Application of Feature Selective Validation to Propagation Path Loss Models for Wireless Cellular Networks

Transcription

1 Applicatin f Feature Selective Validatin t Prpagatin Path ss Mdels fr Wireless Cellular Netwrks Ogundap Olusegun, Obrkhale awrence, Ogunleye Babatunde The Feature Selective Validatin (FSV) algrith is a technique that allws the quantitative cparisn and validatin f data.fsv can cpare large vlues f cplex data and als put the results in a cprehensible fr. In this paper, the FSV technique is extended t the cparisn and validatin f path lss del predictins fr wireless cellular netwrks. The path lss easureents btained fr a base statin in the urban area f Yla, Nigeria were cpared with predictins ade by the COST-231 Hata, ee and COST-231Walfisch-Ikegai dels using the FSV.The results shw a Glbal Difference Measure (GDM) f , and fr COST-231 Hata, ee and COST-231 Walfisch-Ikegai path lss dels respectively. This indicates that the ee Mdel gave a better predictin f the envirnent aking the FSV a useful tl fr fast quantitative cparisn and validatin f standardized path lss del predictins ver an envirnent. Keywrds- Feature Selective Validatin, wireless cellular netwrk, prpagatin path lss dels. Hwever, these dels were frulated based n extensive studies and bservatins in different envirnent. The effect f radi wave prpagatin ipairents varies fr ne envirnent t anther.[2] There is the need t have a fast and reliable eans f exaining the path lss predictins ver ther envirnents t iniize errrs in their usage. The Feature Selective Validatin (FSV) ethd f validating data will therefre be applied t easured and path lss predictins by the three dels under cnsideratin. The study f the path lss predictin behaviur aids effective netwrk planning and ptiizatin f radi resurces. The ai f this paper is t apply FSV t the data sets btained fr easureents and predictins by the COST-231 Hata, ee and COST-231 Walfisch-Ikegai del t deterine their suitability fr cverage predictin and planning in the area. The reaining part f the paper is rganized as fllws: sectin 2 gives the theretical basics needed fr the research wrk. Sectin 3 prvides the ethdlgies used t carry ut the research and the results btained. Sectin 4 discusses the results f the study. Sectin 5 cncludes the research. I INTRODUCTION II THEORETICA BASICS The Feature Selective Validatin (FSV) algrith has been develped t cpare tw sets f data and put the in an bjective fr. FSV allws the autated cparisn f large vlues f cplex data while reliably categrizing the results in a cn set f quality band. [1]Prpagatin dels are used extensively in netwrk planning, particularly fr cnducting feasibility studies and during initial deplyent. They are als very useful fr perfring interference studies as the deplyent prceeds and ptiizatin f radi resurces. Epirical and sei-epirical prpagatin dels have fund favur in bth research and industrial cunities wing t their liited reliance n detailed knwledge f the terrain. Manuscript received Deceber 10, 2012, revised February 25, All the authrs listed abve are with the Telecunicatins and Wireless Technlgies Departent, Schl f Infratin Technlgy and Cputing, Aerican University f Nigeria, Yla, Nigeria. Ogundap Olusegun (e-ail:lusegun.gundap@aun.edu.ng) Obrkhale awrence (e-ail:lawrence.brkhale@aun.edu.ng) Ogunleye Babatunde (e-ail:babatunde.gunleye@aun.edu.ng) The tw epirical prpagatin path lss dels t be used in this analysis are the COST-231 Hata and ee dels, while the sei-epirical prpagatin path lss del is that f the COST-231 Walfisch-Ikegai del. A. COST-231 Hata Mdel This is a ppular del fr predicting the path lss f bile wireless systes f nt re than 10k between the transitter and receiver. The del was first described by Okuura et.al. and Hata fr the predictin f path lss f land bile radi f nt re than 1500 MHz It was later dified by the COST-231 prject t include predictins f path lss up t 2000MHz and the prvisin f crrectin factrs fr urban, suburban and rural areas. The basic equatin fr path lss in db is: [2][3][4] P lg fc 13.82lghbs ah lg h lg d C (1) b h 3.2lg h a (2) Where, fc is the carrier frequency in MHz, d is the distance between the base statin and bile statin antennas in k. C is the area type crrectin factr.

