Extracting analogue signals from noise using a digital computer.
-
Upload
khangminh22 -
Category
Documents
-
view
0 -
download
0
Transcript of Extracting analogue signals from noise using a digital computer.
DUDLEY KNOX LIBRARYNAVAL POSTGRADUATE SCHOOL
LIBRARY MONTEREY, CA 93943-6101
NAVAL POSTGRADUATE SCHOOL
MONTEREY, CALIF. 93940
EXTRACTING ANALOGUE SIGNALS FROM NOISE
USING A DIGITAL, COMPUTER
by
Neil A. J3arrett
Lieutenant, Royal Canadian NavyB. S.E.E. , Nova Scotia Technical College, 1958
Submitted in partial fulfillment
for the degree of
MASTER OF SCIENCE IN ELECTRONICS ENGINEERING
from the
UNITED STATES NAVAL POSTGRADUATE SCHOOLMay 1966
Abstract
It is frequently convenient in data processing to convert analogue
to digital data for computer assimilation. A convenient method of such
conversion has been developed and used in the study of correlation detec-
tion of audio signals corrupted by noise.
A method to use apriori knowledge of the corrupting noise to increase
processing gain has been studied. In the case of detection of a sinusoid in
noise, an additional gain over conventional auto-correlation of up to 14. 5
db has been achieved.
Finally, a signal source located in an unknown random noise field
was detected, classified and located in relative bearing by the cross-cor-
relation of the signals received from two spatially separated sensors.
DUDLEY KNOX LIBRARYNAVAL POSTGRADUATE SCHOOLMONTEREY, CA 93943-5101
TABLE OF CONTENTS
Section Page
1. Introduction 13
2. Statement of the Problem 14
3. Conversion ofAnalogue Signals to Digital Form 17
for Computer Analysis
4. Noise Removal in Auto-Correlation 20
5. Location of a Source by Cross-Correlation 27
6. Conclusion 29
7. Bibliography 31
APPENDIX
A. Analogue to Digital Data Conversion 32
B. Noise Removal in Auto-Correlation Using a Priori 77
Knowledge of Noise
C. Detection and Location of Source in Noise. 130
LIST OF FIGURES
Figure Text Page
1. Flow Diagram of Digitizing Process 19
2. Geometry of Lunch-room Problem 28
Appendix A
1. Block Diagram of Digitizing Process 36
2. Clock. Pulse Development 36
3. The Digitizing Process 37
4. Continuous Sampling Percent vs. Frequency 38
5. 163 Magnetic Tape Capacity vs. Frequency 38
6. Timer Control Word vs. Intersample Delay 39
7. 253 cps. Square Wave Digitally Described 65
8. Digital Auto-Correlation of 253 cps Square Wave 66
9. 253 cps. Triangular Wave Digitally Described 67
10. Digital Auto-Correlation of Triangular Wave 68
11. 253 cps. Sinusoidal Wave Digitally Described 69
12. Digital Auto-Correlation of Sine Wave 70
13. Fluctuations About the Mean of a D.C. Voltage 71
14. Digital Auto-Correlation of D.C. Fluctuation 72
15. Band-limited Noise Digitally Described 73
16. Digital Auto-Correlation of Band-limited Noise 74
17. Noise Generator Output Digitally Described 7 5
18. Digital Auto-Correlation of Noise Generator Output 76
Figure Appendix B, Page
1. F?(oO vs. <* 79
2. Processing Gain vs. (SNR) in 86
3. Processing Advantage of Cross Correlation over 86Auto-correlation vs. (SNR) in
4. Pure Signal Auto-Correlation 112
5. Pure Signal Power Spectral Density, 113
6. Power Spectral Density of Noise, 3750 Samples 114
7. Power Spectral Density of Noise, 18, 750 Samples 115
8. Power Spectral Density of Noise, 37, 500ft Samples H6
9. Auto-Correlation of O DB SNR, 3750 Samples 117
10. Power Spectral Density of ODB SNR After Noise Removal H8
11. ODB Signal After Noise Removal 119
12. Power Spectral Density of -10 DB SNR Before Noise 120Removal. 37 50 Samples
13o Power Spectral Density of -10 DB SNR After Noise Removal 12
1
in Correlation Domain. 3750 Samples
14. Power Spectral Density of -10 DB SNR After Noise Removal 122
in Correlation Domain. 11, 250 Samples
15. Power Spectral Density of -10 DB SNR after Noise Removal 123
in Correlation Domain. 18, 750 Samples
16. -10 DB SNR Singal After Noise Removal in Correlation 124
Domain. 18, 750 Samples
17. -10 DB SNR Auto-Correlation Before Noise Removal. 125
22, 500 Samples
18. Auto-Correlation of Noise. 37, 500 Samples 126
19. Power Spectral Density of -10 DB SNR Before Noise 127
Removal. 22, 500 Samples
Figure Page
20. -10 DB SNR Signal After Noise Removal in Frequency 128
Domain. 22, 500 Samples
21. Power Spectral Density of -10 DB SNR After Noise 129
Removal in Frequency Domain. 22, 500 Samples
Appendix C
1. (a) The Basic D. F. Problem 131
(b) Cross -Correlation of M 1# M, 131
2o Geometry of the Lunch-room Problem 139
3» Cross -Correlation of "O" DB Signal. 3, 950 Samples 140
4. Cross -Correlation of "O" DB Signal. 11, 850 Samples 141
5. Cross -Correlation of "O" DB Signal . 19, 750 Samples 142
6. Auto-Correlation of "O" DB Signah 11, 850 Samples 143
7. Auto-Correlation of Noise. 11, 850 Samples 144
8. Relative Power Spectral Densities of Noise Alone, 145
and Signal Plus Noise
LIST OF TABLES
Table Appendix A Page
I. 160 Program Digitize 49
II Modification to Program Digitize 52
III Program Test 160 53
IV Summary of Test Results 44
Appendix C
I Details of Data Digitized 136
II Frequency Reduction Process 136
LIST OF FORTRAN PROGRAMS
Appendix A Page
1. Subroutine DATA 54
2. Subroutine FINDIT 55
3. Subroutine UNPACK 56
4. Program TEST 58
5. Program SHUFFLE 62
Appendix B
1. Program SIMSIG 90
2. Subroutine PSD 97
3. Program PROCES 100Appendix C
1. Program LOCATE 137
11
1. Introduction .
During the summer of 1965, the author visited the U.S. Naval Elecr,
tronics Laboratory, San Diego, for six weeks. The. experience gained there
suggested several problem areas which would profit from further study.
It appeared that there was a need to analyze large quantities of ana-
logue data collected during various experiments. The volume of the data
almost precluded "hand" analysis and yet no facilities existed to enable a
computer to assimilate the data. This facility was also lacking at the U. S. N.
Postgraduate School Monterey.
The data collected during these experiments at San Diego was often
corrupted by unavoidable locally generated noise. It was felt that if more
could be learned about this self-noise some better method of data extraction
might be possible.
One of the most powerful methods of data analysis is the correlation
technique which may reveal information masked by corrupting noise. These
techniques have been thoroughly investigated over the past 25 years, and a
wealth of literature exists on this subject.
Although continuous time integrations and correlations are possible
mathematically on known explicit functions, physical data often does not
fit a known model. In this case approximations to infinite integrals etc.
,
must be made using either analogue or digital equipment. Analogue equip-
ment tends to be limited in versatility and is expensive, while if a general
purpose digital computer is available, it can be made to simulate the physical
13
situation. Therefore attention was directed toward describing the analogue
data by a digital computer. The use of a digital computer to compute various
correlation functions is the subject of this report.
2. Statement of the Problem.
The correlation function of two continuous time variables is given by
T/2Rs + n<t ) = lim i f sW N (t + T) dt (1)
T-x*>l
-T/2
where s(t), N(t) are defined for -e©<t< oo
When the functional form of s(t) and N(t) are known it may be possible
to evaluate Eq. (l)mathematically. When the form of s(t) and N(t) are known
only as analogue signals Eq.(l) may be instrumented by a multiplier and an
integrator.
The same effect may be achieved by digital means, subject to errors to
be discussed.
The sampling theorem states that a continuous function can be repre-
sented in a finite interval by a finite number of samples of the function. The
rate of sampling must exceed twice the highest frequency component of the
Fourier transform (spectrum) of the function.
By suitably band-limiting a function, samples may be taken at a prac-
tical rate„ The samples if digitally described can then be processed. Equa-
tion (l) implies a memory extending into the infinite past which cannot be
achieved physically. As an approximation the "short-time" correlation may
be formed with some resulting error.
14
If the "short-time" correlation is formed digitally, errors arise from
both the limited number of samples and by the finite accuracy of the indivi-
dual samples. If "short-time" is defined to be some period much longer than
the period of the lowest frequency in the signal, errors of the first kind above
tend to zero as the length of record grows. Measurement error is negligible
if the samples are described to four place accuracy.
The "short-time" correlation is formed as
N-K
Vft (k*T) = TAr^ si
N *+K (2)
i = 1
where
AT - the sampling interval
st , Nl* = sample values of the continuous function iaken at
intervals of T
k = o, 1, 2 K
N = the number of samples available.
In the limit as N goes to»o , andA T ->0
V+ N* (kAT) = RS+N (T) (3)
For finite N
Rs*+N *(k*T) = R&+n(?) (4)
Equation (2) holds only if the data has zero mean. If in Eq. (l) we
assume that s(t) is the desired signal, of zero mean and that N(t) is random
Gaussian noise and that the observed V(t) = s(t) + N(t), then we may form
Ryy(T). This function may in turn be broken down into components:
15
RVvm = RSS^) + %N^ + Wl") (5)
Since the noise and signal are assumed independent the third number
vanishes in the limit T -^°o,
We may approximate Eq. (5) as
Rs *+N *(k*T) = Rg*s*(kAT) + RN *N*(kAT) (6)
in which we wish to determine Rg*g*(k^T).
For very large k = K the second member of Eq. (6) will tend toward
zero because of independence leaving
RS *+N *(K + |at, = RS *S *(Ko +lAT W
J = o, 1, 2, J
This restriction is not desirables since if Rj,j*N*(k4t) were known for
all k, then in Eq. (6)
RS *S *(kAT)= Rs*+N*(kAT) " %*N*(kAT) (8 )
However R^*j^*(k^T) is never known exactly. It can only be known,
since it is a random variable, in an average sense.
If we assume that the normalized correlation function Rj^fcA.T) is
known a priori
RNN(k*T) = ± Rrfrf(kdT) (9)
where R^]^(kAT) is formed in the absence of signal
°^ is to be determined where £_ °< 3r 1
then in Eq. (8) we may write
Rs *s
*(kAT) = Rs *+N *{kAT) - C* RNN (kAT)
Methods of determining <* have been studied for the very limited case
16
of a single sinusoid signal masked by noise.
To gain some experience with the technique of cross -correlation, a
low level signal was placed in a noise field. The output of two spatially
separated sensors was processed to detect and locate the signal. The re-
sults are presented in section 5.
3. Conversion of Analogue Signals to Digital Form for Computer Analysis .
Many analyses of real data may be performed on a digital computer by
first manually sampling the data a sufficient number of times, and entering
the sampled data into the computer.
When the number of data samples desired exceeds a few thousand, some
automatic means of data conversion must be used.
The method should preferably be "off-line, " that is, use auxiliary
equipment, for efficient employment of the main high speed computer. The
Computer Laboratory of the Electrical Engineering Department was equipped
with the necessary items of hardware and these facilities were used to sample
the analog data and form the digital record.
The three main items used were a 12 bit analogue to digital convertor
a Control Data Corporation (C. D. C.) Model 160 Computer having a capacity
of 4096, 12 bit words, and a C. D. C. Model 163 magnetic tape unit.
As now constituted the analogue data is consecutively sampled by the
analogue to digital converter working under control of the CDC 160 computer.
When 4000 samples have been taken, the computer writes the samples
on magnetic tape in a format that is recognized by the main CDC 1604 comf
17
puter. After a delay of 0. 4 second another 4000 samples may be collected,
and so on. Each group of 4000 samples when written on tape is accompanied
by an identifying word so that any block of 4000 samples may later be located.
The maximum sampling rate is 5 Kcs. implying that all signals should be
band-limited to less than 2500 cps. The sampling rate may be varied at will
by manual program entries, as can other convenient parameters.
The system is capable of digitizing only one source at a time. To allow
cross -correlation of two or more sources, a clock pulse input is used to in-
itiate a round of sampling.
At present a simple AND gate is used to sense the presence of a clock
pulse. A more permanent arrangement should make provision for the detection
of a simple clock code, to eliminate the occasional false starts due to noise
pulses, which have occured under present arrangements.
Once the data has been recorded digitally on tape, the data may be
read into the main computer using standard programming techniques. The
subroutines to accomplish this are available as Subroutine Data from the
Computer Facility.
The data when read into the computer is delivered in integer format,
i.e., without a decimal. To convert it to units of volts each data point is
then divided by 409. 6, a factor inherent in the particular A/D converter used.
The diagram below shows the essential steps.
18
SOURCE *\ A/D CONVERTER
CLOCK BIAS
*iCDC 1601* CDC 163
TAPE REELOUTPUT
COMPUTER LABORATORY
TAPE REEL INPUT
CDC 163
EQUIVALENTS/R DATA *—FIND BLOCKUNPACK DATARETURN
MAIN PROGRAM
- CALL DATA-J-REMOVE MEANCONVERT TO VOLTSPROCESS DATA
CDC 1604 COMPUTER
COMPUTER FACILITY
Figure 1. Flow Diagram of Digitizing Process.
The system is presently adequate to describe many physical processes.
Certain signals, such as speech, are beyond the capabilities of the system.
However modifications to the Computer Laboratory hardware to improve per-
formance were not attempted because of pending contractual arrangements to
completely replace the hardware with more sophisticated equipment. The new
equipment will enable two records to be digitized simultaneously. The memory
capacity will be increased by nearly a factor of four, and the sampling rate
will be approximately 20 Kcs. This equipment will.be available for use in
mid 1967.
19
The author has recorded the mixed output of a signal and a noise gen-
erator to provide a vehicle for the study of correlation. Several other students
Q 6j, L 7J have used the system for their own particular needs with no difficulty
experienced to date. Additional details on the digiting process are presented
in Appendix A.
4. Noise Removal in Auto-Correlation .
Correlation techniques have gained considerable favour in recent years
as a method of signal enhancement. Theoretical derivations of the processing
gain to be expected have been stated by Lee£ I J, for both auto and cross-
correlation.
Auto and cross -correlation are defined as follows:
Mr) = lim 4 s T/2
v(t). v(t + r)dt (i)
T^-o -T/2
where V(t) = s(t) + N(t)
s(t) = the signal to be recovered
N(t) = noise,, here assumed Gaussian
i r V1RcCT) = lim -i- J V(t). sMt + rJdt
T->-° -T/2
where
(2)
s*(t) is chosen apriori to have the form of s(t) and is locally
generated
.
Approximations to Eq. (l) and (2) may be generated using sampled
data in which case.
Ra (T)
N /S A±£ ViU). vx a + T) = R
a (r)i = 1
(3)
20
Rc(T") = =£• § Vi (i) - V2 U + T)
.£ ^(T) (4)
i = 1
where Vj = V(t) for t = i^t
It may be shown that Eq. (3) and (4) are of the form
R(£) = Rss(r) + RNdr) + RsN(t) + RNs(r) (5)
Of the four terms in Eq. (5) it is desired to determine only the first,
ARSS (T) which represents the signal contribution. The remaining terms are
undersirabie and success is measured by the extent to which they can be
eliminated.
A AThe cross-terms R
gjyR^ will vanish as N—>«*>, because of the as-
sumed independence of the signal and the noise.
Since the correlation is to be done using a finite number N of samples,
these terms will become small, but may be significant.
Let us assume that a signal is of the form
s(t) = TTE cos wt
N(t) = Gaussian noise of zero mean and variance 6'r,
The variance of the output of Eq. (3) and (4) has been evaluated by
Lee LlL who clearly shows the superiority of cross -correlation over auto
correlation.
In the case of auto-correlation
«l = -fV^'a + <r« (6)
while for cross -correlation
21
This improvement is achieved only when the form of the desired signal
is known apriori. In many cases of interest particularly in the passive sonar
situation, the desired signal form is not known apriori, and consequently
auto-correlation must be used.
In addition to the noise input supplied by the environment, local dis-
turbances may cause noise effects to appear as part of the correlator output.
If these two sources of noise, taken together, are known apriori, and can be
considered to remain stationary over the interval under consideration, then
perhaps some means can be found to eliminate their corrupting effect on the
output of the auto-correlator.
This problem has been studied for many years. In an early work
Lee [_A] states
". . .the correlation of the noise, once measured
can be compensated, (for)"
The details of the implementation of this were not specified.
In a much more recent work Van Trees \_ZJ considers an optimum
receiver in which he considers that the noise is known apriori as a norm-
alized correlation / (T). After correlation of the received data some
constant K times / ( T) is substracted form the correlator output. Deter-
mination of K rests upon knowledge of signal strength and noise strength.
22
Again it is not clear how to instrument the calculation of K.
We will assume that the sample auto-correlation of the signal,
Rs+nC^*) ^as been formed in the presence of corrupting noise. We shall also
assume that the statistics of the corrupting noise are known apriori and are
represented by the normalized correlation function R^N^"^ formed in the
absence of signal.
The object of the processing is to determine a linear combination of
RS+n(T) and R^nM which will enhance the signal and tend to remove the
noise. No claims are made for the optimality of the method presented here,
other than it does improve the signal to noise ratio of the output.
In the following we will be dealing with sampled data functions, not
continuous functions of time. To simplify notation however the symbol T
will be used to denote a sample data function.
The power spectral density P(w), and the auto-correlation functions
are transform pairs, that is
r(7) =J p(w ) e gwT dw (8)
P(w) = \— f R(r) e"Jwr dT (9)
Given one member of the pair, the other may be found. Therefore the
processor could attempt to operate on either the power spectral density or
the correlation function.
From Eq. (5) we get (neglecting cross terms),
23
RSs< T ^ - Ks+nCH-W^) do)
or equivalently by Eq. (9)
Pss (w) = Ps+N (w) - PNN (w) (11)
We have assumed that as apriori information estimates of the norm-
*alized power spectral density function P^^wO and normalized auto-correla-
tion function R^n^ formec* in the absence of signal, are available. At
any given instant there is no assurance that the apriori known noise estimate
represents the true noise in other than an expected value sense.
Then in Eq. (10) and (11) we may write
Rss(t) = Rs+n(t) - <* RnN (T) (12)
Pss (w) = PS+N (w) - °^ PNN(W) (I 3 )
where RNn(^ and PNn(w^ are known functions
O $ ex, ^ I / o< a constant to be determined.
Determination of the <*». in Eq. (12) and (13) becomes the central
problem to be considered.
The following methods of calculating <=K hold only for the case of sin-
usoidal signals in noise. Further study is necessary before any general
method can be stated.
Since the correlation function was produced from the data directly and
the power spectral density indirectly from the data via the correlation, it was
thought advisable to operate first on Eq. (12) rather than Eq. (13).
In the case of a sinusoidal signal, it was thought that finding an
24
^ = °<C' such that RSS(°) > Max RSS('f
)
(14)
would be advisable.
However Eq. (14) is only a necessary not a sufficient condition for
the formation of a valid, physically realizable auto-correlation function.
Thus an instrumentation of Eq. (12) to satisfy Eq. (14) may result in
a correlation function which is the output of a physically unrealizable linear
system. In the time domain there seem to be no simple tests for realizabi-
lity.
Several signals were analyzed using Eq. (12) and (14) at input signal
to noise ratios of and -10 db. No trouble was experienced with the db
case; the signal was almost completely recovered and the noise suppressed.
The -10 db input signal after similar processing was transformed in-fo the
frequency domain. The signal was clearly evidenced by a large spike in the
power spectral density at the signal frequency. The amplitude of the signal
spike after processing showed nearly 6 db greater discrimination between
signal and noise peaks than it did before processing.
The above gain would have been encouraging had not the .power spect-
ral density.' showed some negative contributions.
Negative contributions in a power spectral density, are evidence of, and
in fact are a sufficient condition for, physical unrealizability.
Because of this difficulty no further efforts were made to suitably deter-
mine <X on the basis of the correlation functions. Efforts were then directed
to a determination via Eq. (13).
25
If the signal is a pure sinusoid with no noise present, the power
spectral density shows a delta -function cf (w) at w = wg, and is identically
zero for all other values of w. The perfect processor would give this output
even in the presence of noise.
There are two conditions one may then place on Eq. (13). These are
(a) There must be no negative contributions 0*^ ws= wMax
(b) At least one point must be zero.
This allows an °s to be determined directly in the frequency domain.
Condition;(a) above ensures realizability, while (b) ensures that for any
c*>c\ , condition (a) is violated. On the other hand choice of anc
<* ^ °<c results in the removal of less noise than one might achieve other-
wise, but at least ensures physical realizability.
