EMC Grup/NICT/Japan (Nagaka et al 24). VHP Male is based n the Visible Human Prject and the mdels have been btained frm Brks Air Frce Labratry/USA (Ackerman 998, Masn et al 2). The psture mdels were utputted frm Pser sftware and eventually crrected and manipulated in Matlab (c-peratin with STUK 2 ). The vxel data f the riginal hetergeneus Virtual Family were received frm the IT IS Fundatin/Switzerland 2 STUK: Radiatin and Nuclear Safety Authrity (in Finland).

6 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures 6 Figure 2. Tp rw: VF Male (VFM), VF Female (VFF) and VF By (VFB). Bttm rw: VHP Male, Nrman, Japanese Male and Japanese Female. (Christ et al 2). Fr the hetergeneus mdels the electrical prperties f bdy tissues, cnductivity and permittivity were btained frm the parametric mdels f Gabriel et al (996). Fr the hmgeneus bdy phantms, the tissue density was kg m 3 and the electrical parameters were as in table Results 3.. Detailed nrmalized SAR results fr incming angle, mdel and psture variatin The SAR results shwn in this paper have been nrmalized accrding t reference levels and SAR limits f ICNIRP (998). A nrmalized SAR value is defined as SAR nrm = SAR SAR limit when S inc (f ) S ref (f ) ()

7 62 T Uusitupa et al (a) (b) (c) (d) (e) (f) Figure 3. Psture mdels: (a) sit, (b) examine, (c) arms up, (d) standing, (e) sit n grund, (f) arms ut. All mdels are psitined in free space. where SAR may refer t SAR WB (whle-bdy SAR), SAR g,ht (peak g SAR in head/trunk) r SAR g,li (peak in limbs) and SAR limit is the crrespnding basic restrictin (table 2). S inc is the pwer density f the incming plane wave and S ref (f ) is the crrespnding frequency-dependent reference level (table 2). Imprtantly, if a nrmalized SAR value exceeds, this means that the SAR limit is exceeded at the reference level f expsure, which shuld nt happen (but the latest research, e.g. Kühn et al (29), Cnil et al (28) and Dimbylw and Blch (27), suggests that this can happen in sme cases). A nrmalized SAR has exactly the same meaning fr public and ccupatinal expsure, because the ccupatinal reference levels and SAR limits bth are five times thse f the general public Curves f. Figures 4 and 5 shw at frequencies, 45,, 2, 35 and 5 MHz, fr vertical and hrizntal plarizatin, fr all the

8 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures 63 Table 2. Relative permittivity ɛ r and cnductivity σ fr the hmgeneus mdels (IEC 29), and the pwer density reference levels S ref f ICNIRP fr the general public, tgether with the crrespnding limits fr the whle-bdy SAR, peak g SAR in head/trunk and peak g SAR in limbs. Frequency S ref SAR WB, limit SAR g,ht, limit SAR g,li, limit (MHz) ɛ r σ (S m ) (Wm 2 ) (W kg ) (Wkg ) (Wkg ) mdels. One can see the variatin f as the incming directin is changing. In the gigahertz range f 2 MHz, it is bserved that SAR WB is strngly related t the the area f bdy silhuette presented t the wave 3, i.e. SAR WB is higher when the wave cmes frm the frnt r back and smaller when the wave cmes frm the left-hand side, beneath-frnt r abve-frnt. Hwever, at the lwer frequencies f< GHz, the abve cncept f silhuette area des nt predict directin dependence f SAR WB any lnger. Firstly, fr example, at MHz with vertical plarizatin, VF By Hetergeneus has its maximum SAR WB when the wave cmes frm the left-hand side f the mdel. Secndly, if the frequency is lw, but the whle-bdy resnance is nt effective (e.g. VF Male at MHz r VF By at 45 MHz), the SAR WB (θ) curve may pen up, nt dwn as always in the GHz range. That is, SAR WB may smewhat increase in the few-hundred-mhz regin when θ deviates frm. Furthermre, in VF By, at MHz, SAR WB is ntably higher fr vertical plarizatin, which is explained by the prximity f the whle-bdy resnance Nrmalized SAR fr incming directin variatin. Figures 6, 7 and 8 shw the incming directin variatin results f,sar g,ht and SAR g,li, respectively, fr hetergeneus and hmgeneus VF Male and VF By. In this cmpactfrm presentatin each marker represents an incming directin ( directins ttal) and the plarizatin frequency cases are indicated alng the hrizntal axis: H and V refer t hrizntal and vertical plarizatins, respectively, and the frequency is in MHz. In figure 6,the highest values are fund at 2 MHz, which is due t the lwer reference levels f incming pwer density at, 45 and MHz (table 2) and due t the tendency f the SAR WB decrease abve 2. GHz (where S ref is cnstant). The wrst case is vertical 2 MHz with VF By Hetergeneus under frntal expsure where the limit is exceeded by 54%. At f MHz with bth plarizatins, SAR WB is maximized when the wave cmes directly frm the frnt r directly frm the back, and mre ften the frntal expsure gives the maximum value (figure 6). Because this kind f directin dependence is very prbable fr any bdy mdel in standing psture, in rder t get the wrst-case SAR WB value at a given (high enugh) frequency, it suffices t simulate frntal and back expsure cases nly. It is unlikely that an arbitrary incming directin culd prduce substantially higher SAR WB. Hwever, abve cnvenient rule fr the wrst-case SAR WB des nt hld at lwer frequencies and 45 MHz, where, actually, the maximum SAR WB was never btained under frntal 3 Area f bdy prjectin nt the plane rthgnal t the prpagatin directin f the plane wave.

