UWB on-body radio propagation and system modelling for wireless body-centric networks

Transcription

1 ULTRA WIDEBAND SYSTEMS, TECHNOLOGIES AND APPLICATIONS SPECIAL SECTION UWB on-ody rdio propgtion nd system modelling for wireless ody-centric networks A.Alominy,Y.Ho,X.Hu,C.G.PrinindP.S.Hll Astrct: Given the trend towrds user-centric concept in moile communictions, ody re networks hve received incresing ttention within the wireless personl re community. Reserchers nd designers hve een investigting rdio propgtion chrcteristion nd modelling owing to its importnce in developing nd designing efficient nd relile rdio systems. The pper presents n eperimentl investigtion of ultr-widend (UWB) on-ody rdio propgtion. Chnnel models with respect to lrge scle nd dely nlysis hve een derived from mesured prmeters. Effects of different ntenn types on chnnel ehviour re lso demonstrted. Results nd nlyses highlight the consequences of chnges in ody postures nd positions in ddition to ntenn orienttions on the communiction chnnels. Mesurement dt re lso used in predicting the performnce of potentil UWB wireless systems nd their pplictions in ody-centric networks. It is indicted tht i-phse modultion for UWB odycentric networks performs etter thn pulse position modultion systems. The study lso concludes tht the use of hyrid ntenn types for different on-ody links will provide etter system performnce. 1 Introduction Future communiction systems re driven y the concept of eing connected nywhere t nytime. An essentil prt of this concept is user-centric pproch in which services re constntly ville nd systems provide re-configurility, unotrusiveness nd true etension of the humn mind. Body-centric networks consist of numer of nodes nd units plced on the humn ody or in close proimity such s on everydy clothing [1]. Since low-power trnsmission is required for ody worn wireless devices, the humn ody cn e used s communiction chnnel etween wireless werle devices to form wireless odycentric network. Ultr-widend (UWB) communiction is low-power, high dt rte technology with lrge ndwidth signls tht provide roustness to jmming nd hve low proility of detection [2]. UWB low trnsmit power requirements llow longer ttery life for ody worn units. This leds to UWB eing potentil cndidte for ody-centric networks. UWB offers good penetrting properties tht could e pplied to imging in medicl pplictions; with the UWB ody sensors, this ppliction could e esily reconfigured to dpt to the specific tsks nd would enle high dt rte connectivity to eternl processing networks. The wireless ody-centric network hs specil properties nd requirements in comprison to other ville wireless r IEE, 26 IEE Proceedings online no doi:1.149/ip-com:2546 Pper first received 31st Jnury nd in finl revised form 12th July 25 A. Alominy, Y. Ho, X. Hu nd C.G. Prini re with Deprtment of Electronic Engineering, Queen Mry, University of London London, E1 4NS, UK P.S. Hll is with School of Electronic, Electricl nd Computer Engineering, University of Birminghm, Birminghm B15 2TT, UK E-mil: y.ho@elec.qmul.c.uk networks nd tht is due to the rpid chnges in communiction chnnel ehviour on the ody during the network opertion. This rises some importnt issues regrding the propgtion chnnel chrcteristics, rdio systems comptiility with such environments nd the effect tht it hs on the werer. In the lst few yers, reserchers hve investigted UWB indoor nd outdoor rdio propgtion modelling nd chrcteristion [3, 4]. However, on-ody chnnel chrcteristion hs een presented in few studies [5, 6] for nrrownd chnnels. UWB ody re network chnnel chrcteristion ws presented in [7] only for pre-defined sets of nodes with multi-hopping networking in mind to determine energy nd power requirements, wheres more relistic representtion of humn ody ehviour such s different ody positions nd the effects of vrious ntenn types on UWB on-ody chnnels ws investigted in [8]. The