A RAY TRACING METHOD FOR RADIO WAVE PROPAGATION PREDICTION ON SELECTED LOCATIONS OF SUN-U CAMPUS

Transcription

1 Sunway Academic Jounal Volume 9 43 A RAY TRACING METHOD FOR RADIO WAVE PROPAGATION PREDICTION ON SELECTED LOCATIONS OF SUN-U CAMPUS SOO YONG LIM Sunway Univesity No. 5, Jalan Univesiti, Banda Sunway Petaling Jaya, Selango Daul Ehsan Tel: +60(3) Ext.3213 Fax: +60(3) gacel@sunway.edu.my ABSTRACT Radio popagation pediction is a cucial step to detemine the popagation chaacteistics of any abitay installation fo the implementation of a mobile adio system. In this pape, a ay tacing appoach using image method is adopted to pefom adio popagation pedictions in two selected aeas of Sun-U Campus, namely an empty hallway located on the second floo in the School of Compute Technology (SCT), and its adjacent academic staff ooms. The values of electical paamete such as dielectic constant (elative pemittivity) of diffeent mateials ae alteed assuming the walls that sepaate each academic staff oom ae made of ubbe, glass, plasteboad, and concete. How this vaiation affects adio wave popagation pediction is discussed while the main popagation mechanisms taveling along the empty hallway ae epoted. In addition, the easons fo an appopiate inteval between two sampling points along a tansmitte-eceive oute ae also illustated and explained. Keywods: adio wave popagation, ay tacing, elative dielectic constant INTRODUCTION Sunway Univesity, Malaysia ecently (in the month of Febuay 2011) made available Sun- U wieless access (Sun-U WLAN) to all staff membes and students using thei own devices like smatphones and notebooks thoughout the whole univesity campus. To gain wieless access, staff membes and students will just have to seach fo and connect to Sun-U WLAN WiFi netwok and ente thei egula netwok (PC) login cedentials. This news comes in timely because with the polifeation of notebooks and smatphones in the univesity campus, easy access to the intenet will both enhance students leaning expeiences and make staff s wok moe poductive. Moe impotantly, in this age of infomation and communication technology (ICT), easy wieless access is paticulaly cucial fo constantly binging new excitement into leaning and communication. As a matte of fact, telecommunications, wieless applications, and mobile devices, among othes, ae fundamentally eshaping the way infomation is obtained and the ways people lean and communicate with each othe. It is a well-known fact that wieless eception in buildings has often been spotty, with poo voice quality and slow data ates. This is because adio signal, especially at the high

2 44 fequency bands commonly used fo the thid-geneation (3G) cellula and wieless LAN communications, do not penetate building walls well (Otiz, 2008). To undestand fundamentally what affects wieless eception in buildings, a close examination of the elevant popagation mechanisms is equied. In pinciple, thee ae fou basic popagation mechanisms that impact popagation in a mobile communication system, namely, eflection, tansmission, diffaction, and scatteing (Lim, 2010). Based on the physics of these popagation mechanisms, path loss can be pedicted fo lage-scale popagation models using ay optics with the assumption that the opeation fequency is high o wavelength is small compaed with the typical size of objects in the envionment. Fo adio waves of 2.4 GHz (that of wieless LAN), ay optics is valid fo objects such as hallway and the walls/patitions that sepaate each academic staff oom. Thee is a gowing need to pedict signal levels fo shot popagation paths in the ange of m (Holloway, Peini, DeLyse, & Allen, 1997). This pediction needs aises because fo such shot popagation paths, accuate behaviou of eflecting waves fom the suounding objects like building walls (that ae made up of diffeent mateials) plays a cucial ole in signal pediction. Hence, the eseach findings fo shot paths popagation can benefit applications like micocellula pesonal communications sevices (PCS) deployments in malls and aipots. Besides, it can also contibute to the successful design and implementation of business campuses utilizing wieless pivate banch exchanges (PBX s) and wieless local-aea netwoks (LAN s). Not only that, vehicula communications though uban canyons to neaby elays will also benefit fom the eseach findings of shot paths popagation pediction. LITERATURE REVIEW An examination of the majo eseach in adio popagation in cellula mobile telephone systems shows that theoetical studies, numeical simulations and field measuements have always been closely intetwined. Published eseach findings of these woks ae available mostly fom 1980s onwads although the topic of indoo adio popagation channel can be dated all the way back to 1959 (Hashemi, 1993). It has been found that any ealistic popagation model should take into account a vaiety of factos and its paametes should be validated though actual field measuements athe than simplified theoy. Although published wok on the topic of indoo adio popagation has appeaed as ealy as 1959, it wasn t until the ealy 1980s that the fist wave of eseach in indoo popagation began to take place. Rappapot (1996) in his book has cedited Cox at AT&T Bell Laboatoies and Alexande at Bitish Telecom as the fist to caefully study indoo path loss in and aound a lage numbe of homes and office buildings. Indoo adio popagation is govened by the same popagation mechanisms as outdoo, except that thei conditions ae much moe vaiable. One similaity can be found among the popagation models between indoo and outdoo popagations, i.e., both theoetical and measuement-based path loss models show that the aveage eceived signal deceases logaithmically with distance. This phenomenon is called lage-scale path loss whee signal attenuation is caused by the geomety of objects in the popagation envionments. Fo estimating these lage-scale effects, a common pactice is to expess the aveage lage-scale path loss as a function of sepaation distance between the tansmitte and eceive

