Jurnal Kejuruteraan 33(2) 2021: 317-328 https://doi.org/10.17576/jkukm-2021-33(2)-16
Accurate Path-Loss Estimation for Wireless Cellular Networks
Ibrahim M M Mohamed Electrical and Electronic Department, Faculty of Engineering, Omar Al-Mukhtar University, Al-Beida, Libya
*Corresponding author: [email protected]
Received 27 April 2020, Received in revised form 19 September 2020 Accepted 10 October 2020, Available online 30 May 2021
ABSTRACT
In addition to its rule in achieving acceptable system performance, accurate path-loss estimation leads to realize a simple and precise estimation of cellular networks financial feasibility. In other words, in cellular networks, while accurate path-loss estimation helps to achieve a reasonable cost, inaccurate path-loss estimation could either lead to reduce the cost but degrade the performance, or improve the performance but increase the cost. Thus accurate path loss estimation becomes a very crucial and desirable goal. To this end, different models were introduced in the literature for achieving the aforementioned goal; however, each model sought to be a valid choice in a specific environment. Therefore, our main objective in this paper was to provide an insight to the cellular network designers to simplify the decision-making process. A number of widely used path-loss estimation models that work in different environments, such as Free-space model,T wo-ray model, Okumura/Hata model, COST 231 model, and Indoor model were reviewed and analyzed. This significantly would assist any cellular network designer to select an appropriate model according to its working environment, and thus realize the desirable accurate path-loss estimation. Matlab was used to perform this analysis. The analysis drove us to enter into implicit comparisons between the aforementioned models. Going through these comparisons, it is concluded that although an increase in the level of complexity is encountered when using models with large number of correction factors, more accurate path-loss estimation can be achieved.
Keywords: Cellular networks (CNs); Path-loss estimation models; Free space propagation model; Two-ray model; Okumura/Hata model; Cost 231 model; Indoor model
INTRODUCTION unwanted but received powers produced by the co-channel cells. For acceptable performance, a cellular system should One of the main objectives of cellular systems is to achieve meet a specific value of SNR. I.e. the SNR required for high capacity (i.e., increase the number of user). This acceptable performance may differ from system to another. objective can be realized by creating a distinctive way For example, approximately, a minimum SNR value of 18 that guarantees exploiting the limited frequency spectrum dB or 12 dB is specified for acceptable performance when assigned to the cellular systems efficiently. The concept of advanced mobile phone system AMPS or global system for frequency re-use in which a segment of specific frequency mobile communication GSM is considered, respectively spectrum can be used several times is the key to achieve (Goldsmith 2005). In cellular systems, a wireless channel an efficient exploit of the assigned spectrum. In frequency is mainly characterized by its effects of dispersion and re-use theory, each cell within a single cluster represents a attenuation. While dispersion is analyzed to verify which replica of each of six adjacent and predetermined cells (co- sort of signal distortion encountered during propagation, channel cells) surrounding it. Each co-channel cell belongs attenuation is analyzed to estimate the path-loss. In a wireless to a neighboring cluster. Although a low-power transmitter channel, a signal might be attenuated due to different is commonly employed in each cell, a sufficient distance phenomena, such as reflection, diffraction, and scattering. should be determined for separating the co-channel cells Reflection occurs when a signal that is propagating ina to guarantee minimum interference. Although interference medium passes to another medium with different properties. can also come from the second, third, and higher tires co- Attenuation in this case can be attributed to the fact that channel cells, such interference can be ignored as it often a part of the incident signal energy is either absorbed contributes by less than one percent (1 %) of the total or propagated into the reflecting medium. Diffraction is interference (Jochen 2003). As a type of communication defined as the deviation of a signal from its path. It occurs system, the performance of a cellular system is indicated by when a signal that is propagating in a medium passes into its signal-to-noise ratio SNR. While the term signal refers to a shadow region created by an obstruction. Scattering is the power of an intended carrier, the term noise represents defined as the spreading of energy out of its intended path. the power produced at the receiver due to thermal effect plus It occurs when a signal hits an object whose size is much 318 Jurnal Kejuruteraan 33(2) 2021: xxx-xxx smaller than or on the order of the signalhtt wavelength.ps://doi.org/10.17576/jkukm While respectively.-2021 Pico-cells-33(2)-16 (cells that cover part of a building accurate path-loss estimation helps to realize an acceptable and mainly span from 30 to 100 m) can be included in the system performance and leads to achieve reasonable cost, short-distance prediction models. and scattering. Reflection occurs when a signal that They were initially developed based on the inaccurateis propagatingpath-loss estimation in a medium would either passes lead to to anotherdegrade Indoorsimplest prediction line-of-sight models model; were also introduced referred toto asestimate free the performancemedium with or increase different the properties. cost (Andrea Attenuation Goldsmith, in the path-lossspace propagationin this case. Figuremodel. 2 providesLater, a theydiagrammatic were 2005). thisFigure case 1 (a),can (b),be attributed and (c) clarify to the the fact aforementioned that a part of categorizationdeveloped of based the path-loss on more models. complicated non-line-of Jurnal Kejuruteraan 33(2) 2021: xxx-xxx situationsthe inincident which the signal estimated energy path-loss is either is higher absorbed than theor sight path-loss models. In free space propagation https://doi.org/10.17576/jkukm-2021-33(2)-16 actual path-losspropagated in (a),into almost the reflecting equal to themedium. actual Diffractionpath-loss in model, no obstruct ions due to earth surface or other PREVIOUS WORK AND CONTRIBUTION (b), or loweris defined than asthe the actual deviation path-loss of a in signal (c). from its path. obstacles are encountered during propagation. andDifferentIt scattering.occurs modelswhen Reflection awere signal introduced occurs that whenis inpropagating thea signal literature that in fora TheyUnlike were free initially space modeldeveloped in which based frequency on the and achievingis mediumpropagating accurate passes path-loss in into a mediuma estimation shadow passesregion (Yoo-Seung createdto another Song by