PHASE FUNCTION MEASUREMENT FOR MODELLING RADIOWAVE ATTENUATION AND SCATTER IN VEGETATION BASED ON THE THEORY OF RADIATIVE ENERGY TRANSFER

Transcription

1 PHASE FUNCTION MEASUREMENT FOR MODELLING RADIOWAVE ATTENUATION AND SCATTER IN VEGETATION BASED ON THE THEORY OF RADIATIVE ENERGY TRANSFER Jürgen Richter 1, Miqdad Al-Nuaimi 1, Rafael Caldeirinha 1 Univerity of Glamorgan, School of Electronic, Pontypridd CF37 1DL, UK, jrichter@glam.ac.uk and malnuaim@glam.ac.uk Departamento de Engenharia Electrotécnica, Ecola Superior de Tecnologia e Getão, Intituto Politécnico de Leiria / IT- Intituto da Telecomunicaçõe, PORTUGAL rcaldeirinha@etg.iplei.pt Abtract Modelling radio wave propagation through vegetation i becoming increaingly more important. Many emperical olution have been uggeted in the pat. The theory of radiative energy tranfer offer an analytical method to the problem. Thi theory require however 4 imput paramter, which need to be etablihed experimentally. A meaurement of the angular catter pattern, the o called phae function, can be utilied to determine of thoe parameter. In thi paper 3 different method of meaureing the phae function for meaurement in a real foret environment are preented and compared. One method i preented a the mot uitable one. Keyword RET, propagation through vegetation, phae function, catter medium I. INTRODUCTION Wirele Communication are revolutioniing peronal and telecommunication ervice and the way in which they are utilied. Overall growth in cellular, fixed and atellite communication ytem market in recent year ha exceeded expectation. There i a widepread anticipation that cutomer demand for wirele telecommunication ytem will continue to expand in the foreeeable future. Such ytem rely in their planning, deign and implementation on the availability of radiowave propagation model. Thee model are required to decribe and characterie with ufficient accuracy the interaction of radiowave with the environment, epecially the variou obtacle in the radio path. Prediction tool are highly deirable to radio planner in achieving appropriate coverage planning, determination of the propagation mode and the prediction and control of mutual co-channel interference between exiting and new radio link. In land mobile and wirele fixed acce ytem, tree, ingly or in a group, are uually preent in the radio cell environment, giving rie to both aborption and catter of radio ignal. Effect ariing from both catter and aborption need to be identified and invetigated to improve the modelling accuracy of the mobile radio communication channel, which in turn will contribute to the improvement of prediction tool for radio coverage. Mot of the interet ha been centred at frequencie around the GSM and DCS-18 band, where tree attenuation i conidered a one of the dominant effect influencing radio propagation in rural and uburban area [1][]. In [3] progre achieved in performing analytical tudie baed on experimental meaurement, ha hown that the preence of tree in a radio path, ingly or a a group, ha ignificant influence on the level of the received ignal. Thi i exercied: (i) directly by providing an additional (exce) attenuation to that caued by free pace propagation; (ii) indirectly by cattering which reult in lateral contribution to the received ignal; and (iii) through depolariation of the incident wave which amount to another ignificant and complex factor [4]-[6]. The latter i epecially ignificant at higher frequency band, where the wavelength become comparable to the phyical dimenion of the cattering element of the tree [7]. Propagation mode aociated with aborption, catter and depolariation were addreed in [3][8][9]. For practical application and due to mathematical and phyical complexitie, prediction of propagation in real environment including obtacle and tree, are normally characteried uing approximate model. More accurate model require detailed and accurate electromagnetic repreentation of the obtacle, making uch model complex and highly demanding of ite decriptive data and computer reource. Approximate prediction model, appropriately validated, are very ueful to radio ytem planner and deigner. At their mot baic form, thee model, referred to a empirical, expre the exce attenuation of the vegetation medium a a function of frequency and vegetation depth. More developed form of prediction model would take into conideration more pecific detail of the vegetation medium uch a vegetation denity, patial ditribution of tree, ize and //$17. IEEE PIMRC

