KyungHee University Chapter 4: Bandpass Modulation and Demodulation/Detection Digital Communication 1 -0- Chapter 4 4.1 Why Modulate? Digital Communication 1 KyungHee University -1- Chapter 4 4.1 Why Modulate? KyungHee University Digital modulation : digital symbol : waveform compatible with the characteristic of the channel Why use carrier? ⓐ reduce size of antenna (=3108m/fc) e.g.) fc = 3kHz : antenna span : /4 = 25km fc = 900 MHz : antenna diameter : /4 = 9cm ⓑ frequency-division multiplexing ⓒ minimize the effect of interference : spread spectrum ⓓ place a signal in a frequency band where design requirements are met (e.g.)RF->IF Digital Communication 1 -2- Chapter 4 4.1 Why Modulate? Digital Communication 1 KyungHee University -3- Chapter 4 4.2 Digital Bandpass Modulation Technique KyungHee University General form of a carrier wave s(t ) A(t ) cos (t ) (t ) 0t (t ) s(t ) A(t ) cos[0t (t )] 4.2.1 Phasor Representation of a Sinusoid complex notation of a sinusoidal carrier wave e j0t cos0t j sin 0t Digital Communication 1 -4- Chapter 4 4.2 Digital Bandpass Modulation Technique Analytical form of transmitted waveform j0t e jmt e jmt s (t ) Re e 1 2 2 KyungHee University (cosmt , m 0 ) ( AM ) Analytical representation of narrowband FM(NFM) s(t) Re e j0t 1 e jmt e jmt 2 2 Digital Communication 1 -5- Chapter 4 4.2 Digital Bandpass Modulation Technique KyungHee University 4.2.2 Phase Shift Keying si (t ) i (t ) 2E cos0t i (t ) T 2i M 0 t T i 1 ,..., M i 1,...,M 4.2.3 Frequency Shift Keying si (t ) 2E cos(i t ) T 0 t T i 1 ,..., M 4.2.4 Amplitude Shift Keying si (t ) Digital Communication 1 2Ei (t ) cos(0t ) T -6- 0 t T i 1,...,M Chapter 4 4.2 Digital Bandpass Modulation Technique Digital Communication 1 -7- KyungHee University Chapter 4 4.3 Detection of signals in Gaussian Noise KyungHee University 4.3.1 Decision Regions Two-dimensional signal space (M=2) Detector decides which of the signals s1 or s2 was transmitted, after receiving r =>Minimum-error decision rule chooses the signal class s.t. d r siis minimized distance Decision region Decision rule r Region1 s1 sent r Region2 s2 sent Digital Communication 1 -8- Chapter 4 4.3 Detection of signals in Gaussian Noise KyungHee University 4.3.2 Correlation Receiver Received signal Detection process r (t ) si (t ) n(t ) 0 t T, i 1,...,M Step 1 : Transform the waveform r(t) into a single random variable(R.V.) Z (T ) or R.V .' Zi (T ) (i 1,...,M ) Matched filter (Correlator) maximizes SNR T Z i (T ) r (t ) s i (t ) dt 0 Step 2 : Choose waveform si(t) that has the largest correlation with r(t) Choose the si(t) whose index corresponds to the max Zi(T) Another detection approach (Fig.4.7.(b)) Any signal set expressed in terms of some set of basis functions can be si (t ) (i 1,...,M ) j (t ) ( j 1,..., N ) where N M Digital Communication 1 -9- Chapter 4 4.3 Detection of signals in Gaussian Noise KyungHee University Signal N symbol M Signal N< symbol M Ex) M-ary PSK N=2 Digital Communication 1 - 10 - Chapter 4 4.3 Detection of signals in Gaussian Noise KyungHee University 4.3.2.1 Binary Detection Threshold Decision stage : choose the signal best matched to the coefficients aij (with the set of output Zj(T)) Digital Communication 1 - 11 - Chapter 4 4.3 Detection of signals in Gaussian Noise KyungHee University Two conditional pdfs : likelihood of s1(s2) p( z ) Digital Communication 1 1 p( z | s1 ) p( z | s2 ) 2 - 12 - Chapter 4 4.3 Detection of signals in Gaussian Noise KyungHee University Minimum error criterion for