Wireless Channels Path Loss and Shadowing

Transcription

1 Wieless Channels Pah Loss and Shadowing A. Özgü Yılmaz - METU EE 78 METU AOY 1

2 Wieless channel suscepible o Noise Inefeence Channel impedimens Impedimens change ove ime unpedicably due o Use movemen Envionmen dynamics Channel impedimens Pah loss and shadowing (~deeminisic, lage scale) Mulipah (~saisical, small scale) EE 78 METU AOY

3 Fee space popagaion, line of sigh (LOS) aenuaion An isoopic x anenna wih powe Was Pah Loss and Shadowing Powe densiy a disance d P d P 4d W / m If x anenna has dieciviy G (field adiaion paen) 4d Rx anenna gahes a poion of he adiaed powe popoional o is coss-secional aea. G P 4d A P EE 78 METU AOY 3

4 Fom EMT A G Fiis ansmission fomula P PG G 4 c ( 4fd) Received powe ime EE 78 METU AOY 4

5 Remaks Gains depend on anenna physical popeies. Ohe losses (amospeic absopion) someimes effecive Pah loss P L (4d ) P PG G / P L 1 Usually in db ( P ) ( P ) ( G ) ( G ) ( P ) ( P db db db db L db a ) db Used diecly fo saellie communicaions and adio links EE 78 METU AOY 5

6 Tansmied signal Equivalen lowpass epesenaion of bandpass signals u() complex envelope, equivalen lowpass signal Received signal EE 78 METU AOY 6

7 P PG G c ( 4fd) Wih fee space pah loss c Gl ( ) Re{ 4fd u( ) e jf ( ) Link gain compised of ansmi and eceive anenna gains G G G l } Delay due o he disance aveled by he EM waves EE 78 METU AOY 7

8 Pah loss (usually anenna gains excluded) P P L L, db Gl P P 0log c 4fd 10 c 4fd EE 78 METU AOY 8

9 Example EE 78 METU AOY 9

10 Amospheic aenuaion Oxygen, wae Rain, fog Heigh dependen hp:// EE 78 METU AOY 10

11 hp:// EE 78 METU AOY 11

12 Ray Tacing Mulipah signal componens Many objecs in suoundings Reflecion (EMW on an objec lage han wavelengh) Diffacion (pah obsuced by a suface wih shap iegulaiies) Scaeing (medium densely consiss of objecs smalle han wavelengh) EE 78 METU AOY 1

13 Solve Maxwell s equaions wih bounday condiions Too complex Eveyhing should be pefecly known Simplificaion necessay Ray acing Assume a finie numbe of eflecos wih known locaion and dielecic popeies EE 78 METU AOY 13

14 EE 78 METU AOY 14 Two-ay model x x' Gound eflecion coefficien Gound eflecion coefficien c x x c l, b a l G G G d c G G G f j x x j l j l c e x x e G u R l e G u ')/ ( / ' ) ( ) ( 4 ) (

15 EE 78 METU AOY 15 If ansmied signal is slowly changing in elaion o, ) / ' (, ' 4 l x x x x e G R l G P P j l ' d h h d h h l x x 1 0, 1 1 x x x d h h l x x ' d h h 4 ) ( ) ( ), ( ) ( u u u u

16 Asympoic case d is lage Gl G x x' l d, Fo eah and oad sufaces 0 R 1 Receive powe falls off as Independen of fequency since combinaion of wo ays effecively foms an anenna aay (anenna aay gain does no necessaily decease wih fequency) d 4 EE 78 METU AOY 16

17 900MHz G G R 1 h l 1 50m, h m EE 78 METU AOY 17

18 Example EE 78 METU AOY 18

19 Ten-ay model Fo uban micocells Fla ciy wih 90 degees inesecing linea sees (ecilinea sees) Buildings along boh sides of sees Building-lined sees ac as dielecic canyon o he popagaing signal. Since signal enegy is dissipaed wih each eflecion, moe han 3 eflecions can be geneally ignoed. EE 78 METU AOY 19

20 * Typical powe falloff d Some empiical sudies obained powe falloff popoional o d, 6 EE 78 METU AOY 0

21 Genealized ay acing Diffaced and scaeed ays also aken ino accoun Leads o a complicaed pah loss model. Simplificaions needed Local mean eceived powe (LMRP) Ray acing depends on exac x/x locaions (phase) Only a mean eceived powe usually equied fo link qualiy Cellula sysems uilize LMRP fo powe conol and handoff EE 78 METU AOY 1

