Digital Signal Processing for Communication Systems

Size: px

Start display at page:

Download "Digital Signal Processing for Communication Systems"

Natalie Garrett
5 years ago
Views:

1 Digital Signal Processing for Communication Systems Prof. YONG HOON LEE DEPARTMENT OF ELECTRICAL ENGINEERING KAIST

2 Contents 1. DSP for TDMA (IS-136) Mobile Communication 2. DSP for CDMA (IS-95) Mobile Communication 3. Miscellaneous PRML Receiver for DVD Software Defined Radio

3 Introduction to IS TDMA systems Access Method : TDMA/FDD Frequency Band Mobile to Basestation : MHz ( 25MHz Band ) Basestation to Mobile : MHz ( 25MHz Band ) Channel Bandwidth : 30 khz 25MHz / 30kHz = 832 channels Modulation : π/4 DQPSK Bit rate : 48.6 kbps Pulse Shaping Filter : Root Raised Cosine (roll off factor : 0.35) slot 1 slot 2 slot 3 slot 4 slot 5 slot ms One time slot have 14-symbol preamble and 148-symbol data

4 Digital Traffic Channel Time Slots (bits ) (bits ) SACCH (slow associated control channel) : for power control & handoff CDVCC (coded digital verification color code) : basestation ID number SYNC (synchronization and training) : for synchronization & training CDL (coded digital control channel locator) : control channel location information RSVD (reserved) G (guard time) : for tolerating inter-slot overlapping R (ramp up time) : for delay when power on

5 Mobile Station Block Diagram FACCH (slow associated control channel) : for power control & handoff DVCC (digital verification color code) : basestation ID number

6 Mobile Station H/W Block Modem Vocoder program Assembler Machine code Air MIC ROM RF Baseband codec DSP chip Voiceband Codec Instruction simulator MIGHTI(KAIST) Speaker Micro Controller Layer1 program Layer2 program Layer3 program 80C196NU(Intel)

7 Baseband Modem Block Diagram Channel coded & Interleaved binary sequence π/4 DQPSK modulator PSF D/A 48.6kbps 24.3kbaud 24.3x4kbaud LPF fading channel π/4 DQPSK demodulator equalizer 24.3kbaud timing recovery PSF 24.3x8kbaud A/D (oversampled by 8)

8 Baseband Equivalent Channel Model Rayleigh fading channel with two ray delay spread model transmitted signal delay ( < T ) X φ X φ + + AWGN, η(t) φ φ

9 Baseband equivalent channel ( a () t = 1, a () t = 1, φ () t = 0and φ () t = 0) flat fading channel frequency selective (two ray with τ=0.5t) T(symbol period) Digital channel Transmitted symbol 3-tap FIR filter Channel output Channel model for MLSE equalization

10 π/4 DQPSK MODEM Differential PSK (DPSK) Let the received signal r(t) be: ( 0 θ + θ)) [ ] rt () = Re gte () j w t k where θ k is the signal phase associated with the k-th symbol and θ is the phase offset. r(t) e jw 0 t LPF e j ( θ θ) k e D * j ( θ 1 θ ) k e j θ θ ( k k 1 ) (Differentially coherent receiver) θ = θ θ + At the modulator, we send where θ is determined depending on the symbol at k.

11 e.g. Binary DPSK Symbol at K 0 1 θ e.g. DQPSK and π/4-dqpsk (for IS-136) [1]-[2] π 0 Symbol at K DQPSK 0 π/2 - π/2 π π/4-dqpsk π/4 3π/4 - π/4-3π/4 DQPSK Binary input π/4-dqpsk DQPSK θ π 0 π/2 π DQPSK θ 0 π π 3π/2 π/2 π/4 θ -3π/4 π/4 3π/4-3π/4 π/4 θ 0-3π/4-2π/4 π/4-2π/4

12 Phase Ambiguity in PSK At the M-PSK receiver, the input given by r( t) = g( t)cos( w t 2πi / M + θ ) = Re j w0t [ ( ) ] 2 πi / M + θ g t e ) ( 0 where g(t) is the signal pulse shape and θ is the phase offset (noise-free case). r(t) LPF π θ Phase ambiguity occurs when θ>π/m. To avoid this, at least one reference symbol is required. Sometimes, DPSK with coherent demodulation is used. In fact, in IS-136 coherent demodulation of π/4 DQPSK is employed to avoid phase ambiguity and to employ MLSE. Given a reference symbol, θ can be estimated. This procedure is called carrier recovery.

13 Pulse shaping filter [3]

14 Raised Cosine Filters Cascade of T x and R x filters having the following raised cosine spectrum. e.g. Symbol (Baud) rate = 30ksymbol/sec 1/T = 30k If the roll-off factor r=0.5 (50% excess BW) then the Tx BW is 3/4 1/T = 3/4 30k = 22.5 khz

15 Digital PSF Digital PSF A/D LPF rate = R rate = NR Digital PSF : Interpolator The effect of ADC (ZOH) should be compensated. Often a digital PSF is a cascade of a raised cosine filter and an equalizer compensating for ADC

16 (β = 0.2)

17 Synchronization Symbol synchronization (Timing recovery) Nonlinear spectral line (NSL), Decision directed (DD) (In DD methods, frame sync. Is assumed) Carrier synchronization (Phase and Frequency recovery) Decision directed is popular Frame synchronization Matched filtering (In CDMA systems, code acquisition for symbol timing recovery is essentially the same as frame sync.) Network synchronization Acquisition and tracking Known data sequence (preamble) is generally given during acquisition.