9 64 T Uusitupa et al Nrman Japan Male Japan Female VHP Male Psture sit Psture examine Psture arms up Psture standing Psture sit n grund Psture arms ut VF Male Hmgeneus VF Male Hetergeneus VF By Hmgeneus VF By Hetergeneus VF Female Hetergeneus Vertical plarizatin, MHz. Hrizntal plarizatin, MHz θ: φ: Vertical plarizatin, 45 MHz θ: φ: Hrizntal plarizatin, 45 MHz θ: φ: Vertical plarizatin, MHz θ: φ: Hrizntal plarizatin, MHz θ: φ: θ: φ: Figure 4. Nrmalized SAR WB at, 45 and MHz fr vertical and hrizntal plarizatin. Incming directin f the plane wave, prpagating in free space, is indicated by θ and φ (figure ). r back expsure. But because the values are relatively small there, it is very unlikely that an arbitrary incming directin culd cause the t apprach the limit. Frm figure 6 it is als bserved that in the hetergeneus mdels the values have a clearly decreasing tendency with frequency abve 2 GHz. Because in the hmgeneus mdels this tendency is weaker, hmgeneus phantms becme cnservative fr SAR WB estimatin at 5 GHz (this issue is discussed mre in sectin 3..5). Finally, figures 7 and 8 shw that nrmalized SAR g,ht and SAR g,li never exceed the

10 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures 65 Nrman Japan Male Japan Female VHP Male Psture sit Psture examine Psture arms up Psture standing Psture sit n grund Psture arms ut VF Male Hmgeneus VF Male Hetergeneus VF By Hmgeneus VF By Hetergeneus VF Female Hetergeneus.6 Vertical plarizatin, 2 MHz..6 Hrizntal plarizatin, 2 MHz θ: φ: Vertical plarizatin, 35 MHz. 2 5 θ: φ: Hrizntal plarizatin, 35 MHz θ: φ: Vertical plarizatin, 5 MHz. 2 5 θ: φ: Hrizntal plarizatin, 5 MHz θ: φ: θ: φ: Figure 5. Nrmalized SAR WB at 2, 35 and 5 MHz fr vertical and hrizntal plarizatin. Incming directin f the plane wave, prpagating in free space, is indicated by θ and φ (figure ). limit. Nrmalized peak g SAR values tend t be higher in the head/trunk regin than in the limbs, mainly because the SAR limit in the head/trunk is half f that in the limbs Nrmalized SAR fr mdel variatin. Figure 9 shws the mdel variatin results f,sar g,ht and SAR g,li fr frntal and back expsure. Here, each marker represents a specific mdel in a standing psture (ttal mdels in frntal case, 9 mdels in back-expsure case). At the reference level f expsure, the SAR WB limit is exceeded in VF By at several frequencies. Furthermre, the female mdels are quite