an & Manysimplestseparation research line- ofdistancestudies,-sight model; (distancerelated also to betweenreferred the path-loss totransmitter as free were Hyun-Kyunmeobstruction.dium Choi.with 2017; differentScattering Divya properties. is Kurup defined et Attenuational.as 2011;the spreading Haipeng in introducedspaceand receiver)propagation in the literature.are themodel. only Zaarour Later,contributors et theyall, forproposedwere path - a et al.this 2010;of case energy Mario can out beVersaci ofattributed its etintended al. to2012; the path. A.fact BhuvaneshwariIt that occurs a part when of eta methoddevelopedloss, to estimatefurther based factorsonthe morepath-loss suchcomplicated exponentas antennas non PLE-line height,based-of on al. 2016;thesignal incidentYun-Jie hits Xu signalan & object Wen-Bin energy whose Li.is 2011).eithersize is absorbedmuch smaller or thesight connectivitybuildings path-loss height, between models. and sensors Instreets free deployed widthspace arepropagation in included a determined in propagatedTheytha nwere or onintoinitially the the orderreflectingdeveloped of themedium. based signal on Diffraction wavelength.the simplest areamodel, (Nourthe nonon Zaarour obstruct-line-of etions al.sight 2016).due modelto earthVaishnavis, surfacewhich et allorleads otherpresented to line-of-sightis definedWhile accurate model;as the deviation pathalso- lossreferred of estimation a signal to from helpsas freeits to path. realizespace a practicalobstaclesachieve method aremore encountered accurate for path path lossduring-loss exponent estimation. propagation. calculation Path - propagationIt anoccurs acceptable model. when aLater, systemsignal they thatperformance wereis propagating developed and leads in based a to usingUnlikeloss the Receivedmodelsfree space can Signal modelbe Strengthcategorized in which Indicator accordingfrequency RSSI toand method the medium passes into a shadow region created by an separation distance (distance between transmitter on moreachieve complicated reasonable non-line-of cost, sightinaccurate path-loss path models.-loss whichseparation led to a minimaldistance error as in theeither distance long -estimation.distance obstruction. Scattering is defined as the spreading and receiver) are the only contributors for path- In freeestimation space propagation would eithermodel, leadno obstructions to degrade due the to The predictionmethod was models modeled or mathematicallyshort-distance predictionusing log- of energy out of its intended path. It occurs when a loss, further factors such as antennas height, earth surfaceperformance or other or obstaclesincrease are theencountered cost (Andrea during distancemodels. path lossThey model are (Vaishnaviintended for et al.cellular 2018). systems Minthorn signal hits an object whose size is much smaller buildings height, and streets width are included in propagation.Goldsmith Unlike, 2005) free. spaceFigure model 1 (a), in(b ),which and (frequencyc) clarify et allemploying presented a methodmacro -tocells, estimate and the wirelessmicro-cells, device thathen or aforementionedon the order ofsituations the signal inwavelength. which the therespectively. non-line-of Picosight-cells model (cellss, thatwhich cover leads part toof a and Whileseparation accurate distance path -(distanceloss estimation between helps transmitter to realize and locationachieve in more indoor accurate 2D environmentpath-loss estimation. using the Path widely- receiver)estimated are the pathonly-loss contributors is higher thanfor path-loss,the actual further path- used buildinglog-distance and mainlypath loss span model from 30and to actual100 m ) Receivedcan be an lossacceptable in (a), almostsystem equal performance to the actual and pathleads-loss to in lossincluded models in can the shobe rtcategorized-distance prediction according models. to the factors such as antennas height, buildings height, and streets Signal Strength Indicator (RSSI) measurement. The achieve(b), or reasonablelower than cost,the actualinaccurate path -losspath -inloss ( c). separation distance as either long-distance width are included in the non-line-of sight models, which method showed an average difference of 7.42% from the estimation would either lead to degrade the prediction models or short-distance prediction leads to achieve more accurate path-loss estimation. Path- measurement, which is considered acceptable for location performance or increase the cost (Andrea models. They are intended for cellular systems loss Goldsmithmodels can, 2005)be categorized. Figure 1 according (a), (b), and to the(c) separationclarify estimationemploying in actualmacro environment-cells, and(Minthorn micro et- cells,al. 2018). distancethe asaforementioned either long-distance situations prediction in which models the or Agarwalrespectively. and JagannathamPico-cells (cells developed that cover a parthomogenous of a short-distanceestimated pathprediction-loss is models.higher thanThey the are actual intended path - for Poissonbuilding point and process mainly based span fromframework 30 to 100 to characterize m) can be the cellularloss systemsin (a), almostemploying equal macro-cells, to the actual and path micro-cells,-loss in performanceincluded in ofthe distributed short-distance estimation prediction in models.wireless sensor (b), or lower than the actual path-loss in (c). Excellent performance but Acceptable performance higher system cost c and reasonable cost estimation Cell size is smaller than required Lower system cost but b degraded performance Appropriate cell size a Estimated path loss Actual path loss Cell size is larger than required Excellent performance but FIGURE 1. EffectAcceptable of Accurate performance and Inaccurate Pathhigher-loss systemEstimation cost c and reasonable cost estimationb Cell size is smaller than required Different modelsLower were system introduced cost but in the literature degraded performance Appropriate cell size for achieving accurate apath -loss estimation (Yoo- Indoor prediction Estimatedmodels pathwere loss introduced to Actual path loss Seung SongCell size& isHyun larger-Kyun than Choi.required 2017 ; Divya estimate the path-loss in this case. Figure 2 Kurup et al. 2011; Haipeng et al. 2010; Mario provides a diagrammatic categorization of the path-
Versaci et al. 2012; A. BhuvaneshwariFIGUREFIGURE 1. Effect 1. Effect et of al. Accurateof 2016; Accurate and andInaccurate Inaccurateloss models.Path-loss Path - lossEstimation Estimation Yun-Jie Xu & Wen-Bin Li. 2011).
Different models were introduced in the literature Free space model for achieving accurate path-loss estimation (YooLong--distance predictionIndoor prediction models were introduced to models Reflecting by flat surface Seung Song & Hyun-Kyun Choi. 2017; Divya estimate themodel path(tow-ray-loss model) in this case. Figure 2 provides a diagrammatic categorization of the path- Kurup et al. 2011; Haipeng Widely used et Pathal. -loss2010; Mario Versaci et al. 2012; A. BhuvaneshwariModels et al. 2016; loss models. Okumura/Hata model Yun-Jie Xu & Wen-Bin Li. 2011).