performnce of UWB rdio systems in multipth indoor environment hs een presented in [9, 1]. However, to the uthors knowledge, system-level modelling of potentil on-ody UWB rdio systems with mesured chnnel dt nd dely profile modelling hs not een presented. In this pper, we demonstrte n eperimentl investigtion of UWB on-ody rdio chnnels through mesurement cmpign performed in order to otin sufficient pth vrition dt for effective chrcteristion of the chnnels. Two types of UWB ntenns were used to emine on-ody propgtion inside n nechoic chmer. Pth loss models nd spred dely ehviour in comprison to estlished propgtion models re lso presented. Such mesurements could provide deterministic chrcteristions tht cn e pplied in designing simple nd efficient ody-centric rdio systems. A systemlevel modelling of possile UWB RF trnsceiver, sed on pulse position modultion ( PPM) nd i-phse modultion with mesured chnnel prmeters, is lso demonstrted. IEE Proc.-Commun., Vol. 153, No. 1, Ferury 26 17

2 2 UWB on-ody propgtion chnnel chrcteristion 2.1 Frequency domin mesurement setup A Hewlett Pckrd 872ES vector network nlyser (VNA) ws set on the response mode in the rnge 3 GH 9 GH with intervls of 3.75 MH, t sweep rte of 8 ms, for UWB on-ody communiction chnnels mesurement. Port-I ws used s trnsmit node nd port-ii s the receiver with two pirs of different UWB ntenns to mesure chnnel frequency response (S 21 ). Antenns were connected to the VNA y 3 nd 5 metre long cles. The VNA mesures the mgnitude nd phse of ech frequency components llowing the ese of otining time domin response y mens of inverse discrete fourier trnsform (IDFT). Two sets of mesurements were performed in the nechoic chmer to ccount for deterministic chnnel chrcteristics on humn ody (height 17 cm nd verge width of 35 cm), which resulted in 71 frequency responses for different on-ody chnnels. Figure 1 shows the different ntenn positions plced on the ody nd the mesurement setup. Twenty-two different on-ody scenrios pplied in the mesurement, illustrting possile ody movements nd potentil positions on ody for worn devices, re listed in Tle 1. The minimum mesured distnce is 15 cm, which is lrger thn the wvelength (1 cm) of the low frequency limit in the nd (3 GH), minimising the mutul coupling effect when ntenns re plced ner ech other. Fifteen frequency sweeps were tken for ech sttic on-ody chnnel [8]. All mesurements in this study were tken using effective isotropic rdited power (EIRP) of the order 17 cm R4 R6 R1R2 T 35 cm R3 R5 VNA initil model dt collection sttisticl nlysis chnnel chrcteristion nlysis softwre IDFT postprocessing Fig. 1 Antenn positions for on-ody chnnel chrcteristion cmpign with humn ody dimensions nd mesurement setup including post-processing steps of dbm, pproimtely 3 db less thn tht emitted y cellulr phones. 2.2 Mesured UWB ntenns Two pirs of UWB ntenns (horn shped self-complementry-, nd plnr inverted cone ntenn-) were designed following the descriptions outlined in [11] nd [12]. The plnr ws fricted on RT/Duroid sustrte with reltive permittivity of 3 nd thickness of mm nd overll dimensions of 1 1 mm (Fig. 2). A 3-stge Cheyshev trnsformer cked with mm ground plted on the opposite side of the printed rditing elements is used to provide widend mtching to the 5 O coil cle feed. The ehiits pproimtely constnt impednce nd solute gin cross the UWB nd with mesured gin of to 2.4 dbi. The second type of UWB ntenn pplied in the on-ody propgtion mesurement is the ntenn, which is derived from the volcno ntenn nd the circulr disk ntenn concepts, Fig. 2. It is composed of single flt element verticlly mounted ove smll circulr ground plne with dimeter of 6 mm. The ntenn geometry is very simple with ntenn dimensions of mm nd conductor thickness for the cone round.3 mm. The ntenn provides outstnding impednce nd rdition pttern performnce with gin of 3 to 6 dbi. The impednce ndwidth performnce of the two ntenns is