11 66 T Uusitupa et al VF Male Hetergeneus VF Male Hmgeneus.8.6 θ= φ= θ= φ=2 θ= φ=24 θ= φ=27 θ= φ= θ= φ=33 θ= φ=36 θ=3 φ= θ=6 φ= θ=2 φ= θ=5 φ= VF By Hetergeneus.6 VF By Hmgeneus Figure 6. Incming directin variatin. Nrmalized SAR WB distributins are shwn fr VF Male and VF By (hetergeneus and hmgeneus mdels) in each plarizatin-frequency case. clse t the limit at 2 MHz: fr VF Female, =.94 at hrizntal 2 MHz. VHP Male (5.4 kg) has the lwest SAR WB in almst every case. Calculatin f rati SAR WB,frntal /SAR WB,back fr nine mdels reveals that this is usually between. and.4 at the higher frequencies 2, 35 and 5 MHz. That is, in the GHz range, slightly higher SAR WB under a frntally incident plane wave is typical fr the human anatmy and may result frm the fact that the frnt side f the bdy has mre skin area and rugh surface (tes, feet, genitals, chin, face) than the back side. Nrmalized SAR g,ht and SAR g,li are always less than (figure 9), i.e. the peak g SAR limit is nt exceeded by any mdel in any simulated case at the reference level f expsure. Nrmalized peak g SAR tends t be higher in the head/trunk regin than in the limbs, especially under frntal expsure when, fr example, nse and penis are directly irradiated. In the gigahertz regin, typical peak SAR lcatins are fingers, tes, nse, ears, chin, penis and testicles, because the field can effectively enter inside these kind f bdy parts which have relatively small dimensins. The SAR WB results f VF Male Hmgeneus and Psture standing are very similar (figure 9, tp-left), because the mdels have rughly the same weight and shape. Because they seem equivalent in terms f SAR WB, the psture variatin results f SAR WB (sectin 3..4) are prbably directly usable fr VF Male Hmgeneus, t. Hwever, Psture standing tends t have higher SAR g,ht values, which is very clearly bserved in the frntal-v2 case where its nrmalized SAR g,ht =.92 (figure 9, middle-left). Als, this issue is discussed in mre detail in sectin 3..4.

12 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures 67 VF Male Hetergeneus VF Male Hmgeneus nrmalized SAR g, head/trunk θ= φ= θ= φ=2 θ= φ=24 θ= φ=27 θ= φ= θ= φ=33 θ= φ=36 θ=3 φ= θ=6 φ= θ=2 φ= θ=5 φ= nrmalized SAR g, head/trunk VF By Hetergeneus VF By Hmgeneus nrmalized SAR g, head/trunk nrmalized SAR g, head/trunk Figure 7. Incming directin variatin. Nrmalized SAR g,ht distributins are shwn fr VF Male and VF By (hetergeneus and hmgeneus mdels) Nrmalized SAR fr psture variatin. Figure shws the psture variatin results f,sar g,ht and SAR g,li fr frntal expsure. Each marker in the figure represents a different psture (figures 3(a) (f)). At 2, 35 and 5 MHz, a sitting psture, sit (a) r sit n grund (e), minimizes SAR WB and the regular standing psture (d) gives SAR WB clse t the maximum, which is typically btained by stretching the arms up (c) r ut (f). At and 45 MHz, hwever, a change frm the regular standing psture can increase SAR WB sme tens f percents i.e. abut 2 db. Finally, the set f assessed pstures shuld be representative in the sense that it may be difficult t find a new psture giving significantly higher SAR WB. In general, peak g SAR is mre sensitive t the psture than SAR WB. Furthermre, these particular psture mdels have sme high nrmalized SAR g,ht values in the penis fr vertical plarizatin (figure, tp-right): this happens at 2 MHz with all the standing pstures, and at 35 and 5 MHz with sit n grund. Investigatin f the 3D g-sar distributins reveals that anatmy and rientatin f the penis may have a cnsiderable effect n g SAR in this rgan. Fr VF Male Hmgeneus, cmpared with Psture standing, the g SAR in the penis is abut 5 db smaller (case V2 ), and the lcatin f SAR g,ht is in nse. Als, fr all the standing pstures (figures 3(b), (c), (d) and (f)) at 2 MHz, SAR g,ht is lcated in the penis fr bth plarizatins, but the value is abut 3 db higher with vertical plarizatin Cmparisn f hetergeneus and hmgeneus mdels. SAR values may increase if a hmgeneus mdel is replaced with a realistic hetergeneus ne. Namely, examinatin f the VF Male and VF By results gave a SAR WB increase f.3 db and a peak g SAR

