Wireless Communication Channel Overview

Transcription

1 EC744 Wireless Communicaion Fall 008 Mohamed Essam Khedr Deparmen of Elecronics and Communicaions Wireless Communicaion Channel Overview

2 Syllabus Tenaively Week 1 Week Week 3 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week Week 11 Week 1 Week 13 Week 14 Week 15 Overview wireless communicaions, Probabiliies Digial Communicaion fundamenals Channel characerisics (AWGN, fading) Modulaion echniques Demodulaion echniques (coheren and noncoheren) Source coding echniques Channel coding echniques Mid Term exam (ake home), Diversiy echniques Equalizaion echniques Spread specrum, MIMO and OFDM Wireless neworking: 80.11, 80.16, UWB Ho opics Presenaions Presenaions Presenaions Final Exam

3 Anenna - Ideal - cond. The power densiy of an ideal loss-less anenna a a disance d away from he ransmiing anenna: P a = PG 4πd W/m Noe: he area is for a sphere. G is he ransmiing anenna gain The produc P G : Equivalen Isoropic Radiaion Power (EIRP) which is he power fed o a perfec isoropic anenna o ge he same oupu power of he pracical anenna in hand.

4 Signal Propagaion (Channel Models)

5 Channel Models High degree of variabiliy (in ime, space ec.) Large signal aenuaion Non-saionary, unpredicable and random Unlike wired channels i is highly dependen on he environmen, ime space ec. Modelling is done in a saisical fashion The locaion of he base saion anenna has a significan effec on channel modeling Models are only an approximaion of he acual signal propagaion in he medium. Are used for: performance analysis simulaions of mobile sysems measuremens in a conrolled environmen, o guaranee repeaabiliy and o avoid he expensive measuremens in he field.

6 Channel Models - Classificaions Sysem Model - Deerminisic Propagaion Model- Deerminisic Predics he received signal srengh a a disance from he ransmier Derived using a combinaion of heoreical and empirical mehod. Sochasic Model - Rayleigh channel Semi-empirical (Pracical +Theoreical) Models

7 Channel Models Is almos always linear, and also ime-varian because of is mobiliy. Thus, fully described by is impulse response h(τ, ), where τ is he delay parameer and is he ime. The complex impulse response h(τ, ) is a low-pass equivalen model of he acual real band-pass impulse response. Equivalenly, he channel is characerized by is ransfer funcion which is he Fourier ransform of he h(τ, ): H ( f, ) = h( τ, )exp( jπfτ ) dτ The magniude H(f, ) is changing randomly in ime, so he mobile radio channel is described as a fading channel. The phase arg H(f, ) is also a random funcion of ime.

8 Channel Models Muli-pah channel impulse response )) ( ( ))], ( ) ( ( )exp[, ( ), ( 1 0 f j a h i i i N i c i b τ τ δ τ φ τ π τ τ + = =

9 Propagaion Pah Loss The propagaion pah loss is L PE = L a L lf L sf where L a is average pah loss (aenuaion): (1- km), L lf - long erm fading (shadowing): 0 m ignoring variaions over few wavelenghs, L sf - shor erm fading (mulipah): over fracion of wavelengh o few wavelengh. Merics (dbm, mw) [P(dBm) = * log[ P(mW) ]

10 Propagaion Pah Loss Free Space Power received a he receiving anenna P r = PG G r 4 λ πd Thus he free space propagaion pah loss is defined as: P r GGrλ Lf = Log = Log P (4πd ) Isoropic anenna has uniy gain (G = 1) for boh ransmier and receiver.

11 Propagaion - Free Space cond. The difference beween wo received signal powers in free space is: P = P r1 d1 log = 0log Pr d db If d = d 1, he P = -6 db i.e 6 db/ocave or 0 db/decade

12 Propagaion - Non-Line-of-Sigh Generally he received power can be expressed as: P r d -v For line of sigh v =, and he received power P r d - For non-line of sigh wih no shadowing, received power a any disance d can be expressed as: d P r ( d) = log[ Pr ( dref )] + v log dref 0 m< d ref < 00 m

13 Propagaion - Non-Line-of-Sigh Log-normal Shadowing P ( d) = log Pr d + v [ ( ref )] log r + d d ref X σ Where X : N(0,) Gaussian disribued random variable

14 Received Power for Differen Value of Loss Parameer v Received power (dbm) v =, Free space -1 - v = 3 Rural areas v = 4, Disance (km) Ciy and urban areas

15 Propagaion Model- Free Space In erms of frequency f and he free space velociy of elecromagneic wave c = 3 x 8 m/s i is: L f c / f = 0log 4πd db Expressing frequency in MHz and disance d in km: L f = 0log ( c / 4π) + 0log( f ) + 0log( d) = 0 log (0.3/ 4π) + 0log ( f ) + 0log ( d) db L f = log( f ) + 0log( d) db

16 Propagaion Model- Free Space (non-ideal, pah loss) Non-isoropic anenna gain uniy, and here are addiional losses L ad, hus he power received is: P r Pλ = GGr d > 0 and L 0 (4πd ) Lad Thus for Non-isoropic anenna he pah loss is: L BS f ni = log + 0log ( G ( ) log f ) + 0log ( G r 1 ) 0log ( d) + log ( c / 4π) ( L MU Noe: Inerference margin can also be added ad ) db