equally likely binary signals corrupted by Gaussian noise For antipodal signals, or decide s1 (t ) if z1 (T ) z2 (T ) s2 (t ) otherwise Digital Communication 1 - 13 - Chapter 4 4.4 Coherent Detection KyungHee University 4.4.1Coherent Detection of PSK(BPSK) Coherent detector BPSK example s1 (t ) 2E cos(0 t ) 0 t T T s2 (t ) 2E cos(0 t ) T 2E cos(0 t ) 0 t T T ( E : signalenergy per sym bol) Orthonormal basis function 2 cos(0t ) 0 t T T si (t ) ai1 1 (t ) 1 (t ) s1 (t ) a11 1 (t ) E 1 (t ) s2 (t ) a21 1 (t ) E 1 (t ) Digital Communication 1 - 14 - Chapter 4 4.4 Coherent Detection KyungHee University 4.4.1Coherent Detection of PSK(BPSK) When s1(t) is transmitted, the expected values of product integrator T E z1 | s1 E E 12 (t ) n(t ) 1 (t ) dt 0 T 2 E z2 | s1 E E 12 (t ) n(t ) 1 (t ) dt then, 1 (t ) cos 0t T 0 T 2 E z1 | s1 E E cos 2 0t n(t ) cos 0t ) dt E 0 T T 2 E z2 | s1 E E cos 2 0t n(t ) cos 0t )dt E 0 T Decision stage Choose the signal with largest value of zi(T) Digital Communication 1 - 15 - Chapter 4 4.4.2 Sampled Matched Filter KyungHee University Example 4.1 Sampled Matched Filter Consider the BPSK waveform set s1 (t ) cost and s2 (t ) cost Illustrate how a sampled matched filter or correlator can be used to detect a received signal, say s1(t), from the BPSK Waveform set, in the absence of noise. (e.g. 21000, Ts 0.25m sec, T 1m sec) Sampled MF (N samples per symbol) Digital Communication 1 - 16 - Chapter 4 4.4 Coherent Detection KyungHee University Ex) Sampled MF (4 samples per symbol) sampled s1 z1 (k 3) 2 s1 (T t ) sampled s2 3 z2 [k 3] s1[3 n]c2 [n] z2 (k 3) 2 n 0 Digital Communication 1 - 17 - Chapter 4 4.4 Coherent Detection KyungHee University 4.4.3 Coherent Detection of Multiple Phase Shift Keying Signal space for QPSK(quadri-phase shift keying), M=4 (N=2) For typical coherent MPSK system, si (t ) Orthonormal basis function 2E 2i cos(0t ) 0 t T , i 1,...,M T M Digital Communication 1 - 18 - 1 (t ) 2 cos0t , 2 (t ) T 2 sin 0t T Chapter 4 4.4.3 Coherent Detection of Multiple PSK Signal can be written as Received signal si (t ) ai1 1 (t ) ai 2 2 (t ) 2i 2i ) 1 (t ) E sin( ) 2 (t ) M M (0 t T , i 1,...,M ) E cos( Demodulator i 1 T 1 (t ) i7 Digital Communication 1 T 0 i 8 i5 r (t ) si (t ) n(t ) upper correlator : X r (t ) 1 (t )dt 2 (t ) M 8 KyungHee University lower correlator : Y r (t ) 2 (t )dt 0 ˆ arct an(Y X ) decision - 19 - i Chapter 4 Demodulator of multiple-PSK KyungHee University T upper correlator : X r (t ) 1 (t )dt 0 T lower correlator : Y r (t ) 2 (t )dt 0 ˆ arct an(Y X ) Digital Communication 1 - 20 - Chapter 4 4.4.4 Coherent Detection of FSK KyungHee University Typical set of FSK signal waveforms si (t ) 2E cos(i t ) 0 t T T where i 1,..., M Orthonormal set j (t ) E cos j t ( j 1,...,N ) T T aij (t ) 0 2E 2 cosi t cos j tdt T T aij (t ) E for i j Distance between any two prototype signal vectors is constant d ( si , s j ) si s j 2 E for i j 0 otherwise The ith prptotype signal vector is located on the ith coordinated axis a displacement from origin E Digital Communication 1 - 21 - Chapter 4 KyungHee University 4.4 Coherent Detection Example:3-ary FSK signal i 2i M Digital Communication 1 - 22 - Chapter 4 4.5 Non-coherent Detection KyungHee University 4.5.1 Detection of Differential PSK Non-coherent detection : actual value of the phase of the incoming signal is not required Tx signal: si (t ) 2E cos[0t i (t )] T Rx signal: r (t ) 2E cos[0t i (t ) ] n(t ) (0 t T , i 1,...,M ) T • For coherent detection, MF is used • For non-coherent