Beruflich Dokumente
Kultur Dokumente
I. P. KAMINOW
AT&T Bell Laboratories
Crawford Hill Laboratory
Holmdel, NJ 07733, USA
Fig. 3
a Optical clock synchronised to 33rd slot with 2 ns pulses at
10 MHz
b TDMA frame sequence at photodetector output showing 14
interfering ones and data sequence in 33rd slot. Ones in slot 33 are
raised on the clock pedestal, whereas zeros in slot 33 do not appear
on the clock pedestal
c Recovered data (one is low, zero is high) at monostable multivibrator output 10101000100011101010
References
1
fiber optic local area network using optical processing', J. Lightwave Technoi, 1986, 4, pp. 547-554
3
GOTO, H., NAGASH1MA, K., SUZUKI, S., KONDO, M., a n d OHTA, Y.:
(i)
n= 0
1219
where
'*>-
'-h&
(2)
y(t)fft)exp(-t/k)dt
00
m-1
= I , fi
exp (-t/k) dt
;=o J
(3)
j = 0,l,2,...,m-l
(11)
or
00
y(t) ~ aTJ[t)
(4)
,0 %<*. = fJj
(12)
where
<xT = [ a 0 , a l s . . . , a m _ j "
and
fT(t) = IMtlUt),....
(13)
y(t)exp(-t/k)fo(t)dt=Y(l/k)
(5)
y(t)txv(-t/k)fl(t)dt=Y(2/k)
At this point, integrating ft{z)fj(z) from z = 0 to z = 1, one
obtains
(14)
y(t)txp(-t/k)f2(t)dt=Y(3/k)
^fi(z)fJ{z)dz = - ~ ^
(6)
0 < t < oo
(7)
cxp(-t/k)
exp(-2t/k)
fo(t) = 1
=
y(t)txp(-t/k)fJ{t)dt=Y(j+l/k)
(8)
Y(s) = A/s(s + 2)
From eqns. 12 and 14, one constructs a set of simultaneous
linear algebraic equations for m = 3 and k = 1:
a, + (l/3)a2 = 4/3
(l/2)a0 + (1/3)01! + (l/4)a2 = 1/2
(15)
00
(9)
at = 0
a2 = - 2
(10)
1220
Vol. 22 No. 23
(b) Example 2: Consider the following problem, whose transcendental transfer function1 is given as
Y(s) = (l/s)exp(-s/s+l)
Likewise, taking m = 3 and k = 1, the coefficients are found
by using eqns. 12 and 14 as
a o = 0-9211
a t = 0-768
a 2 = 0-21
Exact
Taylor
Chebyshev
(Chou et al.5)
00
0-4
0-8
1-2
1-6
20
0-3678
0-5009
0-6086
0-6948
0-7633
0-8172
0-3631
0-5007
0-6184
0-7089
0-7746
0-8210
0-3828
0-4937
0-6083
0-7032
0-7749
0-8267
Method: A coupler is characterised by the polarisationdependent coupling coefficients Cx = (/Ju /?2i)/2 and Cy =
(fii - 0$i)A 2 ' 5 where ftx (flO and F2l (0>2l) are the propagation constants for the E\i (Eyu) and 21 (21) modes of the
coupler, respectively. Here, modes are designated according to
Goell's method.10 Since a typical fused coupler is operated at
V P I,2 the mode field in the coupler is approximately uniformly polarised.4 This consideration makes it possible to use
the effective-index method7"9 to accurately determine the propagation constants and hence the coupling coefficients of the
coupler, which can be of arbitrary shape.
The basic principle of the effective-index method is to
reduce, approximately, the 2-D wave equation into 1-D wave
equations which can be solved easily.7"9 We consider a fused
coupler with long axis 2a and short axis 2b as shown in
Fig. la. At x = x,, the thickness of the coupler is given by
2t(xi). The propagation constant for the TEn_x
(JMn_^)
mode of the infinite slab of thickness 2t(x), with index nx
sandwiched between two semi-infinite layers of index n2, is
used to define the effective-index function nx(x). The propagation constant for the TMm-x (TE^i) mode of the infinite
slab of thickness 2a, with index profile nx(x), is then the
approximate propagation constant for the * (^) mode of
the coupler. In many cases, nx(x) being analytically determined
from t(x),7-8 we need to analyse only one inhomogeneous slab
to obtain the solution for the 2-D structure. The coupling
References
1
CHEN, c. F., TSAY, Y. T., and wu, T. T.: 'Walsh operational matrices
for fractional calculus and their application to distributed systems',
J. Franklin Inst., 1977, 303, pp. 267-284
Vol. 22
1319/11
No. 23
1221