3 Sunway Academic Jounal Volume 9 45 n with a path loss exponent, n, i.e., path loss d. Diffeent values of n can be obtained fo diffeent popagation scenaios such as outdoo and indoo envionments, including the heavy multipath indoo cases like a staiwell (Lim et al., 2009). Since diffeent spectum bands have unique popagation chaacteistics and equie the appopiate popagation models, the values of n ae fequency dependent fo both outdoo and indoo popagation envionments. In pactice, these n values ae usually computed fom actual measued data, using linea egession such that the diffeence between the measued and the estimated path losses is minimized ove a wide ange of measuement locations and tansmitte-eceive sepaations. A numbe of popagation models have been developed to pedict signal stength in indoo envionments, such as the log-distance path loss model, the Eicsson Multiple Beakpoint model (Akebeg, 1998), and the Attenuation Facto model (Honchaenko & Betoni, 1993). Inteestingly, these indoo path loss models have all been empiical models that wee developed based on the measued aveage losses in vaious in-building envionments. One advantage with the empiical appoach is that it inheently accounts fo all popagation factos, both known and unknown by actual field measuements. Yet, an empiical model is valid only to simila envionments whee the model is developed. The accuacy of popagation pediction involves seveal aspects such as the accuacy of locations and sizes of buildings as well as an accuate knowledge of the electic paametes of walls and othe objects involved. Tees, lage posts, taffic, and even pedestians in outdoo scenaios and funitue in indoo cases can all influence the popagation pediction esults. The published value of mateials like elative dielectic constant is fequency dependent and may vay a lot fom one to the othe. Fo instance, bicks may be fabicated fom diffeent mateials, and window glass may be metalized and hence vey eflective. Also, because of the vey high value of fo wate (e.g. 70 fo sea wate and 81 fo distilled wate), the wate content of mateials such as bick, concete, and gound has a majo effect on thei dielectic constant. Common values of that have been epoted fo vaious mateials of inteest (Betoni, 2000) ae excepted and listed in Table 1.