A trial and error search routine was carried out to find an ©< which
satisfies conditions (a) and (b) above. The search routine is halted when
ano< is found which results in at least one point of Eq. (I3)having a value
£, 0i€ f= .00001, and no negative points. The time to complete a search
is in the order of . 5 seconds, which is small compared to the time taken to
form the functions originally (70 seconds).
Having formed Pss(w) in ^ <3« (13) the results were renormalized to make
the area under the curve unity, and then plotted.
The o< determined by the search was then used in Eq. (12) to formc
RggCf) and the result plotted. No claims for validity are made for this step
except that having formed RgsC") in this manner:its transform had no negative
26
component..
The noise removal proceedure based on an «k determined from the
power spectral density resulted in an additional 14. 5 db processing gain.
This figure was obtained by considering the ratio of energy under the 50 cps.
comb- width to the total energy under all other comb-widths, both before and
after the noise removal. Before noise removal the calculated input signal-
to-noise ratio was -8. 84 db. After noise removal the equivalent calculation
yielded + 5.62 db.
Another measure of effectiveness should be mentioned. In many cases
the ability to distinguish a signal from noise in the spectral density is of
interest. In the power spectrum the ratio of the amplitude due to signal to
that due to a noise "spike", may be expressed in db. In the -8. 84 db
(nominally -10 db) case, the ratio of signal to highest noise peak was 5.62
db. before noise removal, and 9.05 db. after noise removal, or an improve-
ment of 3. 4 db.
Details of the processing are included as Appendix B,
5. Location of a Source by Cross -Correlation .
The bearing of a signal with respect to an abserver may be determined
by measuring the difference in arrival time of the signal at two or more sep-
arated observation stations.
As an experiment a 200 cps. sinusoidal signal generator was set in
one corner of a cafeteria. Two microphones spaced a distance d units apart
were located symmetrically with respect to the center line of the room, at
the far wall of the cafeteria. Interfering noise was supplied by the noon-day
27
clientele.
The maximum level of signal was limited by the intolerance of the din-
ers. This was determined to be 38 milli-watts measured at the loudspeaker
voice -coil. At this level the signal was inaudible at the microphones.
A stereo recording was made of the signal plus noise, and after slow-
ing the tape by a factor of four, selected portions of data were digitized
using techniques detailed in Appendix A. The resulting digital data was
cross -correlated in the CDC 1604 computer and the correlation function plot-
ted.
The geometry of the problem is shown below.
source
/"
Nioise FIELD
70. 7S
130
Figure 2. Geometry of Lunch-room Problem.
Assuming plane wave propagation, and in the absence of noise, we
may associate with A D in Figure 1, an equivalent amount of time; that is,
28
A T = AD/C where C is the velocity of sound in air.
If the two sensors are cross -correlated a maximum in the function will
occur whenever the time variable of correlation is equal to a multiple, of the
period of the signal.
The signal was located successfully to an accuracy of + 0. 17°. The
frequency of the observed correlation function was nearly 150 cps. , or three
times the expected value of 50 cps. This would indicate that detection was
accomplished on the third harmonic of the source oscillator.
A study of the noise characteristics supplied by the diners was not
made.
Provided that accurate time synchronization between the two data
sources to be correlated can be maintained during the digitizing process,
correlation will provide both bearing and frequency information.
Details of the problem are presented as Appendix C.
6. Conclusion.
A convenient method of digitizing analogue data has been developed
and this digital data used to detect signals masked by noise.
The digitizing system is now operational.
Many interesting areas of research remain to be pursued in the field
of recovery of noisy signals. In particular additional effort is needed in the
area of removal of apriori known noise effects.
The facilities to perform the necessary calculation on a digital computer
are now available at the U.S. Naval Postgraduate School.
29
ACKNOWLEGEMENT
The author extends his thanks to Dr. G. Myers of the Electrical
Engineering Department for his thoughtful guidance and help.
Dr. L. V. Schmidt of the Aeronautical Engineering Department has been
vitally interested in the development of the digitizing process and has lent
his time and his energy to help accomplish this goal.
Mr. E. Jangen of the Programming Staff of the Computer Facility has
been extemely helpful with details of programming and has been of great
assistance.
Finally, I would like to thank the Staff of the U. S. Naval Electronics
Laboratory, San Diego, in particular Mr. W. Littrell and his staff, for the
courtesies extended to me during my visit of June 1965. The experience
gained there largely motivated this work.
30
BIBLIOGRAPHY
1. Lee, Y. W. , Statistical Theory of Communication , John Wiley andSons, New York, London, 1960.
2. Lindgen, B. W. , and McElrath, G. W. , Introduction to Probability
and Statistics , MacMillan Co„ , New York, 1959.
3. Van Trees, H. L. , "Optimum Signal Design and Processing for
Reverberation - Limited Environments ", A. D. Little Inc. Report
No. 1501064, Oct. 1964 a p. 7.
4. Lee, Y. W. , Cheathom, and Wiesner. "The Application of Correlation
Functions in the Detection of Small Signals in Noise " , Mass
.
Inst, of Tech./ Research Lab of Electronics, Technical Report 141,
Oct. 1964. p 7.
5. Fano, R. M. , On the Slgnal-to-Noise Ratio in Correlation Detectors,Mass. Inst of Tech. , Technical Report No. 186, February, 1951.
6. Pope, J. W. , Real Time Estimation and Random Signal Detection ,
Ph. D. Thesis, U.S. Naval Postgraduate School, Monterey, May,1966.
7. Jesberg, R. H. , Data Handling System in Support of Antenna Vibration
Project, M.S. Thesis, U.S. Naval Postgraduate School, Monterey,May, 1966.
31
APPENDIX A
ANALOGUE TO DIGITAL DATA CONVERSION
1. INTRODUCTION.
The process of translating real, analogue data into computer digitized
form has been accomplished in two separate stages.
In the first stage, the analogue data on magnetic tape is sampled at
discrete points of time. These samples are digitized and written in bit form
on magnetic tape.
In the second stage, the digital magnetic tape is loaded on the CDC
1604 peripheral tape units and then read into the computer
Earlier work at USNPG School in computer analysis of analogue data
utilized the "on line" approach, in which the data was fed to the main com-
puter directly from the digitizing computer. This mode of operation is now
impractical because of high computer facility utilization.
The method detailed here results in data being reduced to a form suit-
able for submission as a standard program input.
2. DIGITIZING THE DATA.
Basically the method takes samples until the Core storage of the dig-
itizing computer is full, and then writes this information on magnetic tape
in digital form.
The digitizing computer is a CDC 160 located in the Computer Labor-
otory of the Electrical Engineering Department.
This computer has a 4096, 12 bit memory length, and has the ability
to select various external equipments through the proper external function
32
codes. The lower 4000 cells of the computer were reserved for data storage
and the top 96 cells for programming. Thus the core storage available sets
one quite restrictive limitation on the system. In practice however, 4000
samples were adequate to describe many common processes.
The block diagram of the system is shown in Figure 1. The 160 selects
either the A/D converter, for data input, or the 163 tape unit for data output.
When the A/D converter is selected (EXT 14xx) and the INA (IN put to A) in-
struction is executed, the input analogue signal is sampled, and a 12 bit
number representing amplitude is sent to the 160 and stored. Repetition of
the process in quick succession allows blocks of up to 4000 samples to be
stored. The maximum sampling rate achievable is 5000 samples/second.
To control exactly the time at which samples are taken, the execution
of the INA instruction is prevented until two events have occurred; first, an
enabling switch must be closed to ground (EXF 2410), and second, a pulse
(see Figure 2) must be present at the base of the AND Gate* When these
conditions have been satisfied the flip-flop goes to the "1" state placing -3
volts on the INA line and thus executing the INA instruction.
Further samples may now be taken as the flip-flop will stay in the "1"
state until EXF 2400 is called, resetting FFI to the "0" state. The complete
process is illustrated in Figure 3.
The usual mode of operation is to have each block initiated by a pulse,
with the intersample delay under computer control. Under certain circumstan-
ces, it may be desired to have only the runs initiated by a pulse, with both
33
the block count and the sample count under computer control.
The inter-block timing in this mode is not accurate enough for cross-
correlation. The basic program shown in Table 1 may be modified using
Table 2 to give various modes of operation.
Operating Instructions -Digitize Eq. (3).
Features of Digitize Eg. (3) .
1. Manual Entries
a. 0066 - intersample delay. See Figure 6
b. 0067 - any identifier you wish
c. 0070 - initial run number
d. 0071 - number of runs desired
e. 0072 - number of samples/block £ 7640g
= (4000)1Q
f. 0073 - number of blocks/run
Note that (d) x (f) £ 7608
2. Main Points
a. A block of n, n <- 4000 -,q samples is taken each time a clock
pulse is present and EXF 2410 is made.
b. The inter-sample spacing interval is program controlled and is
variable from approx. 190 Msec, minimum to about 50 7nsec. maxi-
mum.
c. The first four (4) words written on the 163, of each block, are
identifying words in the following order:
(l). Run number.
34
(2). Spare (may be preset to anything)
(3). Block number
(4). Number blocks/run
d. The identifying information when digitizing data must be recorded,
as 1604 subroutine DATA uses this information to locate blocks of
data on digital tape.
e. At the end of each run a check is made to ensure that the storage
capacity of a full roll of 163 tape is not being exceeded. If it is,
an End of File is written and the program halts with the total numb-
er of blocks written in the A register.
f . Program may be stopped at any point to allow cueing data tape,
provided that it is not cleared. Progress resumes when 160 placed
in run condition.
g. Analogue, input is at A/D Channel 1. A voltage reference is est-
ablished by an external source at about -5 volts. Correct opera-
tion is assured if the proper test pattern is displayed. See in-
structions for Program Test 160.
h. A voltage of V volts at A/D Channel 1 is converted to a digital
reading by the A/D converter, equal to -409. 6 x V
3. Operating Instructions
a. Load from cell 0000
b. Run from cell 0000
c. When digitizing is complete program stops at 0124 with current
run number in A register.
35
Signal.
Source*
i,@
G>;N££fiTcR
Monitor!A/0
j
Converterj '{Za
"
|1401 TNA r*—A
IXMrt ENASlCO BY V = -3\/
©O I
R S
\ H MAtviOfiL. Ctocic
*4GO X24IO
CLEAR j SST<rYr
?
1
Tvv
riCiURtr i - Dlock Diagram of Digitizing hardw^r*.
iat ICO fAS
.^nifvr clock- puu^e toTAPS lR£CCKO£R
OUTPUT CLOCK POL.5CFROU W£" 8CCOR.O£«,
/ ' AFTER SATURATION/M-fPLf <
r«CfiX*o«vl
_^An"£R REfit-rJC'RflTlOW
r iSuftt, 2,-* C lock puL^e DE'veLOPg'Meivr.
36
_____ '
<• ] *1 ""
i
~~p_Z —^/Vy/VvjA^1
- ... ... v •.._- .-. », • ,^,.ls,v-v.^w-.'-»>;x* ~?> DATA Tr^CK I (A}
—^^-4^/vA^y^A-— 4 - - ; ,..<V'-Vv,^-^~^sv--\M/.~^Hs>^j^v,«' /^ £?AT"A TRACK i" £S}^^
N PATA TRf.CK (vi
^ CLOCK TRflCKA, Bx. X
Iti
I
c
ExFfcNDEO Analogue- Tape Section
-4-
A /D\ 6
1 6 8
i
tt-to=At
.At ;> o.3 sec.
DIGITAL TAPE
1
w/yA,//
* •.>. K *
>. /• x >
x -A a. y
X V Ik s
K / X X.
.
v\\v
\ ^\\B, >sS
SSNNN
O O O o
o o o o
° p. °
o O 2. o
o © o
C o
«
t! t'
J 4oco L
i ROGram Shuffle* !60^r
>—
>
5
Re- ordfred Digital Tape:x .-
iTOW\\y!
*. x * X
\
,\B ' I A
X X
X.
2. K
_
) X X A X
] > XXX
c c. o o
Dg. o
« o o O
£• <J
S
F IGURE 3 - he Digitizing Process
37
MAX
.£50 .SCO .750 IKC 1-25 l.S I.73T £.0 £..2.5 LSSHANNOM FRE'OUej^Y KcS.
aLNTeR-SAttPLt D£L/*y * K»UM6eft ^rtMPU*/ Block. *- .35" S5C.
Figure- 4-
ATO
60
50
i
30 4
20t
Tma*
10 4-
-J 1 ( 1 4 1 I 1 1 ' > >' ,S UO »'^ 2-0 2«S
SHANNON FREQUENCY Kw,
(63 Magnetictape capacity vs. FasQueNcy
• 5^
38
d. Clear and run from cell 0127 to write E. O.F.
4. Modification-'sum
a. See Table II for various possible modifications.
The limitations of the sampling process are graphed in Figures 4 and 5.
Figure 4 shows the effect of sampling speed on the percentage of the data
that may be digitized. The speed of the data gathering is a variable under
program control, while the speed of output is not. Thus at slower sampling
rates, the sampling approaches continuous sampling, and the amount of the
real time data which may be represented on tape, increases, as seen in
Figure 5.
Figure 6 shows the relationship between the number entered as inter-
sample delay timing word, and the resultant inter-sample delay.
The development of a clock pulse from initial, recording to output at
the flip-flop is shown in Figure 2. Experiment has shown that a total delay
of 100 micro-seconds exists between tQ , the time of initiating the pulse on
record and t^, the time the pulse is delivered. This, period remains station-
ary over a large range of level settings, pulse widths and amplifier gain
settings . Thus the clock pulse may be used to synchronize two pieees-of data,
as in cross -correlation.
Experience has shown that since only the first of a series of adjacent
clock pulses in used to initiate sampling, the recording of a train of pulses,
rather than a single pulse is a great aid in cueing the analogue tape. The
length of the train must be less than the length of time taken to sample one
block.
40
In order to set the CDC 160 up properly for digitizing, a short program,
TEST 160, has been devised (see Table III). In this program, the first cells
-0000 to 0017, contain a program to accurately set the bias voltage. With
no signal in and the bias at -5.0 volts, the value of the conversion should
be 0000 or 7777 . By monitoring D/A Channel 1 with a "scope", deviations
from the zero condition may be detected and adjusted using the voltage con-
trol on the bias source. Small errors in zeroing may be expected, due to line
drift, etc. The effect of this may be eliminated by removing the mean of the
data in the 1604.
The cells from 0020 to 0045 contain a program which has identical
timing to that of DIGITIZE. By observing the input request pulses on the
patch box, between the CDC 160 and the A/D converter, one may adjust the
value of the timing control to give the desired inter-sample delay. Chang-
ing the control word by one number changes the sampling interval by 12.8
microseconds.
In order to distinctly label each block of data written on digital tape,
an identifying series of numbers is first written, followed by the data points.
Since it takes four CDC 160 words to equal one 1604 word, it was convenient
to specify four quantities in the identifier.
1. Current run number-indexes once a run
2. Spare-may be sertt to anything
3. Current block number in the run
4. The number of blocks per run
41
For example, at the 5th run, using two blocks per run, and the "spare"
set at 4321, the identifiers would appear in the 1604 as:
0005 4321 0001 00020005 4321 0002 0002
Precise records must be kept of the identifiers when digitizing data,
as data location on tape is by means of these identifiers.
3. DATA RECALL
The data in digital form may be recalled by calling Subroutine DATA.
S/R DATA in turn utilizes two subroutines, FINDIT, which locates the data on
tape, and UNPACK, a machine language routine which unpacks the samples.
The result is to create the original CDC 160 sample list, word for word in
the 1604.
The unpacked data is presented to the main program in integer format
and appears in either KDATA (M, 1) or KDATA (M, 2) as specified by KLIST.
Integer format output was chosen for the sake of generality as this is
the format of the raw unpacked data.
Normal use is to change the data to floating point format, remove the
mean, and divide all readingsJgya factor of -409. 6. The letter factor converts
all readings to volts.
When reading the tape on the CDC 1604, it is convenient to have the
blocks arranged sequentially on tape to prevent tape rewinding and scanning.
However, when digitizing data for cross -correlation, it is convenient to
digitize all the right channel information, and then the left.
Since it was desired to cross -correlate the left with the right channel,
42
some reordering of data on tape was necessary. Program SHUFFLE does this,
reordering the data so that the output appears as right samples, left samples,
right samples, etc. , in consecutive order.
Once the data has been written on tape in the order that it will be re-
called, the average time taken to recall one block, unpack it, remove the
mean, and convert it to floating point form is less than 2.7 5 seconds.
4. TESTING THE SYSTEM
In order to test the capabilities of the digitizing system, a test tape
was prepared with two blocks per run, and each block of 4000 samples on
each of six known signals.
Run * Signal
1
4
8. OVo peak-peak, 250 cps
square wave
5. volt, 250 cpsTriangular wave
6 volt, 250 cps sine
wave
-8 volts DC.(-3 below reference)
20 cps band-limited
white noise, 7. 07
volts rms.
White noise 7. 07 volts
rms.
Identifiers
0001 0000 0001 0002
0001 0000 0002 0002
0002 0000 0001 00020002 0000 0002 0002
0003 0000 0001 0002
0003 0000 0002 0002
0004 0000 0001 00020004 0000 0002 0002
0005 0000 0001 0002
0005 0000 0002 0002
0006 0000 0001 00020005 0000 0002 0002
The data was analyzed by Program TEST and the resulting analysis is
tabulated in Table IV. The data was recorded on Reserve Tape 18.
43
Table IV. Summary of Test Results.
Run Signal Type # MEAN/EXPMEAN <S
Z/EXP<S2 fc/EXP
1 Square Wave 250 cps 0/-. 132 16/16.365
2 Triangular 250 cps 0/+. 086 2.0/2.09
3 Sine Wave 250 cps 0/+. 027 4. 5/4. 56
4 -8 Volts D.C. -3/-3.025 0.0/4.74xl0"6
5 20 cps WhiteNoise
Bandlimited 0/+. 147 -/2.10 20/29.5cps
6 White Noise 0/-. 094 -/2.2 200/199cps
Mean Recovery Time Per Block ^L 2.75 Sees.
Digitizing Noise is 63 db Below a 3 Volt Signal
ERRORANALYSIS
In the analysis of the data it was desired to determine an estimate of
<J2
, the data second moment, and to determine a confidence interval on the
estimator.
It has been shown [2] that
Ns2 = —-— ^ (Xj - X) 2
, is an unbiased estimator
for if"2 s may be conveniently calculated by:
s2 =
N
To obtain an estimate of the confidence we may place on s 2 , we assume
that in the large sample case, as here, the distribution of s 2 is normal in
.
.-.;*.-*=. . -
44
which case:
Var(s 2 )= _^li
# Pr'Lls2-<f2 1 ]^. kr^ MillJ
<**j
ie 2
*If n^t *
f N^r
where k is the value of the standard normal r. v. which will have a proba-
bility of being exceeded of # .
The confidence interval here was chosen to be 99% or .99 which implies
k^ = 2.57
ie. Pr ? N(Ojl) C 2. 57 7 = 0.99
Thus the confidence range C was constructed:
C = S 2(l -i -)k / —2—\e
fN-i
C indicates the region about s2 in which one has a probabilily greater
than . 99 of finding the true ^"2.
If we form C/s2 we have
-£5- X 100% =S^— o-"A X 100%sZ
1 + k *r~~24 r n-i
In the data analysed here k ^ 2. 57, N = 3900 and hence C/s 2 =
5.42%.
This means that we are confident that the true^" will lie within 5. 4%
of the estimated6"2 = s2 with probability .99.
Note that lim CR
45
In each of the runs it was desired to plot both the signal and its' auto-
correlation function. The units of the signal are convertedto volts by divid-
ing all data samples by 409.6 or .2048 levels
^
5 volts
Sine, square, and triangular waves were analysed to check the scal-
ing accuracy. It was found that the accuracy of measurements exceeded
that of the oscilloscope used to measure the input waves.
The D. C. signal was analysed to observe the amount of digitizing
noise and hum introduced by the process.
The maximum change in the digitized level was about 4 numbers or
fi^l. X 100% = 0.2% of the full scale. Over the, period of 0.8 seconds
the mean square power was 4. 74 x 10 D watts and the rms. voltage was
2. 17 x 10"3 volts. (A. change of one number -2. 44 m volts)
Thus for a 3. 025 volt signal, the noise and hum is 62. 8 db below the
signal.
The presence of a strong secondary peak in R(T* ) at f = 1/6 O**1 sec.
,
indicates the presence of a strong 60 cps component in the noise. Assuming
1 — kthat at T = —: sec. the noise is uncorrected, then R(l/60) = 2 x 10~° or
60
66. 6 db. with respect to a 3. 025 volt signal. Thus the digitizing noise is
3o 3 db stronger than the hum. If hum could be completely eliminated, the
noise power would then be -63. 3 db below a 3. 025 signal or -53. 7 db below
1 watt.
The "white" noise and the band-limited white noise samples were
taken primarily to check the whiteness of the noise generator. Measurement
46
of the — point of the R(T) curve indicates the generator output would be flat6
within 3 db out to 200 cps.
A considerable discrepancy between the rms voltage as measured with
a VTVM, and the calculated value of R(0) and s 2, was observed. Some of
the error is due to the motion of the needle, and some due to the fact that
the averaging time of the meter was about twice that of the correlater. How-
ever the most signifigant factor was a faulty measurement probe.