compred to UWB requirements. Impednce ndwidths of 1:1 nd 8:1 were otined for the nd ntenns, respectively [8]. However, compring the fr field rdition ptterns for oth the imuth plne (Fig. 3, -plne co-polristion) nd elevtion plne (Fig. 3, y-plne co-polristion) of the two ntenns t 3 nd 1 GH shows etter rdition ndwidth otined for the cse compred to. This leds to the fct tht hs etter effective ndwidth cross the UWB nd of interest (3 9 GH). Figure 3 illustrtes the effect of the restriction on ord sie nd ddition of ground plne on the rdition performnce of the t higher frequencies. Phse distortion cused y slight mismtching t these frequencies in the trnsformer circuit cuses chnges in trnsmission responses nd lso rdition pttern with the ppernce of nulls oviously illustrted in Fig. 3 for the cse. Detiled chrcteristion of oth ntenns is presented in [5,6,8]. The ( plnr dipole-like ntenn) nd (verticl monopole-like ntenn) ntenns were chosen for Tle 1: Different 22 ody scenrios nd positions for UWB on-ody mesurement with relted RMS spred dely sttistics (RMS vlues nd men RMS) N On-ody scenrio t rms Men RMS t rms Men RMS N On-ody scenrio t rms Men RMS t rms Men RMS 1 R1 - stnding still, upright R4 - stnding upright R1 - ody turned left R4 - hed turned left R1 - ody turned right R4 - hed turned right R1 - ody lened forwrd R5 - stnding rm stretched R2 - (ck) stnding upright R5 - stnding rm ove hed R2 - (ck) ody turned left R5 - sitting rm long ody R2 - (ck) ody turned right R5 - sitting hnd on lp R2 - (ck) ody lened forwrd R6 - stnding rm stretched R3 - stnding upright R6 - stnding rm ove hed R3 - hed turned left R6 - sitting rm long ody R3 - hed turned right R6 - sitting hnd on lp IEE Proc.-Commun., Vol. 153, No. 1, Ferury 26

3 y 3.6 mm 1 mm 6.8 mm ckside GND 3-stge Cheyshev trnsformer 11.4 mm 29.8 mm l 3 = 6.3 mm l 2 = 6.3 mm l 1 = 6.3 mm 5 Ω w 3 =.37 mm w 2 =.57 mm w 1 =.87mm.3 mm ε = 3 h = mm 1 mm y 76.2 mm 47 mm 29.2 mm 29.2 mm.65 mm GND 6 mm 5 Ω co Fig. 2 Detiled schemtic digrms of pplied UWB ntenns with dimensions in mm Horn shped self-complementry ntenn (), plced prllel to the ody with rditing element fcing outwrds from the ody Plnr inverted cone ntenn (), plced perpendiculr to the ody with ground plte on the ody the reported mesurement cmpign to provide cler picture of the influence of ntenn chrcteristics on propgtion chnnel ehviour. Different ntenn types re epected to hve dissimilr effects on the rdio chnnels especilly in the ody re network where the trnsmitter nd the receiver re in close proimity. Hence, ntenn ner field chrcteristion is needed. Figures 4 nd show numericlly computed ner field ptterns t distnce of 15 cm from the ntenn in imuth nd elevtion plnes, respectively. In the mesurement cmpign, ws plced prllel to the ody with min rditing elements fcing outwrd from the ody. ws plced with the rditor perpendiculr nd the ground plte prllel to the ody. 2.3 Dt nlysis The mesured chnnel dt for different on-ody scenrios were clculted nd processed in oth frequency nd time domin (the setup shown in Fig. 1) to otin initil sttisticl models for oth pth losses nd power dely profiles including spred men ecess dely. Their reliility nd pplicility re investigted ginst estlished empiricl nd theoreticl propgtion models. The verge received power (S 21 )isshowninfigs.5 nd c for scenrio 15 when R5 is plced on the left wrist nd the rm stretched side the ody nd scenrio 12 when R4 is plced on the right side of the hed with hed still fcing forwrd (Tle 1), respectively. Emining the slopes of these frequency responses shows the vrition in slope vlues not only owing to ntenn frequency dependency ut lso the dispersion of on-ody chnnels. Time domin nlysis is performed y otining chnnel impulse responses tht re clculted from the mesured frequency trnsfer functions consisting of 161 frequency points using the rel pssnd technique nd pplying Hmming window nd inverse discrete fourier