Short-distance prediction FreeCOST231 space model model Long -distancemodels prediction
models Reflecting by flat surface model Indoor (tow prediction-ray model) model Widely used Path-loss FIGURE 2. Widely used Path -loss Models Models FIGURE 2. Widely used Path-loss Models Okumura/Hata model
Short-distance prediction COST231 model models
Indoor prediction model
FIGURE 2. Widely used Path -loss Models
319 networks using a variety of scenarios in which a fading where c and f represent the speed of light (3x108 m/sec) wireless channel was considered between the sensors and and frequency in hertz Hz, respectively. Substituting d in the Fusion Center (FC) (Abhishek et al. 2014). This paper kilometer and f in megahertz, free space path-loss can be provides an insight to the cellular network designers to expressed in dB as simplify the decision-making process. A number of widely used path-loss estimation models that work in different L =++32.44 20Logf 20 Logd environments, such as Free-space model, Two-ray model, Pfree[] dB (3) Okumura/Hata model, COST 231 model, and Indoor model were reviewed and analyzed. This significantly assists any It can be observed from (3) that duplication of frequency cellular network designer to select an appropriate model or distance would lead an increase in the path-loss by 6 dB. according to its working environment, and thus realize the In the two-ray prediction model, multi-path propagation desirable accurate path-loss estimation. scenario is assumed in which a number of two propagation paths (direct propagation path and ground reflected propagation path) are considered. A part of the signal power THEORETICAL BACKGROUND is assumed to be received directly over the direct path (i.e., no obstructions between transmitter and receiver) whereas In cellular networks, accurate path-loss estimation helps the rest is assumed to be received after being reflected by to precisely determine the number of cell sites required the ground surface). In this model, assumptions related for providing coverage in a given area. Accurate path-loss to transmitter and receiver antenna heights (base station estimation is highly recommended in the initial stages of height hb and mobile station height hm), the angle between cellular network design to precisely determine the number direct and reflected paths (Δα), and reflecting surface are of cell sites required for providing coverage in a given area, of significance for its validation. In theses assumptions, hb which leads to achieve a simple estimation of financial and hm are assumed to be much smaller than the separation feasibility. Additionally, accurate path-loss estimation distance d, Δα is assumed to be less than π/8, and reflecting leads to realize an acceptable system performance. surface is assumed to be flat. The model is valid for any Various path-loss estimation models were introduced in distance d greater than d̄ which is referred to as the critical the literature; each of which is sought to be a valid choice 4h h distance and given as d = b m . Figure 3 shows a schematic in a specific environment and aiming to achieve accurate λ path-loss estimation. In this section, a review of theoretical diagram of the two-ray model. The path-loss in this case is background of few but widely used path-loss estimation referred to as the two-ray path-loss (LP2-ray). models is provided. Two-ray path-loss is given as (Vijay K. Garg, 2007) 2 d 2 L = which can be expressed in dB as LONG-DISTANCE PREDICTION MODELS P2−ray hbhm
Free space model and two-ray model are the two widely 2 d used long-distance prediction models. In free space model, LdP2− ray [] dB = 20log= 40log − a line-of-sight propagation is assumed. i.e., no obstructions hhbm (4) exist between transmitter and receiver albeit by the surface 20log(hhbm ) of earth. It is mostly applicable for wireless channels in which a transmission system such as microwave or satellite- to-satellite is employed. The path-loss in this case is referred SHORT-DISTANCE PREDICTION MODELS to as free space path-loss (LPfree). Free space path-loss is given as (Vijay K. Garg, 2007). In this section, theoretical backgrounds related to the widely used short-distance prediction models, i.e., Okumura/Hata 2 model, COST 231 model, and Indoor model are presented. 4πd = (1) While Okumura/Hata model and COST 231 model are LPfree λ recommended for path-loss estimation when micro-cells are used (separation distance might reach up to 50 km), Indoor where d and λ represent the separation distance (i.e., distance model is the proper choice for path-loss estimation when between the transmitter and receiver) and wavelength, pico-cells are used (separation distance might reach up to respectively. Given λ = c/f yields few hundred meters, i.e. probably inside a building).
OKUMURA/HATA MODEL 2 4πfd = (2) The Okumura/ Hata model was primarily developed for LPfree c path-loss estimation in typical flat urban environment. Then it was improved to include path-loss estimation in the typical suburban and rural environments. Jurnal Kejuruteraan 33(2) 2021: xxx-xxx https://doi.org/10.17576/jkukm-2021-33(2)-16
urban environment. Then it was improved to 4hbhm given as d = . Figure 3 shows a schematic include path-loss estimation in the typical suburban and rural environments. diagram of the two-ray model. The path-loss in this 320 case is referred to as the two-ray path-loss ( LP2−ray ).