6. RECOMMENDATIONS
No attempt was made to "fix" the installation/ and hence some prob-
lems arose that would not have occurred in a permanent installation.
Noise and hum reduction was only achieved by careful grounding and
the use of BNC type coaxial signal leads. The use of coaxial cable is re-
commended for all digitizing work, for the above reason.
A clock pulse was used to synchronize left and right channel data.
Unfortunately, the act of switching Ampex tape recorders on or off in the
record mode causes transients to be generated on all tracks. This is a
characteristic observed on both the Ampex CP100 unit of E. E. department
and the Ampex FR-100 unit of the Aeronautical Engineering Department.
The transient pulses may be treated as clock pulses by the program,
causing loss of channel time synchronization.
The problem was partially overcome by removing the output of the
pulse generator until the transients were past. This is satisfactory unless
the first clock pulse is close to the transients. A standard interval of 20
47
seconds is recommended between the start of data and the first clock pulse,
as a temporary restriction pending hardware modifications.
A more permanent solution would be to imcorporate a time delay in all
record channels of the Ampex CP100 to allow transients to die out, when
starting and stopping the unit, before energizing the amplifiers. This should
prevent spurious clock pulses from being formed.
Despite great care, occasional spurious pulses appear on the clock
channel, and may cause false sampling to take place. The use of the short
optimal length pseudo-random code as a clock pulse, together with a match-
ed receiving filter would correct this problem.
The system as implemented has a 2. 5 Kc upper frequency and 4000
samples continous sampling capability. Since at best, samples may be taken
every 190 ixsecs. , of which only about 50 Lisecs. are chargeable to the A/D
converter, the most likely place to begin the search for higher sampling rates
is in the computer itself, rather than in a more expensive A/D converter. By
deleting certain functions from program DIGITIZE, the sampling may be made
faster but in no case is it possible to sample at intervals smaller than about
100 Usees.
48
Table 1(a). Program Digitize.
Cell Contents Code Explanations
0000 0101 PTA
1 0603 ADN032 7064 JPI64 Jump to INITIAL.,
3 7500 EXF.00 „
4 2410 2410 Set Enable
5 7500 EXF00 i
6 1401 1401 Call A/D Channel. 1
7 7600 INA INPUT0010 4176 STI7 6 Store sample in- (007 6)
11 2076 LDD7612 3464 5*&r SBN65 Enough Samples, yet?
13 6134 NZF34 If not, go to cell 0047
14 7500 EXF00 #,
15 2400 2400 Clear Enable
16 2074 LDD74 Load Current Run No.
17 4160 STI60 Store in cell 01330020 2067 LDD67 Load Spare ID. .
21 4161 STI61 Store in cell 0134
22 2075 LDD75 Load Current Block No.
23 4162 STI62 Store in cell 013524 2073 LDD73 Load No. of Blocks/Run25 4163 STI63 Store in cell 0136
26 7500 EXF00
11 2111 2111 Call 163 M.T.U.0030 7303 OUTOa Output from
31 0000 C computed L.W.A.32 6102 NZFC2 to
33 0133 0133 013334 2246 LDF46 SetA-013735 4076 STD7 6 Reset running storage address
36 2075 LDD75 Enough Blocks yet?
37 3473 SBD730040 6155 NZF55 If not, go to cell 0115
41 0401 LDN01$.-..•
42 4Q75 STD75 Reset block no.'<
43 5455 AOD55 Update total block count
44 3454 SBD54 Check if capacity exceeded45 6153 NZF53 If not, go to cell 012046 6061 ZJF61 If not, go to cell 012747 5476 AOD7 6 Beqin intersample delay loop
49
Table 1(b). Program Digitize.
Cell Contents Code Explanations
0050 2066 LDD66 Load intersample delay word51 0701 SBN01 If not zero, go back 1
52 6501 NZB01 If zero, go to c&y. 000553 7056 JPI56 0760-M.T. U. block capacity54 0760 0760 Total No. Blocks written
55 0000 C address56 0005 0005 address
57 0003 0003
0060 0133 0133 Address of Current Run No. ID61 0134 0134 Address of Spare ID of ID62 0135 0135 Address of Cur. .Block No. ID63 0136 0135 Address of Tot. .Blocks ID
64 0100 0100 Address of INITIAL
65 C C Address of last word of data
66 M M Intersample delay word67 M M Spare ID anythinq
0070 M M Initial Run No. Set 1
71 M M No. runs desired
72 M M No. samples/block 76408
73 M M No. blocks/run ,
74 C C Current run no. .
75 C C Current Block Number76 C C Running Storage
rAddress
77 C C JUMP CONTROL.0100 4077 STD77 BEGIN INITIAL
101 2200 LDC00 ."
102 0137 0137 Set "A"-0137
103 4076 STD76 Initialize running storage address
104 3072 ADD72 Compute last address
105 4031 STD31 Store in 0031
106 0701 SBN01107 4065 STD65 Store in 0065
0110 2070 LDD70 f
11 4074 STD74 Initialize Run No.
12 0401 LDN0113 4075 STD75 Initalize Block No.
14 7077 JPI77 END INITIAL
15 5475 AOD75 Update Current Block No.
16 5455 AOD55 Update Current Block Count17 7057 TPI57 Go to 0003
50
Table 1(c). Program Digitize
Cell Contents Code Explanations
0120 2074 LDD7 4 Have enough runs
21 3471 SBD71 been done yet?
22 6103 NZF03 If no, go to 012523 2074 LDD74 If yes, display last run no.
24 7701 HLT01 Halt 01
25 5474 AOD7 4 Update current run no.
26 7057 JPI57 Go to 0003
27 7500 EXF00 CallE.O.F.0130 1111 1111 Write an EOF Mark
31 2055 LDD55 Load block count
32 7702 HLT02 Halt 02
33 C C RUN #ID
34 C C SPARE #ID
35 C c BLOCK #ID36 C c BLOCKS/RUN ID.
0137 D D Data Storage
~) D D*r
7776 D D Data Storage
Notes:
1) C is ENTRY PROGRAM COMPUTEDM is MANUAL ENTRYD is DATA STORAGE
2) CDC 1604 Identifier will appear as follows:$
Word 1 RUN # Spare Block # . Blocks/RUN Format 016
Word 2 Sample 1 Sample 2 Sample 3 Sample 4
etc.
51
Table II. Modifications to Digitize
Mod Cell # From To Effect of Modification
0126 7057 7703 Individual blocks clocked at endof a run. To restart run at 0003
0014 7500 03000015 2400 0300
0117 7057 70560126 7057 7703
0014 7500 03000015 7500 0300
0040 6155 6113
0126 7057 7703
Initial block clocked. Subsequentblocks taken as fast as possible.
HLT in 0126. When run completereset FFI manually
Initial block clocked. Subsequentblocks taken immediately. All
blocks have same ID. HLT whenrun complete in 0126. FFI manual-ly reset.
0013 6134 6506 Eliminates variable sampling delay.
Samples possible in this modeevery 120 Usee.
52
Table III. Program TEST 160
Note: M means Manual Entry
Cell Contents Code Explanations
0000 7500 EXF00 Begin Bias -set
1 2410 2410 Set Enable
2 7500 EXF00
3 1401 H*6^/ Call A/D Channel 1
4 7600 INA Input to "A" in Reg.
5 4070 STD7 Store "A" in 00706 7500 EXF00
r-
7 2401 2401 Call D/A Channel 1
0010 7303 OUT03 Output Cell 007 to
11 0072 0072 Cell 0071 INC.
12 6102 NZF0213 0070 007014 0400 LDN00 Load zero's.
15 4071 STD71 Store in 0071
16 6414 ZJB14 Jump back to 0000
17 7700 HLT00 HLT. End Bias -set
0020 7500 EXFOO Begin Timer21 2410 2410 Set Enable
22 7500 EXFOO Call A/D Ch. 1 .
23 1401 1401 v
24 7600 INA e
25 4143 STI43 Store Sample in 0043
26 0300 NOP No Operation-NOP27 0300 NOP These are Time
,
0030 0300 NOP Dummies to
31 0300 NOP Match Program DIGITIZE
32 0300 NOP Timing
33 0300 NOP34 0300 NOP35 0300 NOP36 2042 LDD42 Load Timing Control Word37 0701 SBN01 Subtract 1
0040 6501 NXB01 If not zero, go back 1
41 7044 JPI44 If is zero, jump rto 002242 M M Timing Control Word43 0043 004344 0022 002245 7701 HLT01 End Timer
53
LJa.
rv <C i—cc
CC Lu1
LU XU_ h-O O
x 2^ _- _LU « O <M
q; • O a • - Q LVLU lu >—
i
II ^ ^T zJ a tO u — < X
<x u. o < <jD 2 X O 1—1 X 1— UJ —2T
cOu
c
en < *t
l
XCO < "P —
i
<: 6 a ^ r\j Kf\ 1—
,_,
3•
'~
::
Oi— U 1
Q
*
o X H- Q^
>- LU V to LU O o —1 o o V m a: UJ vO a^lu : 1
11 1: l V J 1/ 1
. x, a ..-
1
C\i --~
• H • ) i ; > n iU.i — -t >- • *—
1
O oo —.
1— \— _J i—
i
r ro o I- 1— Q —t 1— «* h-
ao u Ql x ^ : Li_ O X O tO -
LjJ-- \ LU 21" a 2! UJ \— o MO o >- i— i—i
Q UJX3
L0LU
_J
OnX
U.I
UJLi.
Xi—
<Q
L.J
U,X i-H
tO \X
cc
Xo
X
—
>
I— a X no < O OT < UJ Q CJi tO u 1—i
ti"> —ii
, -r QC 1— C o ;'
c^r X C r. t«i —< •— X < < O 2T -.z O UJ < 3? 1— < i
—
i
<_l CM *—
<
C£ uO (M h- O <c i—
i
i—
i
u_ 1—1 n 1—UL •\ Q o <r 1— X QC lO u ; V <t- Q 3: o <V e o u_ cc < _j <^ U 1 2^ Qd u o Z < Q#* c _J CC O O (— Ll I - t— X U 1 < vO p—
i
< _J vrt— c < 0_ < i— t— ',: • <j ', 0. r-l X h- Ll «\
or , c (— cc i—
i
c a >- *—• i—. "2'. i—i 2 <3 V I—>—i —
•
U o UJ o ;'. X UJ C i ) D UJ X X t« tO h-
X < o x ro [V UJ X UJ •: X 1—1 — 1—1 XX h- i—
i
";• LU ~r i'V o —I O *.""
.
'
3i O r- o h- X Lj_'
«N < vO :5 I 1— UJ Lu _J h- < '3 .-- to<-~\
LU < 2^ v: o — ~X c r-l LU 21 I— tO <]' X Di U to LU V X UJ •^ —1 1— h-*".. V X —i 1— 2c 3 u t-H o t— X I— IX tj X O X to W)2 ^ <C I— UJ UJ X < o <r tO O to cc < t—
<
i—i <c #* 1—
1
i—l
#» ~ X < 1/ tO f— or i— 1—1 t-H D CX ^~ •^ #* yr X Xi
—
i—
t
tO >- h- 0'i 3 (/ i—
i
>— 1— 0..- _J to X <x 2^ sC X C\! LT X fO V V
2T c V i—
i
CO 1—1 >—
.
Q _) LU w X rH ^r f\l (X' •^ CM #* cUJ c u tO ~
^ i—
i
< to to V. 1— X o c ^ ;.- co \/ GO ;r ^c 1—
1
o 1— Q 1—
1
1— C C? (— V 2i X I U J Li. #* «v i t I — —1—1 <— _J 2 UJ to >- <" h- < u i—i u u cc lX ^> X h- r h- o h- < X <— '
J/ CO :lj 1—
1
X 1—
1
_l •
—
Q t^r _J ;• f— LU CO < c 21 t\ 2{* _J X h- < I—<L t. 1—1 Q _J < _l < u.
"-Q. _J <£ , < X (X —- X i: cc i— T> t—i < 2" <
1— o i—
i
a. ~V v. t— vr X UJ 2* -
1
_J o f— h- cc MO «N X #* X i—i IX a c< _J «\ CL < _2. U U < LU 1—
1
•—
1
-C" >—
i
c\; X ^-> Q v. n Vo. a ^-r 3 c _J —1
Ll. LL! Q ^ c t—* — OvJ 1—
1
<f r» ~0 «N
t
—
h- • l/) • • >/ • < h— X UJ c .L. U .—
(
•> c •v 1— CO i.A + \y 10 ^ S C^-—
LU < r-
1
CM C^ <r t^ .-: X r— :uJ < X X X X — <r u i—i »vII •^
2: "Z. Q i—
i
x < h- -x a CO c en i—i O —
.
<t \ <t; c ii e 1 c1—
1
o n/ CO co X o ^r. u tO —• X <)• iTi I'
-
. X X CM X Ql s m 2 fl UJ
I— 1
—
a'. QC L-X' x c < Z —i— 3 — 1
—
— — 1—1 < CO < p^il ac X CO —
\
, I/) 2=" < < < <c _J U 1 i—i D 1— 1— (— H u_ X 1_ X^ o < r- XO x d < O.Cl' < < < <1 U_ 1— II X (M i— 21 CM 1— 1—1
CC X ;- h- ,X UJ !" * 5! X 5. _J i— X X _l <'" 00 2^ CO X 1—
CO ;=r "- <T CD H- cc ,": 1^ >-.- X — 1—1 < _J X —i II 1—
1
X cLj i—
.
O o X Ll. O C o O <; U_ or ;r <T ll_ O CC O cc C Xi/) C u v: tO < u_
<rr\j
CO
u_
lT>
CO
u_
a'
<XJ
CO
X
c
og
U a
C\J
CO
"0
COOvJ
co
u :, :r O d
CO
X O a
C\J
00
U LL
o~CMoc
u u u u x u u u o yuuv
54
Q.
I
oI—t/j
oCM
CO
vO --O
3 O Orv »> «•
O•-
oII
ID
L Q >- Ll
lO.-,
1
—
-
c r-i
Li I T Q —i— ex
i— _j i 1—i ._' CO~ U t— Cs X U C—1 ll) •;1' < — r L ,. r Q i
—
o— c: t— LL ^O'-"" ^r t—
i
h-•J, V L., Q I r—
1
3 — r^ r
<;; 1—
1
Q LU o -) 00 e_J -— L>' < r- u LT i—
:
li e!; V i/ D h- C c C' a c V
>—i u <t <5 h- _i CO H- CO .
«•• o ~-" u Q LU "0 •N .">' «\ aX _i kj i
—
i—l 1—
t
CC} .—i :-
CO o a < i—i •> o i—l : . ~i- —
•
o LU U CO — 1—
1
CO Q co X u t—
t
«v «\ cc i
—
o 1—
1
CC- #\ tfl »>ii
h- — 0? o < —I-s — rt m I— ; r c o ID
/. C\J LU rv u (D V CC o .i ^—
*
H- <:. <«> v o D — u O CO L 1 _J
r c >- LU _1 1— i/j o CO ^~ C rvi Ll Ll>— \ i— 00 _1 •v •~ rv —
<
— .—
i
On —i i—i
— c V ._ >- U. Li
:
< •'0 r~ -— ,--
;
» — CO I—
<
h-4 c I
-••
h- > ..j 1—
1
c <—
i
CO r—i CO b^ -i—
i
~— _J <t 1—1 < CD •.=> — a — * c—l
c < Q_ a- X < •— «v >/ — • o h- h-~2. (— *—
i
<L ;>• r—
t
r—
i
••'.; u i—i i— I —
i
1— X .vt—
i
< c* < a. >—
•
X) c c o i c cr. LL!
Ll £ -: + i—i cc _J — (— — IS, O C <~
V i
—
c c c c — •> n f < c i—i —I(I
:
-...
!
X x T <J"— 1—
1
CI. Q_ 1
—
«* •i
._
-
<r •v/ <j / \ - y i—
i
1°s <-
!
, •- co "i—
<
o ;< IT. r-
:
;<"> '•.. X H-
1
r* 1
—
" <" I— .—
i
I—l .—I rv LU1- •—
1
— w -^ — •---' '
—
2" X) ^^ IT co CO X—
1
I/) ;. J— ;— 1— — '•: ,V i—
.
u ,-' N o Xc X 6 < < .
:",1 LU ;
* q o i- c X •^ i— 1— I—l
a: LU 2 ;-
*> "
; _ r— Ll X —i h- «i ^— 1—
1
I—l ^'.' X (—
CG"
rx• '. X .
~— v >—
i
— ~>3 i—i -. X
j .•—
i
c o o t O v_—
i Ll Ll c 01 Ll Lj LU :jy or. ou c u u. U_ - Ll •—
i
•— !J5 a i—i cr Cr a. CL X'
<r lc :,: i.O p i- CO c r-i CM ro a^ 1—
1
i—
i
.— 1—
(
nj D c "i o o i—
'
t—i r—
1
i—
i
i—i rvj
cc OC ( GO CO CO CO oc: cc CO CO CO a; CO CO
Q
55
<r- <--: oaV ct:
«* 00r—tO xI—
^
1
m: o*v
L:
X _J..' _J':' <~) Uc*
V ^u uo <r.
-J CL
uo cco en0/ _~
< u~> +
II
tO
oor:
I- oto :-:
o ~
to
co
o-j C£ zU Q< Q
Li.
X ^< u
oX< cc :^ <s: 2s —
>
aII oX < <J1 V< -H U:-: i/i o"") (— r-l _J
+ O + -H
ut
toZo
rr uO D< Q< tT
h- O VZ GT •>
ou
-1 rH
I— <:T
X I—Q Li ' <O Z CO iZ
V
<<
re
Vu >-
O CD_l
CD Z
O Li
_JtO _lCt' —
•
O Lj_
+ <r 4
Z
U_
otocv
oto
< Q~> I—
Li. O .'.v.
a: oc o HH- O tOf > + +
ll !
> r-
< +
to
_J\1
<r-<o
0.'
oll
O
n+
to
_j>/
tOtro<
Cl
iV
o
xLUXLU
U_ . t ;
O
+
Z
tO
<U <o
_J
1 Q 10
2T <—
.
h-<-
or vo3 Z
CM i-i
<;X I— oiu < cro o oZ 5:—< t—
in (—
LiJ o_' tO> -< ct:
< u_ >—
'
tO + Li.
IDLU
<OI—2:
QO
— I—
u_l u_co --
— X+ iO
Q —Z' _JLU V\-— •*
x <J-
LU +
z —— ttO H-^ <
i— a'.o v:D~) Z
—• o
cc I—+ to
oa 2"
Ll)
O t-(— x2' LlJ
a \-
0C tOCO 3I—
Li. h-•—
'
x.
X OtO —
'
+ cr
to UJ|— Cl_JV <cn O+ I——
'
z
< Q£Q O
z oCM
UJOr I—
O LLI— _<
tO X+ tO
a —Z H-LU 10l— <—
i
X _JLLi V
Z r\j
O +
tO —<c
•> I—i— <rtn Ci
Z> V
cr Ol—
+ to
< <a. az z3 D
UI— <—
. Cl
x ZUJ Z)
to(X. Cl'
c.
3 CM < * CCH -) H
C\l XCm UJ
X * CMUJ C\. O * rH
CM CM ^O -Jf.
r-r (M if.
CM
r- >-
^: a.:
Ll! J— ~> Z> ID < -J tO < 3 < _' < —q ^; _j »— i—
i o < a' Z <r Q < Ci <m LJ(/)'U)-J J (/)< "- to _J to _J u.
J<D<Jwnj>-M< to to O tO tO O t/j tr; c< Z < Z < Z *-..»-« Z Z Q rr iv: t— Cr' (V i— cv ty t—t/5 _ i/-, _ (^) km >.0 to LU UJ _l _J O tO _J O to _| G tO
Ua.
toc>-
c< <—
i
< ~)CM C<"
—
)
56
Li
tO+
to
-">
Q V1
—
O r—
i
u-.' o
_1 X(\l <I
£
'•
UJ x ..'
X (— o 1 J
~-.
-
_
: < ~-~ Q' T UJ *—* 2 i
—
UJr ~i a t—
*
_ ^ai u U 1 C »-.
\-' DC ,:::' I_
;- o c C c ~l
<f t—
i
h- C :-
C
i— + / + (/. i—• ;v
u: .—
<
t^ ~) U.I ~i CO,-s II 1 4 ii ty UJ z>
+ —
<
+ ~> + or + U)
>;< < * *
o\j <\i r-> f\J
to
":I Z
Li' tO U
tOI——
<f < x- ft U
57
UJ u-J c. L'
' zh- Q. • Q ) UJ UJ•—
1
L. 1/ ( 1 u i— ch~ c 1
1
c < t--. ^-.
—
1
u IT, u -. f\J u Li -1r- Q P\l 1/ ^T X ——
«
U c. iy) l; .— - v'. > O < Pvl-
'
(\l X lTi i— c<: c £ S: k_^•
!—i—
1
C 1
—
rv _j V - •—
i
ill ** ;-"
«— if . o O u _j > Ll 1 LT >^P\j r\i o > '. u h— a Li LT (V Qh- 1— UJ <] •—
i
r- _J • < Q 1—1 i2 U-i X ~z '"•' — o «rr Q r- 1! -r* •> '3
Li!
