Realistic Simulation of a Wireless Signal Propagation in an Urban Environment

Transcription

1 Realistic Simulation of a Wieless Signal Popagation in an Uban Envionment Keyukuma J. Patel Univesity of Southen Queensland Toowoomba, Austalia keyupatel@ieee.og Andew Y. J. Lee Nanyang Technological Univesity Singapoe ayjlee@ntu.edu.au Abstact This pape pesents esults obtained fom a 3D ay tacing electomagnetic model designed to map the signals in a high ise envionment (uban) within the same floo and at the floos above and below that floo. Apat fom using ay tacing, the dielectic popeties of the mateials fom each floo ae included to achieve ealistic signal pofiles. The simulation is done fo scenaios based on a single base-station and multiple base-stations on the same floo and multi-floos. We have efined existing indoo popagating model to obtain ealistic signal distibution in a high ise building. New model include the dielectic popeties of mateials exists in eal envionment. The simulation pogam is witten in Visual C++. Keywods wieless signal popagation; wieless simulation; measuement of dielectic; ay tacing method I. INTRODUCTION Due to the highly competitive natue of the wieless communication industy whee moe than one sevice povide is involved, it is impotant to have a quality gade of sevice fom diffeent sevice povides to addess the need of thei subscibes. It is ecognised that the envionment factos have the most significant contibution to the Gade of Sevice (GoS) of a wieless system []. The GoS equation is calculated as: numbeof blockedcalls _ + * numbeof inteuptedcalls _ GoS () total_ numbeof calls Thee ae seveal simulatos developed using diffeent appoaches: statistical path loss model, statistical model, ay tacing model and exact electomagnetic solution In competing fo sevice, a povide must be able to guaantee a cetain gade of sevice and to do this the povide must be able to assess the popagation chaacteistics of the envionment. Diect measuement is one method of assessing the signal distibution and hence the GoS. But this is not always possible because of its intusive natue and its time consuming equiement. Ray tacing modelling with the dielectic popeties of the mateials in an envionment appeas to be the most ealistic way of assessing the signal distibution fo the coect location of a base station and estimating the GoS. Simulation is moe ealistic since a compact ay tacing algoithm was developed by fo indoo popagation [] and actual dielectic popeties of the envionment mateials measued and calibated fo a close match. The intoduction of the Dynamic Channel allocation (DCA) technique has opened up a new dimension in call outing. This in a sense has made the business of multisevice povides moe competitive because a fee channel is available fo evey povide to access on the fist in best seved pinciple. We have efined ou indoo popagating model to obtain signal distibution in a high ise building. The signal distibution is used in two simulation scenaios epesenting a multi-sevice multi-povide wieless envionment []: ) Multiple netwoks with multiple sevice povides within a single cell aea. ) Multiple netwoks with multiple sevice povides in a multi-floo building scenaio. The main aim of the model is to enable a sevice povide to detemine the location of base stations and also to quickly assess the GoS befoe embaking on the pocess of tendeing fo the sevice. Ou indoo popagation model togethe with ou netwok simulato MobileSim developed in C++ can be used to establish the pefomance unde diffeent loads with multiple sevice povides on the same floo and on multiple floos in a high ise envionment. II. MEASUREMENT OF THE DIELECTRIC PROPERTIES OF MATERIALS In ode to get accuate simulation esults, the dielectic popeties of the mateials of a specific envionment must be used. Fo this pupose we compose a data base of dielectic popeties of mateials which ae measued by us o available in the liteatue. Dielectic chaacteization is an essential step in the simulation of the popagation of electomagnetic waves in a wieless envionment. The dielectic constant and loss facto have a pofound influence on the pefomance of wieless systems. Numeous methods of measuing dielectic constant and loss facto have been developed to date. Some equie expensive equipment, such as an Automatic Netwok Analyze (ANA) and some ely on basic equipment, such as an impedance bidge o a slotted line. The open ended ISSN: x online, pint