22 Empiical Pah-Loss Models Mos wieless sysems opeae in complex popagaion envionmens Canno be accuaely modeled by fee space pop. o ay acing Pah-loss models developed ove he yeas o pedic pah loss in ypical wieless envionmens Lage uban macocells Uban micocells Inside buildings, Neve foge: hese ae jus models! EE 78 METU AOY

23 These models based on empiical measuemens Ove a given disance In a given fequency ange Fo a paicula geogaphical aea o building One mus be caeful in using hese models fo ohe scenaios. Pah loss, shadowing, mulipah all conibue o eceived powe in empiical measuemens Aveaging o emove mulipah effecs Local mean aenuaion ove seveal Repeiions houghou he envionmen Repeiions in simila envionmens EE 78 METU AOY 3

24 Okumua Model Lage uban macocells, 1-100kms MHz Base saion-o-mobile measuemens in Tokyo Pah loss Empiical fomulas Median aenuaion Anenna heigh gains Gain due o ype of envionmen Ohes obained fom Okumua s empiical plos Coecions poposed lae EE 78 METU AOY 4

25 Haa model Closed-fom fomula fo Okumua s model Fequency in MHz, disance in km Coecion faco fo he mobile anenna heigh based on he size of he coveage aea Small-o-medium size ciies Lage ciies f c 300MHz Ohe envionmens K anges fom 39.54(counyside) o 40.94(dese) EE 78 METU AOY 5

26 Haa well appoximaed Okumua fo Haa does no model well he cuen cellula sysems wih smalle cell sizes and highe fequencies, indoo envionmens COST 31 Exension o Haa 1.5-GHz, 1-0kms 30m h 00m,1m h 10m d 1km 0dB fo medium-sized ciies, sububs; 3dB fo meopolian aeas Piecewise Linea (Mulislope) Model C M Empiical measuemens ae fied o piecewise linea funcions EE 78 METU AOY 6

27 EE 78 METU AOY 7

28 Indoo Aenuaion Facos Peneaion hough Walls Floos Objecs Glass, All hese facos significanly affec indoo pah loss Floos Depends on building maeial Aenuaion lages fo he 1 s passed floo (10-0dB) Deceases wih subsequen floos (6-10dB, a few db fo lage han 4 floos) Rappapo has deails 1 Floo 0 EE 78 METU AOY 8

29 Paiion losses (walls) Losses by diffeen sudies vay widely Vey had o make genealizaions FAF: floo aenuaion faco PAF: paiion aenuaion faco EE 78 METU AOY 9

30 Simplified Pah-Loss Model Simplified models necessay fo sysem design K uniless consan depending on anenna chaaceisics and aveage channel aenuaion d 0 efeence disance fo anenna fa field pah-loss coefficien EE 78 METU AOY 30

31 Shadowing Obsacles beween ansmie and eceive Signal aenuaed EE 78 METU AOY 31

32 Bad ecepion Random vaiaion of eceived powe due o blockage fom objecs in signal pah The exac locaions and impac of he blocking objecs ae usually unknown => saisical models The mos common model: log-nomal pdf Gain in db is nomal. Good ecepion EE 78 METU AOY 3

33 P P, db 10log P 10 db db P N(, ) 1 P physically impossible P Log-nomal model capues he undelying physical model mos coecly when db 0 EE 78 METU AOY 33

34 Mahemaical jusificaion fo log-nomal Aenuaion due o an objec Aenuaion due o many objecs CLT afe aking logaihm Shadowing is a andom pocess Assumpion: WSS e i Covaiance beween shadow fading a wo poins sepaaed by disance e i i EE 78 METU AOY 34

35 Decoelaion disance is whee coelaion dops o 1/e of max Shadowing.p. X c Whie noise passed hough a fis-ode IIR file (AR) EE 78 METU AOY 35

36 Combined Pah Loss and Shadowing Combinaion of simplified pah loss Zeo mean shadow fading ceaing vaiaions in eceived powe P P db 10log db d K 10 log10 d 0, ~ N(0, 10 db db db ) Slowly changing Rapidly changing EE 78 METU AOY 36

37 Ouage pobabiliy unde pah loss and shadowing Ouage: even ha he eceived powe falls below a pedeemined powe level Ouage pobabiliy idea can be used o find he cell coveage aea in cellula newoks. EE 78 METU AOY 37

38 Example 900MHz, 3.71, K=-31.54dB, vaianceof log-nomal shadowing 13.9 EE 78 METU AOY 38