13 68 T Uusitupa et al.5 VF Male Hetergeneus.5 VF Male Hmgeneus nrmalized SAR g, limbs.3.2. θ= φ= θ= φ=2 θ= φ=24 θ= φ=27 θ= φ= θ= φ=33 θ= φ=36 θ=3 φ= θ=6 φ= θ=2 φ= θ=5 φ= nrmalized SAR g, limbs VF By Hetergeneus.5 VF By Hmgeneus nrmalized SAR g, limbs.3.2. nrmalized SAR g, limbs Figure 8. Incming directin variatin. Nrmalized SAR g,li distributins are shwn fr VF Male and VF By (hetergeneus and hmgeneus mdels). increase f abut 2 db in the wrst case (figure ). With tissue-layering analyses (Christ et al 26), an even higher increase f peak g SAR may be btained. Figure shwing the hetergeneus-hmgeneus-mdel cmparisn cvers directins and 2 mdels f clearly different size (VF Male and By). Fr predictin f SAR WB, the hmgeneus mdel wrks quite well at and 45 MHz and is always cnservative at 5 MHz. At 2 MHz, the hmgeneus mdel clearly underestimates SAR WB, and the underestimatin wuld be even slightly larger if the hmgeneus mdel had the same weight as the hetergeneus mdel (abut 5% larger, see table ). Based n these results, in the wide band 5 MHz, the accuracy f the hmgeneus mdel fr SAR WB calculatin is abut.3 db and abut 2 3 db fr peak g SAR Field plarizatin and SAR. Peak g SAR may change several db with plarizatin, which can be cncluded frm figures 9 and, fr instance. But als SAR WB depends n the E-field rientatin between and 5 MHz. At MHz, due t the prximity f whle-bdy resnance, SAR WB in VF By, under lateral expsure, is 5.6 db higher fr vertical than hrizntal plarizatin (nly 3.3 db higher frntally, frm figure 4). Between and 5 MHz, SAR WB is plarizatin dependent, t, up t abut 2.2 db fr VF Male, which is illustrated in figure 2: hrizntal plarizatin tends t give higher SAR WB, especially under expsure frm the side, and the rati SAR WB,Hpl /SAR WB,Vpl seems t grw with frequency. Actually, it is bserved frm figure 9 that, abve 2 GHz fr any adult phantm, SAR WB is higher fr hrizntal plarizatin.

14 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures Nrman Japan Male Japan Female VHP Male VF Male Hmgeneus VF Male Heterg VF By Hmgeneus VF By Heterg VF Female Heterg Psture standing Nrmalized SAR WB (frntal) Nrmalized SAR WB (back) Nrmalized SAR g, head/trunk (frntal) Nrmalized SAR g, head/trunk (back) nrmalized SAR g, head/trunk nrmalized SAR g, head/trunk Nrmalized SAR g, limbs (frntal).6 Nrmalized SAR g, limbs (back) nrmalized SAR g, limbs nrmalized SAR g, limbs Figure 9. Mdel variatin. Nrmalized SAR is shwn fr hrizntal (H) and vertical (V) plarizatin between and 5 MHz in the cases f frntal and back expsure. ( Psture standing was simulated nly under frntal expsure.) Inspectin f figures 4 and 5, between and 5 MHz, reveals that SAR WB is mre sensitive t the azimuth angle with vertical plarizatin. The hrizntally plarized E-field is better able t prpagate arund the bdy, i.e. the incming angle des nt have as large an effect n SAR WB as fr vertical plarizatin SAR variatin in db scale The main idea here is t illustrate the SAR sensitivity t the incming directin, mdel phantm and psture in such a way that ne can easily cmpare the SAR sensitivities between different plarizatin frequency cmbinatin cases. Figure 3 shws SAR sensitivity t the incming directin variatin fr fur mdels and 2 6 plarizatin frequency cmbinatins. The variatin cvers directins (figure ).

