Chapter 2 PHYSICAL AND LINK LAYER

Transcription

1 Overview Chaper 2 PHYSICAL AND LINK LAYER Mobile Compuing Summer 2004 Disribued Compuing Group Frequenies Signals Anenna Modulaion Signal propagaion Mulipleing Spread sperum CDMA Disribued Compuing Group Frequenies regulaed VLF LF wised pair 2/2 ITU-R holds auions or new requenies, manages requen bands worldwide (WRC, World Radio Conerenes) Mobile phones 0 km 30 khz R. Waenhoer Frequenies and regulaions Mm 300 Hz MOBILE COMPUTING 00 m 3 MHz MF HF m 300 MHz VHF oa UHF 0 mm 30 GHz SHF EHF 00 µm 3 THz inrared µm 300 THz visible ligh UV Cordless elephones ISM Wireless LANs Europe (CEPT/ETSI) USA (FCC) Japan NMT MHz, MHz GSM MHz, MHz, MHz, MHz CT MHz, MHz CT MHz DECT MHz IEEE MHz HIPERLAN MHz AMPS, TDMA, CDMA MHz, MHz TDMA, CDMA, GSM MHz, MHz PACS MHz, MHz PACS-UB MHz PDC MHz, MHz, MHz, MHz IEEE MHz IEEE MHz PHS MHz JCT MHz AM SW FM Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/3 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/4

2 Periodi Signals Transmiing digial daa: Fourier? g() = A sin(2π + φ ) Ampliude A requen [Hz = /s] period T = / wavelengh λ wih λ = (=3 0 8 m/s) phase φ φ* = -φt/2π [+T] A φ* 0 T Ever (periodi) signal an be represened b ininiel man sinuses and osines g( ) = + an sin(2πn) + bn os(2πn) 2 0 periodi signal n= Bu in wireless ommuniaion we onl have narrow bands Also dieren requenies behave dierenl Modulaion n= 0 superpose harmonis Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/5 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/6 Anennas: isoropi radiaor Radiaion and reepion o eleromagnei waves, oupling o wires o spae or radio ransmission Isoropi radiaor: equal radiaion in all hree direions Onl a heoreial reerene anenna Radiaion paern: measuremen o radiaion around an anenna Sphere: S = 4π r 2 z z ideal isoropi radiaor Anennas: simple dipoles Real anennas are no isoropi radiaors bu, e.g., dipoles wih lenghs λ/2 as Herzian dipole or λ/4 on ar roos or shape o anenna proporional o wavelengh Eample: Radiaion paern o a simple Herzian dipole z side view (z-plane) λ/4 z side view (z-plane) λ/2 op view (-plane) simple dipole Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/7 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/8

3 Anennas: direed and seorized Oen used or mirowave onneions or base saions or mobile phones (e.g., radio overage o a valle) z side (z)/op (z) views / side view (z-plane) [Buwal] direed anenna Anennas: diversi Grouping o 2 or more anennas muli-elemen anenna arras Anenna diversi swihed diversi, seleion diversi reeiver hooses anenna wih larges oupu diversi ombining ombine oupu power o produe gain ophasing needed o avoid anellaion λ/4 λ/2 λ/4 λ/2 λ/2 λ/2 seorized anenna + + op view, 3 seor op view, 6 seor ground plane Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/9 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/0 Signal propagaion ranges Aenuaion b disane Propagaion in ree spae alwas like ligh (sraigh line) Transmission range ommuniaion possible low error rae Deeion range deeion o he signal possible sender no ommuniaion possible ransmission Inererene range signal ma no be deeion deeed inererene signal adds o he bakground noise disane Aenuaion [db] = 0 log 0 (ransmied power / reeived power) Eample: aor 2 loss = 0 log db In heor/vauum (and or shor disanes), reeiving power is proporional o /d 2, where d is he disane. In praie (or long disanes), reeiving power is proporional o /d α, α = 4 6. We all α he pah loss eponen. Eample: Shor disane, wha is he aenuaion beween 0 and 00 meers disane? Faor 00 (=00 2 /0 2 ) loss = 20 db reeived power α = db drop α = 4 6 LOS NLOS disane Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/ Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/2

