In-Building Wideband Multipath Characteristics at 2.5 & 60 GHz

Transcription

1 In-Building Wideband Multipath Chaacteistics at 2.5 & 60 GHz Chistophe R. Andeson 1, Theodoe S. Rappapot 3, Kyung Bae 1, Alex Vestak 2, Naen Ramakishnan 2, William H. Tante 1, Cliffod A. Shaffe 2, and Layne T. Watson 2 1 Mobile and Potable Radio Reseach Goup Depatment of Electical and Compute Engineeing Viginia Tech Blacksbug, VA USA 2 Depatment of Compute Science Viginia Tech Blacksbug, VA USA 3 Wieless Netwok and Communications Goup Depatment of Electical and Compute Engineeing Univesity of Texas at Austin Austin, Texas Abstact This pape contains measued data fo 2.5 & 60 GHz in-building patition loss. Path loss measuements wee ecoded using a boadband sliding coelato channel sounde which ecoded ove 39,000 Powe Delay Pofiles (PDPs) in 22 sepaate locations in a moden office building. Tansmittes and eceives wee sepaated by distances anging fom 3.5 to 27.4 metes, and wee sepaated by a vaiety of obstuctions, in ode to emulate futue single-cell-pe-oom wieless netwoks. These measuements may aid in the development of futue in-building wieless netwoks in the unlicensed 2.4 GHz and 60 GHz bands. Keywods In building popagation, path loss, patition loss, millimete wavelength I. INTRODUCTION Ove the past decade, the maket fo wieless sevice has gown at an unpecedented ate. The industy has expanded fom cellula phones and pages to Pesonal Communication Systems (PCS), wieless local aea netwoks (WLANs), and boadband wieless sevices that can povide voice, data, and full-motion video in eal time [1]. In ode fo the visions of 3d and 4th geneation of wieless communication standads to be ealized, system design enginees must have a thoough undestanding of the wieless channels in which these devices opeate. In ecent yeas, thee has been an inceasing inteest in poviding boadband communications in the 2.4 GHz ISM band and the 60 GHz unlicensed band fo WLANs. In paticula, the popagation chaacteistics of the 60 GHz band povides the pomise of high spatial fequency euse, with low-powe tansmittes opeating in a single-cell-pe-oom configuation [2], [3], called femtocellula [2]. Such a system will povide high data-ate sevices fo densely populated buildings, caying many times moe taffic than cuent wieless netwoks. While spectum in the 2.4 and 60 Most of the available liteatue has so fa concentated on investigating penetation loss into buildings, athe than fom obstuctions inside buildings. Seveal popagation studies show that penetation loss of vaious building mateials inceases as the tansmission fequency inceases. Zhang and Hwang as well as Golding and Livine show how penetation loss in vaious building mateials inceases ove the fequency ange of 900 MHz 18 GHz and GHz, espectively [14], [15]. Additionally, penetation losses fo building mateials at vaious fequencies between 5 and 60 GHz ae epoted in [4], [14], [16] [23], and a geneal incease in penetation loss as fequency inceases can be obseved. These data compae favoably to the penetation losses epoted in Section III of this pape. II. EXPERIMENTAL SETUP A. Desciption of Measuement Pocedue and Locations Eight sepaate tansmitte and 22 sepaate eceive locations wee selected on the 4th floo of Duham Hall on the Viginia Tech campus. The measuement sites wee chosen to be epesentative of a boad ange of typical femtocellula popagation envionments in a wok setting, whee a low powe tansmitte will seve a single oom o potion of a floo. Duham Hall was completed in 1998, with a foundation and famewok made fom steel einfoced concete, with inteio sheetock and concete cindeblock walls, ceamic tile and capeted