FIGURE 3. Two-ray Path-loss Model FIGURE 3. Two-ray Path-loss Model
A correction factor that is related to the mobile The path-loss in dB according to Okumura/Hata model for Two-ray path-loss is given as (Vijay K. Garg, A correction factor that is related to the mobile station2007 antenna) height was included in different forms thestation typical antennasuburban height environment was included is given asin different according to the environment2 considered and frequency 2 forms according to the environment considered and carrier used to achieved accurate path-loss estimation. frequency carrier used to achieve accurate path-loss L = which can be expressed in dB L50 () suburban []dB = The path-lossP2−ray in dB according to Okumura/ Hata model estimation. The path-loss in dB according to hbhm 2 for typical urban environment is given as (Garg 2007; Okumura/ Hata model fcfor typical urban (7) −− Kazunorias et al. 2014) environmentL50 ( urban is )given 2 logas (Garg 2007 5.4; Kazunori et 28 d 2 al. 2014) LP2−ray[dB] = 20log = 40logd − L50 ( urban )[]dB =++ 69.55hbhm 26.16log fc TheL path-loss50 (urban in) [dB] according= 69.55 + to26 Okumura/Hata.16log fc + model for −− 20(44.9log(h 6.55logh ) hd )log 13.82log h (4) (5) the rural environment is given as b m (44.9 − 6.55log h)log d −13.82log h −ah()m − a(hm ) (5) L50 () rural []dB = Where fc is the carrier frequency in MHz, d is the L( urban )−+ 4.78(log fc )2 (8) separation distanceSHORT-DISTANCE between PREDICTIONthe mobile MODELSstation and base 50 station in Km, h and h are the base station and mobile 18.33logfc − 40.94 In this section,b theoreticalm backgrounds related to station antennas height in meter, and a(h ) is correction Where fc is the carrier frequency in MHz, d is the widely used short-distance predictionm models, factor that is related to the mobile station antenna height. i.e., Okumura/Hata model, COST 231 model, and theTable separation 1 provides distance the betweenrange of theparameters mobile station for which a(hm)is given in different forms according to the size of the Okumura/Hata model is valid (Vijay K. Garg, 2007). Indoor model are presented. While Okumura/Hata and base station in Km, hb and hm are the base intendedmodel area. and COST 231 model are recommended for For large cities and at fc ≤ 200 MHz, a(h ) is given as station and mobile station antennas height in meter, path-loss estimation when micro-mcells are used COST 231 MODEL and a(h ) is correction factor that is related to the (separation distance might reach up to 50 km), m 2 The COST 231 model was developed based on empirical Indoorah (model )= 8.29[log(1.54 is the properh )]choice− 11 for path-loss(6a) mobile station antenna height. a(hm ) is given in mm and deterministic approach for path-loss estimation in urban estimation when pico-cells are used (separation different forms according to the size of the intended and suburban areas (L. M. Correia, 2009). It is resulted from distance might reach up to few hundred meters, i.e. area. For largeprobably cities inside and at a fcbuilding). ≤ 400 MHz, a(hm) is given as the combination of Walfisch-Bertoni and Ikegami models (J.WalfischFor large cities& H. andL. Bertoni. at fc 1988;200 F.MHz, Ikegami a (eth mal.) is1991). OKUMURA/HATA MODEL As givencompared as with Okumura/Hata model, a number of ah( )= 3.2[log(11.75h )]2 − 4.97 (6b) mm additional correction factors, such2 as buildings’ height and a(hm ) = 8.29[log(1.54hm )] −11 (6a) The Okumura/ Hata model was primarily separation, streets’ width, and orientation angle (the angle developed for path-loss estimation in typical flat For small and medium-sized cities, a(hm) is given as between the direction of propagation and street axis) have been added to the COST 231 model to achieve more accurate path-loss estimation. Figure 4 shows a schematic diagram a( h )= [1.1log(fc ) −− 0.7] h mm(6c) of the COST 231 model. The path-loss in dB according to [1.56log(fc )− 0.8] COST 231 model is given as (Vijay K. Garg, 2007; L. M. Correia, 2009) 321
TABLE 1. The range of parameters for which Okumura/Hata model is valid Minimum value Maximum value Carrier frequency fc (MHz) 150 2200
Base station height hb (m) 30 200
Mobile station height hm (m) 1 10 Separation distance d (Km) 1 20
correction factor that is related to the size of the city. L L=++ LL L bsh 50[dB ] f rts ms (9) in dB can be given in different forms according to the base station height and average building height as
In the case where Lrts + Lms ≤ 0, the path-loss in dB according to COST 231 model is given as Lbsh[] dB =−18 Log (11 +∆ hb ) for hb≥ h r (14a)
LL50[dB ] = f (10) Lbsh[] dB = 0 for hhbr< (14b)
Where, Lf is the free space loss, Lrts is referred to as the roof- Δhb is referred to as the difference in meter between the top-to-street diffraction and scatter, and Lms is referred to as the multi-screen loss. Free space loss in dB can be expressed base station height hb and the average building height hr and as in (3). The roof-top-to-street diffraction and scatter loss given as (Δhb = hb - hr). ka in dB can be given in different in dB is given as forms according to the base station height, average building height, and separation distance as
Lrts[] dB =−−16.9 10logW + 10log fc + (11) ka[] dB = 54 for hhb> r (15a) 20log ∆+hLmo
k =−≥≤54 0.8h for d50 0m and hh W is the street width in meter, Δhm is referred to as the a[] dB b b r (15b) difference between the average building height hr and the mobile antenna height hm and given in meter as (Δhm = hr ka[] dB =−∆54 0.8hb ( d / 500) for d <500m and hb≤ h r - hm), and Lo is a correction factor, related to the orientation angle ϕ. L in dB can be given in different forms according (15c) o =−∆ < ≤ to the orientation angle as ka[] dB 54 0.8hb ( d / 500) for d 500m and hb h r
o k is given in dB can be given in different forms according to L =−10 + 0.354 ϕ for 0 ≤ϕ≤ 35 (12a) d o[] dB the base station height and average building height as
L =2.5 + 0.075( ϕ− 35) for 35oo≤ϕ≤ 55 o[] dB (12b) kd[] dB = 18 for hhb< r (16a)
oo Lo[] dB =4 − 0.114( ϕ− 55) for 55≤ϕ≤ 90 (12c) 15∆hb kd[] dB =18 −≥for hhb r (16b) ∆hm The multi-screen loss in dB is given as
kf in dB can be given in different forms according to the city Lms[] dB = size. For mid-sized cities, it is given as (13) Lbsh++ k a k d Logd + kf Logfc −9 Logb fc =+− k f[] dB 4 0.7 1 (17a) where b, and d represent the distance between buildings 925 along the radio path in meter, and separation between transmitter and receiver in Km, respectively. where as for metropolitan areas, it is given as
Lbsh and kd are correction factors that are related to the base station height and average building height. k is a a fc correction factor that is related to the base station height, =+− k f[] dB 4 1.5 1 (17b) 925 average building height, and separation distance. kf is a Jurnal Kejuruteraan 33(2) 2021: xxx-xxx https://doi.org/10.17576/jkukm-2021-33(2)-16
where b , and d represent the distance between ka[dB] = 54 − 0.8hb (d /500) for buildings along the radio path in meter, and separation between transmitter and receiver in Km, d 500 m and hb hr (15c) respectively. kd is given in dB can be given in different forms Lbsh and kd are correction factors that are related according to the base station height and average to the base station height and average building building height as