detection, this is not possible because MF output is a function of unknown angle α Digital Communication 1 - 23 - Chapter 4 4.5.1 Detection of Differential PSK KyungHee University Differential encoding : information is carried by the difference in phase between two successive waveforms. To sent the i-th message (i=0,…,M), the present signal must have its phase advanced by over the previous signal i 2i M Differential coherent detection : non-coherent because it does not require a reference in phase with received carrier Assuming that αvaries slowly relative to 2T, phase difference is independent of α as [k (T2 ) ] [ j (T1 ) ] k (T2 ) j (T1 ) i (T2 ) Digital Communication 1 - 24 - Chapter 4 4.5.1 Detection of Differential PSK KyungHee University DPSK Vs. PSK DPSK : 3dB worse than PSK PSK compares signal with clean reference DPSK compares two noisy signals, reducing complexity Digital Communication 1 - 25 - Chapter 4 4.5.2 Binary Differential PSK Example KyungHee University c(k ) c(k 1) m(k ) or c(k ) c(k 1) m(k ) (used here) Sample index k Original message encoder Differential message Correspondng phase 1 Arbitrary setting Digital Communication 1 decoder - 26 - Chapter 4 4.5 .3 Non-coherent Detection of Binary Differential FSK KyungHee University • Just an energy detector without phase measurement • Twice as many channel branches • Quadrature receiver Digital Communication 1 - 27 - Chapter 4 4.5 .3 Non-coherent Detection of Binary Differential FSK KyungHee University Three different cases : 1) r (t ) cos1t n(t ) 2) r (t ) sin 1t n(t ) 3) r (t ) cos(1t ) n(t ) Another implementation for non-coherent FSK detection Envelop detector : rectifier and LPF Looks simpler, but (analog) filter require more complexity Digital Communication 1 - 28 - Chapter 4 4.5.4 Required Tone Spacing for Non-coherent Orthogonal FSK Signaling KyungHee University In order for the signal set to be orthogonal, any pair of adjacent tones must have a frequency separation of a multiple of 1/T[Hz] cf) Nyquist filter si (t ) (cos 2 f i t )rect ( t ) T 1 for T t T 2 2 where rect ( t ) T 0 for t T 2 Fourier transform F{si (t )} Tsinc( f f i )T Digital Communication 1 - 29 - Minimum tone separation:1/T[Hz] Chapter 4 4.5 Non-coherent Detection : Example 4.3 cos(2 f1t ) cos 2 f 2t ⊙ Non-coherent FSK signal : T 0 cos(2 f1t ) cos 2 f 2tdt 0 for orthogonality KyungHee University where f1 f 2 f1 f 2 1 T e.g. two tones f1 10, 000 Hz and f 2 11, 000 Hz orthogonal ? if rate 1, 000 symvols / s, then orthogonal if rate 1, 000 symvols / s, then not orthogonal ⊙ Coherent FSK signal : T 0 0 f1 f 2 1 2T cos(2 f1t ) cos 2 f 2tdt T T 0 0 cos cos 2 f1t cos 2 f 2tdt sin sin 2 f1t cos 2 f 2tdt cos sin 2 ( f1 f 2 )T cos 2 ( f1 f 2 )T 1 sin 2 ( f1 f 2 ) 2 ( f1 f 2 ) sin 2 ( f1 f 2 )T cos 2 ( f1 f 2 )T 0 2 ( f1 f 2 ) 2 ( f1 f 2 ) since Digital Communication 1 Non-coherent이면 둘 다 0이어야 함 - 30 - f1 f 2 0 Chapter 4 4.7 Error Performance for Binary Systems KyungHee University 4.7.1Probability of Bit Error for Coherently Detected BPSK Antipodal signals s1 (t ) 2E cos(0t ) T s2 (t ) 2E cos(0t ) T s (t ) a11 1 (t ) E 1 (t ) 1 s2 (t ) a21 1 (t ) E 1 (t ) 2E cos(0t ) 0 t T T Basis function 1 (t ) Decision rule is 2 cos0t for 0 t T T Digital Communication 1 - 31 - s1 (t ) if z (T ) 0 0 s2 (t ) otherwise Chapter 4 4.7 Error Performnace for Binary Systems 1 PB P( H 2 | s1 ) P( s1 ) P( H1 | s2 ) P( s2 ) P( s1 ) P ( s2 ) 2 1 1 PB P( H 2 | s1 ) P( H1 | s2 ) 2 2 The same PB P( H 2 | s1 ) P( H1 | s2 ) sym m etry of pdf priori 0 a