floos, and suspended panel ceilings. Fig. 1 illustates the building floo plan and identifies tansmitte and eceive locations fo this measuement campaign. Measuements wee gouped into eight diffeent segments, based on tansmitte location, and numbeed based on both tansmitte and eceive location. These measuements look specifically at the wideband popagation effects that may be encounteed in a typical office building, with tansmitte and eceive locations chosen to povide line-of-sight (LOS), non line-of-sight (NLOS), and clutteed popagation envionments. A boadband vecto sliding coelato channel sounde, developed in [24] was used to ecod wideband PDPs at all measuement locations. The channel sounde utilized an 11- bit pseudo-andom noise code unning at 400 MHz, with GPS disciplined oscillatos geneating a highly stable fequency efeence at tansmitte and eceive, poviding the channel sounde with a multipath tempoal esolution of 2.5 nanoseconds. Two diffeent RF chains wee utilized, one fo the 60 GHz measuements and the othe fo the 2.5 GHz measuements. Fo the 60 GHz measuements, the tansmitte and eceive utilized pyamidal hon antennas which had a gain of 25 dbi and fist-null beamwidths of 50. These high-gain hon antennas wee used in ode to ovecome the consideable amount of path loss at 60 GHz, as well as to emulate sectoed /02/$ IEEE. 97

2 Fig. 1. Map of the 4 th floo of Duham Hall at Viginia Tech, with tansmitte and eceive locations identified antenna systems poposed fo millimete-wavelength indoo applications. Tansmitte output powe (as measued at the base of the antenna) was set at -10 dbm. Fo the 2.5 GHz measuements, tansmitte and eceive utilized omnidiectional biconical antennas with a 6 dbi gain. These lowe gain omnidiectional antennas wee used due to thei compact physical size, as well as to emulate 2.4 GHz WLANs opeating with omnidiectional antennas. Tansmitte output powe (as measued at the base of the antenna) was set at 0 dbm. Fo both configuations, tansmitte and eceive antennas wee vetically polaized, and heights wee nominally set at 1.2 metes, with the exception of tansmitte location 4 whee the antenna height was inceased to 2.4 metes. A laptop compute was used to ecod inphase and quadatue delay pofiles, and softwae post-pocessing geneated the PDPs. Measued powe delay pofiles may be found in [24], [25]. B. Definition of Path Loss To measue path loss, the channel sounde ecods wideband powe delay pofiles at 2.5 GHz and 60 GHz. Naowband eceived powe fluctuates ove a small aea due to multipath-induced fading, howeve, aveaging powe ove a local aea yields a eliable estimate fo the local aveage eceived powe independent of signal bandwidth [16]. Additionally, naowband powe can be calculated fom a wideband PDP using the following elationship [26] PR( dbm) = Pcal( dbm) + 10log 10 G τ whee G τ G τ (1) cal cal is the integated powe in a given PDP (aea unde the PDP). The tem Gcal τ is the integated cal powe in a PDP fom a calibation un and is assigned the known input powe P. III. cal SITE-SPECIFIC MEASUREMENT RESULTS A. Lage Scale Path Loss Figues 4a and 4b ae scatte plots of all measued path loss values vesus distance fo the 2.5 GHz and 60 GHz measuements. A Minimum Mean Squae Eo analysis was applied to the measued data to detemine the path loss exponent [16]. The esulting path loss exponent fo 2.5 GHz was n = 2.4, with a standad deviation of σ = 5.8dB and fo 60 GHz the path loss exponent was n = 2.1 with σ = 7.9dB, which ae within the anges fo in-building same-floo popagation epoted in [6], [16], [22], [26], [27]. This wok was suppoted in pat by National Science Foundation Gant Numbe EIA /02/$ IEEE. 98