15 7 T Uusitupa et al Nrmalized SAR WB (frntal) Psture sit Psture examine Psture arms up Psture standing Psture sit n grund Psture arms ut nrmalized SAR g, head/trunk Nrmalized SAR g, head/trunk (frntal) Nrmalized SAR g, limbs (frntal) nrmalized SAR g, limbs Figure. Psture variatin. Nrmalized SAR is shwn fr hrizntal (H) and vertical (V) plarizatin between and 5 MHz in the case f frntal expsure. SAR heterg /SAR hmg (db) SAR heterg /SAR hmg, 22 expsure cnfiguratins in each plarizatin frequency case whle bdy peak g head/trunk peak g limbs H V H45 V45 H V H2 V2 H35 V35 H5 V5 Plarizatin & Frequency Figure. Hetergeneus-mdel SAR relative t hmgeneus-mdel SAR. The range f this rati is shwn fr each plarizatin frequency case. Each such case cvers 22 expsure cnfiguratins: VF Male and VF By with incming directins. Each bar ranges frm lg (SAR min /SAR ref ) t lg (SAR max /SAR ref ), where SAR refers t any SAR key figure. With VFM Hetergeneus, fr example, SAR ref = SAR VFM Het, frntal (pl, f ), (2) SAR min = min VFM Het(pl, f ), inc. directins (3) SAR max = max VFM Het(pl, f ), inc. directins (4) which all depend n the plarizatin frequency cmbinatin (pl, f ). Fr example, at vertical 2 MHz, fr in VFM Hetergeneus, SAR ref =.77 = SAR max and SAR min =.37 (frm figure 6, tp-left, V2 ); and thus, the crrespnding bar ranges

16 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures 7 SARwb Hpl / SARwb Vpl (db) 2 2 θ: φ: 2 SARwb Hpl / SARwb Vpl (db), VF Male Hetergeneus MHz 35 MHz 2 MHz MHz 45 MHz MHz 5 Figure 2. Plarizatin dependence f SAR WB fr VF Male Hetergeneus. frm 3.2 t db (figure 3, tp, case V2 ). Essentially, the db level crrespnds t frntal expsure. Frm figure 3 it is bserved that, fr an adult male, SAR WB varies with incming directin mst in the GHz regin with vertical plarizatin. In the by, SAR WB may vary remarkably als at MHz. In general, the GHz range differs frm lwer frequencies because in the GHz range the penetratin depth in a tissue is smaller and SAR is superficial, and SAR WB is largely dependent n the area f bdy silhuette presented t the incming plane wave. Hwever, in the few-hundred-mhz regin, the penetratin depth in a tissue is larger, SAR is vlumetric instead f superficial and, essentially, SAR WB des nt crrelate with the bdy silhuette area as in the GHz regin. Mdel size matters slightly, i.e. typically, incming directin has a larger effect n SAR WB fr the adult male than fr the by (excluding MHz). Als, typically, the directin dependence f SAR WB is higher fr vertical plarizatin, and at f MHz this is explained by the large azimuth-angle dependence f SAR WB fr vertical plarizatin (see als sectin 3..6). In general, the extent f SAR variatin depends n the frequency and plarizatin. Fr SAR WB the extent stays arund db, and fr the peak g SAR arund db. Figure 4 shws SAR sensitivity t the mdel fr 2 6 plarizatin frequency cmbinatins. The bar diagrams are shwn fr frntal and back expsure f nine mdels: Nrman, Japanese Male and Female, VHP Male, VF Male (hmgeneus and hetergeneus), VF By (hmgeneus and hetergeneus) and VF Female. The reference mdel is VFM Hetergeneus ( db level), and in the frntal expsure case f all mdels SAR ref = SAR VFM Het, frntal (pl, f ), (5) SAR min = min frntal(pl, f ), phantms (6) SAR max = max frntal(pl, f ). phantms (7) Again, each bar ranges frm lg (SAR min /SAR ref ) t lg (SAR max /SAR ref ). Because Psture standing was assessed nly under frntal expsure, it was nt taken int accunt in the bar diagrams, in rder t cmpare tw cnfiguratins: nine mdels/frntal and nine mdels/back. If Psture standing is included in the mdel variatin, the range f SAR g,ht and SAR g,li variatin is increased in sme plarizatin frequency cases (upper limit f