height. ka is a correction factor that is related to kd [dB] = 18 for the base station height, average building height, hb hr (16a) and separation distance. k f is a correction factor 15hb that is related to the size of the city. Lbsh in dB can kd =18− for [dB] h 322 be given in different forms according to the base m
station height and average building height as hb hr (16b)
FIGURE 4. COST 231 P ath-loss Model FIGURE 4. COST 231 Path-loss Model
Tables 2 and 3 provide the range of parameters for which the speed up and simplify their decisions to use an appropriate = − + COSTL bsh231[dB model] 18is validLog( and11 somehb )default values thatfor can path-loss model according to their work environment. This k in dB can be given in different forms according be used in the COST 231 model (Garg 2007). mightf involve us dealing with some complicated models hb hr (14a) thatto employthe city sizeseveral. For correctionmid-sized cities,factors it whenis given calculating as = Lbsh[dB] 0INDOOR PREDICTION MODEL for the path-loss. To avoid fcthis complexity in calculation and k = 4 + 0.7 −1 (17a) save thef [dB time,] the method used in this paper was based on hb
PATH-LOSS ESTIMATIONS FOR LONG-DISTANCE METHOD USED PREDICTION MODELS
As mentioned above, our goal is to provide a technical guide In this part of results, the path-loss is estimated by that would significantly help the cellular network designers considering equations (3) and (5). Figure 6 shows the 323
TABLE 2. The range of parameters for which The COST 231 model is valid Minimum value Maximum value Carrier frequency fc in MHz 800 2000
Base station height hb in meter 4 50
Mobile station height hm in meter 1 3 Separating distance d in kilometer 0.02 5
TABLE 3. Some default values that can be used in the COST 231 model Distance between building along radio path b 20 – 50 (m) Street width W b/2 (m) Incident angle ϕrelative to the street 90o Roof 3 m for pitched roof and 0 m for flat roof Average building height hr 3(number of floors) + roof
TABLE 4. Values of indoor environments’ parameters Indoor environment Residential Office Commercial
LdP ()o[] dB 38 38 38
γ 2.8 3 2.2
Lnf()[] dB 4n 15+4(n-1) 6+3(n-1)
X σ[]dB 8 10 10 graphical representation of (3). It represents the path-loss PATH-LOSS ESTIMATION USING OKUMURA/HATA MODEL versus separation distance at different carrier frequencies, In this part of results, the path-loss is estimated using when free space model is used. equations (5) to (8). The path-loss versus separation distance It is obviously seen from the graph that the path-loss (1 Km to 20 Km) at different significant parameters, such starts to appear at a certain value of separation distance and as carrier frequency (900, 1800, and 2100 MHz), and base continue to increase rapidly with a slight increase in the station height (20, 80, 140, and 200 m) were graphically separation distance. However, it starts to increase steadily at represented. Figure 8 shows the graphical representation of higher values of separation distance. It is observed from the equations (5), (6a) and (6b). graph that the smallest value of path-loss is achieved at the It represents the path-loss versus separation distance lowest carrier frequency. Figure 7 shows the graphical representation of (4). at different carrier frequencies, when the typical urban It represents the path-loss versus separation distance at environment Okumura/Hata model for large cities is used. different base station heights, when the two-ray model is It is obviously seen from the graph that the path-loss starts used. It is obviously seen from the graph that the path-loss to appear at a certain value of separation distance in which starts to appear at a certain value of separating distance the model is valid (1 Km) and continue to increase with the in each case (the_ distance where the two-ray model is increase in the separation distance. Figure 9 (a, b and c) applicable [d>>d]) and then continues to increase with the shows the path-loss versus separation distance at different increase of the separation distance. However, an increase in values of base station heights and carrier frequencies when the base station height leads to reduce the path-loss. typical urban environment Okumura/Hata model for large cities is used. The graph follows the same pattern as in PATH-LOSS ESTIMATIONS FOR SHORT-DISTANCE Figure 8; however, an increase in the base station height PREDICTION MODELS leads to reduce the path-loss.
This part of results is divided into three sections. In the PATH-LOSS ESTIMATION USING COST 231 MODEL first section, path-loss is estimated using Okumura/Hata model, whereas in the second and third sections, path-loss In this part of results, the path-loss is estimated using is estimated using COST 231 model and Indoor model, equations (9) to (17b). We first provide graphical respectively. representations similar to those provided in the previous 324
FIGURE 5. Sequence of Program Operation where Typical Urban Environment Okumura/Hata Model for Large Cities is considered
FIGURE 6. Path-loss vs Separating Distance at Different Carrier FIGURE 7. Path-loss vs Separating Distance at Different Base Frequencies: Free Space Model was Considered Station Heights: Two-ray Model Path-loss Model was Considered sections, i.e., starting by draw the path-loss versus the path-loss versus separation distance at different carrier separation distance (1 to 50 Km) at different carrier frequencies. frequencies (900, 1800, and 2100 MHz), and different base The graph follows the same pattern as in Figures station heights (30, 40, and 50 m). Then, we provide extra 8 and 9; however, it shows higher path-loss values. For graphical representations including further parameters that example, a path-loss value of 183.1184 dB is estimated at might affect the path-loss, such as the building heights (30, 20 Km separation and 2100 MHz, when using typical urban 60, 90, 120, and 150 m), road width (10, 15, 20, 25 m), environment Okumura/Hata model for large cities, whereas and orientation angle (0o, 45o, and 90o). Figure 10 shows a higher path-loss value (195.7259 dB) is estimated at the Jurnal Kejuruteraan 33(2) 2021: xxx-xxx https://doi.org/10.17576/jkukm-2021-33(2)-16
third sections, path-loss is estimated using COST PATH-LOSS ESTIMATION USING COST 231 MODEL 231 model and Indoor model, respectively. In this part of results, the path-loss is estimated PATH-LOSS ESTIMATION USING OKUMURA/HATA using equations (9) to (17b). We first provide MODEL graphical representations similar to those provided in the previous sections, i.e., starting by draw the In this part of results, the path-loss is estimated path-loss versus separation distance (1 to 50 Km) at using equations (5) to (8). The path-loss versus different carrier frequencies (900, 1800, and 2100 separation distance (1 Km to 20 Km) at different MHz), and different base station heights (30, 40, significant parameters, such as carrier frequency and 50 m). Then, we provide extra graphical (900, 1800, and 2100 MHz), and base station height representations including further parameters that (20, 80, 140, and 200 m) were graphically might affect the path-loss, such as the building represented. Figure 8 shows the graphical heights (30, 60, 90, 120, and 150 m), road width representation of equations (5), (6a) and (6b). (10, 15, 20, 25 m), and orientation angle (0o, 45o, and 90o). Figure 10 shows the path-loss versus separation distance at different carrier frequencies.