2 B. ee Mdel This is anther widely used epirical path lss predictin del in bile wireless systes. It was first described using a base statin height f 30.4, carrier frequency f 900MHz, bile statin height f 3, axiu distance between transitter and receiver f 1.6k.Crrectin factrs were then prvided that enabled lnger distances and ther paraeters t be included fr path lss predictin. The set f equatins that define this path lss del are: [5] [6] d fc P lg 10nlg 1.6k 900MHz (3) Where, (4) 1 NewH BS (5) 2 NewH MS 3 3 (6) W 2 NewTransitterPwer 3 10W (7) NewBSG A 4 4 (8) New bile statin gain (9) 5 Where, H BS and H MS are the heights f base statin and bile statin respectively in eters, BSG A is the base statin antenna gain in dbi and is defined as 3 db fr fc >400MHz. d is the distance between the transitter and receiver in eters, f is the carrier frequency in MHz and is the crrectin factr. c C. COST-231 Walfisch-Ikegai Mdel. This is a sei-epirical path lss predictin del fr bile wireless systes f nt re than 5k between the transitter and receiver. The del cnsists f inputs fr publicatins ade by Walfisch et al [7] which prvided fr the ultiscreen diffractin lss and Ikegai et al. [8] that cnsidered an apprxiatin fr the rf tp t street diffractin lss. The del was later dified by the COST-231 prject t include crrectin factrs fr antenna heights. It can be used fr path lss predictin f bile wireless systes up t 2000MHz.The equatins that define this path lss del are: [2][3][8] P rts sd (10) Where is the path lss due t free space, rts is the rftp t street diffractin and scatter lss and sd is the ulti screen diffractin lss. rts f c d lg 20 lg (11) MHz K w f c lg 10 lg MHz hrf h lg ri d bsh K a K d k f c b K f 9 lg MHz 20 (12) sd lg Where, ri and is defined as: lg (13) is the path lss due t the rientatin angle deg 0 0 ri deg deg (14) d is the distance between the transitter and receiver in etres and f is the carrier frequency in MHz b, w and c h are the average buildings separatin, average width f street and height f bile statin respectively in etres. K 54 and K 18. a bsh h bs d 18 lg 1 (15) f c K f (16) 925 D. Feature Selective Validatin The Feature Selective Validatin technique was selected t cpare data fr easureent and predictin due t its ability t cpare, validate and put the in an bjective way. The FSV is therefre an algrith that has been develped t cpare tw sets f bidiensinal data and put the in an bjective and cprehensible fr. It is a sftware that can be fund at [10] The Feature Selective Validatin can be brken dwn int tw ajr cpnents, the Aplitude Difference Measure (ADM) and the Feature Difference Measure (FDM).The ADM and the FDM was then cbined t give the Glbal Difference Measure(GDM).The ADM is a easure f the verall agreeent f the general aplitude trend between the data sets. The FDM is a easure f the verall agreeent f the rapidly changing