17 72 T Uusitupa et al SAR deviatin (db) SAR deviatin (db) SAR deviatin (db) SAR deviatin (db) VF Male Hetergeneus SAR wb SAR g head/trunk SAR g limbs H V H45 V45 H V H2 V2 H35 V35 H5 V5 Plarizatin & Frequency VF By Hetergeneus H V H45 V45 H V H2 V2 H35 V35 H5 V5 Plarizatin & Frequency VF Male Hmgeneus H V H45 V45 H V H2 V2 H35 V35 H5 V5 Plarizatin & Frequency VF By Hmgeneus H V H45 V45 H V H2 V2 H35 V35 H5 V5 Plarizatin & Frequency Figure 3. Incming directin variatin ( directins, figure ). Range f variatin f lg (SAR/SAR frntal ) is shwn. db crrespnds t frntal expsure (see als equatins (2) (4)). SAR WB varies mst in the GHz regin fr vertical plarizatin.

18 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures 73 SAR deviatin (db) SAR deviatin (db) Frntal expsure (9 mdels) SAR wb SAR g head/trunk SAR g limbs H V H45 V45 H V H2 V2 H35 V35 H5 V5 Plarizatin & Frequency Back expsure (9 mdels) H V H45 V45 H V H2 V2 H35 V35 H5 V5 Plarizatin & Frequency Figure 4. Mdel variatin. Range f variatin f lg (SAR/SAR VFM Het ) is shwn in each plarizatin frequency case, fr frntal and back expsure. db crrespnds t mdel VFM Hetergeneus (see equatins (5) (7)). SAR WB variatin is quite similar fr frntal and back expsure. The results fr nine mdels (8 5 kg) are indicated by the bars. In the frntal case als Psture standing was simulated, and in sme plarizatin frequency cases this mdel can change the upper limit f variatin f peak g SAR, indicated by markers + (head/trunk) r (limbs). variatin shwn by a marker in figure 4, tp). SAR WB variatin remains unaffected because the weights f Psture standing and VF Male are rughly the same. Figure 4 shws that the SAR WB variatin is larger in the GHz range and with vertical plarizatin (and at vertical MHz due t the prximity f whle-bdy resnance fr VF By), and the SAR WB variatin is smaller at hrizntal MHz. SAR WB variatin is quite similar fr frntal and back expsure. In general, the extent f SAR variatin is arund db fr SAR WB and arund db fr the peak g SAR. Figure 5 shws SAR sensitivity t the psture fr 2 6 plarizatin frequency cmbinatins under frntal expsure. The variatin cvers six pstures (figure 3). The reference is Psture standing, and again, the bars range frm lg (SAR min /SAR ref ) t lg (SAR max /SAR ref ), where nw SAR ref = SAR Psture standing, frntal (pl, f ), (8) SAR min = min frntal(pl, f ), pstures (9) SAR max = max frntal(pl, f ). pstures () In the GHz range SAR WB varies quite little, but at arund 45 MHz, ne may even expect a2dbriseinsar WB if psture is changed frm the standing psitin. The extent f SAR