FIGURE 8. Path-loss vs Separating Distance at Different Carrier Frequencies: Typical Urban Environment 325 Okumura/Hatta Model for Large Cities is Considered FIGURE 10. Path-loss vs Separating Distance at It represents the path-loss versus separation Different Carrier Frequencies: COST 231Model is distance at different carrier frequencies, when the Considered typical urban environment Okumura/Hata model The graph follows the same pattern as in Figures 8 for large cities is used. It is obviously seen from the and 9; however, it shows higher path-loss values. graph that the path-loss starts to appear at a certain For example, a path-loss value of 183.1184 dB is value of separation distance in which the model is estimated at 20 Km separation and 2100 MHz, valid (1 Km) and continue to increase with the when using typical urban environment increase in the separation distance. Figure 9 (a, b Okumura/Hata model for large cities, whereas a and c) shows the path-loss versus separation higher path-loss value (195.7259 dB) is estimated distance at different values of base station heights at the same separation distance and frequency and carrier frequencies when typical urban carrier when using COST 231 model. environment Okumura/Hata model for large cities is used.FIGURE The graph 8. Path-loss follows vs the Separating same pattern Distance as at inDifferent Carrier Frequencies:Typical Urban Environment Figure 8; however, an increaseOkumura/Hatta in the base station Model for Large Cities is Considered height leads to reduce the path-loss.
Carrier Carrier Carrier frequency frequency = frequency = = 900 MHz 1800 MHz 2100 MHz
a b c
FIGUREFIGURE 9. Path 9.- lossPath-loss vs Separating vs Separating Distance Distance at Different at Different Base StationBase Station Heights: Heights: Okumura/Hatta Okumura/Hatta model for modelTypical for Typical Urban Urban Environment Environment was C wasonsidered. Considered.
FIGURE 10. Path-loss vs Separating Distance at Different Carrier Frequencies: COST 231Model is Considered same separation distance and frequency carrier when using 30 m to 60 m leads to a noticeable increase in the path-loss, COST 231 model. whereas an increase in the average building height (from 60 Figure 11 (a, b and c) shows the path-loss versus to 90 m, from 90 to 120 m, or from 120 to 150 m) leads to separation distance at different values of base station gradual increase in the path-loss. heights and carrier frequencies when COST 231 model is Figure 13 (a, b and c) shows the path-loss versus used. separation distance at different values of road widths and The graph follows the same pattern as in Figures 8, 9, carrier frequencies when COST 231 model is used. Although and 10. The worth-mentioning point here is that identical the graph follows the same pattern as in Figures 8, 9, 10, path-loss values are estimated at the same separation distance 11, and 12, it reveals that there is an inverse proportional (nearly 27 km) but at different base station heights (30m and between the street width and the path-loss, i.e., as the width 50 m). Another worth-mentioning point is that the highest of the street increases, the path-loss decreases. Figure 14 (a, path-loss is always estimated when the base station height b and c) shows the path-loss versus separation distance at is 40 m. Figure 12 (a, b and c) shows the path-loss versus different values of orientation angles and carrier frequencies separation distance at different values of average building when COST 231 model is used. The graph follows the same heights and carrier frequencies when COST 231 model is pattern as in Figures 8, 9, 10, 11, 12, and 13. The observation used. The graph follows the same pattern as in Figures 8, 9, here is that an increase in the orientation angle from 0o to 45o 10, and 11. An increase in the average building height from leads to a noticeable increase in the path-loss; however an Jurnal Kejuruteraan 33(2) 2021: xxx-xxx Jurnal Kejuruteraan 33(2) 2021: xxx-xxx https://doi.org/10.17576/jkukm-2021-33(2)-16 https://doi.org/10.17576/jkukm-2021-33(2)-16
Figure 11 (a, b and c) shows the path-loss versus 120 to 150 m) leads to gradual increase in the path- Figure 11 (a, b and c) shows the path-loss versus 120 to 150 m) leads to gradual increase in the path- separation distance at different values of base loss. separation distance at different values of base loss. 326 station heights and carrier frequencies when COST station heights and carrier frequencies when COST 231 model is used. 231 model is used.
Carrier Carrier Carrier Carrier Carrier Carrier frequency frequency = frequency = frequency frequency = frequency = = 900 MHz 1800 MHz 2100 MHz = 900 MHz 1800 MHz 2100 MHz
a b c
FIGURE 11. Path-loss versus separation distance at different base station heights: COST 231 model was FIGUREFIGURE 11. Path-loss 11. Path versus-loss versus separation separation distance distance at different at different base stationbase station heights: heights: COST COST 231 model231 model was considered. was considered. considered.
Carrier Carrier Carrier Carrier Carrier Carrier frequency frequency = frequency = frequency frequency = frequency = = 900 MHz 1800 MHz 2100 MHz 2100 MHz = 900 MHz 1800 MHz
a b c a b c
FIGURE 12. Path-loss vs Separation Distance at Different Building Heights: COST 231 Model was FIGUREFIGURE 12. Path-loss 12. Path vs-loss Separation vs Separation Distance Distance at Different at Different Building Building Heights: Heights: COST COST 231 Model 231 M wasodel Considered. was Considered. Considered.
Carrier Carrier Carrier Carrier Carrier Carrier frequency frequency = frequency = frequency frequency = frequency = = 900 MHz 1800 MHz 2100 MHz = 900 MHz 1800 MHz 2100 MHz
Jurnal Kejuruteraan 33(2) 2021: xxx-xxx a b c a https://doi.org/10.17576/jkukmb -2021-33(2)-16 c
FIGURE 13. Path-loss vs Separationo Distanceo at Different Road Widths: COST 231 Model was increaseFIGURE in F theIGURE 13. orientation Path-loss 13. Path vs - angleloss Separation vs fromSeparation Distance 45 toDistance at90 Different at Different Road RoadWidths: W idths:COST COST 231 Model231 M odelwas Considered.was Considered. leads to reduce the path-loss. Considered.
