where (u’ = / + u) and u’ > /. Accordingly, if we set 
YS aes Ci = hy’, 
1=0 
Eq. 6.8 will be 
SS Gi ea =) lbp hn @ ey (6.9) 
1=0 u'=0 
Since €, is a pure random process, C; and h, can be estimated by the least squares 
method. 
From Eq. 6.9 
€,* = YY: Gyee »y [pips Lope (6.10) 
1=0 u'=0 
(* = conjugate) 
and 
ioe) io 2) 
Et+s = » C) Vis = i hy’ Xt+s—u'- (6.11) 
1=0 u'=0 
Taking the products of both sides of Eqs. 6.10 and 6.11, and then taking the statistical ex- 
pectation of each term 
R_(8)= > CG; > GC RyG)— », GY) hy. Ra(s—w' +) 
1=0 1=0 1=0 u'=0 
-> C1 >, hy Ry(s—1+u') 
1=0 =u’ =0 
+ > hy >, hy Rus). (6.12) 
u’=0 u'=0 
Then, by the Fourier transform, 
204 