4 Aenuaion b objes Real World Eamples Shadowing (3-30 db): eile (3 db) onree walls (3-20 db) loors (20-30 db) releion a large obsales saering a small obsales diraion a edges ading (requen dependen) shadowing releion saering diraion Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/3 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/4 Mulipah propagaion Ees o mobili Signal an ake man dieren pahs beween sender and reeiver due o releion, saering, diraion Channel haraerisis hange over ime and loaion signal pahs hange dieren dela variaions o dieren signal pars dieren phases o signal pars quik hanges in power reeived (shor erm ading) signal a sender signal a reeiver Time dispersion: signal is dispersed over ime Inererene wih neighbor smbols: Iner Smbol Inererene (ISI) The signal reahes a reeiver direl and phase shied Disored signal depending on he phases o he dieren pars Addiional hanges in disane o sender obsales urher awa slow hanges in average power reeived (long erm ading) power Doppler shi: Random requen modulaion shor erm ading long erm ading Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/5 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/6

5 Mulipleing Muliple hannels (k) in our dimensions spae (s) ime () requen () ode () hannels k i k k 2 k 3 k 4 k 5 k 6 Eample or spae mulipleing: Cellular nework Simpliied heagonal model Signal propagaion ranges: Frequen reuse onl wih a erain disane beween he base saions Goal: muliple use o a shared medium Imporan: guard spaes needed! Eample: radio broadas s s 3 s 2 Can ou reuse requenies in disane 2 or 3 (or more)? Graph oloring problem Eample: ied requen assignmen or reuse wih disane 2 Inererene rom neighbor ells (oher olor) an be onrolled wih ransmi and reeive ilers Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/7 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/8 Carrier-o-Inererene / Signal-o-Noise Frequen Division Muliple (FDM) Digial ehniques an wihsand a Carrier-o-Inererene raio o approimael 9 db. R D Separaion o he whole sperum ino smaller requen bands A hannel ges a erain band o he sperum or he whole ime + no dnami oordinaion neessar Assume he pah loss eponen α = 3. Then, + works also or analog signals wase o bandwidh i rai is disribued unevenl inleible k k 2 k 3 k 4 k 5 k 6 whih gives D/R = 3. Reuse disane o 2 migh jus work Eample: broadas radio Remark: Inererene ha anno be onrolled is alled noise. Similarl o C/I here is a signal-o-noise raio S/N (SNR). Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/9 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/20

6 Time Division Muliple (TDM) Time and Frequen Division Muliple A hannel ges he whole sperum or a erain amoun o ime + onl one arrier in he medium a an ime + hroughpu high even or man users preise snhronizaion neessar k k 2 k 3 k 4 k 5 k 6 Combinaion o boh mehods A hannel ges a erain requen band or some ime + proeion agains requen seleive inererene + proeion agains apping + adapive preise oordinaion required k k 2 k 3 k 4 k 5 k 6 Eample: Eherne Eample: GSM Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/2 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/22 Code Division Muliple (CDM) Cokail par as analog or mulipleing Eah hannel has a unique ode All hannels use he same sperum a he same ime + bandwidh eiien + no oordinaion or snhronizaion + hard o ap + almos impossible o jam lower user daa raes more omple signal regeneraion Eample: UMTS Spread sperum U. S. Paen , Hed K. Marke (a.k.a. Lamarr or Kiesler) and George Anheil (942) k k 2 k 3 k 4 k 5 k 6 Spae muliple: Communiae in dieren rooms Frequen muliple: Use soprano, alo, enor, or bass voies o deine he ommuniaion hannels Time muliple: Le oher speaker inish Code muliple: Use dieren languages and hone in on our language. The arher apar he languages he beer ou an iler he noise : German/Japanese beer han German/Duh. Can we have orhogonal languages? Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/23 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/24