The graph follows the same pattern as in Figures 8, The graphCarrier follows the same pattern as in FiguresCarrier 8, Figure 13 (a, b and c) shows the path-loss versus 9, and 10. The worth-mentioning point here is that Figure 13 (a, b andCarrier c) shows the path-loss versus 9, andfrequency 10. The worth-mentioning point herefrequency is that = separation distancefrequency at = different values of road identical= 900 pathMHz -loss values are estimated at18 the00 MHz same separation distance2100 MHzat different values of road identical path-loss values are estimated at the same widths and carrier frequencies when COST 231 separation distance (nearly 27 km) but at different widths and carrier frequencies when COST 231 separation distance (nearly 27 km) but at different model is used. Although the graph follows the base station heights (30 m and 50 m). Another model is used. Although the graph follows the base station heights (30 m and 50 m). Another same pattern as in Figures 8, 9, 10, 11, and 12, it worth-mentioning point is that the highest path-loss same pattern as in Figures 8, 9, 10, 11, and 12, it worth-mentioning point is that the highest path-loss reveals that there is an inverse proportional is always estimated when the base station height is reveals that there is an inverse proportional is always estimated when the base station height is between the street width and the path-loss, i.e., as 40 m. Figure 12 (a, b and c) shows the path-loss between the street width and the path-loss, i.e., as 40 m. Figure 12 (a, b and c) shows the path-loss the width of the street increases, the path-loss versus separationa distance at different values of b the width of the street increases, the path-loss versus separation distance at different values of decreases. Figure 14 (a, b andc c) shows the path- average building heights and carrier frequencies decreases. Figure 14 (a, b and c) shows the path- average building heights and carrier frequencies loss versus separation distance at different values when COST 231 model is used. The graph follows loss versus separation distance at different values whenFIGURE COSTFIGURE 14. 231 Path-loss 14 model. Path -vslossis Separationused vs S. eparationThe Distancegraph Distance follows at Different at Different Orientation ofOrientation orientation Angles: Angles: anglesCOST COST 231and 231 Modelcarrier M odelwas frequenci wasConsidered. es when the same pattern as in Figures 8, 9, 10, and 11. An of orientation angles and carrier frequencies when the same pattern as in Figures 8, 9, 10, and 11. CAnonsidered. COST 231 model is used. The graph follows the o o COST 231 model is used. The graph follows the increaseincrease in the in orientation the average angle building from height 45 to from90 leads 30 m to sameIt represents pattern theas inpath-loss Figures versus8, 9, 10, separation 11, 12, anddistance 13. at increase in the average building height from 30 m same pattern as in Figures 8, 9, 10, 11, 12, and 13. reduceto the 60 path-loss. m leads to a noticeable increase in the path- differentThe observationcases of indoor here environment. is that an Inincrease each case, in thethe path- toPATH 60 -mLOSS leads ESTIMATION to a noticeable USING increaseINDOOR PREDICTIONin the path - FigureThe observation 16 shows here path is -lossthat aversusn increase separation in the loss, whereas an increaseMODEL in the average building orientation angle from 0o to 45o leads to a loss, whereas an increase in the average building loss orientationincreases with angle the increasefrom 0 ino theto separation 45o lead sdistance. to a A height (from 60 to 90 m, from 90 to 120 m, or from distance at different number of floors when height PATH-LOSS(from 60 toESTIMATION 90 m, from USING 90 toINDOOR 120 m , or from separationnoticeable range increase lies between in the 30 path to 100-loss; m howeverwas used anin the In this part of results, the path-loss is estimated commercial,noticeable increase residential, in the pathand- loss;office however indoor an PREDICTION MODEL analysis as it represents the separation distance over which using equation (18). Figure 15 shows the graphical environments are considered. It is obviously seen the indoorthat an model increase is valid.in the The separation lowest path-lossdistance isleads estimated to In thisrepresentation part of results, of equation the path-loss (18). is estimated using in thean caseincrease in which in thethe commercial path-loss. indoorFurthermore, environment an is equation (18). Figure 15 shows the graphical representation used, whereas the highest path-loss is estimated in the case increase in the number of floors leads to gradual of equation (18). in whichincrease the inoffice the path indoor-loss. environment is used.
CONCLUSIONS
In this paper, a number of widely used path-loss estimation models that work in different environments were reviewed and analyzed, which significantly would assist any cellular network designer to select an appropriate path-loss model according to its working environment. The study FIGURE 15. Path-loss vs Separation Distance: Different drove us to enter into implicit comparisons between Cases of Indoor Environments are Considered the analyzed models.
Commercial Residential Office
a b c
Figure 16. Path-loss vs Separation Distance at Different Number of Floors: (a): Commercial Indoor Environment, (b): Residential Indoor Environment, and (c): Office Indoor Environment.
It represents the path-loss versus separation distance at different cases of indoor environment. Going through these comparisons, there were some In each case, the path-loss increases with the significant observations that might be exploited increase in the separation distance. A separation advantageously. For example, results in Figures 7 range lies between 30 to 100 m was used in the and 9 confirmed that an increase in the base station analysis as it represents the separation distance height leads to reduce the path-loss. Although this over which the indoor model is valid. The lowest observation can be exploited advantageously, i.e., it path-loss is estimated in the case in which the gives the opportunity to achieve a lower path-loss, commercial indoor environment is used, whereas it might affect the cost effectiveness of the cellular the highest path-loss is estimated in the case in system. On other words, the base station height which the office indoor environment is used. should be chosen such that it is comparable to the
Jurnal Kejuruteraan 33(2) 2021: xxx-xxx https://doi.org/10.17576/jkukm-2021-33(2)-16
increase in the orientation angle from 45o to 90o leads to reduce the path-loss.
Carrier Carrier Carrier frequency frequency = frequency = = 900 MHz 1800 MHz 2100 MHz
a b c
FIGURE 14. Path-loss vs Separation Distance at Different Orientation Angles: COST 231 Model was Considered.
PATH-LOSS ESTIMATION USING INDOOR PREDICTION Figure 16 shows path-loss versus separation MODEL distance at different number of floors when In this part of results, the path-loss is estimated commercial, residential, and office indoor using equation (18). Figure 15 shows the graphical environments are considered. It is obviously seen representation of equation (18). that an increase in the separation distance leads to 327 an increase in the path-loss. Furthermore, an increase in the number of floors leads to gradual increase in the path-loss.
CONCLUSIONS
In this paper, a number of widely used path-loss estimation models that work in different environments were reviewed and analyzed, which significantly would assist any cellular network designer to select an appropriate path-loss model according to its working environment. The study drove us to enter into implicit comparisons between FIGURE 1FIGURE5. Path- loss15. Path-loss vs Separation vs Separation Distance: Distance: Different Different Cases of Indoor Environments are Considered Cases of Indoor Environments are Considered the analyzed models.
Commercial Residential Office
a b c
FIGUREFigure 16. Path-loss 16. Path vs-loss Separation vs Separation Distance Distance at Different at Different Number Number of Floors: of Floors: (a): Commercial(a): Commercial Indoor Indoor Environment, Environment,(b): Residential (b): Residential Indoor IEnvironment,ndoor Environment, and (c): and Office (c): O fficeIndoor Indoor Environment. Environment.