trnsform (IDFT) [3]. Time domin pulses for different sweeps were used to verify tht chnnels were sttic during mesurements for the specified on-ody link in ddition to verifiction in frequency domin [8]. Figures 5 nd d present impulse responses for on-ody scenrio 15 nd scenrio 12 (s defined in Tle 1). It cn e seen tht most energy is received vi the direct pth with some multipth reflections t the lte time. The chnges in chnnel chrcteristics owing to different ntenn ( nd ) properties cn e lso found in these Figures. The min dissimilrity is tht more strong echoes nd ringing effects pper in the cse. This cn e eplined y the fct tht the hs nrrower ndwidth nd more resonnce frequencies within the mesured nd, which IEE Proc.-Commun., Vol. 153, No. 1, Ferury 26 19

4 y (3 GH) (3 GH) (1 GH) (1 GH) Fig. 3 Comprison etween mesured nd fr field rdition ptterns t 3 GH nd 1 GH Inset: Picture defines the ntenn coordintes Aimuth plne (-plne co-polristion) Elevtion plne (y-plne co-polristion) y y (3 GH) (3 GH) (1 GH) (3 GH) y Fig. 4 Comprison etween computed ner field ptterns for nd t 3 nd 1 GH t 15 cm distnce from the ntenn Inset: Picture defines the ntenn coordintes Aimuth plne (-plne E-totl) Elevtion plne (y-plne E-totl) increses oth the ringing nd pulse width tht directly ffects dt rte. Power dely profiles ( PDP) were produced y verging ll impulse responses nd determining the noise threshold. The PDP is chrcterised y the first centrl moment (men ecess dely t m ) nd the squre root of the second moment of the PDP (RMS dely spred t rms ). The RMS dely spred provides figure of merit for estimting dt rtes for multipth chnnels [13]. The men ecess dely is defined s vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PN 1 t i jhðt i Þj 2 i ¼ t m ¼ u ð1þ t PN 1 jhðt i Þj 2 i ¼ 11 IEE Proc.-Commun., Vol. 153, No. 1, Ferury 26

5 reltive power, db normlised mp frequency, GH ecess dely, s( 1 8 ) reltive power, db frequency, GH c normlised mp ecess dely, s( 1 8 ) d Fig. 5 Mesured chnnel frequency responses nd impulse responses for on-ody scenrio 12 (R4-receiver on the right side of the hed with ody stnding still) nd scenrio 15 (R5-receiver on the left wrist with rm stretched side) Chnnel frequency response scenrio 15 Impulse response scenrio 15 c Chnnel frequency response scenrio 12 d Impulse response scenrio 12 nd the dely spred t rms s vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PN 1 ðt i t m Þ 2 jhðt i Þj 2 i ¼ t rms ¼ u t PN 1 jhðt i Þj 2 i ¼ ð2þ where h(t i ) is the time domin impulse response otined from mesurement dt. Dely prmeters were evluted for intervl of 8 ns, since echoes fde rpidly fter this period. Tle 1 shows summry of the RMS dely spred nd men RMS for oth mesurement sets ( nd ) for vrious onody propgtion scenrios. As epected, the men dely owing to propgtion link etween the trnsmitter (T) nd the receiver t the ck (R2), Fig. 1, for scenrios 5, 6, 7 nd 8 (Tle 1) link is the highest where non-line-of-sight (NLOS) chnnel nd propgtion round humn ody on the surfce (creeping wves) re the min propgtion chnnels. Men RMS dely for links with smll distnces nd where oth ntenns re plced on the sme prt of the humn ody in the ntenn cse is smller thn RMS spred dely for the sme links with (cses R1, R2, R5). This cn e relted to the plcement of oth ntenns of which stronger surfce wves re lunched in the cse. For other links where line-of-sight (LOS) components ( free spce wves) re dominnt, τ rms,ns fitted (5.7) fitted (1.9) T/R seprtion, m Fig. 6 RMS dely spred vrition ginst distnce for oth nd produces etter dely performnce. Figure 6 shows the RMS dely spred ginst distnce for oth mesurement cses while Figs. 7 nd show the cumultive distriutions of the RMS dely spred nd men ecess dely, respectively, fit to log norml distriution. Other different empiricl distriutions were pplied, however, log norml proved to e the est fit with more devition for chnnels, which is predictle owing to the rndom ehviour of the on-ody chnnels tht could led to modified fit model for more etensive links. IEE Proc.-Commun., Vol. 153, No. 1, Ferury