3 (a) (b) Fig. 4. Scatteplot of all Measued Path Loss Values (Excluding Antenna Gains) on a Single Floo of a Moden Office Building fo (a) 2.5 GHz with Tansmitte and Receive using Omnidiectional Biconical Antennas with 6dBi Gain, and (b) 60 GHz with Tansmitte and Receive using Pyamidal Hon Antennas with 25dBi Gain and 50 Fist-Null Beamwidth B. Patition Based Path Loss Analysis Using measued penetation losses, Dugin, et. al., Nobles, et. al., and Kalsson, et. al., have developed indoo popagation models at 5.8 GHz, 17 & 60 GHz, and 5.0 GHz, espectively, to pedict path loss based on the numbe and types of obstuctions encounteed between tansmitte and eceive [4], [22], [23]. These models can also be used to chaacteize the site-specific natue of emeging femtocellula systems, and may be used in ay-tacing algoithms to pedict netwok coveage and thoughput. In popagation analysis the path loss exponent, n, is useful fo pedicting lage-scale popagation effects. Howeve, the path loss exponent model is inadequate at pedicting sitespecific popagation effects, such as eflection, diffaction, o penetation losses caused by a paticula building layout, constuction mateials, funitue, etc. A moe efined model uses patition-dependant attenuation factos [4], [28], which assumes fee space popagation (n=2) with additional path loss incued based on the numbe and type of objects (such as walls o doos) intesected by a single ay dawn fom tansmitte to eceive. Then, the path loss is given by the following [4] 4π d PL( d ) = 20log10 + ax a + bx b + λ whee ( ) d in metes fom the tansmitte,,, K (2) PL d is the path loss in db at a paticula distance Xa Xb K ae the attenuation values in db fo the patitions between tansmitte and eceive, and abk,, ae the numbe of times the ay intesects each type of patition (i.e. a intesections with patition X a, b intesections with patition X b, and so foth). Fo measuement data at a paticula site, the only unknowns in (2) ae the individual patition attenuation factos, Xa, Xb, K, which may be found by applying a Minimum Mean Squae Eo analysis of measued vesus pedicted path loss using the pocedue descibed in [4]. By looking at the building floo plan (Fig. 1), patitions that existed between each tansmitte and eceive link wee placed in five sepaate categoies: 1. Dywall: 2 sheets of standad ½ inch thick sheetock wallboad. 2. Office Whiteboad: Standad office dy-ease melamine whiteboad, attached to ½ inch thick plywood backing. 3. Clea Glass: 3mm thick clea glass which is untextued and uneinfoced. 4. Mesh Glass: 3 mm thick clea glass which has been einfoced with intelacing 24 gauge wies configued in a ectangula gid with openings of ½ inch ½ inch. 5. Clutte: Objects that encoached into the fist Fesnel zone but did not diectly block the LOS signal fom tansmitte to eceive. Clutte includes office funitue such as chais, desks, bookcases, and filing cabinets, in addition to soft patitions that did not extend to the ceiling. A summay of all patition attenuation factos at 2.5 GHz and 60 GHz is shown in Table I, with the attenuation values epesenting loss in excess of fee space, i.e., loss induced by the patition in addition to the ideal fee space path loss (n = 2). Additionally, to ensue that patition attenuation values could be compaed in a meaningful way, all attenuation values wee nomalized to db pe centimete of mateial thickness. IV. CONCLUSIONS This pape pesented the esults of a measuement campaign and detailed analysis of in-building 2.5 GHz and 60 GHz wieless channels. Measuements wee analyzed in context with site-specific infomation, and esults include patition loss values fo a vaiety of mateials encounteed in an office o laboatoy building, and ae compaable to values published in the liteatue /02/$ IEEE. 99