2 coaxial pobe [3], a ecent development, has become popula because it is easy to use with an ANA. Its majo advantages ae in the speed and the boadband natue of measuement, enabling the investigation of tempeatue and time vaiation effects as well as any fequency esonance effects. Howeve, thee ae seveal dawbacks with this method. The sample must be homogeneous and optically flat to obtain good esults. The sampling size is vey small. It is not usually accuate fo low loss dielectic mateials. Thee is also the poblem with lift-off whee minute ai gaps between the pobe and the dielectic mateial can influence the accuacy of the measuement. We investigate a waveguide technique of dielectic measuement. It consists of an open waveguide teminated by a shot cicuit plunge. When the mateial is vey lossy, the shot cicuit becomes ineffective because signal neve gets to it. In this case we have an infinite line measuement. As shown in Fig., a shot cicuit at the open end of the loaded waveguide will detemine the adequacy of the infinite line assumption. Accoding to Altschule [4], the dielectic popeties ae then given by: i tan κ ( D DR ) ε + λ λ i tan κ g ( D DR ) c + + λ λ g c () Whee κ π/λg, λg guide wavelength, λc cut-off wavelength, VSWR measued by any impedance mete, κ ( DD R ) epesents the phase shift in the minimum position elative to the efeence shot cicuit and ε epesent the dielectic popeties. Fo medium and low loss mateials, it is not pactical to use the infinite line technique because the sample size could become unacceptably long; we use anothe method of measuement by using the shot cicuit to povide a coect length fo an open cicuit. With an ANA, the dielectic chaacteization can be simplified to the point of placing the shot cicuit ove the sample in the waveguide and then shifting it to a distance equal to one quate of the guide wavelength to ceate an open cicuit as shown in Table. TABLE I. OPEN CIRCUIT FROM A SHORT CIRCUIT Waveguide Equivalent o/c length (mm) WR WR WR Ou method uses only one pot of the netwok analyse and afte calibation; a shot cicuit then an open cicuit ae placed behind the sample to measue the coesponding admittances Y sc and Y oc. The advantage of this method is to use only one pot, thus educing calibation time and most impotantly, not involving solutions of a tanscendental equation. Figue. Waveguide sample holde It is accuate fo both low loss and high loss mateials as discussed by Altschule [4]. The dielectic popeties of a mateial ae given by the poducts of Y sc and Y oc. Yg YSC. YOC ZSC. ZOC (3) λ ε λc Yg YSC. YOC λ λc (4) Whee, Y sc admittance of the sample when it is teminated by a shot cicuit. Y oc admittance of the sample when it is teminated by an open cicuit. λ c cut off wavelength and λ is the fee space wavelength. We then measue the shot cicuit (Y sc ) and open cicuit (Y oc ) admittances. Whee, a width of the ectangula waveguide and Y ε G ε + i B ε. Let Y ε g + jb, ε' and ε'' can then be calculated fom equation 3 as: ( ε ) Ι λg Gε + a λg + a B λg + a ΙΙ ε ( ε ) TABLE II. TYPICAL RESULTS FOR MATERIALS Mateials Dielectic Popeties Concete.96 i.66 Wood.8 i.494 Nylon Glass 5.95 i.333 (5) (6) ISSN: x online, pint

3 Metal infinity Wet Bick i.3 Plaste Boad. - i. Polypopylene.3 - i. The pesent method is not boadband but it has seveal advantages ove the open ended and open shielded coaxial pobe in chaacteising the dielectic popeties of building mateials: ) The sample size is much bigge, which can account fo any inhomogeneity and povide a bette aveage chaacteization. ) The sample suface need not be vey flat as in the case of the open-ended pobe. 3) Method is simple to calibate and measue. 4) It is not expensive. III. ELECTROMAGNETIC PROPAGATION SIMULATION USING RAY TRACING METHODS The envionment used in ou simulation is a typical office (in a cente of the building thee is a povision of lift, stais and toilets and they suounded by compatments in each floos) in Hong Kong. The fequency is.8ghz, and the mateials in the envionment consist of concete floo, nylon capet, wood panel, glass windows and doos. Layout desciption is given in Fig.. Figue 3. Simulated 3D Signal distibution of a typical office The path loss pofile can be seen in Fig 3. It shows high signal loss in aeas whee thee is a highe density of building objects density. The meeting ooms and coido eflect lowe signal loss. Fig. 4 depicts output esult of signal stength veses distance in the typical office envionment at.8 GHz. As distance inceases signal stength educes. Fig. 5 shows that ou simulated powe distibution (continuous line) is in ageement with ideal Rayleigh (dotted line) fom - 5 db up to 35 db and beyond Gain (db), nomalized to m level Signal Stength Distance (m) Poly. (Signal Stength) Figue 4. Signal Stength vs. distance in the typical office (.8 GHz) Figue. Floo layout of a typical office in Hong Kong Pobabilty abscissa not exveeded, % powe, db (elative to aveage) Simulated powe distibution Ideal Rayleigh Figue 5. Simulated cumulative distibution function of powe in the typical office 3 ISSN: x online, pint