2 orientation of branche and leave, and eaonal factor relating to foliation and moiture content [1][11]. II. THE RADIATIVE ENERGY TRANSFER THEORY (RET) An invetigation forming a main part of a collaborative project in the UK, i concerned with the development of a narrowband prediction model for attenuation and catter from vegetation. An analytical model baed on the theory of radiative energy tranfer i utilied to predict the attenuation curve and directional pectra due to propagation of radio ignal through vegetation [4][1][13]. Uing the RET, the vegetation medium i modelled a a tatitically homogeneou random medium of elementary catterer d which i characteried by the aborption cro ection per unit volume σ a, the catter cro-ection per unit volume σ and the catter function of the medium p ( ˆ, ). The catter function, referred to often a the phae function, i characteried by a narrow forward lobe and an iotropic background, where the unit vector ŝ ' and ŝ indicate, for each catterer, the incidence direction and catter direction repectively, a hown in Figure 1. The model conider a plane wave incident from an air half pace upon the planar interface of a vegetation half pace. The baic equation of the radiative energy tranfer theory i expreed in term of the pecific intenity I. d S' dw ' Figure 1: Scattering from a homogeneou random medium of catterer d The pecific intenity I at a given point within the vegetation medium comprie of the um of a coherent component, which i reduced in intenity due to aborption and catter of the incident wave, and an incoherent (diffue) component, due to the cattered wave. σ. I A 4 d π 4π ( rˆ, ˆ ) + ( σ + σ ) I( rˆ, ˆ ) = p( ˆ, ) I( rˆ, ) Ω' Each catterer i aumed to have a directional catter profile, or phae function. A the contituent of the tree are dw S (1) relatively large relative to the wavelength at micro- and millimetre wave frequencie, the catter function i aumed to conit of a trongly cattered forward lobe, which can be aumed to be Gauian in hape, with an iotropic background level. Each catterer i aumed to have a directional catter profile, or phae function. A the contituent of the tree are relatively large relative to the wavelength at micro- and millimetre wave frequencie, the catter function i aumed to conit of a trongly cattering forward lobe, which can be aumed to be Gauian of width, β, with an iotropic background level. The forward lobe take the form in Eq., β ( θ / ) f ( θ ) = e () β where θ = co 1 ( ˆ, ) forward lobe., and β i the beamwidth of the The cattering pattern i given by Eq. 3, where α i the ratio of the forward cattered power to the total cattered power. p ( θ ) = αf ( θ ) + ( 1 α ) The parameter σ a, σ, α and β are pecific to a vegetation medium and may be etimated for the medium under tudy by comparion of experimental and predicted reult. The relative magnitude of σ a and σ are decribed in term of the albedo W which i given in Eq. 4. W ( σ + σ ) σ a (3) = (4) A. Solution for the Tranport Equation For the olution of the tranport equation (Eq. 1) the pecific intenity at a given point within the vegetation medium, I, i divided into two part: ` I( z, θ ) = I ( z, θ ) I ( z, θ ) (5) ri + d where z i the ditance into the medium Thi repreent the um of a coherent component, I ri, which i reduced in intenity due to aborption and catter of the incident wave, and an incoherent (diffue) component, I d, due to the cattered wave. It i alo convinient to plit I d into two part: I 1 and I. I 1 i determined primarily by the forward lobe of the catter function and I by cattering into the iotropic background, o that: I = I ri + I 1 + I (6) A prediction model for determining the exce attenuation due to vegetation uing the RET require four parameter to be determined. Thee are σ a, σ and the two parameter decribing the phae function. The phae function i

3 decribed by the angular width of it coherent forward lobe β and the ratio of forward cattered energy relative to the total cattered energy α. Appropriate meaurement can be ued to determine the four parameter, which may ubequently be compiled in a databae claifying the parameter againt plant pecie, eaon, vegetation denity and ignal frequency. Meaurement of the phae function for a vegetation medium ha proved to be crucial in the etablihment of uch a databae. Thi paper conider method for phae function meaurement and identifie one meaurement technique, which for outdoor ite, ha proved ucceful in etablihing the phae function and it parameter α and β. III. PHASE FUNCTION MEASUREMENTS In order to determine the parameter α and β the phae function meaurement need to be conducted over a range of angle large enough to cover the expected beamwidth β. Some area of the incoherent back-catter alo need to be covered, o that with the magnitude of the forward lobe over the back-catter α can be determined. Modelling vegetation with the RET lead to it generally being modelled a a homogeneou cattering medium. Therefore when meauring an angular pattern over a relatively mall range of angle within a homogeneou medium, three different technique for meauring the phae function can be enviaged, which are ketched in Fig.. point at the interface, while the receiver i tationary inide the medium ( moving the tranmitter technique, Fig. b). The third method conit of rotating the receiver in dicrete increment about it vertical axi at