4 Numbe of Measuements at Each Fequency TABLE I. PARTITION LOSSES (LOSS IN EXCESS OF FREE SPACE) AT 2.5 & 60 GHZ ON THE 4TH FLOOR OF DURHAM HALL, VIRGINIA TECH Dywall Office Whiteboad Clea Glass Mesh Glass Clutte Mateial Thickness (cm) GHz 60 GHz Aveage Measued Attenuation (db) Measuement Standad Deviation (db) Nomalized Aveage Attenuation (db/cm) Aveage Measued Attenuation (db) Measuement Standad Deviation (db) Nomalized Aveage Attenuation (db/cm) A pseudo deteministic method fo detemining the eceived powe in an envionment whee tansmitte and eceive ae sepaated by vaious obstuctions is given by [4] N 4π d PR( d) = PT + GT + GR 20log10 aixi λ (3) i= 1 whee P ( d ) is the eceived powe in dbm at a paticula R distance d in metes fom the tansmitte, P T is the tansmitte powe in dbm, and a i & X i ae the numbe and attenuation value (chosen fom Tables IV o V) fo the i th obstuction intesected by a line dawn fom the tansmitte to the eceive. Based on this wok, the patition based channel model woks well fo shot tansmitte-eceive sepaations, povided thee ae a small numbe of multipath scattees in the envionment. If a significant amount of the eceived powe comes fom multipath, then the patition based model loses its physical significance. One dawback to the patition based model is the need fo site-specific infomation, consisting of a floo plan that identifies the composition of all walls, doos, and othe obstuctions; howeve, futue geneations of wieless netwoks may waant such detail and accuacy [29]. REFERENCES [1] T. S. Rappapot, A. Annamalai, R. M. Buehe, and W. H. Tante, Wieless communications: past events and futue pespectives, IEEE Communications Magazine 50 th Annivesay Issue, pp , May [2] G. Vannucci and R. S. Roman, Measuement esults on indoo adio fequency euse at 900 MHz and 18 GHz, IEEE 3 d Intenational Confeence on Pesonal, Indoo, and Mobile Radio Communications, pp , Octobe [3] D. Molkda, Review on adio popagation into and within buildings, IEE Poceedings Micowaves, Antennas, and Popagation, Vol. 138, No. 1, pp , Febuay [4] G. Dugin, T. S. Rappapot, and H. Xu, Measuements and models fo adio path loss and penetation loss in and aound homes and tees at 5.85 GHz, IEEE Tansactions on Communications, vol. 46, No. 11, Novembe [5] H. Xu, Teestial adio wave popagation at millimete-wave fequencies, Ph. D. Dissetation, Viginia Polytechnic Institute and State Univesity, May [6] H. Xu, V. Kukshya, and T. S. Rappapot, Spatial and tempoal chaacteistics of 60-GHz indoo channels, IEEE Jounal on Selected Aeas in Communications, vol. 20, No. 3, pp , Apil [7] W. J. Tanis, II, and G. J. Pilato, Building penetation chaacteistics of 880 MHz and 1922 MHz adio waves, Poceedings IEEE 43 d Vehicula Technology Confeence, Secanus, NJ, May 1993, pp [8] A. F. de Toledo, and A. M. D. Tukmani, Popagation into and within buildings at 900, 1800, and 2300 MHz, Poceedings IEEE 42 nd Vehicula Technology Confeence, Denve, CO, May 1992, vol. 2, pp [9] E. H. Walke, Penetation of adio signals into buildings in the cellula adio envionment, Bell System Technical Jounal, vol. 62, no. 9, pp , Nov [10] Y. P. Zhang, and Y. Hwang, Time delay chaacteistics of 2.4 GHz band adio popagation channels in oom envionments, Poceedings IEEE 5 th Pesonal, Indoo and Mobile Radio Communications, The Hague, Nethelands, Septembe 1994, vol. 1, pp [11] R. Davies, M. Bensebti, M. A. Beach, and J. P. McGeehan, Wieless popagation measuements in indoo multipath envionments at 1.7 GHz and 60 GHz fo small cell systems, in Poceedings IEEE 41 st Vehicula Technology