Signal Processing and Linear Systems-B.P.Lathi copy

Numerical convolution c an also b e p erformed from t

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: s e jrrlok for r = 0, 1, 2, . .. , No - 1. T he F ourier series for a n No-periodic signal f[k] c onsists o f o nly these No h armonics, a nd c an b e expressed as i:a. N o-I f[k] = L n _ 21r "0-- v rejrrlok No r=O (lOA) T o c ompute coefficients V r i n t he F ourier series ( lOA), we multiply b oth sides of (1004) by e -jmrlok a nd s um over k from k = 0 t o (No - 1). N o-I L N o-I N o-I f[k]e-jmrlok = k =O LL k=O V rej(r-m)rlok (10.5) r =O T he r ight-hand sum, after interchanging t he o rder o f s ummation, r esults in N o-I ~ Vr [ NO-I ~ e j(r-m)rlok (10.6) T he i nner s um, according t o E q. (5043), is zero for all values of r t m . I t is nonzero w ith a value No only when r = m . T his fact m eans t he o utside s um h as only one t erm V mN o ( corresponding t o r = m ). T herefore, t he r ight-hand side o f E q. (10.5) i s e qual t o VmNO, a nd N o-I L =- 21r N o-I L f [k]e-jrrlok 1 0.1-1 Fourier Spectra o f a Periodic Signal f[k] T he F ourier series consists of No c omponents T he frequencies o f t hese c omponents a re 0, n o, 2 n o, . .. , (No - l)no w here n o = 21r/No. T he a mount o f t he r th h armonic is V r. We c an p lot t his a mount V r ( the F ourier coefficient) as a function o f n . S uch a plot, called t he F ourier s pectrum o f f[k], gives us, a t a glance, t he g raphical p icture o fthe a mounts o f v arious harmonics of f[k]. I n general, t he F ourier coefficients V r a re complex, a nd t hey c an b e r epresented in t he p olar form as (10.10) T he p lot o f IVrl vs. n is called t he a mplitude s pectrum a nd t hat o f L Vr vs. n is called t he a ngle (or phase) s pectrum. T hese two plots t ogether a re t he frequency s pectra o f J[k]. K nowing t hese s pectra, we c an r econstruct o r s ynthesize f[k] according t o E q. (10.8). Therefore, t he F ourier (or frequency) s pectra, which are a n a lternative way of describing a signal f[k], a re i n every way equivalent (in t erms o f t he i nformation) t o t he p lot o f f[k] a s a function o f k. T he F ourier s pectra o f a signal c onstitute t he f requency-domain d escription of f[k]' i n c ontrast t o t he t ime-domain description, where f[k] is specified as a function o f t ime ( k). T he r esults are very similar t o t he r epresentation o f a continuous-time periodic signal by a n e xponential Fourier series except t hat, generally, t he c ontinuous-time signal s pectrum b andwidth is infinite, a nd c onsists of a n infinite n umber o f exponential components (harmonics). T he s pectrum o f t he d iscrete-time periodic signal, in c ontrast, is b andlimited a nd h as a t m ost No c omponents. Periodic Extension o f Fourier Spectrum N o-I L a nd r =O No (10.9) No k =O N ote t hat i f 4>[r] is a n N o-periodic function of r , t hen f[k]e-jmrlok = V mNo ~ =- O bserve t hat D TFS e quations (10.8) a nd (10.9) are identical (within a scaling constant) t o t he D FT e quations (5.18b) a nd ( 5.18a).t Therefore, we c an c ompute t he D TFS coefficients using t he efficient F FT a lgo...
View Full Document

This note was uploaded on 04/14/2013 for the course ENG 350 taught by Professor Bayliss during the Spring '13 term at Northwestern.

Ask a homework question - tutors are online