4 46 Table 1. The Values of Relative Dielectic Constants fo some Common Mateials Mateial Relative Dielectic Constant ( ) Glass Wood Gypsum boad 2.8 Chip boad 2.9 Dy bick 4 Dy concete 4-6 Aeated concete 2-3 Limestone 7.5 Mable 11.6 Gound 7-30 Fesh wate 81 Seawate 81 Snow Ice 3.2 METHODOLOGY The ay tacing using image method povides a simple and accuate way fo detemining the ay tajectoy between a tansmitte (Tx) and a eceive (Rx) (Iskande & Yun, 2002). The image method utilizes the images of the tansmit antenna location elative to all the sufaces of the envionment. The coodinate of all the images is calculated and aftewads ays ae taced towads these images. Fist and second ode eflections can be calculated vey efficiently without sending ays to all diections. The dawback of this method is that calculation time gows exponentially when the ode of the calculated eflections is inceased. Within the scope of this wok, howeve, the image method is an ideal appoach fo detemining the main popagation mechanisms. Figue 1 pesents a simple eflection suface to illustate the basic idea of how the image method woks. Fo the scenaio in Figue 1, LOS (line-of-sight) is the path between the Tx and the Rx. To calculate the eflection fom the suface, the image of Tx with espect to the suface is identified and is denoted as Tx. Note that the distance fom Tx to the suface and the distance fom Tx to the same suface ae equal ( d1 d2 ). Next, by connecting Tx and Rx, the intesection point on the suface (P) is the eflection point whee eflection occus. Fo multiple eflections, multiple images with espect to the elevant sufaces will be detemined in a simila way and the coesponding ay paths can then be obtained.

5 Sunway Academic Jounal Volume 9 47 Rx LOS Tx P d 1 d d 2 1 = d 2 Tx Figue 1. Illustation of the Image-Based Ray Tacing Method Two locations have been selected fom Sun-U campus fo ay tacing simulation pupose, namely SCT academic staff ooms and the adjacent hallway. In these two locations, the adio links fo ay tacing simulation have been established in a staight line manne, and they ae denoted as the paths fom Tx 1 to Rx 1, as well as that fom Tx 2 to Rx 2. Figue 2 pesents the layout of the SCT academic staff ooms and the adjacent hallway, with indication of the locations of the two tansmittes (Tx 1 and Tx 2) and eceives (Rx 1 and Rx 2). Figue 2. Layout of the SCT Academic Staff Rooms and the Adjacent Empty Hallway (not to scale)

6 48 SCT Academic Staff Rooms Buildings have a huge selection of patitions and obstacles that fom not only the intenal but also the extenal stuctue (Rappapot, 1996). Fo instance, in an office building envionment, soft and had patitions can be commonly seen. While the fome ae patitions that may be moved and do not extend to the ceiling, the latte ae fomed as pat of the building stuctue. On Sun-U campus in paticula, the buildings walls and had patitions ae made up of vaious mateials such as bicks, plasteboad, and glass. It is woth pointing out that one distinct featue about Sun-U campus is that quite a significant numbe of the building walls ae glass walls. This is especially the case fo those at o nea the foye like the info cente, secuity cente, the Sunway College office (then Mancheste Business School s office), Sunway TES, Lancaste Univesity s office, and LeCodon Bleu Institute of Culinay Ats. The LeCodon Bleu Institute of Culinay Ats fo instance, is a moden, newly-completed 2-stoey full glass building to house the eception office of the Sunway LeCodon Bleu Institute of Culinay Ats. This pape will examine, via ay tacing simulations, the effects of building walls (that ae made up of vaious mateials) on adio popagation path gains acoss multiple patitions in the SCT academic staff ooms. Figue 3 and Figue 4 show two pictues of the envionments concened. Figue 3. Pictue of the SCT Academic Staff Rooms and the Adjacent Hallway

7 Sunway Academic Jounal Volume 9 49 Figue 4. Pictue of the Foye on Sun-U Campus Hallway Adjacent to the SCT Academic Staff Rooms The othe objective of this pape is to investigate, again via ay tacing simulation using image method, the main popagation mechanisms tavelling along an empty hallway adjacent to the SCT academic staff ooms. This is a diffeent popagation scenaio wheeby no obstacles ae assumed pesent along the hallway fo simulation pupose. Neithe was thee any movement of people o stuff duing the simulation. The goal of this wok is to undestand how signal tavels fom one point to anothe. In othe wods, the behavio of the EM wave as it makes its way fom the tansmitte to the eceive at vaious inteval (e.g. 3 m, 1 m, 0.02 m, m) is unde scutiny. RESULTS SCT Academic Staff Rooms The eceived signal stength aising fom a specific tansmitte placed at a cetain place will vay with locations within any given building. As a signal passes though walls that ae made up of vaious mateials, diffeent levels of signal attenuation esult. The exact natue of this signal vaiation depends pimaily on the shape and constuction of the building as well as the geomety of the adio link. To depict the vaiation esulting fom a signal passing

