Digital Signal Processing Reference
In-Depth Information
Again, note that Equation (E.13) is equivalent to
H k e j 4 k ¼ H Nk e j 4 Nk ;
1 k M
(E.18)
Substituting (E.17) in (E.18) yields
H k e jðN 1 Þk p =N ¼ H Nk e jðN 1 ÞðNkÞ p =N ;
1 k M
(E.19)
Simplification of Equation (E.19) leads to the following result:
H k ¼ H Nk e jðN 1 Þ p ¼ð 1 Þ N 1
H Nk ;
1 k M
(E.20)
Since we constrain the filter length to be N ¼ 2 M þ 1, Equation (E.20) can be further reduced to
2 M
H k ¼ð 1 Þ
H 2 1 k ¼ H 2 1 k ;
1 k M
(E.21)
Finally, by substituting (E.21) and (E.17) into (E.16) , we obtain a very simple design equation:
(
)
k ¼ 1 H k cos 2 p kðn MÞ
H 0 þ 2 M
1
2 M þ 1
;
0 n 2 M
hðnÞ¼
(E.22)
2 M þ 1
Thus the design procedure is simply summarized as follows: Given the filter length, 2 M þ 1, and the
2 p k
ð 2 1 Þ
specified frequency response, H k at U k ¼
for k ¼ 0 ; 1 ; / ; M , FIR filter coefficients can be
calculated via Equation (E.22) .
 
Search WWH ::




Custom Search