Confeence, St. Louis, MO, 1991, pp [12] T. Manabe, Y. Miua, and T. Ihaa, Effects of antenna diectivity and polaization on indoo multipath popagation chaacteistics at 60 GHz, IEEE Jounal on Selected Aeas in Communications, vol. 14, No. 3, pp , Apil [13] S. E. Alexande and G. Pugliese, Codless communication within buildings: esults of measuements at 900 MHz and 60 GHz, Bitish Telecom Technology Jounal, vol. 44, No. 10, pp , Octobe [14] Y. P. Zhang, Y. Hwang, Measuements of the chaacteistics of indoo penetation loss, IEEE 44 th Vehicula Technology Confeence, Stockholm, Sweeden, June 1994, vol. 3, pp [15] L. Golding and A. Livine, RLAN-A adio local aea netwok fo voice and data communications, Poceedings IEEE GLOBECOM vol. 3, pp , Novembe [16] T. S. Rappapot, Wieless Communications: Pinciples and Pactice, 2 nd Edition. New Jesey: Pentice-Hall, [17] B. Langen, G. Lobe, W. Hezig, Reflection and tansmission behaviou of building mateials at 60 GHz, IEEE 4 th Intenational Confeence on Pesonal, Indoo, and Mobile Radio Communications, The Hague, Nethelands, Septembe 1994, vol. 4, pp /02/$ IEEE. 100

5 [18] L. M. Coeia and P. O. Fançês, Estimation of mateials chaacteistics fom powe measuements at 60 GHz, IEEE 7 th Intenational Confeence on Pesonal, Indoo, and Mobile Radio Communications, The Hague, Nethelands, Septembe 1994, vol. 1, pp [19] K. Sato, T. Manabe, T. Ihaa, H. Saito, S. Ito, T. Tanaka, K. Sugai, N. Ohmi, Y. Muakami, M. Shibayama, Y. Konishi, and T. Kimua, Measuements of eflection and tansmission chaacteistics of inteio stuctues of office building in the 60-GHz band, IEEE Tansactions on Antennas and Popagation, vol. 45, No. 2, pp , Decembe [20] M. Lott and I. Fokel, A multi-wall-and-floo model fo indoo adio popagation, IEEE 53 d Vehicula Technology Confeence, Rhodes, Geece, May 2001, vol. 1, pp [21] K. Sato, T. Manabe, J. Polivka, T. Ihaa, Y. Kasashima, and K. Yamaki, Measuement of the complex efactive index of concete at 57.5 GHz, IEEE Tansactions on Antennas and Popagation, vol. 44, No. 1, pp , Januay [22] P. Nobles and F. Halsall, Indoo popagation at 17 GHz and 60 GHz measuements and modeling, IEE National Confeence on Antennas and Popagation, pp , [23] P. Kalsson, C. Begljung, E. Thomsen, and H. Böjeson, Wideband measuement and analysis of penetation loss in the 5 GHz band, Poceedings IEEE 50 th Vehicula Technology Confeence, Amstedam, Nethelands, Septembe 1999, vol. 4, pp [24] C. R. Andeson, Design and implementation of an ultaboadband millimete-wavelength vecto sliding coelato channel sounde and inbuilding measuements at 2.5 & 60 GHz, Mastes Thesis, Viginia Polytechnic Institute and State Univesity, May [25] Site-specific system simulato fo wieless system design [Online]. Available: [26] M. J. Feuestein, K. L. Blackad, T. S. Rappapot, S. Y. Seidel, H. H. Xia, Path loss, delay spead, and outage models as functions of antenna height fo micocellula system design, IEEE Tansactions on Vehicula Technology, vol. 44, No. 3, pp , August [27] D. M. Matic, H. Haada, R. Pasad, Indoo and outdoo fequency measuements fo MM-waves in the ange of 60 GHz, Poceedings IEEE 48 th Vehicula Technology Confeence, Ottawa, Canada, May 1998, vol. 1, pp [28] R. R. Skidmoe, T. S. Rappapot, A. L. Abbott, Inteactive coveage egion and system design simulation fo wieless communication systems in multiflooed indoo envionments: SMT Plus, IEEE 5 th Intenational Confeence on Univesal Pesonal Communications, Cambidge, Mass. USA, Octobe 1996, vol. 2, pp [29] T. S. Rappapot and S. Sandhu, Radio-wave popagation fo emeging wieless pesonal-communication systems, IEEE Antennas and Popagation Magazine, vol. 36, pp , Oct /02/$ IEEE. 101