probability P( z | s2 )dz PB KyungHee University ( a1 a2 ) 2 2 1 1 z a2 dz exp 2 0 ( a1 a2 ) 0 2 u u a1 a2 2 1 Q( X ) 2 Digital Communication 1 - 32 - u2 z a2 x exp 2 du, u 0 Chapter 4 4.7 Error Performnace for Binary Systems u PB u a1 a2 u2 1 exp du Q 2 2 2 0 a1 Eb , a2 Eb , Eb : signal energy per binary sym bol ( a1 a2 ) 2 Since 02 N0 2 (n(t ) : white noise with PSD PB KyungHee University 2 Eb N 0 N0 2 Rn ( ) 2 Eb u2 1 exp du Q 2 2 N0 Digital Communication 1 - 33 - N0 2 ( )) Chapter 4 4.7 Error Performance for Binary Systems KyungHee University Another approach (1) Digital Communication 1 - 34 - Chapter 4 4.7 Error Performance for Binary Systems KyungHee University Another approach (2) 2 BPSK s2 Ed 4Eb Eb BFSK s2 s1 Eb Eb Digital Communication 1 Ed ( 2Eb ) 2 s1 Eb - 35 - 1 Chapter 4 4.7 Error Performance for Binary Systems KyungHee University Probability of bit error for several types of binary systems TABLE 4.1 Probability of Error for Selected Binary Modulation Schemes Modulation PSK(coherent) DPSK (dfferentially coherent) Orthogonal FSK (coherent) Orthogonal FSK (noncoherent) Digital Communication 1 PB 2 Eb Q N0 E 1 exp b 2 N0 Q Eb N0 E 1 exp b 2 2 N0 - 36 - Chapter 4 4.8 M-ary Signaling and Performance KyungHee University 4.8.2 M-ary Signaling(M=2k k:bits, M=# of waveforms) ① M-ary orthogonal k↑ BER↑ BW↑ M-ary PSK k↑ BER↑ Shannon Limit -1.6dB k=∞ same BW ② (R, Eb/No, BER, BW) : fundamental “trade-off” Digital Communication 1 - 37 - Chapter 4 4.8.3 Vectorial View of MPSK Signaling KyungHee University ① (M=2k↑, the same Eb/No) bandwidth efficiency (R/W) ↑, PB ↑ ② (M=2k↑, the same PB) bandwidth efficiency (R/W) ↑, Eb/No ↑ Digital Communication 1 - 38 - Chapter 4 4.8.4 BPSK and QPSK : the same bit error probability General relationship BPSK Eb S / 2 2W N0 N R QPSK S W NR Digital Communication 1 KyungHee University QPSK = two orthogonal BPSK channel (I stream, Q stream) Magnitude (A) I stream ( A/root(2) ) Q stream ( A/root(2) ) Power/bit Half Half Bit rate Half Half - 39 - Chapter 4 S REb 4.8 M-ary Signaling and Performance If original QPSK is given by R[bps], S[watt], each BPSK: N WN0 KyungHee University Eb S 2 W S 1 N0 N 0 R 2 N 0 R Same BER, BW efficiency : BPSK=1,QPSK=2[bit/s/Hz] Eb/N0 vs. SNR Eb N 0 ( Normalized SNR ):the most meaningful way of comparing one digital system with another Eb E Eb S W S WT S WT S 1 b N0 N R N 0 N log 2 M N k N0 N k log 2 M k where W : Detection BW , R : data rate, WT 1(typical ) T T Effect of normalized SNR : noise increases as k increases Digital Communication 1 - 40 - Chapter 4 4.8 M-ary Signaling and Performance KyungHee University Fig. 4.34 : M-ary orthogonal signaling at PE=10-3 in dB(decibel, nonlinear), factor(linear) k=10 (1024-ary symbol), 20SNR(factor)→2SNR per bit(factor); each bit require 2. Eb S [dB] [dB] 10 log k N0 N Eb N [dB] 0 Eb N [dB] 0 Eb [dB] N0 Eb [dB] N0 : k 1( BPSK) S [dB] 3[dB] : k 2(QPSK) N S [dB] 4.77[dB] : k 3(8 PSK) N S [dB] 10[dB] : k 10 N S [dB] N Digital Communication 1 - 41 - Chapter 4 4.9 Symbol Error Performance for M-ary System(M>2) KyungHee University 4.9.4 Bit Error Probability vs. Symbol Error Probability for Multiple Phase Signaling Assume that the symbol(011) is transmitted If an error occur, (010) or (100) is likely→3bit errors Gray code : neighboring symbols differ from one another in only one bit position PE P E ( for PE 1), PE sym bol erroer probability log2 M k Note that PE 1 (1 PB )k . BPSK vs. QPSK PB PE PB for BPSK PE 2 PB for QPSK Digital Communication 1 - 42 - Chapter 4