3 features between the data sets. The GDM is an verall single figure gdness-f-fit between the tw data sets being cpared. The ADM, FDM and GDM are then cpared n a pint-by-pint basis t give the ADM i, FDMi and GDM i. This allws the user t analyze the resulting data in se detail, prbably with the ai f understanding the rigin f the cntributrs t pr cparisns. A lwer scre eans better agreeent. The ADM i, FDMi and GDM i can be used t create histgra f the nuber f pints in varius agreeent categries. These histgras are referred t as ADM C, FDMc and GDMc. The current agreeent categries are excellent, very gd, gd, fair, pr and extreely pr. The FSV interpretatin scale is shwn in Table 1. [9] Table 1: FSV Interpretatin Scale FSV Value (Quantitative) FSV Interpretatin (Qualitative) ess than 0.1 Excellent Between 0.1 and 0.2 Very gd Between 0.2 and 0.4 Gd Between 0.4 and 0.8 Fair Between 0.8 and 1.6 Pr Greater than 1.6 Very pr III IMPEMENTATION AND RESUTS The signal strength easureent was carried ut in the urban area f Yla, Nigeria. The area cnsists f buildings whse average height is abut seven flrs (24). The signal strength easureents were cllected thrugh drive tests with the aid f Ericssn Test Mbile Syste (TEMS) alng the pen rute f the wireless bile netwrk base statin. The TEMS was cnnected t a laptp with the aid f a Glbal Psitining Syste (GPS) fr tracking the distance between the base statin and the bile statin. The height f the receiver was abut 1.5.The Path lss fr each base statin was cputed using the COST-231 Hata, ee and COST-231 Walfisch- Ikegai Path lss dels. The FSV sftware was then used t cpare the easureents and predictins ade by the dels. The FSV cparisn f the three dels with easureent presented in histgra frs are shwn in Figures 1 t 9. The suary f the ADMc, FDMc and GDMc results are shwn in Table 2. Table 2: FSV Suary Values Ttal Hata ee Walfisch Values ADMc FDMc GDMc IV DISCUSSIONS OF RESUTS The plts in Figures 1 t 9 shws the ADMc, FDMc and GDMc results presented in histgra fr fr the cparisn between easured and COST-231 Hata, ee and COST-231 Walfisch-Ikegai dels using the FSV sftware. The GDMc results which is a cbinatin f the ADMc and the FDMc indicates that the ee del gave a better predictin f the envirnent as shwn in Table 2. This is crrbrated by the GDMc plt fr the cparisn between easureent and the ee del in Figure 6 which has st f the cparisn results within the excellent range than the COST-231 Hata and Walfisch-Ikegai dels. The suary f quantitative values fr the ADMc, FDMc and GDMc cparisns are presented in Table 2. V CONCUSION This paper presents the applicatin f the FSV tl t cpare easureent and path lss predictins using the COST-231 Hata, ee and COST-231Walfisch-Ikegai dels. The result shw that the ee del has a GDMc f which indicates a better predictin f the envirnent than the and prvided by the COST-231Hata and COST-231 Walfisch-Ikegai dels respectively. The FSV tl therefre prvided a faster quantitative cparisn and validatin f standardized prpagatin path lss del predictins fr the envirnent. Reference [1]. V.Rajaani et al, Intrductin t Feature Selective Validatin (FSV), in the prceeding f the Internatinal Sypsiu n Antennas and Prpagatin Sciety,pp ,2006. [2]. COST 231, Digital Mbile Radi Twards Future Generatin Systes, COST 231 Final Reprt, Ch 4, pp [3]. J.S Seybld, Intrductin t Radi Frequency Prpagatin, published by Jhn Wiley and Sns, pp [4]. J.D Parsns, The Mbile Radi Prpagatin Channel,2nd ed.,published by Jhn Wiley, West Sussex, 2000,pp [5]. N.Blaunstein, Radi Prpagatin in Cellular Netwrks, Artech Huse Nrwd, MA, 2000, pp [6]. W.C.Y ee, Mbile Cunicatin Design Fundaentals, 2nd ed., Jhn Wiley, New Yrk, 1993, pp [7]. J. Walfisch et al. A Theretical Mdel f UHF Prpagatin in Urban Envirnents, IEEE Transactin n Antenna and Prpagatin, Vl. 36, N. 12, pp , [8]. F. Ikegai et al Prpagatin Factrs cntrlling Mean Field Strength n Urban street IEEE Transactins n Antenna Prpagatin, Vl. 32, pp , [9]. Duffy, A. Martin, et al. The Feature Selective Validatin (FSV) Methd, in prceedings f the IEEE Sypsiu n Electragnetic Cpatibility, Chicag, USA, Vl.1, pp , [10]. FSV_Tl_2_0 Tl:

4 Figure 1: ADMc Results fr Measured and Hata Mdel Cparisn Figure 2: FDMc Results fr Measured and Hata Mdel Cparisn Figure 3: GDMc Results fr Measured and Hata Mdel

5 Figure 4: ADMc Results fr Measured and ee Mdel Cparisn Figure 5: FDMc Results fr Measured and ee Mdel Cparisn Figure 6: GDMc Results fr Measured and ee Mdel Cparisn

6 Figure 7: ADMc Results fr Measured and Walfisch Mdel Cparisn Figure 8: FDMc Results fr Measured and Walfisch Mdel Cparisn Figure 9: GDMc Results fr Measured and Walfisch Mdel Cparisn