7 Spread sperum ehnolog Frequen Hopping Spread Sperum (FHSS) Problems: narrowband inererene and requen dependen ading Soluion: spread he narrow band signal ino a broad band signal using a speial ode sender i) reeiver iii) P P ii) iv) Side ees: o-eisene o several signals, and more ap-proo Alernaives: Frequen Hopping or Dire Sequene P P v) user signal broadband inererene narrowband inererene P Disree hanges o arrier requen sequene o requen hanges deermined via pseudo random number sequene Two varians Fas Hopping: several requenies per user bi Slow Hopping: several user bis per requen + requen seleive ading and inererene limied o shor period + simple implemenaion + uses onl small porion o sperum a an ime no ver robus requen hopping has overhead Eample: Blueooh Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/25 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/26 Code Division Muliple Aess (CDMA) (Media Aess Laer ould as well be in Leure 3) As eample or Dire Sequene Spread Sperum (DSSS) Eah saion is assigned an m-bi ode (or hip sequene) Tpiall m = 64, 28,... (in our eamples m = 4, 8, ) To send bi, saion sends hip sequene To send 0 bi, saion sends omplemen o hip sequene Eample: MHz band wih 00 saions FDM eah saion a 0 khz band assume ha ou an send bi/hz: 0 kbps CDMA eah saion uses he whole MHz band less han 00 hips per hannel: more han 0 kbps CDMA basis Eah saion s has unique m-bi hipping ode S or omplemen S Bipolar noaion: binar 0 is represened b (or shor: ) Two hips ST, are orhogonal i S T = 0 S T is he inner (salar) produ: S T = m Noe: S S =, S S = Noe: S T = 0 S T = 0 m i = ST i i Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/27 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/28

8 CDMA basis 2 CDMA basis 3 Assume ha all saions are perel snhronous Assume ha all odes are pair wise orhogonal Assume ha i wo or more saions ransmi simulaneousl, he bipolar signals add up linearl Eample S = ( ) T = ( ) U = ( ) Chek ha odes are pair wise orhogonal I S,T,U send simulaneousl, a reeiver reeives R = S+T+U = (+3,,,,,, +3, ) To deode a reeived signal R or sender s, one needs o alulae he normalized inner produ R S. R S = (+3,,,,,, +3, ) ( )/8 = ( )/8 = 8/8 = b aiden? R S = (S+T+U) S = S S +T S +U S = = Wih orhogonal odes we an sael deode he original signals Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/29 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/30 CDMA: How muh noise an we olerae? CDMA: Consruion o orhogonal odes wih m hips We now add random noise o beore we reeive he signal: R = R + N, wheren is an m-digi noise veor. Assume ha hipping odes are balaned (as man + as ) I N = (α, α,, α) or an (posiive or negaive) α, hen he noise N will no maer when we deode he reeived signal. R S = (R+N) S = S S +(orhogonal odes) S +N S = = How muh random (whie) noise an we olerae? (See eerises) Noe ha we anno have more han m orhogonal odes wih m hips beause eah ode an be represened b a veor in he m- dimensional spae, and here are no more han m orhogonal veors in he m-dimensional spae. Walsh-Hadamard odes an be onsrued reursivel (or m = 2 k ): The se o odes o lengh is C0 = {( + )}. For eah ode ( ) Ck we have wo odes ( ) and ( ) in Ck+ Code ree: C0 = {( + )} C = {( + + ),( + )} C = {( ++++ ),( ++ ),( + + ),( + + )} 2 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/3 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/32