8 50 though walls made up of fou diffeent types of mateials, Figue 5 is plotted fo the case when Tx 1 is put on one end of the SCT academic building while Rx 1 is moved acoss the SCT academic staff ooms (see Figue 2) in a staight line manne. The fou mateials unde scutiny ae ubbe wall, glass wall, plasteboad wall, and concete wall. Of all these fou types of diffeent walls, plasteboad wall is the one that eflects the tue condition in the existing building. In addition to these fou simulations of distinct walls, line-of-sight (LOS) ay is also plotted assuming thee ae no walls blocking the tansmitte-eceive path as the signal fades away afte leaving the point souce. This additional simulation will seve as a geneal guideline to compae how signal will attenuate passing though walls made up of diffeent mateials. In Figue 5, the five scenaios descibed ealie cay diffeent values of elative dielectic constant. In fee space (LOS scenaio), dielectic constant has a value of 1. Wheeas fo the fou diffeent walls, the elative dielectic constants ae: ubbe wall, = 2.5; glass wall, = 5; plasteboad wall, = 6; and finally, concete wall, = 9. Table 2 ecods all these afoementioned values. Table 2. Values of Relative Dielectic Constant of Vaious Walls Wall Relative Dielectic Constant ( ) Rubbe 2.5 Glass 5 Plasteboad 6 Concete 9

9 Path Gain (db) Sunway Academic Jounal Volume Vaiation of Signal Stength Diect Ray (LOS) Rubbe Glass Plasteboad Concete Distance between Tx & Rx (m) Figue 5. Signal Attenuation acoss the SCT Academic Staff Rooms (Wall Tansmission Effects) In Figue 5, the ays ae plotted by consideing only the tansmitted ay (a single tansmission coefficient fo each ay) passing though the wall sepaating each of the SCT academic staff ooms. When only a single ay is consideed in the simulation, be it a diect ay, eflected ay, o tansmitted ay, the cuve will appea smooth unless a mixtue of the ays (e.g. diect + eflected ay; o eflected + eflected ays) ae shown. Additionally, to show the wall eflection effects that esult fom a mixtue of diffeent ays, Figue 6 is plotted using the total ays tavelling along the empty hallway next to the SCT academic staff ooms, which consists of diect ay, left-wall singly-eflected ay, ightwall singly-eflected ay, left-wall doubly-eflected ay, and ight-wall doubly-eflected ay. Fou cases of the total ays ae plotted in Figue 6, which illustates the diffeent eflection effects fom the fou distinct types of walls, namely, ubbe, glass, plasteboad and concete walls.

10 Path Gain (db) 52 5 Vaiation of Signal Stength Rubbe Glass -30 Plasteboad Concete Distance between Tx & Rx (m) Figue 6. Signal Attenuation next to the SCT Academic Staff Rooms (Wall Reflection Effects) Hallway Adjacent to the SCT Academic Staff Rooms In this section, a diffeent scenaio is pesented, which is diffeent fom the signal attenuation acoss the SCT academic staff ooms descibed ealie. In Figue 5, only one ay is examined at any given time. But fo this section (simila to Figue 6), up to seveal odes of ays ae examined all togethe at one time. This includes both the LOS ay as well as the eflected ays of diffeent odes fom both the left and ight walls. These suounding walls ae assumed as ubbe wall (elative dielectic constant, = 2.5) fo the simulations in Figues 7, 8, 9, and 10, espectively. Since diffeent wall mateials will cause diffeent eflection coefficient values, which in tun will esult in some changes of the analysis esults, only a single wall mateial is selected fo the simulations to analyze the effects of the inteval between two sampling points. Fo this wok, the Tx and Rx ae moved to sit along an empty hallway adjacent to the SCT academic staff ooms. In paticula, Tx 2 and Rx 2 ae put at the cente of the empty hallway, with Rx 2 being moved away gadually fom the Tx in a staight line manne, as illustated in Figue 2 ealie. The esults of the signal attenuation along this empty hallway along the eceive oute ae plotted in Figue 7, at an inteval of 0.02 m. In addition to Figue 7, Figue 8 is plotted with the inteval between any two sampling points along the eceive oute inceased fom 0.02 m to 3 m. Othe paametes emain