6 cumultive proility (mes) (lognorm) (mes) (lognorm) pth loss, db (mes.) (lest Sq) (mes.) (lest Sq) RMS, ns distnce, m Fig. 8 Mesured nd modelled pth loss for nd with proposed dul-slope model for cse cumultive proility men ecess dely, ns The pth loss of the chnnel is clculted directly from the mesurement dt using verging over the mesured chnnel frequency responses t ech frequency point [4]. The men pth loss referenced to distnce d o (reference distnce is 1 m s clirted for VNA mesurement) cn e presented s fitted line model for oth cses. For PL db ðdþ ¼86:5 þð4:4þ1 log d d o ð3þ for 15 cm d 1 cm nd for PL db ðdþ ¼7:3 þð2:7þ1 log d d o ð4þ for 15 cm d 1 cm (mes) (lognorm) (mes) (lognorm) Fig. 7 Dely prmeters cumultive distriution fit to log norml distriution RMS dely spred for on-ody chnnels (t rms ) Men ecess dely (t m ) where ¼ 4.4 nd 2.7 re the pth loss eponents for nd models, respectively. Figure 8 presents the mesured vlues nd modelled pth losses ginst distnce for oth ntenn cses. It cn e concluded from Fig. 8 tht on-ody chnnel pth losses cn e pproimted using the conventionl dul-slope model. It cn e clculted tht t the UWB nd, the fr field will e t distnces equl to nd greter thn 37 cm for, which grees with tht from the mesurement in Fig. 8 with rekpoint round 37 cm. The non-reflecting environment in the nechoic chmer leds to the high eponent of pth losses, however, for, good omni-directionl rdition cross UWB rnge results in the lower loss eponent vlue owing to multipth reflections from humn ody nd clothes. In ddition to distnce chnges nd different ody positions contriution to chnnel pth loss, nother importnt fctor is ntenn orienttion considering oth trnsmit nd receive ntenns. For emple, n dditionl 2 db loss will occur when two s re orthogonlly oriented on the ody. 3 UWB rdio system performnce in ody-centric environment 3.1 System-level modelling The wireless trnsceiver systems used in ody-centric networks hve to e idelly miniturised, lightweight, ehiit low-power opertion for longer lifetime nd e esily integrted with dily clothing. The different prts in the system need to shre common wireless protocol to enle the estlishment of re-configurle communiction medium. Sufficient dt rtes nd controlled power levels re required y the rdio trnsceivers to enle flwless connectivity etween ody-distriuted networks. Mny currently eisting short-rnge wireless technologies provide communiction medium nd cle replcement technologies for different trnsmission types. To design suitle efficient rdio interfce for the wireless ody-centric network, the understnding nd integrtion of eisting stndrds re required in order to ring to light the min res in which new techniques re required to meet the hrsh nd demnding communiction environment. The UWB rdio front-end could e used for the sme pplictions trgeted for other short-rnge wireless systems, however, t higher dt rtes, lower emitted rdio frequency (RF) power nd less compleity in trnsceiver designs. Since ccurte performnce prediction of UWB systems requires the knowledge of the propgtion chnnel ehviour, system modelling of potentil UWB rdio trnsceivers for ody-centric network is proposed. Figure 9 shows lock digrm of the rdio system modelled using Agilent DSP Designer TM to investigte the effects of on-ody chnnels on UWB systems using different pulse modultion techniques. 112 IEE Proc.-Commun., Vol. 153, No. 1, Ferury 26