FigureIt represents 16 shows path-lossthe path versus-loss separationversus separation distance at to reduce the cost by employing base stations with 30 m differentdistance number at differentof floors cases when ofcommercial, indoor environment. residential, and heightGoing instead through of 50 these m when comparisons, the separation there wereis less some than or officeIn indooreach case,environments the path are-loss considered. increases It withis obviously the equalsignificant to 27 Km . observationsHowever, base that stations might with be 50 exploited m height are seen thatincrease an increase in the inseparation the separation distance. distance A separation leads to an recommendedadvantageously. for use For at furtherexample, separations. results in Figures 7 increaserange in thelies path-loss.between 30Furthermore, to 100 m wasan increase used in inthe the and 9 confirmed that an increase in the base station numberanalysis of floors as leadsit represents to gradual the increase separation in the distancepath-loss. height leadsDECLARATION to reduce OF the COMPETING path-loss. INTEREST Although this over which the indoor model is valid. The lowest observation can be exploited advantageously, i.e., it path-loss is estimated in the case in which the None.gives the opportunity to achieve a lower path-loss, commercial indoorCONCLUSIONS environment is used, whereas it might affect the cost effectiveness of the cellular
the highest path-loss is estimated in the case in system. On other words,REFERENCES the base station height In thiswhich paper, the a numberoffice indoor of widely environment used path-loss is used estimation. should be chosen such that it is comparable to the models that work in different environments were reviewed Abhishek Agarwal., Aditya K., & Jagannatham. 2014. Distributed and analyzed, which significantly would assist any cellular estimation in homogenous poisson wireless sensor networks. network designer to select an appropriate path-loss model IEEE Wireless Communications Letters 3(1):90-93. according to its working environment. The study drove us A. Bhuvaneshwari., R. Hemalatha., T. Satyasavithri. 2016. to enter into implicit comparisons between the analyzed Semi Deterministic Hybrid model for Path Loss prediction models. improvement. Proceeding of 2nd International Conference Going through these comparisons, there were on Intelligent Computing, Communication & Convergence some significant observations that might be exploited (ICCC-2016) 336 – 344. advantageously. For example, results in Figures 7 and 9 Andrea Goldsmith. 2005. Wireless Communications. Cambridge confirmed that an increase in the base station height leads University Press. to reduce the path-loss. Although this observation can be Constantino Perez-Vegay., Jose Luis Garc´ıa Gy., & Jos´e Miguel exploited advantageously, i.e., it gives the opportunity to L´opez Higueraz. 1997. A simple and efficient model for indoor achieve a lower path-loss, it might affect the cost effectiveness path-loss prediction. Meas. Sci. Technol 8 (1997):1166–1173. of the cellular system. On other words, the base station Divya Kurup., Maria Scarpello., Günter Vermeeren., Wout Joseph., height should be chosen such that it is comparable to the Kristof Dhaenens., Fabrice Axisa., Luc Martens., Dries Vande surrounding buildings. Result in Figure 11 confirmed that Ginste., Hendrik Rogier., Jan Vanfleteren. 2011. In-body path identical path-loss values are estimated at same separation loss models for implants in heterogeneous human tissues using distance (nearly 27 km) but at different base station heights implantable slot dipole conformal flexible antennas.EURASIP (30 m and 50 m) when COST 231 model is used. This can Journal on Wireless Communications and Networking Article be exploited advantageously, i.e., it gives the opportunity ID 51 328
F. Ikegami., T. Takeuchi., & S.Yoshida. 1991. Theoretical Minthorn Phunthawornwong, Eakkachai Pengwang & prediction of mean field strength of urban mobile radio. IEEE Rardchawadee Silapunt. 2018. Indoor Location Estimation of Trans. Antennas Propagat 39(3):299–302. Wireless Devices Using the Log-Distance Path Loss Model, Haipeng Ding., Zhengyuan Xu., Brian M. Sadler. 2010. A Path Loss TENCON. Proceeding of IEEE Region 10 Conference, South Model for Non-Line-of-Sight Ultraviolet Multiple Scattering Korea 28-31. Channels. EURASIP Journal on Wireless Communications and Nour Zaarour, Nahi Kandil, Sofiène Affes & Nadir Hakem. 2016. Networking Article ID 598572 Path loss exponent estimation using connectivity information Schiller, J. H. 2003. Mobile Communications. Pearson Education in wireless sensor network. IEEE International Symposium Limited. on Antennas and Propagation (APSURSI), Fajardo, Puerto Walfisch, J. & Bertoni, H. L. 1988. A theoretical model of UHF Rico propagation in urban environments. IEEE Trans. Antennas Vaishnavi Dharmadhikari, Neha Pusalkar & Pradnya Ghare. Propagat 36(12):1788–1796. 2018. Path Loss Exponent Estimation for Wireless Sensor Kazunori Uchida, Naoto Hadano, Masafumi Takematsu & Junichi Node Positioning: Practical Approach. Proceeding of IEEE Honda. 2014. Propagation Estimation by Using Building International Conference on Advanced Networks and Coverage and Floor Area Ratios Based on 1-Ray Model Telecommunications Systems (ANTS), India 6-19. Combined with Okumura-Hata Model. Proceeding of the Garg, V. K. 2007. Wireless Communications and Networking. 17th International Conference on Network-Based Information Elsevier. Systems 555 – 560. Yoo-Seung Song & Hyun-Kyun Choi. 2017. Analysis of V2V L. M. Correia. 2009. A View of the COST 231-Bertoni-Ikegami broadcast performance limit for WAVE communication Mode. Proceeding of the 3rd European Conference on systems using two-ray path loss model. ETRI Journal Antennas and Propagation. EuCAP Berlin. 39(2):213-221 Mario Versaci., Salvatore Calcagno., Fabio La Foresta., & Biagio Yun-Jie Xu & Wen-Bin Li. 2011. Propagation path loss prediction Cammaroto. 2012. Ppath loss prediction using fuzzy inference model of multi-sensor network in forest. Proceeding of system and ellipsoidal rules. American Journal of Applied Advanced in Control Engineering and Information Science Sciences, Science Publication 9(12):1940-1943 Elsevier 2206 – 2210.