> < >— Li
~
'
oO U—
,
o -~ o£> < 1/ i i
~ .J < i-O JT 1—
1
-j•— IT 1: o u Q. h- ^~ D ~-- 1—
I
D• ^ — CL U K 111 i
- >/"
L0 < <t ^'
r-H h- D <o — C 1 1 . 3 i—
i
-n 1
-~ "• «^ 1— Q r-
o t\l o (/ a a. > T~ ]
-'•* .o o ^~ lr
) ixj i 3- *
_l Xi-< i—l —
1
> "= D ^ u 1—
1
^H o _l e -rc ~- •— Q. LP, v0 1 1 1
1 u 1/ • l0 UJ o '<r i— '. \
I-H i—
i
.
1 D >—
i
UJ t- 0* r-l 1—1 u < ii
«t 1— 1 a ex. r\i (\j P"
_J r> *.-< •> " Do . i—
l
>1. -
i c c a.) H- Q 11 r- f\l "'» ,
'
r-H Ijj u c C C C c •
- \ W) h- - -0 #> c?; (0 1—** Q •N • C c c; c c . to —1 ^, r-H -t i—i c ij
-1
•-i X<r —i — •• 2 <r o • (V r-H i. ^ <if\ ** C\l rv (xi r
'.J Pv) . L _J .'' V •^ .^ r\i a Ll r-H "j
ii—
•
1 r\i c c C < "i. h— < r-i J— »— r-l XiT, CM 1—1 c c c c c 1
:
"
U : O Lt.'
u :•i w 7" X
<r ^ >— < c c a c c < h- /< '«\ >—i i/i • -r
\ C _J c <s< >-. r q Q >: r—
,
CC o /.
LU c < (M G c - c r V. "7 H- UJ i—i r—
1
< a ;^' U r-H
\ <'
21 ( o ' c c. < o > LO c- C: *< < +<J, <t C c c i_ c c c c : h- <t r-l w r a iorv — i- c C" c c c c h— c C r-H < • c i—\ "I i/) c fy -J- UJ Mil Pvl 1— —
i ::. : l<Si h- •* c (XI m -i- IT V-O ' ^ h- -H
; c .—
1
r1" i—
i
i-H Lli —1i-H <r •— c o c '. > P\J »—
1
1— UJ 1—, w- i- X\ Q r c c '
CO h- < »v [ ,_' r\l h- •: '0
m X .. -» t—
i
c . c c'
1—
1
1—
.
W LO '-?J cc X a i-' r_ O
\ #* r-H < c1
' _l — D (L.1 1—
1
i
.
~Z. t—
i
, £ 2!!
o ~ >-- i— c r O Q c C\ U '
1 ;"^ —
1
•\ LU cc a i
<j- <\J L, *r T c z . 2 ; CL r; I LO Ll. vt x O a X\ •'- r- i" <: • < <:• f— <'
.
'
L. — i—, * <i i—
.
c c
o o LL X o h- < •—i i—
i
k0 -_-r» 1— r\j ;-" (/• •^ w -. . -
"V c ,-. #^. C\ r\i r . Psl P\ 10 :r. X ^~ 1—
1
r—
1
2T .—
i
#^ D •— <T.
r-H r 1" <A < c ' —
1
U
!
""?; j- -
U hi LO C i— CO II
"N. <J- l~ LO c > ll r- i—
<
•v' a r- rtr
-1 p — i— e X)i—i -— V 1— f\J c C
' o K I !/"• i—h •s X c •—1 X ^^a. .—
i
u_ < <•V ••i •^ Q c • LJ 1 h- >c tn _ !XJ y *> h- I-H LL' r-
:a 1— -~ \V >—
.
,"T . --, .-i r-H r-i 0^ c./. ._.' 3 m «* : ~-~ •«;
:
2 X CC XC *— c C D Li •v O #* j -— UJ •f;
'
S' U i lO <X 1— C ( <; •7 i-H c 1 •
?" rr LTi U 'i m -^ ,--,
:
r-H i .:, O -j :•:
l/j >/ r— h- / c i 1 1i—i i
—
* H c -1 —
•
r :*
•> i—i "V _j
a LlJ »— '-. U . .T3' ;.• u • ^ i—
1
*v * " X '
#* ro #^ 1—1 «^ «> <r .^
i.-:_* r. ... _J 1 c - c 1 L ~
: o i- X "T- X ^ r- ii X < II i\j COi— o ^
-
1; 1 < c c c ( ) rv; D CO r—
;
r\ .'. rvj vO i—i pfl v0 (1—1 c . O 3 X
(— '"1-1 r- —
i
LiJ c c C O c o U~) U _J — w - — — >— lO — — v.
" a o O <ij i <C lO (V .- < CT
' c c c c c /_ —-< (— i-i < I— 1— i— h- •—i
—
1— r- r-H .—
i
r-H .
-''
re • (V 2T. #> > c ; a_ :•: u r-T < t < < <J < ;
:
ll
L< Ixi O UJ -^ '. w UJ r-l f\l m <r lTi ^o ,— <t Ll. i—i > > " '
: .£ 2; a 1—
1
Q Q < X< •« O •
P>•
' 3 a r- c c o c c u h- U <t c ; t ;>;
I
V (V fV LV < < < 1— <<r cc .— r
—: o >-
~ c c c c < < X cr C c O C UJ c UJ Li < :
*s - a. fi ^« u '.I h- c c c c c c U- o Lli r~ _J ii; Li. !j_ LL U_ LL Ll.' U_ u_ Qi c o; a O X#s _i r- i—j
a «Ni—
i
C\l rO <r LH vO r~ c
o . <--
; o r-H
o H i ; : co u r-l r-i i—
i
r-H i—
i
r-i r— 1 r-Huuuu l; o u u u u u u
58
1/
1
1 —' — !--
1
O3 _ •
lO1
•
.—
i
i
—
o.; 1
X ,_^
_l _J <t CM ") < ——
I1 at < •;
ct:,—
<
V --- < X ^#* •% *x V0 »-. — i—
• t— — X •— QL \ u-.- u -'}
. •1
";
•' — <-,
— — >—
1
e .
—
~) o CM Ll.
• 1— -J ~3 iM 1/3 u 4-
t—' \/ «— •x X 1 Z3 ^
| r> • ; *> ~i X lOO : 1 X 1— — Ll •- r—
<
i C*
XII
U.- O ;_ CM
Q XCM
X -'' CM — h- Ii„ .'-^ i/3 i
/-, G i0— — •x :— s'_ *» — — ^ X ": r- X LOr—
I
- •
-
_J 1
~> r-( o.i -— i- _l l
*v *v '1 1 Q —
'
« *< r- II <[ r> CM1
"")) •— > — <t ~'> ~) •• — -." > D CMw •
—
•" .-
i «— r- — ~— ~
,
• X , ^ ^ 1/
II.' - o <jT ' ' • r
—
-
•v D ^ CM LOI
—
1
—
— *: ;— -- 1- c J- H < X ~31 i
—
II to il
u •:
.- -> < '/ .,' < I- < »— OJ i"• CM
—I —3 1 X :i Q II
"
I 1• < c ^ ',') — -— ll
—
1
X X -_. i/3 — X X Q X 1— I: 1 ;. lO u. — s*V : II • —
I
^-^ J X v. <r 't'i
- h- -— oi.
.
1
cm r-»
+•> T_.
••
i
:
'
; _J• ~3
'; (MCM
CM < V •••) l>
•u
—:
«* r— I'- c > > U >— -~ ~1/
"" o ^> C I
'
t—i tf!~•> •* ~3
ll:'"
-. i 1/3 lO LO f) "? ,* • i
~— — — .'
;
>9 • li ii
*v Q r-< II X -—
i—
i
.—1.
.
II X/ < u~>--'
i . 1 CM II .—
•
r-H u 1 CM : —i— <—< —
1
•N 1— l-1 1
• ) ".>. (Y a X < or;
'
_J 1 yj 1/3 -J —.
:
. 1 X •1 1/ CM <l > Q ^T•'" X o
:x/
<
)
2) 1/ ,- r-i II i.;
1
,.-)-
' ^'•' •^ r—
' •N r— — X •v • r r-l 1— — •: «3 X x; • r X -. X er— X —
. i/3 V. ,-. p
1— i/3 r <
II> —
I
—
•
—
.
— "P.
,
X ~ Ll Q X • Ll
H ' ' D —t—
i
— -— _J CM ^-> ;;.• cH o 1 1— fM +
. ff* "V «• r-1 .' X v: «\ *s r- ' .: ''• •x ~> r-H — cr— •> r— "0 •* - r "J
-) • ^~ •"-) o - !j_ •
O :! — ""
)
<£ '1..-, tC — — "
)r-i X — 5 • r
I
—
•— i—i
*—i u h- .
-
-.-- : m* • — *- .-1: o IjJ >•", -"I rv — cr.
.«* ..' C i— X X D .' — o- X 1
-- i- ~> '".
—
~^i
;V. 10 G X r-i
•)" •—
i
ii J —
i
•. . i u
-yLTl < •
.1- —~ X 1
—
.
,-. /) CO i—
i
— '.". Q ! X '
Ii < ~— •v:
• < <3 X, X i—I -•
.":' — — LO Ur—
(
—i —i : < (\J C •^ .- h- 1—
j
- X i < < Ll * LO Lr
( . -^ «vII x v. —
1 ,.
' X 1- Lf > r-H II ll x '-~ ".ll
--• Hi CO t— r- ^- • .,
i— 1' t -— C* —
-
i i- r- i
*
u .
;
:—
.
—
-
+ '"«1 — •—
;- X Q. r 1 n.
r—
1
.•
. r r- <—i i— ;«* a IU -^ ll CM CM 1—
1
X lt, r-H r-H \ 1 O • i-H
_J .
1—II
«- ;:~
1. - .—1 -; o «\ »* + .. 1!•* i—
i
-. r- '.0 CM tr 2»
h- 1—
1
—
)
~i- ~~
. I'- .- ~-
Ji/ i/: l—> < "") ~» ~> ~
r— „".' II~
3
—3 D O 10 II 11 U ) u
; _j V — in ~-ll— T
'
. ; i o II _J »— ~-~ U ;_ — "u . r 1/3 r-H or II D
i—
i
_i — h-'
i \ -r ii.--. r-H < r-H o ii
1 u. O'l <r «
:
II u i—i i—
H
;r. <: •_, *— <c o •-- (.9
o- < I
•"-H i—
1
1— r-i 1— D r- 1- < CM 'il V r—
i
t— r— 1—
1
II h- fv H i
—
1— CM < 11' 1— Li C
CL u 1— .— —- : < < —
\~ 1/ m : l
^«- < •- i-H "Z. -T -r i
—
— o u sZ 7-
'
Lr" r-s r a o i <; • (/: ?_\ u_ Q c . c C c 'r < LO >— < C —
- h—
'
Q X l/.i kj > X (.;-
_"• l-H 1—
i
c X X Ij (/ l..- 1 X X U LO j U: I/' Li. U _J
r~ CM <!" ^J <t- C" lO inS\ lO 1
ri—
i
1-^ lTi' LP r—
i
i-H m
U L ' U L^ u
59
(\J
i n— , CLO 1—
c
»»1
•
—
1
^Ll <\J'•
Li.
yu o^ u
1—
1
<\
lO C\J
i/)
•> LO•—
i
#*
DC <\J
DC:': <£X.#\ X
r- (NJ "S I- oO LiJ I—~ Q <« -4 _JvO •> ll
O VO !_-_'
o o ce-H z o
•-* ui— i—
.<: z cc_ .-. c
a a.>-
I—
Ll
C
r\i
QX-~ LT,
in if ;i
«-
C O LUHO C 3: : ii
C-H r-4 C •— -J Cz h- •
I— H- \0 cZZ< JH IIhhk j >DC Cr < < C Za. a. o. . u c u
<<QX+
lO
II
D
<QXII
X<
a 1—
1
<o«. •* *> #*
r- <f" .0
r- ii «. r>
CM >—
i
( r\j
•"• » *. *V
r\ — c NJ
0s •
—
* t*
rH -—Li i
rH#*• u _J *»
<r .j 1— j-
#\i
—
i—
<
«*
c — l— o«'
i
—
—
.
c>
c i—i (N o# — _l *\
!i •v; j 1 J
_J f- -) _)
t— ( • : t—
_JuLT<3
QO
Oc
<r<M *.-< lO•> I—
Li
XII
_1 l:J"
^ —) i—
i
:
--1 11
X '
'• c
L0n
<t —cr t-
h- _i >-<
r-, + h-
O ii oo ~> u
<?
LO uLLJ < r-l
u -
<
< *1. 1
•
-":
•— »»
v- _j rvj
«• i oII (J u
1--.I
1"T
«N 1 f) <|
— •> II
— LO -I
— h- l-l
u u a_i »-i <i— i _ii—
i
lOI— .-
'
.— «\
«— i/) j> .
•> I— II
r- - u r
ox •
c .
I.i 3 H•
if •-;
-' " _J•>
i— _J r-\
LO _l ••
iz <r oo<!" u *II o_J '.A •>
LU CMr r
j
-— #n
<l cv r\i
LiJ
.(
O
LO
r.
' u
i— i—
X _J
LO <I
o u— '0
LV I—
O LO<
_J _J_l *< c
LiJ
O 2:
VJ UJ
r-
u u
60
CM :
- i- .'" h- X c t— I— 1—
a !^ 1- r •: LL .-i- 1
—
•
1 u !U > LUi— : < (X! _J CM > rv 1— '
JV f\J'-.
I
s- ''.:
-. i—
i
_l c CO c~ > fV (X) o c» < lIj ; ro v -_
'
:.:
•1 < .—
1
C < <f i—
l
"
t-H _J :- .•
i—i < * <j;
h c/" u • o '-1 i—
<
":.. a 1 1 [^ OQ a r~- n• o D H o h- r— D V- a \— t—
; <_i
'
'
-rtv i-H
I
-
•T cr
I—
UJ ;7'
-U
; !
o_j -' (V z
1
1
2*
CM C\) CM <X CM CM CM c
.
(' CM . CM a >—
i
CM < U CM' u u_ CM ex-
c c c i c C" C c u c I— cj C u rLl <c C C
—
c c c c c c <. r— c ex c C"
h- c Lu u c cc . c c c_ c c c 0; i
! c a u c < c • _J c 1 T c .j ,'*
I-H 1—
1
1i rH r—
,
1—
1
1—
1
IS) C -T 1—1 o .-i-H c^ a. .-1 ;.•' C 1—
1
.--i 1—
1
c
c c c ,; c ; c h- U >—
i
'.' o- c c<~. u i
;
^ c o I '. O c .:.' UJc c. > ' c o c Q _J c m I—. I CJ3 :
I-H c:--^
a. . ^ ( '3 H-c c_ <" •
i-• c c c C c Cl 1 C -J ;
'c
..: i- L Q c u i—
i
c —
i
—
i
c c. c C c c c c 1
.
c 2.'; a c Q ^-. •r • iLl c Q c;. _J c '0 "I"
r Q C c o c c cr CM < c o c\ c -—
i
CXI J o _i r_: > o [JJ CM a o ^"
C c c c o c c C<~; 10 c •—
<
• < c 1— Q c i
"V I— c '•;
c c ( c c "_ ," • ! c 1- u -• c
'
i—
i
• -'_ r1—
.
f\l-
-? i Pi -0 1
—
;- r • <M -1 X .•- -J CM vl- D CM in •\~Ml '/) o fV CXI ;_"
c c c . c c .;: v c ._) '.T: c : C ~ .'' c >— 3 c 00
c c c c c (.' c C —1 - c U c
' r—
1
I ^^ -J .—
i
kj ( > r—
1
i—i
c c c c C c c u • u ( • Ll c .- - ;-
- a L, I C" ._
1
I—"
!
—
-
O v H . .— ')I
s- 1— D> ! D ! r >
u-"v > <.,' CC •
'"'-'- " .
,
> ,y : >•Q<
(\! c\ CM cm (X1 CV ex -; CN .-, c\ CM u CM CM >x O CM iX iX: u CM r- CM l/J
c c (c c c r ; c
."1—
1
;
"
f- u '•
;
I>—
>
c
c c c c c 1 c r- c o •I i :
,-, c _J c .*. c i
*
C • < UC c c c I c C c c c c • O "
-'._ O c _; c1
C . tv C LUi—
l
r—;1—
1
r-'
I—I ,.., ,—
i
Q r- : 1—
.
o i-H o 1—
1
i—i i-H > ,-
i—
I
4" 1—
1
i>. r—
1
c/J
c c r- c •
c C CM c_ h~ c o c .
/': .
I .y c Q (• c ) c O
c c c c ( c c. C' C c '•--
c: rv o —s o c Cf"• < 1 u •- r~ c i. i i'.
c c c ^ c •' c >. • c » c c c — CM r 1 c c : C • c c uC C c c V ;
• J
.
c a c 1/ r" • c lO V c 1 ex ( j UJ c i. c 1—
I
<r c c c c c C c • c i c • c Q c < o • c • c c O ^ c. i 'i cc c c c c • c I— c Cl f h- c 1 c 1— I h- c r- c —
i—
i
c i—
•
cc c c c. c C r- c .- c c r— c Q. c_ f— V _ 1 c h- G. c— c 1—
-
c 3 c 5 c _•
r—
i
ex r^ <r in vD c- 1 i1—
'
(.A ex U CM ro I. in <i- Ll vt u 1 u^ 7 O vO cvC 0'
I
h-i
C>
f _
1
o
(V
<i
—
_l
c rx
:(V
<n
'-
CV.
•
co
61
oooo
\r\
II
cmv.
IL)
10cm\cO
to
co
O
CD
XoCD
LUccCC<CO
zLU
UJa:
oUJ
LU Oi-m z< —USI— _l LU
•— ccO 3:cO OLU LIa. cc o< L.l Zh- > —o >
Z u <O LU I
cr< LII— UJ CL<r cd <r
Q
OLl I— i—O 7C
CO LUcO ^ >•±C U LUO
O -J
_J CDco
cca.
~ cr x
o
LL!
_JLl_
L_
XCO
<• CC
lj e>* o< Ql«• c.
ZOto
LU
O
CCLU h-ccc <.
O H-LJ <Q" Oo crI- CO
CO >-•-. CO
x t-,
< <CC LU
o
or: ~Q —•— CC• O.
cm —
CCQ
orX
CMLUacr
o
lo
cO LU— ax <
• vcr: uLJ Ocr: COo
I—LJ X> 111
ft zk—
X IDU ZL! —
<
LO I—
Z h-O Ll
U ID
Q
Cr:
Li J
Qoro
i— z
Xo
•
I/' •
b/ •
u •>
o —_J CMCO —
X >a co
ZLU
x 6
* _JCL «<
O 2O l-U L_
(Nj co—
i
Lu LJLJ Cl QQ. < << K- I—
_J < << x ux — ——
'
c oo o oO _l _J_J
cO COI/) >-' i-h
O _J QLj C Li
cr x crLJ LiO C_ Qcr: x crO L_: O
OcO ~sr 2:u •—
O _l_J oCO —
•
m
x co —Q < Z* or:
— LJ OHi Mw < «_J I—o
»- i_:
Li_i QUJ Cl <C. < h< H-I— I—
Cl Di— < ax cr: t-
Cl V. X2 co C
<CD
ocr:
CL
CO
oLu
fM Z
O QL_ —
'
cO •—
•
<> CO
i— -O X h-—
• a z< LJto o :-:
Z LJX < i-o <>- 1.0 I—o to
< CCI— LU LU< rvj XO h-
L_
L. O >C '01
\CLO <.>» OLU O LUU _J -"
LJ CO _Jft a.a <r >:
LJ uOi— L^. IUI.J)2-; >— s. •?—(—>-«
L Vc: v^>
OS -J
co< _i \I— Ll
1
(/)
<i z: ljCi Z _l
< Q.Ll X ?JO U <(/) I-V Ll. U-
U LU OC _i
-j aCO L i LU
io h- s:
>- o z>- 2:Ll i— CO
c0
<
<L
X
<
-2)
O c0I—
ijj —.
U XLU LU
<
— LJ L0
CD — GL,
' —i
a cos: -« u.X o2- _JQ c0
i— —< Cl
>-
<r
CfCt
<r
zt—
i
o_Jo
<-
r> • <
.-: l_-J z o< <i— x d:U U Lli
C' CD'
J- 'fO X X-i o z<r cr lj
XL0 LU I—^ X
i~ CO
C: Z-^ o z
Ll cr:
O LU
LUZ CT
oa <
6ct:
.— CL
Z LU:::: x< i—u
L0 D.
