Digital Transmission through the AWGN Channel ECE460 Spring, 2012 Geometric Representation Orthogonal Basis 1. Orthogonalization (Gram-Schmidt) 2. Pulse Amplitude Modulation a. Baseband b. Bandpass c. Geometric Representation 3. 2-D Signals a. Baseband b. Bandpass 1) Carrier Phase Modulation (All have same energy) 1) Phase-Shift Keying 2) Two Quadrature Carriers 2) Quadrature Amplitude Modulation 4. Multidimensional a. Orthogonal 1) Baseband 2) Bandpass b. Biorthogonal 1) Baseband 2) Bandpass 2 Geometric Representation Gram-Schmidt Orthogonalization 1. Begin with first waveform, s1(t) with energy ξ1: 1 t s1 t 1 2. Second waveform a. Determine projection, c21, onto ψ1 c21 s2 t 1 t dt b. Subtract projection from s2(t) d 2 t s2 t c21 1 t c. Normalize 2 t d2 t 2 where 2 d 22 t dt 3. Repeat cki sk t i t dt k 1 d k t sk t cki i t i 1 k t dk t k where k d k2 t dt 3 Example 7.1 4 Pulse Amplitude Modulation Baseband Signals Binary PAM • • Bit 1 – Amplitude + A Bit 0 – Amplitude - A M 2k M-ary PAM M-ary PAM sm t Am gT t m sm2 t dt T A 2 m T g 2 T Binary PAM t dt Am2 g Fixed Rb 1 k Tb kT 5 Pulse Amplitude Modulation Bandpass Signals Baseband Signal sm t X Bandpass Signal sm t cos 2 f ct 2cos 2 f ct um t Am gT t cos 2 f ct m 1, 2, ... , M Um f Am GT f f c GT f f c 2 What type of Amplitude Modulation signal does this appear to be? m 2 u m t dt Am2 t gT2 t cos 2 2 f c t dt Am2 2 Am2 2 gT t dt 2 g t cos 4 f t dt 2 T c 6 PAM Signals Geometric Representation M-ary PAM waveforms are one-dimensional sm t sm t m 1, 2,..., M t where 1 g gT t 0 t T sm g Am d d d m 1, 2,..., M d d d = Euclidean distance between two points 0 7 Two-Dimensional Signal Waveforms Baseband Signals • Are these orthogonal? • Calculate ξ. • Find basis functions of (b). 8 Two-Dimensional Bandpass Signals Carrier-Phase Modulation 1. Given M-two-dimensional signal waveforms sm t , m 1, 2,..., M um t sm t cos 2 f ct 0 t T 2. Constrain bandpass waveforms to have same energy T m um2 t dt 0 T sm2 t cos 2 2 f c t dt 0 T T 1 1 sm2 t dt sm2 t cos 4 f ct dt 20 20 s m 9 Two-Dimensional Bandpass Signals Quadrature Amplitude Modulation um t Amc gT t cos 2 f ct Ams gT t sin 2 f ct m 1, 2,..., M 10 Multidimensional Signal Waveforms Orthogonal Multidimensional means multiple basis vectors Baseband Signals • Overlapping (Hadamard Sequence) • Non-Overlapping o Pulse Position Mod. (PPM) sm t A gT t m 1 T / M where m 1, 2,..., M m 1 T / M t mT / M 11 Multidimensional Signal Waveforms Orthogonal Bandpass Signals As before, we can create bandpass signals by simply multiplying a baseband signal by a sinusoid: um t sm t cos 2 fct 0 t T Carrier-frequency modulation: Frequency-Shift Keying (FSK) 2 b cos 2 f ct 2 m f t m 0,1,..., M , 0 t T T um t mn 1 s T u t u t dt m n 0 sin 2 m n f T 2 m n f T 12 Multidimensional Signal Waveforms Biorthogonal Baseband Begin with M/2 orthogonal vectors in N = M/2 dimensions. , 0, 0,..., 0 0, , 0,..., 0 s1 s2 s s s M /2 0, 0, 0,..., s Then append their negatives sM 2 1 s , 0, 0,..., 0 s M 0, 0, 0,..., s Bandpass As before, multiply the baseband signals by a sinusoid. 13 Multidimensional Signal Waveforms Simplex Subtract the average of M orthogonal waveforms 1 sm t sm t M T M s t k 1 k 1 s sm t dt 1 s M 2 0 In geometric form (e.g., vector) 1 sm s m M M s k 1 k Where the mean-signal vector is 1 s M M s k 1 k Has the effect of moving the origin to s reducing the energy per symbol 2 s sm sm s 2 1 1 M s 14 Multidimensional Signal Waveforms Binary-Coded M binary code words c m cm1 , cm 2 ,..., cmN m 1, 2,..., M where cmj 0 or 1 for all m and j. For example: c1 1 1 1 1 0 c 2 1 1 0 0 1 c3 1 0 1 0 1 c 4 0 1 0 1 0 In vector form: sm sm1 , sm 2 ,..., smN m 1, 2,..., M where smj s / N m and j 15 Optimum Receivers Start with the transmission of any one of the M-ary signal waveforms: g M 2k symbols having k -bits sm t , m 1, 2,..., M g Transmitted within timeslot 0 t T g Corrupted with AWGN: r t sm t n t r t sm t n t Demodulator Detector Sampler r t r r1 , r2 ,..., rN Output Decision r sm t 1. Demodulators a. Correlation-Type b. Matched-Filter-Type 2. Optimum Detector 3. Special Cases (Demodulation and Detection) a. b. c. d. Carrier-Amplitude Modulated Signals Carrier-Phase Modulation Signals Quadrature Amplitude Modulated Signals Frequency-Modulated Signals 16 Demodulators Correlation-Type k 1, 2,..., N rk r t k t dt T 0 sm t n t k t dt 0 T sm t k t dt n t k t dt T T 0 0 smk nk r sm n Next, obtain the joint conditional PDF f r | sm 1 N 0 N /2 1 N 0 N /2 N 2 exp rk smk / N 0 k 1 exp r s m 2 m 1, 2,..., M / N0 17 Demodulators Matched-Filter Type Instead of using a bank of correlators to generate {rk}, use a bank of N linear filters. The Matched Filter Key Property: if a signal s(t) is corrupted by AGWN, the filter with impulse response matched to s(t) maximizes the output SNR Demodulator 18 Optimum Detector Decision based on transmitted signal in each signal interval based on the observation of the vector r. Maximum a Posterior Probabilities (MAP) P signal s m was transmitted | r m 1, 2,..., M P sm | r f r | sm P s m N f r | s P s m 1 m m If equal a priori probabilities, i.e., P sm 1/ M for all M and the denominator is a constant for all M, this reduces to maximizing f r | sm called maximum-likelihood (ML) criterion. N D r, s m rk smk 2 minimum distance detection k 1 D r, s m 2r s m s m C r, s m 2r s m s m 2 2 minimize maximize (correlation metric) 19 Example 7.5.3 Consider the case of binary PAM signals in which two possible signal points are s1 s2 b where b is the energy per bit. The prior probabilities are P s1 p and P s2 1 p. Determine the metrics for the optimum MAP detector when the transmitted signal is corrupted with AWGN. 20