Continuous Time Signals All signals in nature are in continuous time x(t ) t From Discrete Time to Continuous Time A continuous time signals can be viewed as the limit of a discrete time signal with sampling interval TS 0 x(nTS ) TS 0 x(t ) From Discrete Time FT (DTFT) … We saw the DTFT of a discrete time signal X DTFT ( ) j n x ( nT ) e S n 1 x(nTS ) 2 jn X ( ) e d DTFT F Substitute 2 F 2 F TS and obtain: S TS X DTFT 2 F TS FS / 2 j 2 F nTS x ( nT ) e TS S n 1 x(nTS ) 2 TS X DTFT (2 F TS ) e j 2 F nTS dF 2 FS / 2 … to Continuous Time FT Now take the limit so that TS 0 nTS t discrete time -> cont. time FS sampling freq -> infinity ...T S ...dt sum -> integral Then we obtain the Fourier Transform X ( F ) FT x(t ) j 2Ft x ( t ) e dt x(t ) IFTX ( F ) j 2Ft X ( F ) e dF Fourier Transform We want to represent a signal in terms of its frequency components. Define: Fourier Transform (FT) X ( F ) FT x(t ) j 2Ft x ( t ) e dt x(t ) IFTX ( F ) j 2Ft X ( F ) e dF Example of a Fourier Transform Take a Rectangular Pulse t x (t ) rect T0 1 T0 / 2 T0 / 2 t j 2FT0 / 2 j 2FT0 / 2 e e X ( F ) e j 2Ft dt j 2F T0 / 2 T0 / 2 F sin FT0 T0sinc F F0 Example of a Fourier Transform F X ( F ) T0sinc F0 t x (t ) rect T0 F0 1/ T0 0.1 0.08 T0 0.1sec 0.06 1 T0 / 2 T0 / 2 t 0.04 0.02 0 -0.02 -0.04 -50 -40 -30 -20 -10 0 10 20 30 40 50 F(Hz) F0 10Hz 1 T0 1sec 0.8 1 0.6 T0 / 2 T0 / 2 t 0.4 0.2 0 -0.2 -0.4 -50 F0 1Hz -40 -30 -20 -10 0 F(Hz) 10 20 30 40 50 Properties of the FT: 1. Symmetry If the signal x(t ) is real, then its FT is symmetric as X ( F ) X * ( F ) since X ( F ) x(t )e j 2Ft * j 2Ft dt x(t )e dt X * ( F ) Example: just verify the previous example Symmetry of the FT Magnitude has “even” symmetry | X ( F ) || X ( F ) | F Phase has “odd” symmetry X ( F ) X (F ) F Properties of the FT: 2. Time Shift FTx(t t0 ) e j 2Ft0 X ( F ) since FTx(t t0 ) j 2Ft x ( t t ) e dt 0 j 2Ft0 j 2Ft ' x(t ' )e (let t ' t t0 ) e dt ' In other words a time shift affects the phase, not the magnitude Bandwidth of a Baseband Signal • A Baseband Signal has all frequency components at the low frequencies, around F=0 Hz; • Bandwidth: the frequency interval where most of the frequency components are. | X (F ) | B B F What does it mean? If you take the signal at two different times with t 1/ B t and then x(t ) x(t t ) x(t t ) j 2Ft j 2Ft X ( F ) e e dF B j 2Ft j 2Ft X ( F ) e e dF x(t ) B 1 since | Ft | Bt 1 t t For Example: x(t ) 70 1 | X (F ) | 60 0.5 50 0 40 30 -0.5 20 -1 10 -1.5 0 10 20 30 40 50 60 t (msec) 70 80 90 zoom 100 0 0 0.5 1 1.5 2 2.5 3 3.5 4 F (kHz) B 1kHz 1 / B 1m sec 0.4 0.2 0 0.1m sec -0.2 -0.4 -0.6 31.2 31.4 31.6 31.8 t (msec) 32 32.2 samples spaced by less than 0.1msec are fairly close to each other 4.5 5 F (kHz) Computation of the Fourier Transform • Whatever we do, physical signals are in continuous time and, as we have seen, they are described by the FT; • The FT can be computed in one of two ways: 1. Analytical: if we have an expression of the signal (like in the example); 2. Numerical: by approximation using the Fast Fourier Transform (FFT). Fourier Transform and FFT Consider a signal of a finite duration x(t ), 0 t T0 with Bandwidth B . Then we can approximate, by simple arguments, T0 M 1 0 n 0 X ( F ) FTx(t ) x(t )e j 2Ft dt x(nTS )e j 2FnTS TS 1 (say at least an order of magnitude smaller) B M round(T0 / TS ) where TS Fourier Transform and FFT Using the FFT: • Take an even integer N M . Then compute the N point FFT of the sampled data, padded with zeros: X [k ] FFTx(0),...,x(M 1),0,0,...,0, k 0,..., N 1 • Assign the frequencies: X k X k N T X [ N k ], k ,...,1 S 2 FS N k 0,..., 1 TS X [k ], N 2 FS N negative frequencies positive frequencies Example Take a sinusoid with frequency F0 10kHz and length T0 5m sec Let the sampling frequency be FS 200kHz Example X=fft(x, N); F=(-N/2:N/2 -1)*Fs/N; plot(F,fftshift(20*log10(abs(X)))) 50 | X ( F ) |dB dB 0 -50 -100 -80 -60 -40 -20 0 kHz 20 40 60 80 100 F (kHz) Example (Zoom in at the Peak) Max at F=10kHz | X ( F ) |dB 0 dB -20 -40 -60 Sidelobes due to finite data length -80 7 8 9 10 11 kHz 12 13 14 F (kHz) Complex Signals All signals in nature are real. There is not such as a thing as “complex” signal. However in many cases we are interested in processing and transmitting “pairs” of signals. We can analyze them “as if” they were just one complex signal: x(t ) a(t ) jb(t ) a(t ) x(t ) Complex Signal b(t ) Real Signals j Amplitude Modulation: Real Signal You want to transmit a signal over a medium (air, water, space, cable…). You need to “modulate it” by a carrier frequency: s (t ) xAM (t ) s(t ) cos(2FC t ) 2 2 1.8 cos(2FC t ) 1.6 1.4 1.2 1 1.5 1 0.5 0.8 0 0.6 0.4 -0.5 0.2 0 0 50 100 150 200 250 300 350 400 450 -1 -1.5 0 50 100 150 200 250 300 350 400 450 500 Amplitude Modulation: Complex Signal However most of the times the signal we modulate is Complex x(t ) a(t ) cos(2 FC t ) s(t ) a(t ) jb(t ) b(t ) sin(2 FC t ) Re{.} e j e j 2FC t Notice now that the modulated signal is real and it contains both signals a(t) and b(t). FT of Modulated Signal x (t ) s (t ) See the different steps: x(t ) Re{.} e j e j 2FC t X ( F ) FTx (t ) e s(t )e j j 2FC t j 2Ft e x(t ) Rex (t ) x (t )e FT x (t ) * dt e j s(t )e j 2 ( F FC )t dt e j S ( F FC ) 1 1 x (t ) x * (t ) 2 2 * j 2Ft * j 2Ft dt x (t )e dt X * ( F ) e j S * ( F FC ) FT of Modulated Signal Put things together: x (t ) s (t ) | X (F ) | Re{.} | S (F ) | B x(t ) e j e j 2FC t F X (F ) Usually B FC FC 1 j 1 e S ( F FC ) e j S * ( F FC ) 2 2 FC F