O Za
LULJ ClX <
<X •
3 co!U
CO X
I— QXa. Cii— zX —
>
O L_
LU H-X o
1-
<(V
oaLUx
D I-
c0 I—
X oI— c >
Xlj or
< <
<
s: lj< a-< <D t-O~c za o
i
LU
LU>LJ >
o <X U i
ClLl Z Z <-oou
mm3r-H
O
co h-—
<
<^ ^— X —
i
ro *r
•— W* rrL_l v0 LUa: —
i
h-rr. < O»-. (~ #» X
o_ —
«t o- X
•> vO
L o —
X X Xr'^. ol i—i
co XCC oLi —— zL Oh- CO'
LU lj" ; Or. o z>— <i. UJ
u.
Li X <O f-
I.L. <co o acr:
•— cr: x< L.L) OQ Q(V
CV I—Ij. o zO CO
o. uor lj
rr i—u
CO QIj- Za —
-j-
•-< LU- C^
<CO I—
±c ouO QX l.U
cr. cr\ crco xLU vj
-J Ua o< CCCO O
crll oro LJ
cr' >-
LU h-CO —
•
X <z a
X
>/
ro K-i
«- ijj
x i—ro <•> U
v0 Ortf -J
o•> o
I—LUI— LU
< XU COo <x
i-
o o
IX CMX
Ll:
< a.z <X h-
C C Or-I H3. I X X X X X•.O cm cn cr vO CT> Oro -4- cm c\j ro r-i ir\
< < < < < <r < < < <: :•
.:•: >. x :: :.: x co
a .- q o •"• cr cr or lc cc —.
. o o o o o o o •>
L. L_ U Ll X X X X Ll X XCM C - <f- lO. CO O CM
U U U U U U U U U U U i_» U U U l J U C ; U .cj U U
62
oo
-.0
CO _
»—
*
X#* ZX >—
i
co (—«* ^»
vO 6rH i j
o#\ .;*_
D!— O"*
< •
—
»
C\) <^/
*—
LU _l
0- Q
LlI
o <u
o o;
1
1
_JD. o7) 1—
uu LU
o _jCO
n" <OQi (—r oUJ z
•—
1
UJ0T CL< <a 1—
o f—
1
X Xv0 cCO \T\
< <cco
CO
X
lo
(V
or
<r
)
CM
o
c
:• r-
O
D '
;
- - — — -— C#\ •* > o >-
-1 t\)
;. i'~
h-l
—
.:
.
-T •^j* U I~3 *
.- ".' co ._ 4-. •7
"'•- c\j
1-1 t-H CM •> •^ #* •* — ii
':c I— LU .— r—{ ..
:
— -) _iQ. 2 a —
I
"', ">.—
i
—•—
i
•-' — r« Cv — CC kO-0 </- LY i— > o . J- > Q fO
CL .
., -j- I— T ' .--
.
CO -- •
ii
-~<f * c
) >• ; .?: c
.
|
s[i
;
#* — •—[j -1 a
~-'_J r—
1
i r,~ •— <f i—
1
_
: -
h-M c + ;;•:. j >—
i
«* — c^rv-*
I
—
<t r 7" —
-
rH >- «\
a. #\ V, , ;* } ~> r-H <j" CC
—
.
#* •N•-:'
ol— ~>
II
~i
)r—
1
o LU ~--
'
'
~— •- - . ;
' I— •— <j- #»
\ ^ D .-; •* *—1 a.
*• r—
I
c •.r r N <— r—
,
CO'
r-H —. ;i"
.--'_ II I—
I
Li X 11•••' 1—
<
^ ~) D«s —} •^ _J 00 i U 'j h- ~
-
"-)
I—
I
t\j c i—
.
' *r q; .—i CC, >-< H- i:' cc (AJ o UJ "2L a o
o CM i
;
r— 2; 1— CM —
>
Ll. 5 >—
<
I—<c -:
;,-. r7 i
—
iT; U_ -— —UJ c ;
.
.
O a; 6 X C ir; Zj Ll_ i^ v_^
cc c . :
ex
a u a Cj < cc
4-
<f
U.I
rv
O
Vu
<xu
CM1—
r, E'• O
10 >" ~) rH -' — U i
i—i (V; •-» •- i—
i
~)
>-'i Id .. — l^
. - ^.
I
«• H- >: Hi C-) C O r-i — — rH UJ LU
'H — rH rH _J C«"l ~_l —J
z: q i— u z >- > o •- ll—
- v-t :-l LL *-> Q:' £V i- !—-.- — H-i u. 3 :•". :: z Qijj Ll r> X Uj u ; O O Zq: —• a. co or 5 5 o fvj 1.1 j
. _ :,-
-i.
• X
M L-:
— r\j
UJa<
a
~)
r- *^
<1" 2IT
>-T* rv
LD vO1—
t
LLI <tu~ ••
W) r r—
(
< — O in1—
i
iS •V
-— *• >0
'.e >- c- <tr>- <T —
€ ^->
0. Ll CO 1—
1
""'. <T -^
x ~- «- >-~> — r~- C i
rH <r 1^
o *._
• h- rv• -
—
r—i C\| ~2L
II lO >—
Vc<^ OiXl —
O L0
c <CO CO O rH O- [» v0
co <r co co co co coC\J
CM
0- —. fvUJ Z- Qu X —
D u. ll.CDmhvo r~- 00
^ «d- <t-
u u u u u u
63
m
OCM
(— O *-
vO C O<J- O o
i
—
o ' 9
o feft '.-! co
h-z: ^_^
o.
3 ii— —
. ;— _J
KD v:.^ s
te •-:
li
tP • >I
r-i
n - / ll
>- r>~
i—
1
s •
.
j •> o :-~
•N ^ Ll . D — _J : 1 -o u .
'
D — sj" ;~) -/. v£ ^o f—
4
~) :,' #1 c «\ *• — t\J r> «s m\ Q tf •;
"*•* <r h- — — 3 _J •—
•
U ' 1— IJJ
O (Y sO t—
'
r-Hi
—
Q i-H vf) •o CO .
I— Q c* '^y *- ^ f—
i
1— •1 •v o
>—
i
<o o >- >_ > #* d > C'on •v <t rv <f fV ^L 6 _jlO cm — ->
;
.
I•N •— r-l
1 ;— ^~ ,.:•
j— i\
— i—
i
—- LO T' —
.
— u_CO t—
1
V •" : LT LO '-t— -~ 'T H —: >_/ 2>
m c iv- •> ~— — *• «\ t—l ;
;' / — ~ •— #* . .''
* 1—1' >- ~ —
.
—~ r- .-r lO — -_. — >- > — ;•• —CNJ LO »—
*
(
' v r—
1
<—, LO .. '0 _J 1—
1
1 V UJ _-.*
lO L0 (V — «< «\ •^ — <c Q 1
•>" ' — [ ^.
^* Q. 1
"• rf i—i ••• —
1
. _J•N
•-'c »— «— ;
,
-,. —
; QD w 1 'l
<— ;— —,! | 1 r —
i
'.-'
S <—
.
9> IT 1 .• r~i"
II
—
\
—
>
— «v
3 (M CM V CO O —
i
•> ~)1—
t
-—
i
-— — O :/ _.• C u ^L i—
i
~i — 1—
1
|— ~) i—l!
»-- — r-l J -J
Q Q -,- ^v " — 1 (2 — •— D , _i ^ t—
1
''"''
O 2 Z t— U i o 2 rz >- >-1
, !— »—
4
Li_ ~Z '•-'
1— i—
i
i—
i
1—
1
!L_ Li ^J h- 1—1 •—
i
-:1 Ll. -• >—i 2s K —i UJ
-: "^j •— >—
(
, i
~— ~7* s '.
:. i j«— — •—1 ;." Q ;
o UJ UJ u_ a:"~
, Li. - UJ UJ U_ 3 U_ j. Tk c Z UJ ~'j-l
o (V or '— Q i'i- •-1 O C£. o_- £ 2 ..:' *— !—
'
Q u U_ (X UJ
o> cm co c <j- r-i vO CO <r LO 1—
1
r\i vO a r- lO 0-"
<f lO LO. lO vO LO ^0 LO LO lO LO vO vD ^O lO LO vO rv
<f NJ" vj <f -J- sto O f\ r\ rv rvc C c; C C Cc c c c o o1—
1
<f r-H <t 1—
1
<i-' C\| o OvJ c CMo o G Q o oc C C O cf\i r\ f\i OJ rx1 CMCO c~ re o
-
^ rq
c o c C c O O oc H c f— c c I— -
<t- <l LO LO vOc ;- c ;. c c :
' rc '• c C - c cc i c c c
cc c 6 C
-r • <r vj" • <f <r • <f>
.r^ CM u fM fM u CM
c h- C c •l— C c I— Cc L C c u. C c UJ o— #» <? 1—1 «> cf 1—t » <tc • f\ ^ • rv c • CM
' o ' c c u c c u O'
h~ ^ c t— c c t— cc i-H Ll. r— r— LU r-H H UJ r—
1
-•- - f\ C\J rv CMc ' c C C c.
c c C c Cr— r—i rv CM CO CO
xu
I
64
cccsCS)
1
1
•%
CSS
«
.
c
c>>pa
,
OS-
<spa
302
CMCS
398i
882 084 806 883
CScs
cs.
—
*
*j-nt;&xj^AaW
I
CS1 Jt
IC1
t
"C^K^, —
>
X i>cale-20 unitg/inoh. l Seals- a unit/inch.Figure 7- 253 cps square wave, 8 volts p-p. 200 usee
sampling.
65
NRTflU
Volts
X Scale-20 units/inch. i Scale- $ units/inch.Figure 6- Au to-e©relation of figure ? over* 3900 samples
66
Voi-TS
X Scale- 20 units/inch. * Soale-1 unlt/inoh.
Fi<mre 9- 253 cps sawtooth wave, 1+.8 volts p-p.
200 usee, sampling.
fi7
TfiU
t
V
X Scale- 20 units/inch. Y Scale-1 unit/indh.Figure 10- Auto-correlation of Figure 9 over 3900
samples. >
68
X Scale- 20 units/inch. Y Scale- 1 unit/inch.Fifrurc 11- 253 cps sine wave at 6 volts p-p.
200 usee, saiHpiiag.
&9
TALI
tVOLTS
J
X Scale- 20 units/inch. Y Scale- Z unit/inch.Figure 12- Auto-correlation of Figure 11 .
39oQ samples .
i
70
10
TO
pa jb.= 10 &* 4«7*icT6.
J1 9. oz.tr x
-
10TO
. TO
TOTOr<5
1 ir •
i
B0Q
TO
398J82 1
I
i
88 1! e>e>6 886
••
I1
' i
TO
TO
1 IU LJ Li
10
TO
-
i
_* ^ Waco, -*-
TO
02
IC1 iVOLrS
4
X Scale- 20 units/inch. Y Scale- 0.00£ units/inch.Figure 13- Fluctuations about the mean of a D.C. wave*
VI
("2
*
i «CcO«^.7^-6
PS 1
vz~
CS5 \ /
m %m ^wi mi &er eas ^
i
(S
i
ca \
> ^t ypum-ct *ar
CO
aJ
r
VO«-T<
1
-6X Scale- 20 units/inch. Y Scale- 2x10 units/inch,
Fisure ll|:= Aufee-eerpelaSien ef Figure 13 •
3900 samples .
ji
72
tVOLTS
4
X Scale- 20 units/inch, Y Scale- 1 unit/inch.Figure 15- Noise generator band-limited at 30 cps»
73
-W2
?(0)=r Z<\ CS7
eioa Na§2_^^" w*
15.88 ?tfA*o
tTflU Vfc«-TS
i
J<^. cerrors fkeoueNcy Of &C Ft*-re# /s Zf.Sst/a*.
X Scaler 2Q units/iaife, Y ©oale= 1 unit/ineH.Figure 16- Auto-co ^relation of Figure 15.
3900 samples .
74
eg
t;
cs>
•
OSC5/u
302 eiQa 90Q\ y aw\ / \ m \ zm \
ca
•
i
r>
*• ?K4jUj
i
(30(TV
1 fe
IG3 1
X Scale- 20 units/inch. Y Scale- 2 units/inch,figure 17- Noise generator output. £oo usee, sapling.
r>CS>ca
CJ, R(o)=*.2c >49(SC^j
r—
1
wcs>
1 i
.1 *
©<^> ~*—
^-v.TO y-^"^~ jr ^^^^ >-^"'"""% £
ddQ 308 V / 902 "—
'
Wi ~~— 096 \_aaff ^—-^
CJ&P
1
4- TTUUtJ <rC3CS
i
W<S , , , __ !,.„ .
t
5<$. cur- orr fXewetfcY or sure* k /?? cps.
X Scale- 20 units/inch, Y Scale- 1 unit/inch,tfigua?e ||a Autp=^^^l^|^n @# Figure 17»
3900 samples •
76
APPENDIX B
NOISE REMOVAL IN CORRELATION,
USING A PRIORI KNOWLEDGE OF NOISE
Assume that as apriori information we are given the expected form of
the interfering noise, represented by a normalized correlation function
Rjjjj(IcaT) formed in the absence of signal. We assume that when the re-
ceived signal is corrupted by noise, the statistics of the received noise are
described by this priori knowledge. Specification of the statistics of the
noise by the correlation function implies that the noise distribution is ad-
equately described by only it's first two moments. We shall assume that
the observed date is of mean zero.
Knowing the noise, it was shown that under certain assumptions we
could write
RgsOcAT) - Rg+N(kAT) - ^%N (kAT)' (l)
where %+N^ A ^ *s ^e °t>served normalized correlator output
/*
Roc(k^T) is to be determined.
R?TM (k a T) is known apriori
^SS'
lNN'
Obviously we must determine °^ before Rgg(k A'f) may be determined.
Since the correlator produces a correlation function the author's first thoughts
were to find aK using the observed and known correlation functions.
In the general case, the form of the correlation function of the signal
is not know. For the class of input signals having a non-periodic correlation
77
function, any choice of m, which forces the output to be periodic, ie max
R = R(0), would be fallacious.
Therefore we shall only consider the class of signals having periodic
correlation functions.r
Let us consider the function
F(<X) = ](Rs+n(0) -^N (°)) " MaxC T*o
which may be written
fRs+N (T)- <*RftN (T) (2)
1
A(3)FH ) = Max J(i -k )
-
Since the functional forms of Rs+n(7~) and R&N (T),are not known explicitly
no further analytic evaluation can be made of F(°* ).
Certain features of F(°^ ) are known however in the case of periodic cor-
relation function. If F(<* ) < 0, then <* is too large,, and the output is not
a valid correlation function since R(T * 0) > R( T = 0). If F(°<) > 0, not
enough of the noise correlation has been subtracted, and the output will bek
excessively noisy for small T«
The best choice of <* would then seem to be ^ c such that F(o( c) = 0.
The determination of c< must be done iterativ^ly since no explicit
analytic formulation for the function. 3 is known.
78
Consider Figure 1 below
> <*
Initially set o<p= 0, o<
n = 1, and <*! = —* —
Evaluate F(°< ) and decide
a) oep= 0^
1if F(°C.) >
b) o<n= c<
iif F(*
4)< JD
c) Terminate if |f(«^.)| < £
Then o(i°<p + <Xn
i+1
. i.
^
Having found occsuch that - € 5 F(°( ) £ + € the portion of Ryy( fj due
to signal may be estimated by substitution in £q. (l) 4 that is
RssW = %*i^) - °t^(t) (4)
At this point we have a combinational ratio which causes a periodic out-
put. However there is no guarantee that the new correlation function formed
in Eq. (4) is the result of a physically realizable linear system. This is be-
cause the condition that the correlation be no larger for any T other than
T = is a necessary condition and not a sufficient condition. A sufficient
condition for realizability is that its transform, the power spectral density,
have no negative portions.
79
Rather than attempting to determine ^ in the correlation domain, it is
more fruitful to do a similar search in the frequency domain. The obvious
advantage is that physical realizability is easily satisfied.
Transforming Eq. (l) we have
Pss (k *$ ) = P|+N(k^ ) - ^ PJN (kA* ) (5)
where the asterick indicates normalization.
An iterative search, similar to the previous search, results in determina-
tion of e*c subject to physical realizability constraints.
This is ensured by terminating the search when for any k, the value of
Eq. (5) is less than .00001.
If any value greater than this is used then the contribution will be neg-
ative and realizability is violated. If a smaller value is used, more noise
will be left than in the case ®< =<*c .
This value of o< = ^ c was then used in Eq. (5) and in Eq„ (l).
The results of Eq. (5) were renormalized to have, unity area by the trap-
ezoidal rule, that is
i
L_1
-±-(kPl + kpL) + 2 j5 kPi = 1 (6)
2
from which we haveL-l
k = 2/ (p x+ Pl + 4 £ p^ (7)
2
The results of Eq. (l) were renormalized by a factor 1/ Rss(0).
To evaluate the results assume that the correlation of the desired signal
formed in the absence of noise Rss(k^ t) is available for comparison.
80
An estimate of the noise power in the correlated' output may be found by
evaluating
No = T^f *< ( ^SS^^T) - (1 - o<c) 'R^
S(U T))2
(8)
The signal power in the correlator output may be evaluated several ways.
One method is to assume that the scaled version of the pure signal, (l - k ) •
Rgs( T - 0), represents the peak value of the periodic output. In the case
of a single sinusoid signal, the output power would be given by ((l - <* )•
Rss(o))2/2.
The input signal-to-noise power ratio may also be determined in several
ways. One method is to compare the variance of the noise with that of signal
plus noise to give
(S/N)INpuT= (^ - i* )/<?N
2•
(9)
An alternate method to determine S/N ratios and processing gains achieved
is via the normalized power spectral density, obtained from the transform of
the correlation function. This method is felt to be more accurate than deter-
minations made from the correlation function because of the difficulties
encountered in separating the signal components of a correlation function
from the noise components.
Determination of signal to noise ratio from the normalized power spect-
ral density proceeds as follows.
The energy under the normalized density curve, with respect to a unit
change of index i, is given by
81
M+l- J X di = 1 (10)
Using the trapezoidal rule we may evaluate this as
1 J?E = 4 (Xx+XM+1 ) + 2 S. Ki + 2XS = 1+ € (11)
i=*2
I*S
where Xg is the component due to the signal.
The desired output, Y, is given in terms of frequency rather than unit
change of index and is given by
Yi = kXj/E (12)
where k = 2 ^ tM
** = k- 1
A t = interval between samples in seconds
M = the number of frequency estimates
Substitution of Eq. (L2) in Eq. (il) yields& •
^E = N+ S JL\3)
Mwhere N = i- 0^ + XM+1 ) + 2 £L Xj
Zi=2
i*S
S = 2 E Yg a f
The ratio of energy due to signal S, to that of noise, N is
A
S = 2E ^s = 2Y_s (14)
N E - 2£Yg 1-2'lJ Af
or in decibels
(S/N)db = 10 log! (S/N) (15)
82
2. Comparison of Auto and Cross-Correlation
No method of processing can improve the processing gain given by cross-
correlation, which is taken to mean comparison of a signal plus noise with
the completely specified input signal. This fact arises because of the a-
priori | information assumed to be available.
When in general the desired signal is not known, auto-correlation, that
is comparison of a waveshape with a time shifted replica of itself, may allow
detection of the signal. In no case does it do it as efficiently as cross
-
correlation with an uncorrupted signal.
If the noise can be assumed to be completely specified by it's correla-
tion function, then it seems reasonable to assume that results approaching
that of cross -correlation might be obtained by suitable removal of the known
noise effects.
Lee [_lj has developed models to estimate the processing gains possible
for both auto and cross correlation, for the case of a single sinusoid *
Emsin(wt+ 9) corrupted by noise of zero mean and variance^-^
For a single sinusoid input the desired output of the correlator ^SS^) =
E2 cos w T where E = Em/ fl^*
At the input to the auto -correlator, let N± be tryS input noise in rms
value, and let
Si - -g- - -f Mbe the input noise-to-signal ratio. This definition while not conventional
leads to compact formulation.
83
The output noise in rms value may be shownQl] to be
Noa „ ^44
+2^*.|^)]1/2
(17)
where N is the number of data samples considered. . For sinusoidal input, the
rms signal output ^u«.
Soa = E 2/ fT ) (18)
At the output of the correlator the signal to noise ratio R is given by
Roa = 20 1og1Q
|2i(19)
Noa1-
Substitution of £q. (13), (14) and (15) into Eq. (16) yields
Roa = 10 lo9l0 ; j 4J,j ; 2fi4
(20)
If in the case of cross -correlation we assume the signal plus noise is
correlated with a local signal Em sin w t, the output noise in rms value
may be shown to be \_1 J.
Noc = JJL f.E2 p]
1/2(21 )
and the output signal rms
soc =~Tj
==k^ ^22 ^
[iven by
Roc = 20 log 4*- db (23)
The gain in db, Roc , is given by
soc.0 Noc
This may be written using Eq. (16)
Roc = 10 log ^ (24)
1 + 2^i2
r"
Inspection of Eq. (20) and Eq. (24) clearly shows the superiority of
cross -correlation especially for large ft (ie weak signals).
84
The superiority is expressed by the difference
G " Roc - Roa = 10 lo9l0 i + iJl2 '
' (25)
i The concept of processing gain is perhaps more useful for comparison
of any proposed method with these two standard methods.
Let us define processing gain as the difference^ in signal to noise ratio
at the output and the input of the processor.