19 74 T Uusitupa et al SAR deviatin (db) Frntal expsure (6 pstures, db: "Psture standing" ) SAR wb SAR g head/trunk SAR g limbs H V H45 V45 H V H2 V2 H35 V35 H5 V5 Plarizatin & Frequency Figure 5. Psture variatin (six pstures, figure 3). Range f variatin f lg (SAR/ SAR Psture standing ) is shwn in each plarizatin frequency case fr frntal expsure. db crrespnds t Psture standing (see equatins (8) ()). variatin stays arund.7.7 db fr SAR WB and arund.6 6. db fr the peak g SAR, i.e. the variatin is larger fr peak g SAR than fr SAR WB. 4. Cnclusin In-huse parallel FDTD and SAR cdes were run in an HP supercluster t perfrm 72 freespace plane-wave SAR simulatins. The bjective f the research was t btain SAR variatin data due t incming directin, human bdy mdel and psture variatin, within the frequency range 5 MHz, fr tw rthgnal plarizatins. Detailed results f SAR were pltted at, 45,, 2, 35 and 5 MHz fr vertical and hrizntal plarizatin f the electric field. Frm these illustratins ne can see the effect f the incming directin, the mdel phantm and the mdel psture n SAR. The results were shwn in a nrmalized frm t directly reveal whether r nt a SAR value is prperly belw the ICNIRP SAR limit (basic restrictin) at the reference level f expsure. The limit f whle-bdy averaged SAR was clearly exceeded in Virtual Family (VF) By in the GHz regin. Als VF Female was clse t the limit (but belw) at 2. GHz which is the frequency giving the highest fr the hetergeneus mdels. Thus, at arund 2 GHz, the SAR WB limit culd be exceeded even by an adult having relatively lw weight ( 5 kg) and well-absrbing bdy shape. The peak g-sar limits were never exceeded in the studied cases, but with the hmgeneus-male psture mdel the peak g SAR in the head-and-trunk regin (in penis) was merely.8 db belw the limit at 2 MHz (mdel standing arms up) and.36 db belw when the arms were dwn. With a crrespnding hetergeneus psture mdel SAR might increase in sme cases, because the VF By and VF Male results shwed that replacing a hmgeneus mdel with a hetergeneus ne may lead t a SAR WB increase f.3 db and abut a 2 db increase f peak g SAR. The sensitivity f SAR t certain variables was illustrated in the db scale separately at each frequency, frm MHz t 5 GHz, and fr vertical and hrizntal plarizatins. Firstly, the incming directin variatin cvered directins fr fur mdels. Fr the adult male, SAR WB varies with directin mst in the GHz regin with vertical plarizatin. Fr the by (height.7 m), SAR WB may vary remarkably als at MHz. At and 45 MHz, peak g SAR varies much mre than SAR WB. Typically, the directin dependence f SAR WB is higher fr vertical than hrizntal plarizatin, and at f MHz this is caused by the large azimuth-angle dependence f SAR WB fr vertical plarizatin. Secndly,

20 SAR variatin study frm t 5 MHz fr 5 vxel mdels including different pstures 75 the phantm variatin cvered ten mdels (8 5 kg). SAR WB varies with mdel mre in the GHz range and fr vertical plarizatin than at hrizntal MHz. The by in the mdel set causes a high SAR WB variatin als at vertical MHz (prximity f whle-bdy resnance). SAR WB variatin with phantm is quite similar fr frntal and back expsure. Thirdly, the psture variatin cvered six pstures. In the GHz range SAR WB varies quite little, but at arund 45 MHz, ne may even expect a 2 db rise in SAR WB if psture is changed frm the standing psitin. The variatin f peak g SAR with psture is clearly larger than the SAR WB variatin. The results shw the clear difference between the GHz regin and the few-hundred-mhz regin. At arund t 5 GHz, ne btains maximum SAR WB (r clsely maximum) with a standing psture under frntal expsure, but at r 45 MHz, SAR WB may rise 2 db when changing psture r incming directin frm the default standing-frntal cnfiguratin. Nrmalized SAR levels are lw at and 45 MHz, thugh, due t lw reference levels. Furthermre, the effect f plarizatin must nt be neglected: near the whle-bdy resnance plarizatin has a strng effect n SAR WB, but als in the GHz regin the effect may be mre than 2 db. Between 2 and 5 GHz fr adults, SAR WB is higher fr hrizntal than fr vertical plarizatin if the incming directin is in the azimuth plane. Hrizntal plarizatin gives higher SAR WB especially under expsure frm the side. Finally, the cmputed data can be utilized in many ways, fr example, t adjust SAR estimatin frmulas f base-statin antennas, r t extract crrectin factrs and uncertainty estimates fr SAR assessment. Als, this research suggests that the ICNIRP (998) reference levels vilate the basic restrictins fr SAR, i.e. incming pwer density r field strength that is belw the reference level may lead t a SAR greater than the basic restrictin. The present ICNIRP guidelines shuld be revised by lwering the current reference levels especially at arund 2 5 GHz. Acknwledgments The research was cnducted with funding frm the Mbile Manufacturers Frum and the GSM Assciatin. The high-perfrmance cmputing resurces were prvided by CSC (Finnish IT Center fr Science). References Ackerman M J 998 The Visible Human Prject Prc. IEEE Caputa K, Okniewski M and Stuchly M A 999 An algrithm fr cmputatins f the pwer depsitin in human tissue IEEE Antennas Prpag. Mag Christ A, Klingenböck A, Samaras T, Giceanu C and Kuster N 26 The dependence f electrmagnetic far-field absrptin n bdy tissue cmpsitin in the frequency range frm MHz t 6 GHz IEEE Trans. Micrw. Thery Tech Christ A et al 2 The Virtual Family the develpment f surface-based anatmical mdels f tw adults and tw children fr dsimetric simulatins Phys. Med. Bil. 55 N23 38 Cnil E, Hadjem A, Lacrux F, Wng M F and Wiart J 28 Variability analysis f SAR frm 2 MHz t 2.4 GHz fr different adult and child mdels using finite-difference time-dmain Phys. Med. Bil Dimbylw P 997 FDTD calculatins f the whle-bdy averaged SAR in an anatmically realistic vxel mdel f the human bdy frm MHz t GHz Phys. Med. Bil Dimbylw P and Blch W 27 Whle-bdy-averaged SAR frm 5 MHz t 4 GHz in the University f Flrida child vxel phantms Phys. Med. Bil Findlay R and Dimbylw P 25 Effects f psture n FDTD calculatins f specific absrptin rate in a vxel mdel f the human bdy Phys. Med. Bil