17 Propagaion Model - Mechanisms Reflecion Diffracion Scaering Source: P M Shankar

18 Channel Model- Plan Earh Pah Loss - Ray Reflecion In mobile radio sysems he heigh of boh anennas (Tx. and Rx.) << d (disance of separaion) d d d Direc pah (line of sigh) h b h m d r Ground refleced pah From he geomery d d = [d + (h b - h m ) ]

19 Channel Model- Plan Earh Pah Loss - cond. Using he binomial expansion Noe d >> h b or h m. d d d1 + hb 0.5 d h m Similarly d r d1 + hb d h m The pah difference d = d r - d d = (h b h m )/d The phase difference φ = π λ h b d h m = 4πh b λd h m

20 Channel Model- Plan Earh Pah Loss cond. Toal received power P r = PG G r λ 4πd 1+ ρe j φ Where ρ is he reflecion coefficien. For ρ = -1 (low angle of inciden) and. 1 e j φ = 1 cos φ + j sin φ Hence 1 e j φ = (1 cos φ) + sin φ = (1 cos φ) = 4sin ( φ / )

21 Channel Model- Plan Earh Pah Loss cond. Therefore: λ π π λ = d h h d PG G P m b r r 4 sin Assuming ha d >> h m or h b, hen 1 << λ π d h h m b sin x = x for small x Thus = d h h PG G P m b r r which is 4 h power law

22 Channel Model- Plan Earh Pah Loss cond. Propagaion pah P loss (mean loss) r hbhm L PE = log = logggr P d Compared wih he free space = P r = 1/ d In a more general form (no fading due o mulipah), pah aenuaion is L PE = log G 0log log h m + G r 40log 0log d h b d L PE increases by 40 db each ime d increases by

23 Channel Model- Plan Earh Pah Loss cond. Verical Horizonal P r ~ 1/d 4 P r ~ 1/d Free space S Loyka

24 LOS Channel Model - Problems Simple heoreical models do no ake ino accoun many pracical facors: Rough errain Buildings Refecion Moving vehicle Shadowing Thus resuling in bad accuracy Soluion: Semi- empirical Model

25 Sem-iempirical Model Pracical models are based on combinaion of measuremen and heory. Correcion facors are inroduced o accoun for: Terrain profile Anenna heighs Building profiles Road shape/orienaion Lakes, ec. Okumura model Haa model Saleh model SIRCIM model Oudoor Indoor Y. Okumura, e al, Rev. Elec. Commun. Lab., 16( 9), M. Haa, IEEE Trans. Veh. Technol., 9, pp , 1980.

26 Okumura Model Widely used empirical model (no analyical basis!) in macrocellular environmen Predics average (median) pah loss Accurae wihin -14 db in urban and suburban areas Frequency range: MHz Disance: > 1 km BS anenna heigh: > 30 m. MU anenna heigh: up o 3m. Correcion facors are hen added.

27 Haa Model Consolidae Okumura s model in sandard formulas for macrocells in urban, suburban and open rural areas. Empirically derived correcion facors are incorporaed ino he sandard formula o accoun for: Terrain profile Anenna heighs Building profiles Sree shape/orienaion Lakes Ec.

28 Haa Model cond. The loss is given in erms of effecive heighs. The saring poin is an urban area. The BS anennae is mouned on all buildings. The effecive heigh is hen esimaed a 3-15 km from he base of he anennae. P M Shankar

29 Haa Model - Limis Frequency range: MHz Disance: 1 0 km BS anena heigh: m MU anenna heigh: 1 m

30 Haa Model Sandard Formula for Average Pah Loss for Urban Areas L pl u = log ( f ) + ( log h ) b log d 13.8 log h b a ( h ) (db) mu a a ( ) [ ( )] h = 3. log 11.75h 4.97 ( f 400MHz) db a ( ) [ ( )] h = 8.3 log 1.5h 1.1 ( f 00MHz) db mu mu mu ( h ) = [.1log ( f ) 0.7] h [ 1.56 log ( f ) 0.8] db mu mu 1 mu

31 Haa Model Average Pah Loss for Urban Areas cond. Carrier frequency 900 MHz, BS anenna heigh 150 m, MU anenna heigh 1.5m. P M Shankar

32 Haa Model Average Pah Loss for Suburban and Open Areas Suburban Areas L pl su = Lpl u Log f Open Areas L pl o = Lpl u 4.78(Log f ) 18.33Log f 40.94

33 Haa Model - Average Pah Loss S. Loyka, 003, Inroducion o Mobile Communicaions

34 Improved Model Haa-Okumura model are no suiable for lower BS anenna heighs ( m), and hilly or moderae-o-heavy wooded errain. To correc for hese limiaions he following model is used [1]: For a given close-in disance d ref. he average pah loss is: L pl = A + v log (d / d ref ) + s for d > d ref, (db) where A = 0 log(4 d ref / ) v is he pah-loss exponen = (a b hb + c / hb) hb is he heigh of he BS: beween m and 80 m d ref = 0m and a, b, c are consans dependen on he errain caegory s is represening he shadowing effec [1] V. Erceg e. al, IEEE JSAC, 17 (7), July 1999, pp

35 Improved Model Terrains Model Type A Type B Type C parameer a b c The ypical value of he sandard deviaion for s is beween 8. And.6 db, depending on he errain/ree densiy ype Terrain A: The maximum pah loss caegory is hilly errain wih moderae-o-heavy ree densiies. Terrain B: Inermediae pah loss condiion Terrain B: The minimum pah loss caegory which is mosly fla errain wih ligh ree densiies