11 Path Gain (db) Sunway Academic Jounal Volume 9 53 unchanged. In Figue 9, howeve, the inteval is educed to 1 m, which is a value between 0.02 m and 3 m. On top of that, Figue 10 is also plotted to investigate the signal behaviou when the inteval is futhe educed to a small numbe, m. 10 Vaiation of Signal Stength Diect Ray (LOS) X: Y: LOS+1 Ref. (left) -40 LOS+1 Ref. (Left&Right) LOS+2 Ref. (Left&Right) Distance between Tx & Rx (m) Figue 7. Signal Attenuation along the empty hallway (Inteval = 0.02 m)

12 Path Gain (db) Vaiation of Signal Stength Diect Ray (LOS) LOS+1 Ref. (left) LOS+1 Ref. (Left&Right) LOS+2 Ref. (Left&Right) Distance between Tx & Rx (m) Figue 8. When the Inteval is inceased to 3 m

13 Path Gain (db) Sunway Academic Jounal Volume Vaiation of Signal Stength Diect Ray (LOS) LOS+1 Ref. (left) LOS+1 Ref. (Left&Right) LOS+2 Ref. (Left&Right) Distance between Tx & Rx (m) Figue 9. When the Inteval is educed fom 3 m to 1 m

14 Path Gain (db) Vaiation of Signal Stength Diect Ray (LOS) LOS+1 Ref. (left) -40 LOS+1 Ref. (Left&Right) LOS+2 Ref. (Left&Right) Distance between Tx & Rx (m) Figue 10. When the Inteval is futhe educed to m DISCUSSION SCT Academic Staff Rooms It can be obseved fom Figue 5 that the signal attenuation fo the LOS case is the smallest, with signal dopping fom db to db, o a total attenuation of db. Howeve, in the absence of the LOS ay, when thee ae walls standing in the way between Tx 1 and Rx 1, the signal attenuation becomes geate as the value of the elative dielectic constant of the mateial of the wall inceases. This can be seen in Figue 5 that shows that signal attenuation fo ubbe wall, glass wall, plaste boad wall, and concete wall ae db, db, db, and db espectively. The diffeence of the total signal attenuation between the LOS case and the concete wall is db. Hallway Adjacent to the SCT Academic Staff Rooms Fom Figue 7, which shows the signal attenuation along the empty hallway depicted in Figue 3, it can be noticed that when multiple eflections ae included in the simulations, fading occus due to multipath popagation.