4 Fig. 6 shows simulation esult output between log distance and powe db efeence up to m. It is obseved that as powe inceases log distance inceases. Powe db ef. to m level Log distance Figue 6. Distance powe law Fig. 7 shows simulation esult output between T-R sepaation (m) and RMS time delay spead (ns). It can be seen fom Fig 7 that as distance inceases RMS time delay inceases. Ove 35 metes, RMS time delay inceases by 8 ns. ms time delay spead (ns) T-R sepaation (m) IV. Figue 7. RMS time delay with distance SIMULATION PARAMETERS AND PROPAGATION MODEL A. Multiple Sevice Povide Simulation in the Same Aea The simulato MobileSim fo multiple sevice povides on multiple floos ae used to investigate the pefomance of the tele-taffic system when multiple sevice povides shae the fequency spectum in the same aea. Spectum planning is not compulsoy in a DCA netwok. Fo a maximum of caie fequencies in opeation, input taffic paametes of 6 Elang and 7 Elang ae used to simulate the GoS of the teletaffic system. The numbe of sevice povides ae vaied between and 7 and ten caie fequencies ae available in the simulation. The simulation studies the effect of GoS with the change in taffic loads and the numbe of floos used. The esults ae plotted in Fig. 8, depicts the GoS of the system degading logaithmically as the numbe of sevice povides inceases. Unde a constant taffic load of 6 Elang, the wieless system can only suppot.% GoS with a single sevice povide. Inceasing the numbe of sevice povides will exceed.% GoS; call congestion and call dopouts incease. With two sevice povides allocated the same spectum, the GoS inceases to % and 6% fo a taffic load of 6 and 7 Elang espectively. GoS (%) Figue GoS vs No. of Opeatos No. of Sevice povides 7 elangs 6 elangs Pefomance in a multi-sevice povide envionment Using seven sevice povides simultaneously, the GoS degades to 7.8% and 74.3%. This study concludes that multiple deployments of sevice povides within the same aea will place a temendous stain on the wieless system esouces. Futhe spectum planning and fequency allocation is needed to esolve this poblem. B. Pefomance of Multiple Sevice Povides Ove Multiple Floos Unde this simulation scenaio, we investigate the pefomance of the DCA algoithm in a multi-floo application. The ETSI [] has limited the study to thee floos, but we extend it up to seven floos, as seen in Fig. 8 and 9, which is moe ealistic scenaio in an uban envionment. Two caie fequency caies ae used with the taffic at 6 Elang. The GoS is pesented in a scenaio with inceasing numbes of floos, with an independent sevice povide on each floo. A maximum of seven sevice povides ae used in the simulation. GoS (%) Figue GoS vs No. of floos No. of floos unsynchonised synchonised The pefomance of DCA (GoS) with up to seven sevice povides, one on each floo 4 ISSN: x online, pint

5 V. CONCLUSION We have used ou popagation model with ealistic dielectic popeties of the mateials in a specific envionment to obtain the popagation chaacteistics of the envionment as shown in Fig. to Fig. 7. We then use the data to input into ou simulato MobileSim to obtain the pefomance of a DCA in a scenaio fo multiple povides and multiple floos. In a multiple floo multiple sevice povide scenaios; DCA equies a minimum of 4 caies in ode to maintain a GoS of.95 at 5.6 Elang of taffic. Fo the GoS to be maintained egadless of the numbe of floos added, the numbe of caies has to be inceased by the numbe of floos occupied by a sevice povide. If the sevice is synchonised among the multiple sevice povides, it pefoms 5% to 5% bette. Including dielectic chaacteistics of the envionment in the simulations impoves the GoS of the system by a magin of.5% to 6.4%. The simulation is executed fo scenaios based on a single base-station and multiple base-stations on same floo and multi-floos. In a single base-station, no handove is equied. In multiple base-stations, when uses move fom cell to cell, handove occus and ae taken into account. Ou simulations include the dielectic popeties of mateials fo mateials in the envionment. In many cases they ae measued diectly. The simulation tool fo DCA deployment can be futhe extended to include othe wieless standads such as 3G as shaing of the spectum becomes unavoidable. REFERENCES [] ETSI Technical Repot Radio Equipment and Systems (RES); Digital Euopean Codless Telecommunications (DECT), A guide to DECT featues that influence the taffic capacity and the maintenance of high adio link tansmission quality, including the esults of simulations, ETR 4, July 99 [] A. Lee, V. N. Tan, The Pefomance of Wieless Systems using Dynamic Channel Allocation and Electomagnetic Popagation Simulation, In Poceedings of Intenational Confeence on Compute, Communications and Contols Technologies: CCCT 3, Olando, Floida, USA, vol. IV, pp. 9-96, 3 [3] A. Lee, V. N. Tan, Dielectic Chaacteisation of High Loss and Low Loss Mateials at 45 MHz, Ampee Confeence, UK, [4] K. Altschule, Handbook Of Micowave Measuements, Vol II, Dielectic constant, Suche and Fox Eds, Polytechnic Pess, ISSN: x online, pint