a fixed location and performing a ignal can ( canning technique, Fig. c), the latter method i the eaiet to implement from a practical viewpoint. A. Meaurement Site Meaurement to ae and validate the three method were carried out at a ite located at the Glebeland Park at Newport in South Wale. The meaurement were conducted on a group of Sycamore tre at the edge of the car park hown in Fig. 3. Thi provided a clearance between the tranmitter and the vegetation of more than 5m (Fig 3a). The tree were on average 1m high and had a foliage pread of about 6m and with no appreciable gap between canopie. The receiver wa poitioned behind the interface at a poition m into the vegetation, Fig. 3b how a photograph taken at the receiver location. The ground wa flat at both end of the interface. Both antenna were placed at 6m high from ground, reulting in an illuminated area centred at the mid canopy region of the tree. (a) (b) (c) Figure : Technique for meauring the phae function (a) Circular motion of receiver inide the medium; (b) the tranmitter moving in a circular motion at ditance from the interface with receiver tatic inide the medium; (c) rotation of the receiving antenna about it vertical axi, inide the medium. The firt method conit of moving the receiver in a circular arc inide the vegetation medium whoe radiu accommodate to 3 tree width ( moving the receiver technique, Fig. a). The econd method i imilar to the previou one, except that the tranmitter i now moving in a circle with a focal (a) (b) Figure 3: Group of tree, Glebeland park - Newport: (a) photograph a een from the tranmitter location, and (b) a een from the receiver end inide the vegetation medium. The tree were about 1m high with a foliage pread of 6m. B. Meaurement Reult Reult of phae function meaurement performed on the group of tree in Fig. 3 are preented here. Fig. 4 how the reult obtained for the moving receiver technique (Fig. a). The reult obtained for the moving tranmitter technique (Fig. b) i hown in Fig. 6. The reult for the canning (rotating receiver) technique (Fig. c) i hown in Fig. 7, here the thick blue line repreent the average ignal level received during the azimuthal can/rotation of the receiver, while the fine red line repreent the modelled phae function uing the RET. The ignal level i normalied to the free pace meaurement (reference level)

4 obtained at the interface in the radio path, where the tranmitter and receiver are aligned, o that the curve how the degradation of the phae function caued by the vegetation obtacle in the radio path. The received ignal normalied to the maximum received ignal i plotted a a function of canning angle. The meaurement receiver i embedded in the vegetation medium at a height of 6m and at a ditance of m from the air interface. The total depth of vegetation in the radio path wa about 15m, being equivalent to ½ tree width. The reult demontrate the unreliability of the meaurement data in identifying the phae function. The main reaon wa attributed to the altered meaurement geometry, due to radio terminal movement inide the vegetation acro everal metre along the arc indicated in Fig. a for the moving receiver technique and everal 1 of metre for the moving tranmitter technique in Fig. b. In a practical foret environment the medium i ufficiently in-homogenou to give rie to the preence of puriou ignal component. Meaured Data Model Normalied Scattering Pattern p( ), db φ Normalied Scattering Pattern p ( φ ), db Rotation Angle of the Receiver φ, º Figure 4: Meaurement reult for vegetation uing the moving receiver technique (r = m), normalied to maximum db Rotation Angle of the Receiver φ,º Figure 6: Meaured and modelled phae function for vegetation uing the canning technique (r = m) (normalied to maximum - db) Normalied Scattering Pattern p( ), db φ Rotation Angle of the Receiver φ, º Figure 5: Meaurement reult for vegetation uing the moving tranmitter technique (r = m), normalied to maximum - db IV. CONCLUSIONS The paper demontrate the complexity of the phae function meaurement and the difficultie encountered in enuring accurate poitioning of the meaurement et up. From theory the moving receiver technique can be argued to be the bet method. Reult preented in the paper how, that it i rather difficult to implement. Similarly, the moving tranmitter technique alo yielded inconcluive reult, and therefore no attempt wa made ubequently to extrapolate phae function information from thi data et. A ummary of the phae function information may be found in Table 1. A broadening of the phae function with increaing vegetation depth can clearly be een. Remarkably, a relatively mall value for α wa obtained from the canning meaurement poibly due to a highly aborbing medium. Here it i ueful to note that he vegetation volume wa quite dene with no gap in between tree. In concluion, the (rotating receiver) canning technique i recommended for meauring the vegetation phae function.