9 CDMA: Random odes CDMA: Random odes 2 We anno have more han m orhogonal odes. Marin Cooper (Moorola, righ) sas wih UMTS ou ge a mos Mbps, he Swiss newspaper Sonnagszeiung adds bu when ou have o share a ell wih 2 [6?] ohers, ou ge a mos 64 kbps. We said: 00 saions wih less han 00 hips per [saion] Idea: Random odes are almos balaned and almos pair wise orhogonal Wih k oher saions, and m hips m R S = m S S + m (k random odes) S = ±m + X, where X is he sum o mk random variables ha are eiher + or. Sine he random variables are independen, he epeed value o X is 0. And beer: The probabili ha X is ar rom 0 is small. Thereore we ma deode he signal as ollows: R S > ε deode ; R S < ε deode 0. Wha i ε R S ε?? Eperimenal evaluaion (righ): For 0.4 k = m = 28 deoding is orre more 0.3 han 80%. Bu more imporanl: 0.2 Even i k > m (k=..500), he ssem 0. does no deeriorae quikl. 0 righ wrong Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/33 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/34 CDMA: Problems CDMA: Summar Some o our assumpions were no aurae: A) I is no possible o snhronize hips perel. Wha an be done is ha he sender irs ransmis a long enough known hip sequene on whih he reeiver an lok ono. B) No all saions are reeived wih he same power level. CDMA is piall used or ssems wih ied base saions. Then mobile saions an send wih he reiproal power he reeive rom he base saion. (Alernaivel: Firs deode he bes saion, and hen subra is signal o deode he seond bes saion?) + all erminals an use he same requen, no planning needed + redues requen seleive ading and inererene + base saions an use he same requen range + several base saions an dee and reover he signal + so handover beween base saions + orward error orreion and enrpion an be easil inegraed preise power onrol neessar higher omplei o reeiver and sender Eamples: Third generaion mobile phones, UMTS, IMT C) We sill didn disuss how o ransmi bis wih eleromagnei waves. Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/35 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/36

10 Modulaion and demodulaion analog baseband digial signal daa digial analog 0000 modulaion modulaion radio ransmier radio arrier Digial modulaion Modulaion o digial signals known as Shi Keing 0 Ampliude Shi Keing (ASK): ver simple low bandwidh requiremens ver susepible o inererene 0 analog demodulaion radio arrier analog baseband signal snhronizaion deision digial daa 0000 radio reeiver Frequen Shi Keing (FSK): needs larger bandwidh Phase Shi Keing (PSK): more omple robus agains inererene 0 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/37 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/38 Dieren represenaions o signals Advaned Frequen Shi Keing For man modulaion shemes no all parameers maer. MSK (Minimum Shi Keing) A [V] [s] A [V] I = A sin ϕ ϕ bandwidh needed or FSK depends on he disane beween he arrier requenies Avoid sudden phase shis b hoosing he requenies suh ha (minimum) requen gap δ = /4T (where T is a bi ime) During T he phase o he signal hanges oninuousl o ± π ϕ* [Hz] R = A os ϕ Eample GSM: GMSK (Gaussian MSK) ampliude domain requen sperum phase sae diagram Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/39 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/40

11 Advaned Phase Shi Keing Modulaion Combinaions BPSK (Binar Phase Shi Keing): bi value 0: sine wave bi value : invered sine wave Robus, low speral eiien Eample: saellie ssems QPSK (Quadraure Phase Shi Keing): 2 bis oded as one smbol smbol deermines shi o sine wave needs less bandwidh ompared o BPSK more omple D (Dierenial ) 0 00 I I 0 R R 0 Quadraure Ampliude Modulaion (QAM) ombines ampliude and phase modulaion i is possible o ode n bis using one smbol 2 n disree levels, n=2 idenial o QPSK bi error rae inreases wih n, bu less errors ompared o omparable PSK shemes Eample: 6-QAM (4 bis = smbol) Smbols 00 and 000 have he same phase, bu dieren ampliude and 000 have dieren phase, bu same ampliude. Used in 9600 bi/s modems I R 000 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/4 Disribued Compuing Group MOBILE COMPUTING R. Waenhoer 2/42