_E2 >
ie Gpa
= Roa " 10 1og10 ^2 db -
I
Nfi
2N
= 10 log 10 £* 10 log 10 (26)
1 + 4jV + 2 f i
4,; ,
4+ 2 ^i2
E2 I
and Gpc = RQC - 10 log 10 —j- db'
= 10 logN 3 i2 - C=? 10 log
10 f- (27)
1 + 2 fiZ
Finally the difference in processing gains D_ '
ac
Dac " Gpa " Gpa = 10 lo9l0 <2 + 1 i2) dl
?(28)
The above forms are quite accurate forj1
^ >3. However the quantity
Ni
i i- -q -raus is not the most commonly used description of signal to noise
ratio. Accordingly equations (26) and (27) have been recast in terms of in-
put signal to noise power ratios in db, and plotted in Figure 2. Eq. (28)
was similarly treated and is shown in Figure 3.
Figure 3 conveniently shows the extent to which auto-correlation proces-
sing is able to emulate cross -correlation processing.
85
5 -JO -IS
Xnput SnR xm Db.
Figure %.'< R*ocess»*G Gmt-i vs. SNR
-S" -lo -15
Tnpot 5Nft xh Db-zo
Figure 3: Processing Aov/wrtfcS vs. SMR
-zsr
86
Figures 2 and 3 are for 3750 samples. >
3. Experimental Results
The results obtained for the case of unity signal to noise ratio are ex-
cellent, as may be seen by comparing Figures 5 and 10. Figure 5 is the
power spectral density of a sine wave formed without noise. Figure 11 is
the power spectral density of the db signal to noise ratio input after the
noise has been removed. A processing gain of 26 db was attained, as com-
puted from the correlation function. This is mid-way between the performance
of an auto-correlator and a cross -correlator.
Since results were so encouraging, the -10 db signal to noise case was
then considered. Several problems became evident that previously were mask-
ed by the strong signal component.
Determination of ©< becomes difficult at low signal-to-noise ratios due
to the small signal component in the auto-correlation function, and led to
the formation of physically unrealizeable correlation functions, as evidenced
in Figures 13, 14 and 15. Inspection of Figure 12, the power spectral den-
sity of the -10 db signal plus noise before noise removal, shows a strong
component near 50 cps. It's transform, the correlation function (not shown)
is difficult to interpret.
This suggests that the signal-to-noise ratio measure of quality may not
be the most meaningful for this type of signal detection.
A more meaningful measure is the ability to distinguish a signal com-
ponent in the power spectral density from a large noise spike. This ability
87
may be expressed in db as
A = 10 log, n£§- for i ^ s (2 g)iU max yi
If A is computed before and after noise removal, the increase in A is a
direct measure of increased gain.
In other words, at low signal-to-noise ratios the problem shifts from
one of interpreting a correlation function, to that of evaluation of the psd.
During the course of experiment it became evident that the spectrum of
the noise may vary quite considerably from block to block. Since it was
desired to remove the expected noise spectral density, more than a single
block of noise must be used. By averaging the correlation function of noise
over ten blocks much smoother spectra resulted (Figures 6, 7 and 8). The
average noise correlation over ten blocks was used for the -10 db case.
This of course implies stationarity over the period of time represented by 10
blocks of noise.
The claim that averaging separately formed correlation functions is
equivelent to an increase in integration time, when the process is ergodic,
should be verified. Examination of Eq. (26) and Eq. (27) leads to the con-
clusion that G^ should increase by 3db when the number of samples is
doubled, ie
A Ni4 Gpa=10log 10 IT (30)
where N\ and N are the number of samples used.
This goal was not reached, but was closely approached, falling short
by about 1 to 1 . 5 db.
The results obtained for the -10 db signal to noise ratio input are shown
88
in Figures 17 and 19, before noise removal, and in Figures 20 and 21 after
noise removal. Noise removal had the effect equivalent to increasing the
input signal to noise ratio by 14.46 db. Discrimination against roise peaks
was increased by 3. 42 db.
89
C .
in <— ""*"
CO ^ —
.
'6
1— •- V —
.
U_ ~z X •—
i
cO\—
1
z — —X — ^- fM _. „to eg •N "—
'
t— \ \s/ «- — U 1 Z C^ c<~
• * C CO la- U_' 1U _ »_c *» O h- 2^ D o oro c a u. — Ll 1 < 1—1 <o C <J" —
' t— i— II Xo m — X < < or -- a II II
•-H — < to I— 2 t—
<
cO•* D H- O _J
—
i— •
—
o <r < Z UJ o i—
i
1—1 <rvj h- Q — o —• _i \— jr 2 U#* or x «v —
'
to a o -?•LU l5 Z ^r
-* * #v —- #» —
i
1— LU 1- —
i
t—
Lf> — — — cr — lO o Q <r cO ZII o O fM a r< u h- h- —
i
_iIT\ lO fM r-H rv < 5" UJ LU •
<t r-H _ *_ < CY <i X cO Ll. orV. ~— f\l UJ u o z CO ^-H fV —UJ fM h- -J — — »—
i
l— o o <fV. > Z (— (- — ^t: D • u z • c_
CO #* U 1—
i
z * -— O LU cO >—
1
z fM • zfM «— O 1— U 1
— o Z i-H ,-> c; Q <J r-H t—
1
\ O 1—1 1—1 o >: Li O I— •—
i
O < UJ UJ ^, zCO LP. ** »• >— or or co >- o z 1— or UJ
">"" * z +—
(
r—l — — - o u_ c 1— 1—1 u X X Z'V. — c o •* z it z o cr. <r i-H II
-^
CO h- CM O ,~ w- Li •— or h- a LU h- *II
V to — LT x * — c X u 1- cr < <f z ;"
o- LU rH — <r ^~ < h- c c < _J •
<t CD i— h- :" h- "< LU cO Ul < fM z zV, •N Z Q. *- <J Z X •
;,i or c a < h- i—
I
UJ uo —
-
uj q: c0 >-i >—
1
u UJ CO 1— cr < UJ X XV o Q < CC 0. or u. o rO a_ \— < Ui Q or •^ 25 3;f—
1
lO •— CO *-~ O —:' O _l :" z o i
—
UJ XN. r—
1
• O' < X — c CD 3 i— <; <r or —i r-H CO•-H — c-^ #-.
1
—
CD •> U CO Z z S U_ _J cO Ll. #1 f— cr•> —
~
Z f\j — <.""' ^~ *— cr or z < O Ixl ^~ i—i st 3!)
~i—
i
-p* •> C c Z l- i—i u_ (— >,—
or yO z — h- • a u— O or ( X :• 2 — < a <T 2» O cr or — 1—
1
i-i ^0 Z fM h- urr >«r c O •s C m c* ^- i—
.
Z '3. X o >- V o <J o t—
(
UJ i-H x ol_; tf"\ i-H — C" ~-~ < 1 C!_'
>— CO 3; o u u 1—
<
# or o o UJ oX o f—
I
— O <r <.n 1— (./
i a ;;:: u_ UJ 1— > Ll c o «' X C * 1—1 » f-
O _i — V in — < < 1— i.r >— 3 O _i z ;.-j f-< _J 1- X c^ or x X h- ^cr. CQ or u i—
I
< _J c u fN ^ — CL o UJ u_ • oc X c- #• or r- fM 10 aO •—
4
o o —
•
1— X \C 1—
1
oc ~ • !>" 5 >—
<
O <L •^ v0 LlJ > 9- or t—i—
i
»- :.•: _J cO < •< •v^ X < — - O «a _l UJ • <r o <^ 1—
1
— r^ vO ~* >—i z>Z co <; or ': CO c — lt w c z <ir co o or CO _i ~g.
'"l-H O <r i
—
i-H c^ Ll o>: h- UJ .—
•
cr- ^ o UJ V CO i—
i
or u <r u. \— < « #* i—
.
—i •> —~ *~~ o o I—\~ M < o u < a. 0. II u. i— z _j or X X r X X i-H a i X * X XI— co c z ~ - Z Z in Z _l i- —
<
cO o z 1—1 u_ CO I/' O CO r>i X i-H C X < m X X in mLl. v o o o O — u < < —
-
— cO o >—
<
Ll.' Q ^ «— CO ro <f 1—1 v0 in CO c •4- <tOT j. t—
»
—
1
—1 —< or _J Ll 1 c •> or o u CO <J) > or
or <r z CO CO LO lo <r <-f 0' ^~ — o ?_! —
1
t—i H- 1— a 1— f— h- 1— l— h- i— i— h- V- 1— t—
< • or O ^ Z z Z CO > u. w CO X II cO Z UJ < <r < <r < < < < < < < <r
CD a a 2 UJ UJ UJ LU OT —
1
UJ C 2 c'i CC —
•
u Q >—
•
or z 2^ 5.-' 7F 5 ~~>.> s: *> 3 51 "5 ^:
• o jt-r *v ;r «E! ** 3 D. >—
i
— ;~ X z CT; or i— LU or or or cr or or or or or or or orUJ a: O »—
i
y—i —
i
—i -— o >- a z: S D <r z >— a D. UJ X O c O o o o o o o o o C)
^ • a u Q o Q o o Ul ^- UJ »— 77 7 z z »-< i—i cr l— U. LL LL lJ_ U- u_ U_ u. u_ Ll u_ u.* _J i-H f~y
a •- or c (M fTi o> c: fM r- o i—
i
1—
1
fM en
c Z O c o O O --I i-H i-H v0 ^o h- 1^- h-
o f— o c o o o o O o o O O C5
u1
u.
1 U i- J u c. u U u w u U u i~> u•—H I—1 I-H T—1 1—1 r-\ i—
1
i-H 1—
1
i-H r~' r—
-
90
^T;
_ .-^ _ ._ — — ^~ —
.
^> v. o-
"V V. V. V. X Li 1 X V, x — N. vO u_
rO en CO CO c^ 2' CO CO <t v. CO • II
^ " MhH •- 1 C ~Q O
1—1
u.
Q.
OII
• II 11 r-
O UJ DC1/) II II II 11 CO II II II LL CO O 71
a. LO oc in—. i_ _J LO ii >—
<
y •N •** -— '
—
z r- -.-• _J h- r- -~ LU X •— — -— X X X• *— u s: <£ - lO U_ < cm X X X vD >t
3 K M [V PC 3 OC •—XI— > •v «— — MD — vO • • •
LU CL Q. O Q C X < _J CO DC CM X X • "> — — X • • O CM
1— •—••—• z ,_ O X W '. UJ <£ t-—i vO vC C X X X \C Cm r-l r-l r-4
i— LO CO a LO X II • • r-H ~) vO vO • Ll Ll. Ll UJ Ll.
• • • • a x ;-" • UJ i
—
(T- Ll '—
"
r-l rH O_l O O • 3 «- m _J LO Z Ll LL. O O LL. 11 II ll II
a. Lit uj ^ •
h- z * O CL X II II II OC
r-- (— ~,- —
1
^T z> z 11 II 11 11 Ll UJ Ol ll
X Z 1— < Q. lO V O < <; / 00 CO Li 3 lil rf"
< ~ O UJ DC h- * f- u u LO L; j^* c. a DC O __-.» C
i— a _j m C X' CO U. O UJ X X —
1
u _J ill — a boc a 0. —
«
• O DC — _J co a r- DC -rio? •JD Cl Z z —
*
_j ft .—
.
X CO z X •^ DC >1 7* *"! Z — O • _j r- —* •—
'
Xz uj uj < QC LU < cO X 1— .—
.
0> X UJ r-
<
LO t—
«
to O _l O <T 1— X < <r CO
uj re co '•: O > 0- CO < Q. r-H CM ~P — Z UJ r-4 z —
)
O LO c •—
'
X OC u i_ Z 1.1 .! r- a #v v> 6 CL DC co ro LL. Cu
5 JJO r c 0- O -J -1 t—> X < G. U U u UJ •— t- c c e II
-J -J.Z c u. •— D U V' CO • 3- 3- D: Li; t— CO II 1/ 10 •— «i zUj _< r-. II —
'
I— 51 O LO •* a- LU O O U. ry, || , O r- r-. •—
•
zz "** 3> z* 1— u. < < vO u. > n. Q. X X CO z (— <- (/I LO
6 " <Q O Q 1—
1Jl/)H Li 1
• n <r -J o_ :-." •—
1
X 11 DC LO LO (X.
LO f— LU < 2 a^ _i X LU 1— <j: j£ a CO UJ lu LU
LO cO < ^y V'. z DC < O Ll. < u 0' *~^> z DC DC U DC 1— •—
'
Z U L. (O
O Q . O. O O UJ OC h- LO O 11 z >—
1
a. 19 O O — Z —• '—- x O "»"
l/) •» •> • + —i O < UJ H- — <v 1—1 O z »—
•
U. LL. X CO X O <£. CO DC DC L^
Q, r-l rH r-l 1— 1—
I
U Q i— .X cO X Q LO .' r_. lO 3* vO O. CL -i_
• II II 11 r-H u < O LU vT' f— 0: OC G • L^ O Cr—1 1 II 3 u. u_ ll. je: 1—
1
CO — • u Q £ O UJ UJ X UJ O LU LJ LU O O zII Z h- DC O >-, I— X CM < u U.I U LU M M CD 1— U_ 1— I— t- U_ UJ
—• lj :•" < I— 1— <I U. vTJ r-l u. I 1— t- r- U_ Lu < < < <r t— 1— 1—
H
•^
1/7 OC —« OC DC co dc DC DC W. _J O • Ll <3 <i' << < _. —< r< _l II _J _J _i < <r I— r--
CL CL 0- O C: z u. UJ UJ UJ '_:_ CM II UJ — ** > 5 \- \- O D _o _0 _' -":^** < •
^_ •-. CD r.r.! 00 DC r>" ,—
1
O' _l vD >—
—
—
'
1—
.
. r7* 1
—
U — U LJ Lx •— —
1
DC cvO \. 5 2 O DC <r LL <i' • 1— I— i— 1— — UJ iu u _J r-H _l _J _J h- 1— LJ 1—
1
U_ U- I.L. U_ LL • O D D X to O LU II CO u O LO lO C^ LO O Q G < < — < < < c/5 CO h- Ll_
aLLZ •
' z z a U - .'
DC I/"
LL
UJ Li u LU
O•—
1>—
1 Ll_ u cm c_;
>u U LH LU
X X X X X • X X X T X X .L. X r zz X X -j..» XXX X X X I X X X X
m in in lTi in X r. m in ji .-h LT, 4 X X r- >: t—
1
•—1 CM CM X r—1 r-l —
1
r-l X CO C cc co CO CM
<r <r <}- <]- <r <<• iP <!- <r <r <r ^ -1 <r r~- — CM c^ C"' (< CO c CO (O ro CM O CO 0"i IT' C<v
. c>^ CM r-l
1— (—(—•— 1— 1- 1— t- 1— f— H- 1 r- H- 1— 1— (— t— I— r- t— 1— 1— 1- 1— 1
—
r~ r- r- r- 1- 1— 1— 1—
< < <r <5: s: x 2Q' DC (V DC
<Ol D'
< < < •-' <r
ft: a. DC DC
<DC
<
DC
<,0;
<ft
<DC
<
DC CL
<D'
<
DC
< < <x rbi xa: oc oc
< <DC (X.
<DC
< <X XDC DC
<
CL
<DC D'
d O c c OOO O O O C O O O O O O c O O O c O O O O O OU_ U_ a_ U. U. U. U. LL u u_ u_ U. Ll LL Li_ Ll Li_ Li LL u Ll LL LL. LL. U_ li_ U. Ll, Ll. u u_ LL U. LL
<f \s\ o r- co
r— r— t— r— r—
o c. o o o
.
c ,_, tM ,-, 4- if> o i-H CM C<- C ri C cf^ r~- CT; C> O CXI CO -T in vO 1^ CO O^ O r-i
p- a; CD CO CO CO CO O O O "1 (M Nf\J NM CM CM (O cc. CO CO CO C^ cc CO (^ --f O
o b C r C " O O O ( O C C C C C O O O C O C co C C O C) O r-
,_, ,_ ,_, ^ rH ,—
:
r- CM CM CM CM CM CM CM CM CM CM; CM CM' CM CM CM CM CM C\ CM CM' CM
91
X<
enooo
OTxto
10
oTCO IT;
_0 u. i— O U!O *r .
>—
1
<r •— (TO' #* T~ i— i— z a o0 ZUJ r—
1
00 »-: _i l-H LU UJ u.o or m O t—i i— i? UJ or or > ">
UJ o z u Q O" u y < Q z u. < <*•
2c t—
i
« z U) Q K-
<
or or oO—* z L0 oo z z
C' C (\J »—
1
a. f— Q. c; o »> 0.. 1—1 i—
i
UJ u.. - «C" «:. i— '- Q-1 2 t—
'
ft H-H *£_ >—
i
<i- >—
i
o. 2r 9- e ou_ — z o •^ LO •* •^ •^ oo y oO 3; \£\ r- cnj r*
1!
l0 •—
•
z —
-
I—
1
^ «I
s- vO I
s- oo o r #v ^ CNJ CNI o o3: Z a: "— CNJ z en <)- c. r- I
s-Is- f—
1
in 1—1 ro o iO CNJ CNJ— O O Q. CM I— 1— r
Is-
Is- r-l o o O CJ0 CNJ CNJ CNJ CNJ
CNJ <c — —
'
t— *r. «c_ CNJ C' o i-H r-l t—
i
J— O o O H- 1—."•. r\j 1
—
fefl Z UJ •—1Is-
i—
i
M (— z i—t CNJ CNJ 1— h- z z«v Z •~
<j" UJ r_ or. C z 1— 1— 1— —
i
z Z" 1—1 1—
1
H O » _J X Q —
i
a (—1 1— i— —
<
z 2! z or l— 1— K- 1—
<
*—
*
or orT'l 1—
I
i—
i
Li < '—
H
#^i7 z or —
<
>—
<
t—
-
a Z z z O' or x aII h- I
'' (V X o-~> C* o~ -~ w 1— —
i
>— a or i.' or ^-< 1—
1
i—i a a„-<r ii C>T •v or -— DT *i. z or or Cl a Q. t« or or rr m to
«* vO _i ~ p• 00 a Z Q — i/) >—I a CL k^ Cl a. a tA tA~ + LU r. o LJ £V X ~- X I-H I'- or te w to i— L'.
1 U\r _ IN or —
•
+ Q oc .—
i
m 1— LL Cl w; tfi i— < bO eo to z Q >— 00'
~— jr or ~~
r\j U ' T" '' \— jv*l' t—
1
~i— 3: i— >^—
1
Lll oO Z00 o — "-•. oO b Z 3 UJ X b^ V.-' c >— u < _j u0 J-! z or or UJ UJ
1—
1
cO II u t- «£_ 11 ^~ s z UJ z CL- on 2 W: c>- •—1 or or UJ oO >—
•
•—I z Ll :" sCNJ [V CNJ cnj z CM O —
', «. O *• -< O z 1—1 a. fX a. C Q o i—t Cl z 00 l/) ^r <v0 w — 7^" _l vO •^ rr
*~ Z Z t—
i
t—
H
r-H *v z z z •—
•
•—
<
f—
i
z >—
<
s: oO - -UJ UJ 2T 2
(XI II r—
1
* < n i—
i
•— *ii •* II CNJ
U^ 1—
1
(X II ,—
I
W 2£ cnj ii «N !•• C '• -
'
CNJ z c CO 1—
1
1—t i—. i—
i
t-A i-H i—
1
o <—
i
0> c> c en co ao5 o X ~> c> UJ o * ~) 1
1 Lu i—
i
3 o o o I
s- J^_ ^ r— I
s-Is-
I
s-Is- r— I
s-Is- CNJ C CNj CNJ
•—i 1—
1
I
s- i-H Z5 t—
1
C: I 1
) O H o 4- 1—
1
o o c c c.- O c o c o O o o O o o(—
\—fcA
c Q+r\l
Z oO Wo
I
s- +CNJ
Z a.
O—
t
oc
1—1 Oo h-
•"* i—
i
1—
1
1—1 <—i —
i
1—1 i-H 1—
1
i—
i
i—
(
.—
<
CNJ i—i CNJ CNJ
_J Z 1—
1
I
s- <r 5: t— UJ i—
*
I
s- O X f— Q r—
1
Q 1—
1
z Q Q Q Q o Q Q Q /—
\
a Q o o Q O Q_J —
i
II LU ii z or II <. II Z o < < >—
1
< < < < < < <i < < <i < < < < <f <l
< OT 1—
1
O (Y rH o "0 <—
i
o U. i—
i
c > UJ O UJ o DJ UJ UJ UJ UJ UJ h Lli UJ UJ UJ UJ UJ UJ UJ UJ U.
u CL o s: uoc
x 2£ Q Or 5 u
C 3
I
s-
< OT o or
.—
i
oo
o a
oo
or or fi or or (Y or or or or or or or or o: or
u u
92
(XI
<;~> ^<h- orX c<.i r-H
it <D t—< <h- aX v<̂r
X i-H
rH1—
'.'> O.—I
—
«
CO •N
cO >t UJi—
t
r-l _jzz. — c — _i+ c 1—
1
O <1—1 < #* <-, ^.j —+ ">"
_J i-H _J LUV -— < u_ —
1
u_ cr •
X i-H £< v: O \C•> + O » r—
1
9 "^g
<M V (.0 O 1
—
h~ O *sj
ro u CO 0; CO CO ^ 1
o < 1—
1
C_ *—H r- —
<
r- — <Mcm CD
O ^^CO
UJ c_J UJ
-«.N.