5 It ha proved to be reliable in etablihing good correlation between the phae function parameter and the type of vegetation under conideration. Furthermore, the optimal ditance to carry out uch experiment hould lie well within the forward coherent component region, typically to 3 tree width. A vegetation depth meaurement prior to the phae function meaurement, to verify the coherent region limit i thu recommended. Table 1: Phae function parameter obtained from bet fit of experimental and theoretical reult for variou technique outlined previouly Model Parameter Free pace at interface Phae Function Meaurement Inide the medium Scan Scan Moving RX Moving TX β 3dB (º) n/a n/a β (º) n/a n/a α.9.1 n/a n/a Work i currently being carried out to invetigate the effect of deconvolving the radiation pattern of the receiver antenna uing of the radiation pattern the meaured phae function. Even with highly directional antenna, which hould only have a HPBW of a few degree there will be contribution received through the ide lobe of the antenna. Thi will ignificantly alter the pattern of the meaured phae function thu leading to an overetimation of the magnitude of the iotropic part of the function. Preliminary reult howed that the etimated α value will increae a a reult of the deconvolution. Thi will be demontrated in the full paper. ACKNOWLEDGEMENT The work preented here wa conducted in the framework of a conortium coniting of 4 organization: QinetiQ, Rutherford Appleton Laboratorie, the Univerity of Portmouth and the Univerity of Glamorgan. The work wa funded by the Britih Radiocummunication agency contract reference: AY388/ The author wih to thank the conortium member for their fruitful input and vital contribution. REFERENCES [1] Draft COST 31 Final Report, "Digital Mobile Radio: COST 31 View on the Evolution toward 3rd Generation Sytem", COST 31 TD(96)4-A, Turin, April 17-19, [] DTI, "Multimedia Communication on the move", A Conultation Document from the Department of Trade and Indutry, July 1997 [3] CALDEIRINHA, R. F. S.: 'Characteriation and Modelling of Re-radiation Function from Single Tree Illuminated by Radiowave at Micro- and Millimetre Wave Frequencie', PhD thei, Univerity of Glamorgan, 1. [4] AL-NUAIMI, M. O. and HAMMOUDEH, A. M., " Meaurement and Prediction of Attenuation and Scatter of Microwave Signal by Tree", IEE Proceeding on Antenna and Propagation, Part H, Vol. 141, No., [5] AL-NUAIMI, M. O. and STEPHENS, R. B. L.: 'Meaurement and prediction model optimiation for ignal attenuation in vegetation media at centimetre wave frequencie', IEE Proceeding on Microwave Antenna and Propagation, Vol. 145, No 3, June [6] AL-NUAIMI, M. O. and HAMMOUDEH, A. M.: 'Attenuation function of microwave ignal propagated thorugh tree', Electronic Letter, 1993, 9, (14), pp [7] STEPHENS, R. B. L., AL-NUAIMI, M. O. and CALDEIRINHA, R.: "Characteriation of depolariation of radio ignal by ingle tree at GHz", Fifteenth National Radio Science Conference, Helwan, Cairo, Egypt, B1 1-7, Feb. 4-6, [8] CALDEIRINHA, R. and AL-NUAIMI, M.: "Co-Polar and Cro-Polar Meaurement of the Re-Radiation Signal at GHz from a Tree and their Analyi in the region around the Null", IEEE Proceeding of the Sixteenth National Radio Science Conference, Ain Sham Univerity, Cairo, Egypt, INV4, Feb. 3, [9] CALDEIRINHA, R. and AL-NUAIMI, M.: "Analyi of the Re-radiation Function of Single Tree and Idealied Structure at GHz", AP Millennium Conference on Antenna & Propagation, Davo, Switzerland, Pap. 46, April 11,. [1] CALDEIRINHA, R. and AL-NUAIMI, M.: "An Accurate Model for RCS of Single Leave ", 4th European Peronal Mobile Communication Conference, EPMCC1, Vienna, Autria, February th-nd, 1. [11] CALDEIRINHA, R. and AL-NUAIMI, M.: "A Novel FDTD baed method for prediction of bitatic RCS of ingle leave and tree", 11th International Conference on Antenna & Propagation, ICAP1, Mancheter, UK, April 17, 1. [1] JOHNSON, R. A. and SCHWERING, F.: 'A tranport theory of millimeter wave propagation in wood and foret', CECOM-TR-85-1, Fort Monnouth, New Jerey, 1985 [13] ISHIMARU, A: 'Wave propagation and cattering in random media', Academic Pre, Vol. 1, Chap. 7, 1978.