21 76 T Uusitupa et al Findlay R and Dimbylw P 26 FDTD calculatins f specific energy absrptin rate in a seated vxel mdel f the human bdy frm MHz t 3 GHz Phys. Med. Bil Gabriel S, Lau R and Gabriel C 996 The dielectric prperties f bilgical tissues: III. Parametric mdels fr the dielectric spectrum f tissues Phys. Med. Bil Hirata A, Asan T and Fujiwara O 28 FDTD analysis f bdy-cre temperature elevatin in children and adults fr whle-bdy expsure Phys. Med. Bil ICNIRP (Internatinal Cmmissin n Nn Inizing Radiatin Prtectin) 998 Guidelines fr limiting expsure t time-varying electric, magnetic and electrmagnetic fields (up t GHz) Health Phys IEC (Internatinal Electrtechnical Cmmissin) 29 Technical Cmmittee 6, Human Expsure t Radi Frequency Fields frm Handheld and Bdy-Munted Wireless Cmmunicatin Devices Human Mdels, Instrumentatin and Prcedures, Part 2: prcedure t determine the specific absrptin rate (SAR) fr mbile wireless cmmunicatin devices used in clse prximity t the human bdy (frequency range f 3 MHz t 6 GHz) IEC Ed., Draft IEEE (Internatinal Cmmittee n Electrmagnetic Safety) 22 IEEE recmmended practice fr measurements and cmputatins f radi frequency electrmagnetic fields with respect t human expsure t such fields, khz GHz IEEE Std C Kühn S, Jennings W, Christ A and Kuster N 29 Assessment f induced radi-frequency electr-magnetic fields in varius anatmical human bdy mdels Phys. Med. Bil Laaks I, Ilvnen S and Uusitupa T 27 Perfrmance f cnvlutinal PML absrbing bundary cnditins in finite-difference time-dmain SAR calculatins Phys. Med. Bil Masn P, Ziriax J, Hurt W, Walters T, Ryan K, Nelsn D, Smith K and D Andrea J 2 Recent advancements in dsimetry measurements and mdeling Radi Frequency Radiatin Dsimetry ed B J Klauenberg and D Miklavcic (Nrwell, MA: Kluwer) pp 4 55 Nagaka T, Watanabe S, Sakurai K, Kunieda E, Watanabe S, Taki M and Yamanaka Y 24 Develpment f realistic high-reslutin whle-bdy vxel mdels f Japanese adult males and females f average height and weight, and applicatin f mdels t radi-frequency electrmagnetic-field dsimetry Phys. Med. Bil Nagaka T and Watanabe S 28 Pstured vxel-based human mdels fr electrmagnetic dsimetry Phys. Med. Bil