15 Sunway Academic Jounal Volume 9 57 In this scenaio, fist and second odes of eflections fom both the left and ight walls ae consideed in the simulation using ay tacing of image method. The two suounding walls act as eflectos fo the empty hallway whee the tansmitte and the line segment (between Tx 2 and Rx 2 in Figue 2) ae located. Since the walls suound both Tx 2 and Rx 2, multiple paths ae ceated fo which a tansmitted signal can tavese. As a consequence, the eceive sees the supeposition of multiple copies of the tansmitted signal, each tavesing a diffeent path. Each of these signals (diect ay, left-wall-eflected ays, and ight-wall-eflected ays) expeiences diffeences in attenuation, delay and phase shift while taveling fom the tansmitte to the eceive. Hence, at the eceive, constuctive and destuctive intefeence occus that amplifies and attenuates the signal powe espectively. Stong destuctive intefeence (deep fade) occus at location m along the eceive path. As fo the simulation esults that involve only the diect ay, signal attenuates gadually fom 4.32 db to db as it moves away bit by bit fom the tansmitte (o a total of db in signal attenuation). Judging fom Figue 7 to Figue 10, it was obseved that the inteval between two sampling points of any given tansmitte-eceive path should be caefully chosen so that the signal can be intepeted pecisely. Fom the fou intevals that ange fom m to 3 m (0.002 m, 0.02 m, 1 m, and 3 m), it was obseved that 0.02 m is the ideal choice in this scenaio as the answes convege at this inteval. A futhe eduction to m bings not much diffeence to the ay tacing simulation esults, since Figue 10 and Figue 7 ae almost identical. But when the sampling inteval is set at a high numbe, such as 3 m in this scenaio, the esults obtained ae a geneal one, with no show of deep fading along the eceive oute. These esults might not eflect the tue condition of the signals actual behaviou. CONCLUSION In this pape, a ay tacing using image method has been adopted to un simulations fo adio popagation pedictions at two selected locations on Sun-U Campus at 2.4 GHz. Fom this wok, it was shown that multiple eflections fom the suounding walls along an empty hallway poduce many paths by which the signal can popagate fom the tansmitte to the eceive. This popagation phenomenon wheeby a signal tavels fom a tansmitte to a eceive though multiple paths (ays) is called multipath. The effects of multipath can be good, and/o bad, depending on the applications. Fo instance, fo naowband systems, the sum of phasos that causes fading is a bad effect that might esult in tempoay failue of communication. Likewise fo wideband systems, multipath that might cause inte-symbol intefeence is also a bad effect. Nevetheless, fo multiple-input multiple-output (MIMO) system, multipath is desied because it can lead to multiple paallel channels. To analyze the multipath effects accuately, the inteval between any two sampling points along the tansmitte-eceive oute ought to be chosen caefully. Explicitly, the inteval cannot be set to an uneasonably high numbe (two consecutive sampling points ae sepaated too fa fom one anothe); othewise it might esult in an inaccuate intepetation of the esults obtained. Fo ay tacing simulations acoss the SCT academic staff ooms, the eceived signal shows a systematic decease with distance, but with a vaying amount of total signal attenuation, that in tun depends on the mateials of the associated wall/patition. The

16 58 simulations and analysis fom this wok shows that signal attenuation ises as the value of the elative dielectic constant of the mateial of the wall inceases. In othe wods, a signal can penetate ubbe wall moe easily with less attenuation than it does concete wall. In a nutshell, the findings of this wok have added moe insights to shot popagation paths that have seen steady ising of its accuate pediction needs especially in ecent yeas. One may conclude that adio waves popagation within a building is stongly influenced not only by the layout of the buildings but by the constuction mateials as well. In the pesent scenaio of SCT academic staff offices, the plaste boad patition exhibits a gadual signal loss totaling to db ove a distance of 25 m along a tansmitte-eceive oute at 2.4 GHz. REFERENCES Akebeg, D. (1988). Popeties of a TDMA picocellula office communication system. GLOBECOM 88 IEEE Global Telecommunications Confeence 28 Novembe 1 Decembe 1988, Vol. 3 (pp ). doi: /GLOCOM Betoni, H. L. (2000). Radio popagation fo moden wieless systems. Uppe Saddle Rive, NJ: Pentice Hall. Hashemi, H. (1993). The indoo adio popagation channel. Poceedings of the IEEE, 81(7), Holloway, C. L., Peini, P. L., DeLyse, R. R., & Allen, K. C. (1997). Analysis of composite walls and thei effects on shot-path popagation modelling. IEEE Tansactions on Vehicula Technology, 46(3), Honchaenko, W., & Betoni, H. L. (1993). Mechanisms govening popagation between diffeent floos in buildings. IEEE Tansactions on Antennas Popagation, 41(6), Iskande, M. F. (1992). Electomagnetic fields and waves. Pospect Heights, IL: Waveland Pess Iskande, M. F., & Yun, Z. (2002). Popagation pediction models fo wieless communication systems. IEEE Tansactions on Micowave Theoy and Techniques, 50(3), Lim, S. Y., Yun, Z., Bake, J. M., Celik, N., Youn, H. S., & Iskande, M. F. (2009). Popagation modeling and measuement fo a multifloo staiwell. IEEE Antennas and Wieless Popagation Lettes, 8, Lim, S. Y. (2010). Radio wave popagation measuement and modeling in wieless communication envionments (Unpublished doctoal dissetation). Univesity of Hawaii, Manoa. Otiz, S. (2008). The wieless industy begins to embace femtocells. Compute, 41(7), Rappapot, T. S. (1996). Wieless communications: pinciples and pactice. Hoboken, NJ: Institute of Electical & Electonics Enginees.