18 For the IS-136 system Timing recovery : NSL with oversampling by 8. Carrier recovery: Phase offset is compensated by the MLSE. Frequency offset compensation is sometimes required. Frame synchronization : matched filtering with 28 sync. Bits.

19 Timing recovery d(nt) π/4 DQPSK modulator output PSF D/A fading channel A/D (oversampled) PSF kt r(nt/k) r(nt) timing recovery Nonlinear spectral line (NSL) method [4] r(nt/k) 2 g(nt/k) narrow BPF center freq. = 1/T f(nt/k) Peak detector

20 Baseband equivalent channel oversampled by 32 (k=32) flat fading channel frequency selective (two ray with τ=0.5t) T(symbol period) NSL for BPSK in flat fading NSL for π/4 DPSK in flat fading channel (IS-136 preamble) received data, r(nt/k) square output, g(nt/k) T(symbol period) 1 received data, r(nt/k) BPF output, f(nt/k) T(symbol period)

21 NSL for BPSK in two ray Rayleigh fading NSL for π/4 DQPSK in two ray Rayleigh fading channel (IS-136 preamble) received data, r(nt/k) square output, g(nt/k) T(symbol period) 1.5 received data, r(nt/k) BPF output, f(nt/k) T(symbol period) - Due to the fading, there are no ISI-free positions. Equalization for compensating ISI is essential.

22 Carrier frequency offset compensation (carrier recovery) d(nt) π/4 DQPSK modulator output Digital channel X r(nt) e j2π fnt Carrier offset estimation X e j2π fnt ü Carrier frequency offset estimation with the help of preamble sequence. principle of carrier offset estimation for AWGN [6] r(nt) X γ (nt) X ρ(nt) 1 2πT arg( ) ü f 1/d(nT) T * j2π fnt j2π fnt r(nt) = d(nt) e + η( nt) e π η( nt) γ ( nt) = e + e d(nt) j2 fnt j2π fnt j2π ft ρ( nt) = e + noise related term( ϕ( nt))

23 Equalization and Demodulation Use of the differential decoder (detector) before equalization is not recommended, because the differential decoder introduces nonlinearities to the channel. Differential decoder (detector) r(n) Equalization X T * MLSE (maximum likelihood sequence estimation) outperforms the linear and decision-feedback equalizers which are associated with symbol-by-symbol detectors.

24 MLSE via Viterbi Algorithm (VA) [5] MLSE via VA ü Channel estimation SYNC preamble For M-ary symbol and L-tap channel, the number of states for VA is M. In π/4 DQPSK, signal constellation at even and odd time instants are different, but at a given time M=4. Therefore, we define 16 states assuming L=3. In this case, however, the state vectors at even and odd time instants should be different. L- The channel is estimated from the SYNC Preamble by using the RLS algorithm.

25 e.g. An MLSE equalizer with 4 states (M=2, L=3) State 0 (-1-1) T r() 1 ( 00,) ü T r( 2) ( 00, ) ü T r() 3 (,) 00 ü State 1 (-1 1) T r() 1 ( 01,) ü T r() 3 (,) 20 ü State 2 ( 1-1) T r( 2) ( 12, ) ü State 3 ( 1 1) State 0 State State 0 State 1 ( jk, ) : Hypothesized input vector when the transition from the state j to the state k occurs. T T T e.g. (0,0) = [-1,-1,-1], (0,1) = [-1,-1,1], (3,2) = [1,1,-1]

26 State vectors for π/4 DQPSK (M=4, L=3) 0 1/4 1/ /4 1/ /4 1/ /4 1/4 1/2 0 3/4 1/4 1/2 0 3/4 1/4 1/2 0 5/4 1/ /4 1/ /4 1/ /4 1/4 3/2 0 7/4 1/4 3/2 0 7/4 1/4 3/2 1/2 1/4 3/4 0 1/2 1/4 3/4 0 1/2 1/4 3/4 0 1/2 3/4 3/4 1/2 1/2 3/4 3/4 1/2 1/2 3/4 3/4 1/2 1/2 5/4 3/4 1 1/2 5/4 3/4 1 1/2 5/4 3/4 1 1/2 7/4 3/4 3/2 1/2 7/4 3/4 3/2 1/2 7/4 3/4 3/2 1 1/4 5/ /4 5/ /4 5/ /4 5/4 1/2 1 3/4 5/4 1/2 1 3/4 5/4 1/2 1 5/4 5/ /4 5/ /4 5/ /4 5/4 3/2 1 7/4 5/4 3/2 1 7/4 5/4 3/2 3/2 1/4 7/4 0 3/2 1/4 7/4 0 3/2 1/4 7/4 0 3/2 3/4 7/4 1/2 3/2 3/4 7/4 1/2 3/2 3/4 7/4 1/2 3/2 5/4 7/4 1 3/2 5/4 7/4 1 3/2 5/4 7/4 1 3/2 7/4 7/4 3/2 3/2 7/4 7/4 3/2 3/2 7/4 7/4 3/2 Time n=0 n=1 n=2 n=3 n=4 n=5

Design of a Transceiver for 3G DECT Physical Layer. - Rohit Budhiraja

Design of a Transceiver for 3G DECT Physical Layer - Rohit Budhiraja The Big Picture 2G DECT Binary GFSK 1.152Mbps 3G DECT M-ary DPSK 3.456 Mbps DECT - Digital Enhanced Cordless Telecommunications Overview