1- Q • * Q _J #S 1—1 ~- •v i-H <r 1—1 r-H ^~ »-2.' LtJ o O CO j- X •> (XI X •^ •* •> r-H r-H
i-i J— o II t— QQ X < cc >-• 1— <f <f < C •— 1-. *• r-
£V ."- LP cm U_ ;-; T> > r .:
'
Z ^~. * i-H i— • •>— — l-H l-H
D- •— •N > 1—1 5«* C" LJ UJ *-
1
#^ O #"l O X <r < — >—
Ol cr 11
<!' X ~s ~?~ ^ I— r-
1
Q a ^» h- i-H C 11 < l— r— < <*A 0. < (XI CO »* O 2T 2? •v —
1
•— CM ^^ «' 1-1 J_>~ < <' 1— r-til II CO o r—
(
— u 1 C- •*1
1— LU r- Li_ ID d < <V _J or O 2! q .—i c 7
' <— W. O c O LO r v: X Q QX _J s/ • <-.'.'
1II c !— 1—
<
c•
•v •ZL LU l-H c r-H 11 11 X X•> < CM u. \
t/ c x CM '."-' — 1—1 c- c- ~-^ Cj <— i-H (/. to II
•~> —o> in r-H M II o < '.'' _j i-H UJ <t
—
c r-H c- •—I i- -1 <T U. \ / 1—
1
r-H i-H + +r~- s * c l-H — • >" c Q H- c. O 1— I— 1 n U.I 5 U #- ^
o o *—
i
I—
1
_J LTV V o il -O r-H 1—
«
r—
1
< < i-H r-
1
c- cm Z II <r <L 3. i— O c O 1—
1
1—4 '-' ?"
I—) 2H r— <' c \ '
II ID O Q II II C' _l 11 UJ Q _J~
H- _J • CM •w «-• CO Dt— hH c — i-H <C O F- h- i
—
Lt_ 1— 1— O h— a 1— U. •—
.
CD O < < CO cOQ < _J 2T 1— i—
!
cv: > 1— .—
1
2T ^1_ ;7 CO 1 ^ ^ ^ i— CO ^-i i:T _l V^ t— _J II .—I h- r- II 11
< r- _l i—
>
t—
1
<r < X 1—
1
l-H Lit »—
1
_l — -1 i—l 1—
1
»— UJ _J »— 7? _J O /> < < f-i 21LU <£ <- ry
-; c CO r\j <x. O < (V Q _J <c L. Q Ol O _J LL O <L u_ C < ~- ~) O O Q cO Dor o u Q. c
C
cO O o. V. U D
r-cc
a C'
co
Oi-H
r—
1
O
U
C7;
CJ \r- in O X X CO (/;
OCMO
u u u u «.^ u
93
C\J
< <y -y
<s>
I 1
CO CO•* •* G
1—1 CM CM UJ—
«
00 CM to >rH CO C CO OCO »* 1—1 r* 5I/) 2 — — z UJ— < rH •\ CM < crLl_ LLl rH > CO CO UJh~ 2! <t <c CM CO 7^; CM ^7a: X 51 cm ~— >: << <O •* O cO 1—
1
!J_ •^ 2: LUco
Z I/)
1
rH *r—cr ^
5f_
11
^ V.cr
1/ C\ _ Oc CM
I/)
r—
1
1
—
•""> •— C v£i r~ .^ <<!' z z * r-l • 11 2* zs: LU < — O CM UJ < CO
Q LU >- — <; Q LU 1—•—
'
—
<
2[ \ —» <» 5" i—i ^_ _Jto r—
1
O rHX
1• ^_ ,
i
l
CMO CO
rHX
1
O>
to 1—
1
rH c ,;
rH CMI-- ^^
•s. *CM
cO rH >-0
CM O•— «N rH r\ 4 2 2" CM f\l — * t-i # 1—
1
*> r-c O >—
1
rH u —
'
*- #* Oi t—1 CM 1-1
• CO r— w > <^ X i—i 1—
1
• »— — O h-
—
O1—
1
r-H z <f <f 1—1 \~ O <£ — •— >—
i
•^ c rH -^r X < UJ
X 1rH >—
1
h- CM < • 2" < <r — h- X • rH •—
1
<r 1— h-<; X #v CT X < CO 1 X O c h- 1— < < < rH «t cr
•}' < cr'.'' < 1—1 a < Q < X II
* < < t— Q 5C 1 CO Q. •. Q LUX 2! '/ X II
-— 5 CM 1—
1
O c < X X X rH X >V* X rH . *> ~— Z c* II a -
' V X c v. < -—
f
to- — 2I"" •rf <—
1
rH II rH — .-H CO II II II X + y-
^r rHII II O
X 3 -s •—II— > rH il —
>
vO
—
»~ 1 X O -— <U< CO rH rH • vO 1—1 1—
<
< h- c 2E CM CM to 1—
1
CM CM + j>" CO v- rH • vO 1—1 CMy 00 1
1—1 ( ,•V CM 2 I
«N UJ O c #* —
i
^v 1rH #^ LU ~*5T
ii co ;>'II —
1
rH >—
1
CO LU X "v. y _) rH rH •—
•
•—
<
CM CO II CM |STII rH LTl i—i D 6
II Cr- r—
1
(\J —
-
O < CM «*» 2* • <t" — — CO CO cr. <MJ CN] «— 2 22: 11 C) < h- O <£ f} —
1
2 O <; 1—
1
O c O < < if) II 2 CO O < 1— O < h—
1
X <: Z ^>" Z rH r— 11 rH 2! 1— i
—
1— II 1-1 »— 1— II < II z i" 2 rH h- f- <<r LU 1—
1
~-» O —1 < 2 — X II < :^ 5; < < CM "; |j_i CM «— O 1—
<
< 2" r-
i* 5. CO u. 1—1 cr O Q II U. <: O G_ C C O 3 O O c C^ . SL CO LL. —
1
rr Q Q C) <X X CO H—
«
rH
OrH
a.
rH
O X1—
1
CM
>- 1—
1
CO
CM'
O
Q 2 X
\l
O CO
CMCM
e
O X X I/} u
f—i
<rc
X co r— CO
CO
rH
a.
CM
rH
a X
CM
O
UvOCM
O
Q
u u u u
94
u
UJ
02Ou'0
cO
o
<
h-
6
to\—
XI/)
o
3
<
inI—
I
II
C vO• I—
I
c: oII HD O00 Q
U
U
I
_J
II
— .J
f cO~> r-— U.cl' 1—
I
H X< V"r.-, \/x>!-- ^H
(X D•> <•
s< I—— X< <i- :
r
< xQ \X 2
+ v~,
7: II
II _l
v7; <!
-
c-
r-H
c
l/-> _j< •>
_J cc> ^
c. -~ ai•^ r- CM #*
CO \ '>"vO
c m »-» •N •— • •-»
r vD CM CO r-H CM CM CMr—
i
•^,—
I
-r 7>*.—
1
«^ r-H
II r^j • ^ -— o II » (M •^
rr r r- i
—
u T l> * l>I" CNJ ii a. » II O II
D IS >~< Ctl CO —
.
_ «N l~l
*' c •^
-I u ^^#*
•
n
oo <t-
*>
"—
o a.
•—
'
o —
'—
o UJ i— • r>' LU o LUvl) ^ j u m < _l •^ _J# o h- ^-< CO h- LU r—
r-H
11 UJ ^ Q.a)
L\'
(—_l
r—
—4
-— t~I _l •-• #\ a «^ •— ^-^ •—
•
—
«
•^ i— -— 1— u r-H ~- 1— -—_
) —
-
—
:
•^ ^ o o * >~
i
•*
— I—
1
h- I
s- i~i a> r- r- * r-02 «— —
<
i—
i
-vi—
i
_J t-H
< u •N O CL _J X o UJ oCD _J CO ^H >—
4
r\j <r V.J r-H aj r-H
1
h- <_j Q LU O CO Q
<_i a
-—i
—
^ < _l (\l >—
•
d <f *^ <~) i—
i
c LU 1— cO UJ c: UJ^- ^-> • Cr. >—
i
X tfl d: #« o:V- r t— u Lu: o
~r r- o +» >—< ->•* Qi — C) o <-e
h~ f—
i
—
-
~
.
^H f D Z> (M ">
— 02 ( CM #• —
>
1 i— Q_ a S" «— r. —r-H — r-H r-H u'i vt) 00
-?r< vO ^i! vCJ
--- ',< 4T* './' > 1— Ll 1 w ^ r-H *v 2! •N
D — O o- or t—
1
Ljl Q 5 ^»" r-H (V r-H
< X <r ii•> II •—
c
—4 r\J ct II II** II
I
—
x u. t—
.
L0 •—
<
X «\ <f i^ ; :. •-H CO —
a x [V V I— «* lO ' O u •> •« 1— »>
x v^ 1 — ii ^ o »- r\i UJ — ts! -— u. —o 00 x • fcf) —
•
>— •—
1
2 |— a.
v. i—
i
>—
1
^-1 •—
1
l/l •i
—
•~l .—
1
LU w X -_. II < \— co — C. — X —or
, i—
i
u. Ll- -^ 1
"
U CO LlI P" h--•
r— vO p- U! l0 UJ•—I ii >—1 >,'c + •— _J v.' _) -J »—
H
rv 0. + T _) V _l"- X -1— — -^ U" 1— r» 1— &> Ll 1
;• UJ tr (M <f h- «• H-i")
—
<
oo ~) -) X — CO —i Q a "< < ^"II — ro ^—4
+ <4 u_ O --' —
-
<t 1— '-h- o »* 6 .—I cc II _J 1— 21 h-
(— .:'_ 13 02 li ._ ")
»--. —
1
LO ;-/"'. o. CNJ o: CM U. •—1 3 >-H
U 02 ••> •* < <! D' »— z — cO •/; vO •• -— ;:-' m «— 2? ~-
I.
'
r-H 1— co <1 ^ -— «- Ll a: C\J X ii r-H •v < *• -^ •^
>-H Li: o II 02 02 _J r^- 2 r- Ir -^ r-H CO (M II r-H to _i h- "s» o
~
)
U M • ~>II ii UJ UJ i—
i
<c ^H < o o <t *? ~> o IJ 1 r—
1
< r-H""". <--
•
o — " 3 7) H c or o ^r cO <—i — r- i—
i
~ ^1 tc c 02 oII
,* _J ll CM "-) ~) ZT H t—i Q r-H ii a. 1— fcfl CO + *T-' r—
1
o r-H
-< <•'0:: o —
<
-— i—
i
—i O t— h- < vl; H- CM r-H cu 1- r < O ._y 02 K- t— II o _J o 22 _j ^ T r-H r~ za 5 1— n Q _l QI'.. c aj < <r; ^ 2T c < _l <c u .j >—
i
cc II •—I II^-~ c < _J — <
\~M c a a: o r— CO o O o UJ < Ll; G < or O ^H o cr r-H c c UJ < H- UJ~i
r-
1
<ci < a 02 02 ucm
O
uvD
OC
>; c LJ ~ (X. •—
<
LJ G. LL.
<Xi
Q a. *r.
a-
2 cr u CO CCr-H
CMU
95
Ha<
uCL
o_Jra
o
z
Q.
LLi
i/; zh- UJu_ ox *•
a- >•
o »2: h-11 <X (-
_l
b"- UJa
u : *•
LO Z«c «-x
U:
<ZII
ro
Li I
c
o
><LU
o oCM zX Li.
u oX 5:a x
(V
uCO UJ
<r ac U)
CMa:
I— UJX 3v-h Odc a.a
r-1
ft cm
CM I II HrH LH CM II
o ^- >: ~>
—1
•—
I— tp oI— < cz 5: r-i r~-
hh or 11
0: O h c0. U_ > Q
ro
oex1
o_(X<:
CO
Or>
XuDCL
fcfl
CM
II
I— vOa +QC CM< xCO 11
D_' rv.i
1—1 if-
C
_)c.
CO*t
vOtfv
—
-
<\J
CM r>
.—
1
f\l
* #v
r- c11 •*
•— < ^~d •^
—1—1 c-.- OUJ *t
_J u
r-r-H
o
<UJ
O <r
I-H L0
— u.
+ —
"
IT —u
X _J<r I—II 1 1
_i I—U_ •-
<r »>
_j r-
o
s.
r'j
o
II
1—
I
<t' -XO \tM I/)
X •
Z >Z) <Q
1—
I
LU
<f h-
o u.cm -j;
2! <C•—
'
COcl' orc
CM
OT<COa:
or-
xuzXQ.
t.0
2: vc— +or cm
< :
.
r-H CO II
CM Q' fM
t— OsJ —
'
Z X. h-
0T i-h Ca :. u
a
c
II Qc
•
C UJx or
cm x >ni—t CM' II
O m CM
2; >. 1
H
1—
1
oi n
o: O .-1
o_ u_ >:
CM
Li V,
CM
l~~ a ^~
II 0: #•
Ci 11 - 2>— O' —1—1 #\ <t h-— h- 4- aLlJ u en_i M r-H <h- c CD—1 1—1 X orK- »l (J
—
—i h- c— z '5—I
1 1—1 Cl Or- or r^^1 a «}
1—1 <f X1—
i
•^<J" u
UJ O zO _j .—
1
D< H- ^~ aLU —1 h- -x
or 1— £' • tn1—
1
i—i l—i 13te r c or > ^~
1/ h- a < CM—
•
i— Z ~~
vO u. LiJ w s0 •s
C- 1—
I
Q -Hi-H X 1—
1
C\J cr :-~
II w *> 1—1 UJ n
t—
1
c :r O •— ">"
c 27 r> r-H u_ #>
.J i-< l0 1— ^1)
u. fv a n. +a. or oj
oCM
II r-« W-> O UJ
r- •-* xr- + zO I— CM •-<
r- z >: 1—»— 11 z
O 0- <H OO CL il u
r-cr-
H- — Or W=3 + r-t Id II
—
1
i-h < o: r\j or rj— h- n_ vo •• «— ?:•> z or 11 ih »
r~- UJ — cm ^h n t-i te—
1
C Q I" CM ~~>C. UJ r-t
O •— «-0 — r- ^ xh 11 a •-' — '."• + z oI— < H h (Vi w z
O z Jh2 r~ z 2E 1— »-'
< uj _j 11 or m 11 z 3 oUJ C. < i—i C O LK •—
i O LU zo: m u 2: u_ o a 5: u k yj
•T
oc o>r-l O
96
\LU
•
>I—
GI/) <
3< LU
or1
1
Q >-
~ZL
2J,
!
i
2* <L "-" X 1
— o Q ,_ —
.
O— (XJ <r, \ •1— 1—
1
u a. c0 \ ou
LU UJ l' UJ v0 \ , „ ^^il
/"!_ _lCO O t:
• \ -') Xa
*v >—
<
(^ i—
.
Ll, .a KJ Ll!
1
—
i— o «2 a o £V • •— c<^ cr2£ i—
"
•— '0 i2L;~
i; C\J O • LL—
>
•. li o i- Cl. i
—
i—
i
1 u Cvj 1—1
— CO £/ :• D <^ iLl Ci — r-l i— s-
Cl oo <r
UJ_i
ou
O H-
O<
[Vh~
ll
U. LP,
•LL <
OCO
*si r. h- CL Ll :— _J L5 U i o ii y~ • ^ _J
UJ — 2> cr z a. CD :
—
Q — r-H i— r—
H
— in_J ~~^ i— <T c >—
i
-V O Li. tr ^-i ~- '— _l CJI
—
< < CO cO <c Ll c^ o CL U ^ -^ ^ cl l0 LL.>~ -— +—
1
>—I
—
r— LT a i— C^ < X U 21" rr U_ <T .—
1
!•"
1
—
•v — u_ < t 1 LU •- -- en L 1 <c \ 'i' (.0 ^—
<
—i—
i
£V I—I c cO I h- o <)' D •< X C; t— —
•
XI u CM* C «— CT *t~ Li to ;.i .L_ Cl •—
'
< ro < < C_ <c >!'- II
i
—
c < C co Ll CT. •—• <. b • h- • "S _J * •-« _J V.
U. lt> CJ O jr: j5 rvj l0 II •— cj- U. < o :
;
+ 2" <—<
.— o ~
2! u cv U_ Ci. • 2L' in r-C- c tO f— "Z. O ,"' .—
.
ID ->Li
' #* r- <f *—i ">ii < T
--< UJ o_ •—
i
Uj >— O O O rV ^£ X u lO -
)
t—
i
5** l£ > LU h- —
<
jI > > < en LL il CO _i U_ r7 *~
V. Ll. G. _*^ <3 I— G. >- T" ». T* a II < ^L X "
.
#v •* X ~~i— _J <r •c u_ a_ i
—
vD in in LD 1—
»
l0 •— 0. ll
i
—
•— CO 6 UJ ... u JV i—i —
i
O LU*~~
^H • .—
i
CL i—i li —i
—
<E c i—i n' f" t—
1
C >—
i
00 «s o ** nr 2'' w 1—
1
U M— c ct i
—
0. LL I—-i r—
1
CL i/ <v „". —. r—
1
lO LU fcr) c0 ;~ — u :•"'
_' '.- o < CD V *
—
<
£ —
1
o • u u_ • ~<^ z> r—
1
<r aLU — u_ —
i
y c_ (— CO o o Z) ^-1 CO LU •> a^ 6 >~ yi l0 2" 1-H CO H=
c X ..— Z-' CL •— ,
—
Q Q u
.
L0 X LL a <. o^ _J -r#* «•• to £ cj U Li'
•7h- -- CI II —
.
r-i >:< •tl'. LV r-<
<c -— 1— fV < o C u i— 1
—
< ,, ii ) *• li L 1 I: 2 tfl _j CO ;-' LU Z?w o ^V* O a >-^ c U UJ I
;
—
< r- m "f" ^r li_ to ii ;
-'- COG o LU U </)
/. LO U) ,— ~r LJ (— • !_• 1
—
i'j_ X ii — l0 i CO 6 ii
CO Lf\ ^£ «— UJ _i u -3»LU < r-l o 1
•— _l I ; ;u < a v:a -^ L
-.
1 o lL' ." < i -: c ^-. C.i Ll i—
1
u i fafl a. Si 10 t—t c0 w1 2:
< crj h- 51
'
i
—
;— < < ~Si—
i
•—, X LL 1»- c0 _\ u c-0 ;|< 1 CO
LU '.• *~
>—i*
C_' — * —
i
!\ _L C * — CD Ll «. CJ ! (-) in 1
if Z <c <_ h- Z! f"> t— d. u_ t—
«
0^-- ~1 CO X x i—
1
• lO fcf) Ovi + 'l; • — feft'
i—
<
o in CO 00 in .—
i
CO — O O ll J c—
i
LU o Vi— •—
.
CD II i; ii II li ii ii ."Z ~-- ~- — —. .-. — X U r—
i
_J o c0 _l 3 •^ r—
1
3 io Ll~ 1—i
—
l
—
h~ i- i— I— il lO • u 0^ 2T II X CO
q -zr i—
<
<. I— >'r
—
LU i— i— c <x. < < < < < II CJ r\j 1—
1
II u i—
i
CJa: ll _J 2." _l :>' u Cj ~> y -r 1' S s. Q 1j__ il LO ^- i-H II 1— *-* C II
[j_l ;E ! l— Ll \~ k—. i^ cc r :: £V (V ci. Di ri z> i—i _J II _i _J ^r. 1—1. 2 ^
% >—*^C _i C. <— or a Ll o O o O G C >—
<
L0 l/j 10 o ,v . L0 CO c ~~ •— COCO C U 1 ! 1- a. — u_
o
Ll. L^
lO
Ll U.
O CCO
O
L_ <olO
IlJ *^ Q <t u CJ
f\i
in
X LL CJ
U U C ; C C^ C./ u CJ u u U CJ u C' c. u
97
KD
oO !• LU
1 • Xu
•
x "; LJ ll r-
••
h- -or
. 1
;
•
u -^*
LU 1— 3:1 c -'
V h- L «\ ^' D LUz ^0 ;- <r O i-nI
LO OO KD ^/ cv U —Ir—
1
>!< a :
' U _J-z _J I J J X • < <oO v: I X LU r
II 00 c LU tj (V_J —
i
U • a LO z cr LU OZ z + < >—c Ll LU .7;
oO 0000 1—
1
a i-i >- ^J_ cQ
•>
w^~
_Jto
u.,
LUr-
UJ •
LU1—
1
I u _l :,: LU 1 Lj <j <^r •
CO — + \ / V LL. r^ -yrD _J I _
u <c 1—
1
? LO —1 LU 2 DII >!; oO 00 u 1- < < U LO 1
_l ISJ u II + < 1— _J UJ 1
OO < * r-H rH y Z u < —<
u + _J _l LOu
>- OLj- J—
U u oOa -h
fcfi bu
COII
zOO \ O' >- z
-—
,
LUD
LJ +
V <}' 1—
1
: Z LD Z) r-H C • 11 Nz II _l -. 00 UJ cc + >- LU 1— M <oO 5: z 11 O LU 1
• LD rv' • 7Z \II _j oO v: 1—1 r—
1
Z h- ~-- or LL CC O_J LO 2 _l V _) LU 1
• X LU • lO •
V < tfl oO v: ;.