7 input dt spreding code timing nd coding pulse shpe genertor PA mplitude, V recovered dt it slicer correltor su-system LNA mesured on-ody chnnel + ntenns mplitude, V Fig. 9 Block digrm of the UWB rdio system modelled for performnce nlysis of potentil trnsceivers tht could e pplied in ody-centric networks Vrious modultion schemes re possile for UWB signlling such s pulse position modultion ( PPM), mplitude shift keying (ASK), on-off keying (OOK), phse shift keying ( PSK), nd frequency shift keying (FSK). The performnce of PPM nd i-phse modulted UWB signls in multipth environment ws investigted in [9] nd [1] for different dt rtes nd under different requirements. The system model presented here pplies 1 Mit/s dt rte nd opertes in low signl-to-noise rtio (SNR ¼ ) for performnce nlysis under mesured on-ody chnnel dt. 3.2 Performnce nlysis Both trnsmitted nd received signl spectrum were emined to determine the effects of oth modultion technique nd chnnel chrcteristics on the system performnce. The effect of mesured chnnel responses on received UWB pulses in oth ntenn cses ws lso nlysed nd investigted. Figures 1 nd 11 show the trnsmitted nd received pulses for on-ody chnnel scenrio 1 (Scenrio 1, receiver plced t R1 on the chest with ody stnding still) using PPM nd i-phse modultion, respectively. The vrition in time delys etween the different received pulses highlights the ntenn effect on the chnnel ehviour of the on-ody links for the received pulses. More echo nd noise components re introduced in the cse of pulses mplitude, V mplitude, V mplitude, V time, s ( 1 8 ) Fig. 1 Trnsmitted nd received UWB pulses pplying PPM modultion in nd mesurement cses (scenrio 1 R1 plced on the right side of the chest with ody stnding still) c mplitude, V time, s ( 1 8 ) Fig. 11 Trnsmitted nd received UWB pulses pplying i-phse modultion in nd mesurement cses (scenrio 1 R1 plced on the right side of the chest with ody stnding still) received under the compred to those with the nd this is due to the wider ndwidth of the ntenn (s discussed erlier). However, the improved pulse shpe received in cse is trded off y reduction in energy received. In regrds to the influence of modultion techniques on UWB pulses, i-phse provided etter roustness to multipth components in comprison to PPM cses. As the most referred to receiver design for UWB systems, RAKE receiver is modelled to collect the pulses energy in multipth component directions. At signl-to-noise-rtio (SNR) ¼ two RAKE fingers were modelled to collect the direct pth component nd the second finger to collect one multipth component with verge dely window of 8 ns for the investigted scenrios. The pplied on-ody chnnels re scenrio 1 (R1 on chest ody stnding still), scenrio 8 (R2 on the ck with ody lening forwrd), scenrio 14 (R4 on the right side of the hed withhedturnedright)ndscenrio16(r5ontheleft wrist with rm stretched ove hed). Different it error rte (BER) vlues re otined for the vrious investigted scenrios nd for the modultion schemes pplied, Fig. 12. BER r1- stnd still r2- lening forwrd r4- hed turned right c -PPM -iphse -PPM -Biphse r5-arm ove hed -iphse -PPM -iphse -PPM Fig. 12 BER t specified on-ody links (R1 with ody stnding still, R2 with ody lening forwrd, R4 with hed turned right nd R5 with rm ove hed) for oth modultion schemes PPM nd i-phse with two rke fingers IEE Proc.-Commun., Vol. 153, No. 1, Ferury

8 Differences of round 6% in BER mesures re found when PPM is used in comprison to 1.5% for the i-phse modultion system etween nd mesured chnnel responses. This implies tht i-phse modultion cn e dopted in designing initil potentil UWB trnsceiver designs for wireless ody-centric networks since it provides etter performnce of UWB systems even for higher dt rtes, which lso gree with results otined in [1]. 5 Acknowledgments The uthors wish to thnk Mr. John Dupuy for his ssistnce in ntenn mesurement nd friction nd lso for his help with propgtion mesurement setup. Also thnks to Miss W. Dong for her help with simultion nd mesurement nlysis. Thnkful cknowledgment goes to the reviewers for their constructive comments nd suggestions. 6 References 4 Conclusions UWB on-ody propgtion chnnel mesurements were performed in n nechoic chmer to study the deterministic chrcteristics of the rdio chnnel on humn ody. On-ody chnnels depend minly on spce wve nd surfce wve propgtion; therefore the two UWB ntenn types were chosen to study these propgtion contriutors in detils nd lso to present verticl nd plnr UWB ntenn chrcteristics. When oth the trnsmitter nd receiver re plced on the trunk (front side of the ody) in the mesurement stronger surfce wves were ville compred to the cse. This cn e demonstrted in the vritions of dely prmeters etween oth ntenn cses for different on-ody chnnels. The chnnel pth loss ws modelled s function of the logrithmic distnce, however, it ws shown tht conventionl dul-slope model cn e otined to provide pth loss model s function of distnce. Received power ws proven to e dependent not only on the ntenn positions nd distnces ut lso on the frequency dependency of oth ntenns nd on-ody chnnel. Mesured RMS spred dely nd men ecess dely dt fitted to log norml distriution tht provides tool for empiricl on-ody propgtion modelling tht cn e comined with ny other models representing different environments, specilly with fitting results presenting slightly similr distriution coefficients for oth ntenn cses. The mesured nd nlysed dt were used for evluting the performnce of potentil UWB trnsceivers tht could e pplied in the wireless ody-centric networks. For very low trnsmitting power nd low SNR, i-phse modultion provided etter system performnce in comprison to PPM for potentil UWB on-ody wireless communictions. 1 Bernrdhrd, J., Ngel, P., Hupp, J., Struss, W., nd von der Grun, T.: BAN-ody re network for werle computing. 9th Wireless World Reserch Forum Meeting, Zurich, Switerlnd, July 23 2 Foerster, J., Green, E., Somyulu, S., nd Leeper, D.: Ultrwidend for short- or medium-rnge wireless communictions, Intel Technol. J., 21, 5, (2) 3 Chong, C., Kim, Y., nd Lee, S.: UWB indoor propgtion chnnel mesurements nd dt nlysis in vrious types of high-rise prtments. Proc. IEEE Vehiculr Technology Conf. (VTC24- Fll), Los Angeles, USA, Septemer 24 4 Ghssemdeh, S.S., Jn, R., Rice, C.W., Turin, W., nd Trokh, V.: A sttisticl pth loss model for in-home UWB chnnels. IEEE Conf. on Ultr Widend Systems nd Technologies, UWBST, Bltimore, USA, 22, pp Alominy, A., Owdlly, A., Ho, Y., Prini, C.G., Nechyev, Y., Constntinou, C.C., nd Hll, P.: Body-centric WLAN for future werle computers. First Int. Workshop on Werle & Implntle Body Sensor Networks, London, UK Imperil College, 6 7 April 24 6 Nechyev, Y., Hll, P., Constntinou, C.C., Ho, Y., Owdlly, A., nd Prini, C.G.: Pth loss mesurements of on-ody propgtion chnnels. Proc. 24 Int. Symp. on Antenns nd Propgtion, Sendi, Jpn, Aug. 24, pp Zsowski, T., Althus, F., Stger, M., Wittneen, A., nd Troster, G.: UWB for noninvsive wireless ody re networks: chnnel mesurements nd results. Proc. IEEE Conf. on Ultr Widend Systems nd Technologies, Reston, Virgini, Nov. 23, pp Alominy, A., Ho, Y., Prini, C.G., nd Hll, P.S.: Comprison etween two different ntenns for UWB on-ody propgtion mesurements, IEEE Antenns Wirel. Propg. Lett., 25, 4, (1), pp Bstids-Pugo, E.R., Rmire-Mireles, F., nd Muno-Rodrigue, D.: Performnce of UWB PPM in residentil multipth environments. IEEE 58th Vehiculr Technology Conf., VTC Fll 23, Olndo, Florid, 6 9 Oct. 23, Vol. 4, pp Chung, W.C., nd H, D.S.: On the performnce of i-phse modulted UWB signls in multipth chnnel. IEEE Vehiculr technology Conf., VTC Spring 23, Jeju, Kore, April 23, pp Sitou, A., Iwki, T., Honjo, K., Sto, K., Koym, T., nd Wtne, K.: Prcticl relition of self-complementry rodnd ntenn on low-loss resin sustrte for UWB pplictions. 24 IEEE MTT-S Int. Microwve Symp. Fort Worth, Tes, June Suh, S.-Y., Stutmn, W.L., nd Dvis, W.A.: A new ultrwidend printed monopole ntenn: the plnr inverted cone ntenn (), IEEE Trns. Antenns Propg., 24, 52, (5), pp Rppport, T.S.: Wireless communictions: principles nd prctice (Prentice Hll, 1999, Ch. 3, 4 nd 5) 114 IEE Proc.-Commun., Vol. 153, No. 1, Ferury 26