'
<T 0^ + 21 _j 3: rH
2: >;<~;* LO lV Q 1
« ;.- Lli _i O II
IT) tf
z1—
1
_l_i oO
H b 1
s 300 _J
<r >- •
1— Oro
w _j 00 ~-"*r 5- -A u • u_ >- <; «C Ll — II 5
^ >; cr oO - ' 00 00 ,—
1
1— a. i-H ;n
l V h— <—, 00 — LJV lO ! II 1 Z in lH »- 2^ 0^; + q; :~ O '.Z r-H
I/" u -^ _l V 00 -~ V t— '. u 1
• r- O LU —
'
^u * lo *?z oO r- C O —
1
-,- (/ tr «•.. £ CM • Q Z)
II
—
i (-0 U c.O in 1— h- -— ,1 UJ UJ r-H u + Ll-~ ^H < ;>"
_J ; _J i—
i
>;- «v <3"
• 5 —l 11--~ O 1-1 _J I— U.
V •* LO oO fcfl _j .-H I-- Q -r 00 . :o / /"j —
>
r-H — II < >;<
LO (M o u V _J Lf> 13 1
',
1 oO a. a: i—
»
\: »-- z; < CV ih r—M u II + >
('c 1—
1
00 v./ LU #* I", c—
1
LL •— X \ {._} r- <^ _J c _i _J u LO 3 <0 •;•'. 1 LU • ^->
• X -^ >-H >- .J u c h-** tft • 00 IS- II u -^ m ..; L_ x a ^ •> + -,: r- O 3 Lli • _J
CM IT. I—) u u l-l II H-
1
' —'i: <—
1
H- < — CM \'
Li I . -i. O n. LUII o LP w II II _J i h- — U_ LL :'- \/ \i UJ L^ U3 X cs II 3 u • <^" U X 00 II QV • II r—
1
_J V v / z iNi \ \ (VI L'
00 D oO 1- .
"
• 5Z i/; J c QL X LO — '!<
c c M _J LO oO 00 o 7." c IP Q u u '1—
1
<3 1 Q < 2! II U. ll~'. >—
1
O0^ II p < oO U u u u 'l • • II II —< >- O: 1 II vD II
•—
1
> r- r— I—. ^^ •
m ;> u •^ 1—
1
O 1—
1
II :
—
•—> i LU •;: O
.
I— C LJ U <r CM
3 --'II II V ^ V ^ CL CY t— Z) 10 r/ <il Z <. < 1— LU . II
O 1/ LL (VJ M — LO u — — a. G O or, O 0' O UJ MH LL 1L q:. lv NQ < |-H Q O X u u U x < LL 21 < Q <^ LJ X Ll X X O Ll <or, <r in vO r- <x; if\ ^Oin m m LO lO lO m ^O vD
u vj u u u LJ U u
98
/
I_
'
i—
-
o+
,-
;
l
II
. i
r*
o ..".
• •
(
4\
I 1 -
-'
m
(.1 l
1—
1
•-
•
1
UJ _
u —
-r'
cn
Li_ h
LU —
•
X -.<
s. —rsj — 7.
2 -/ Cv ^_^ ;.
—
i—( x ii
II ;;.— .::
v/ rvj •—i i~;
< -;- zcc ii V —
•
V D -
O — '"
Q X I— U
I—.r'
D
LUQl
<
.
— ~i
h-•
X --.
I—. ..t•
«N .•
|
,— r*
'_ <t <» *-*i L
1 9* o u~> *,: • .
'
. .. "1 _J r-l UJ «-
U : > NA ,-.
' —' r-' :.
'-;
- ,—
,'.i_ o. a. a ~3
1' | _J rH —l|- ~i
r— V. .' • -
)
—I—
I
x- a. <
;: I- II J ~_ w Q ~)
a. •- 2 <
3 I
- i I
—
LiJ
V ir—
n
O U .
I— —i_J ~) ;-
_j — ::
C 2
vO o -:-
C\J
•"J-
u u u
99
XCO
O co <m I—
r
^y~ ^H II o O •
co- _J Q ll COrn m — UJr-H •> X Q ^/. 1—.
~— »— CO to w I—CO (M •» a z 1_1— rH X ~- r~ Q COU_ — CO \ s . ; co z^ I— •l CM i—
i
0. LUX — CM •— i-> o ac0 » •—
i
D X ^Jxr — II >- <T LJ ii _j
• • CO t— or (— or -~ <o ,— ._ z i— w < ^ or.o O > V or LU ~_- i—o m o z CO CO uo rH Z X o t— Ulf—t — UJ IO _l U_ a•t O •> #v X < —
.
i/)
o Ul — X CO Xr-H or c: CO •N > CO or h-t» u_ m #* X m \T LU O<t •* rH » CM W 3 rHin ^ ^- • 9\ O Q u O r^ r- r-~ r-II O CO o r^ CO -~ LU II o. o o o o r«- om m a rH • 0_ \ "^ ^-~ rH rH rH rH O t-<t- H CO Ll o « *—
*
V o rH\ — a ll r—
1
r- ^>" w. z _J o o •O O oUJ or » _J Ll. • 0' O u < 1— h- h- I— o os < »-. < II c LU U' II — r-C' m c _l c—
1
h- or —
«
CO vO o o o o wCM or m h- -r LL LU r-H LL M? _.' o o CD c9 O O\ » r-H — < • Q o o — o r-' O r-~
UO ^- _^ o h- c X CM U IXJ i—
i
#> (O fcf) »r> rO (f> CMH o z m < LL,1 CO 1— II I— U^ -^ ^. — *">"
\ in o_ .—
<
or » ._J (— — w <r o -~ CM — oj ni •^
CO t—i i/j —
»
a h- u. r~- 0. .'.' v: _i r—t CM ^ cm :-: s rH\ — n KD —
•
U • 1— —
i
— u UJ •^ Z> •t « >".
ON z io <— no LU c 3 ,
- or ii or <J r rH •> rH rH II
<f z a CO r< .—
i
o 0. <c — or o rH J." rH ^ ;; ^r
V fV «. O or Ll CO \/ O rH ^r II V II II V
o •> -^ c- a o • :..rr. — u ,
*^ iT ;r ii "Si s .—
.
< — C ~ i— X -a to CO"~
«> ^ •• * 5To m — CO r
—
CM Ll.' U Q Q o •* ^~ •S r^ r-v w> ^ODO in i—\ O t— • < <;' * 5 II (O LLi rH ~^ •s ^~ rx ;;
r Z +I—1 w m U_ o X -~ X CXI •—
i
CO — a ^V »v 2; ^« >" i_ ^^ a (XI
X CO — 2 <—t •— rH 0l X a • I— >- O %/ cr CM — CO — z or CO >~
o UJ c0 C — X u_ _l r- _j CO < m — to O r-H C CO o z o < a II
CD u CO CO 2 CO o <r • •t U_ _J (Y r—
1
z Ll o CM r-< co CO z co no CO CM
o or a. (Sl ^ » — O <~ w- _J (Y »• z u LU • vO V ** or 0. or a or a 7-"
^_ O" X r-l X r-
4
X CD X X X < X rH or ii or r—
(
II r-H I—
1
a z z y _l o X co u. ph < CM 1—
1
<f O II '.O — a II CM II c IOI— o o O < rH I—
1
(XI II C\J M5 * f—1 in U vf. u V UJ _l ' ~> l—l o o o o o o u. Ulh- r~r I » »—
4
»—
1
LU — — — •~- *— «-< — •>^ w tf <N| n — D rr — o o o c o o rH z> 3LU <r CO CO '0 Or 1— h- I— _J \— y~ t— 1— I— .—
.
CO — z Z v-i c te r—
1
CM CM CM CM CM CM + ^r Z'OT or z z z < <£ <^ •— <r <: < <J < c 1 O ^ a •—i C: o o r\i l"H >—
1
01 a LU UJ UJ UJ 2: ^rJ5~ M 75~ 3c >_ ^r 5; • o CO — co r- Q 0", r-H CO 1— Q Q a q o o % h- K
< • o tt jr ^> o_ or or or or Oi lY or or o or CO Q z < II < <. < <r < < ii z ZOf Ll q: i—. .—
.
»—
1
> o o o X o o o C) o II in O I/-
) CO O LU C t—< O o u LU LU LU Ul U r—\ o c* « Q o a c H- u_ Ll Ll vO u_ Ll u. U_ LL u rsj O or a. u or o "> O LD or or or or or or 2! o *^'
* _J rHa » o •—
i
CO <f m ^o r- 00 CO »— i CM co <f in vO r- rH co z o O o o o o o C) o o O o o o o o C oo r- <-J CM CM CM CM CM CM CM CO rH r—
i
rH rH rH rH i—
i
CO ou
1
IL
1 U u
100
Qz<
ocmLUNtO
N
in
C
O
2
or.
i
i j j ("5 ->
z o +Q o U)i/) tfl .
Q. i—tO
orii
a < _J —z LL) — i/"'
< _J rvl i/)
u 1— _J or.
\ < •>
U <r 1- •>
+ 2! Z O'Ll. to LlJ \/ • OII < X c ^ uu_
23> ii _J
<or
O otA *-"* X
UIU
•T
• Ll 1—
u1—
U_ 15 or : o *—4 < —Iii O <r 'O —
-
rf (\l \— —
~
u_ <• — z LU i
#> _l —
.
0/ _i LO o — —
i
h- _J — ID U.I _iIII — < • II Z c ;
r" _l Z ZsT to «^
Oi O 1
—
to 1— o <f -J •—
'
< 3 —
i
o 3 zI.L. Ll
l> oLUD
LUV
toa r\i
•*
cc ^v LJ '.0 rsl
I—
I
•
pzor.
Q LL Z •—' >;< <t • ,,:> C5 * _J II z >r
Z a" —i 1— tfi i <t <j* i—i L'^ i—
i
_J < ro < •— _l<r tfi LU 1— <
c i i—
i
__ c+ o + 1— <
1rv tA ^^ o
O•
<;1
co vO r—
)
La J o LU • o o — c c -4- Z • '' :-/ vO — O m bo to vD r—
|
— •
H- C r\:1
1— u Z o • o _i vt c *« V m */ » (\J _l z •— rsl * a CNJ 1 _iLl_ • .—
1
_J < </) II f\l •—
i
— •> r ro II c ll _J i—
;
^- vO r—4 C 1 r-l •— »— UJ_ O # II X f -* LU ~ ' v. ^ *c o c C: 1— -i I— < » z Q f\! f • U ! #> 1— * o^ NX 11 r—
1
— 3 <i n or <J -~- •—
i
a o c <r <i — :~ •V i—( Q_ to r—
1
CM vt i\; 1—
1
* — < -—
i
tO U_ II _l Ll "V <r Z ii to <)- • .— ^: c N/ <t- II tO a. II + —
i
II —
.
_J ro _JV _J — Lu 0: si 0_ (/» o _i o —
>
1 LU r—
i
r\j o _J Q M _J — <l _J _J _l ~— q: <w to 3 o i
—
i
_J ~,r tO + to — «~ ~^i w 1 W <M to LU + _J -— < -^ N ii s:
a CNJ h- Z~ r- <r < 1— o o_ <r a _J _J 2* >- I— O n i— — in «— *S. r^- N II-- or
'D O o u. »—
i
CO or z • < O ii — »-' r—
1
•—
i
Ll: _j 21 _J 1— i—
i
ii 3 t—
i
c;> N c or o II— _j o
• c<^ 1—1 I— c _J O o 1—
1
1—1 — ro — M rsl 11 l— V < v: < Z-' Cf1 — a — 0^ + • c rr —
»
_J — zt- m X z ( 1 _J z o o II II _i •w •
—
>- Z ~— n -^ II >— _j "S" M ro rsi z —I — a UJUJ II O u. o f\J <T 1—1 LV a a _l o ~ U_ u_ u C Lu Z" u. a. (X O — 6 II O n n LU o ^^ 5: —
i
LLCO u O V LJ
o
1 L; u a < <Lf\
o
O n.i
CM
O o oen
<
o
<
o
a
o
Q
i—
i
u < Cmo
l- or O M IS to
r*-
oCO
o u U LJ u u LJ u
101
f-to
<— _J1— #>
tO c< **
_J UJ*> •n
.—1 vDA •n
CO (\j
*» »v£) C\J
*« #. er\J <\) (NJ o.—) »• r—
1
#*
•< O •k <fr- •> r- *II vO n o•> O O—
' oH- O«-- i— <— LU•v •—
i
#* a.m r> lT\ < vOO _J o z _j cCO LL' C\i o
cf\
vO n < Q h- • 1— "->
r^; < _j < < o *i ->t—
i
UJ ** LU -J o *— ^-
II ex. o X LLI 5 or (Dfsi tfi Cd #v CL >—i
E W3
Oy=> X.O i—
1
wto
wvO
5:•- a vO
U tOCO —
)
o •• o o - _l CO O w• LU 1—
1
1.1 J r-l .—
i
<r I— c\ MDi II x n ii *ri Ll. i
X O U_ 1—
t
u. a. t—
<
a i—i 1— ^»>—
t
CO X #* #* ^E •v •—
•
X 2" ~)
Ll a <f -^ ^ ~~ to to • i—
i
»-— v£) * 11 1—
1
NJ tf* >—
1
V 0. vO o1—
1
r\j -- a _l — t\ — O *• to a <^j UJ•— i—
i
_j LJ LU 1— M h- h- LU a. N- r—
i
c
o ». •«— h- CD t—> JE tO I—
<
1— _l tr *| LL1—
1
i—
<
o 3 < —
•
— Lii — 3 ~-- ~*> r—
1
•
w, ii•—
i
0- _l #* 3 ro r. D. 35 to .—
i
II r-s. _J to ;s- lT\ < X LP. J." < LU c ~) co ii b tf> o or <r o o a X cm (M• 00 — u f\i Q II r\j u o o
1—
1
o _J O _i \— H o h-II c*~ — 2* II o _J LU o ^ _i Zf ^ ro -v
X c < Q < _l ro < < _j —
i
>— i—i
— O »—
i
IX O LLI < '< LU C£ << (X a: C cru_ O tO
aO
O a: U —
1
or Q u a a c
o
— -< Q
VJ U u
102
CM
XX
ccc
ccc
CO —1 C\J
LU IU <» h- 0- 1— 1— O LU5: 2: O >J- r^- •—
*
r- —. —
1
5 -HN co :sN-l »-» jjj O O 10 O CO CO —< 1— i- t— —t- H- * C II II 11 II O 11 II II >_ _< _ >- 1—
i- >e a-LU H- -> O U _I —O 2 JZ< U >»—
' LU CO LU _l CO
-OC <— <0_IOOO~)0cO UJ ClOlOUMNNUlU
LU:>
—
.
<f
• o»—
vOv. o01 o>•
<j- o I— H-• H w UO Ij — LUr-l » LU OU (/I
2.'
CO o
II
oO H-_l u
2?cuLUto
I J. LlI
+ O-- y»<N O
uLUQII
OI—uLUQ
2! I- I— Oen
rsi lu lu\— ^ cc*—
I
I—I I ! J
•< t— u.i-H Ll.
1—
t
h- LU
2T J^h
— U— UJLi. Qj— •>
<; mO
uLiO
O Q.
oII
oOOli uI T t- UJvo co <r oM (NJ O tfl
orawoI—L.'
UJo
I
LU UJ
O <DC 1.CO CYD OCO U.
<r 11 f-5: h- toor u 11
Oli-'XLL O X
oII
oh-LJLl
C
D
o zU LU
en m r-H r\i <}-
U
103
ot—
«
COSI
CO
iot^-or.t\jr-H-j-^£)onn:)C<0<j-<f<—cr-OvOmcorMOO-
r- oo c h o m o ^ m<rmcocor^oco.-ir--vOvO-JONOvO^^ih-m m. O •—< <r <r CM cm vO cm co
C\J .—I *X) r"- CM <f LO.
CM f-
vC O- <l
^ocM(>oov£)r-HaNr-ir-Hr--^tc>oocoooc\jv£)cornocot>mr--vo>£>co<i-cr>coo-<-"o ooo\jo>frcovor-
<CYOocY •
a COUl
2 _J•—
«
a •
sa < COuj CO Q*<" ZqE o oo in uLL. r-~ UJ
1 CO COLU 1
Qi • Oa en
_J u< —
<
• z s:CO CDLU — cU co co 00a: LUCL CC U.
D Os 0_< cODi u. ao O LU
o i—
or Z COa. o
—• ^c
o 1— •—
•
h- <r
_l CO
< LU COh- DC O< a: ccO o u
u <1—
1
D o oQ_ 1- <Z D LUi—i
, < CC
I I I I I I I I I I
^0(^^nc^H<5ln^oD^ on r- co en cm vo O o i-h <m <m^co<to<finHO\cor~-criHpjfciooHi^(%j(M<rhvO o <f O vO vO m m cm r— !<•• r\j a vO fi c\i \0 *o co cm covO cc o oo \D en <t f. ai (<) m it. o <—i <r r*- c\j r— <f co <hr~-0(Mvor~-coG^f—tcocooor" co co co in r- r- co cm cminvOH(\imin(Mi^cor^f>Mr--M c o o >n r- o •—
<
co r- r~- vO co <r o cm o c a-' r-i c- c> cr- ua co vo vO co mI l I l l i I I I I
corner. ovjOr-Hooor-r-ocor-c vO o r~ >n co cm cmcoocMO^or-<for-Lnr- a c\j <j- c ^o vO o cm co mr-a'G->£>ocooc\jr-ovO<Mf\i«x)oi~~c\jomr^voCOGO--HOOjr--0<f<rOCOCOCMr---C<rO<rcOoj r-i i—
i o co cr- c c kq c. cm <r a f\j c o o co <]- co covi-o- vo.—i-.-±<MOcoa-r--vOvOcc <r on o vt cn mm oor- m co <t 0> cm 0> o o —i o m cc m r- ^o ^o co <r a> (M
I i I 1 I I I I
hH^NcOOOvfcfi^axOOvOC^OMOO-tCOco vo c<"- o cm m co o c\i vo m m. a cocr-vtvOcOi-Hcoomi-HCocriOcr>m\or--oj-<t a. a c \0 cm r- r—i o co <—
i
r-- a; vc <f- o <t r-- o co c •—< r- or r- r c cr» m c cr- c
CM a: cc vO cr co i\i c o- <r cc oc t~ c it cv <) o vO co cccoh-mcoo>—iOrH<j- o ^o r-- m co o m co co <f co <f
co co cr r-H cr. c o cnj (.^ <f co ir. r- h^co<K>cjooI l I I I i I I I I
CO c vo co co vO cr m o co cm <r cc cm o .—
i
<r cm ^o <rv000CM00Mv0C0v0H(10CJ'Oa.Cf C" CO vO m cm cm cmmoo-'h-<tor--vOooir>croooi— o m r- r- <J- o coc<" a- cr- co cm <j- <t <j- o. m co a oo co <r m r— <t- oo cm voc (Mcrr-r— inincomcocMcocT r^ <r m ro 0- co in mr- <r o m co in co <r -
o i-h c o o cm or-t cm it. vo r- c
I I I I I
J- co vc r-- r- m co co cr cm o
II II
10 o o o cr* oO O i-H
I I I
o c> a- <r <J" 10 cm cm a^ <( 10 r—< (<". co 10 c in <t lo <r >—
1
O \D co in 1—1 <f ro^-iOCMOc^ooco c--hoocmvO^O
C <r r-~ -—1 co r~ <o c> <f vO c\i m m co cc; ^h cm m co ro cm
o c<~' in co in cm ro co co <)- in co r~- cm co r-- c<^- o 00 i—i r-
c h e-' r- o N (M n h vo m m c<"i in O'l^ o tr co cm t^
cc: r^Or-ioocM-crOoococr-invOtn cr<fcoco<rc\ic--'a-(MO><j-covO*hiocorf'Cj> r-Joooc\jo<t
— I I I I I I I I I I I
104
cc ro c c co cm lt\ c\; Lf <: co nc f > i\ og im co (\j r- 1 .—
i
O a' <} vO <f c c c c c c c c c c c c c c c c•—
'
c C\J iT\ o c c c c c c c c c c c c c c c c ccO oj o c c c c c. c c c c c c c c c c c c c c5 C r-i c c c c c c c c c c c c c c c c c c ci—
i
c fl c c c c c c c c c c c c c. c c c c c coo • c o c c c c c c c c c c c c c c c c c c c>-
<J~> •1 1 1 1 1 1 1 1 1 1 i 1 1 1 1 1
< LL! >or _J u oc <f vO f\J <f a r- r-H o vC vf <*• <r f CO CO C* •—< CM ,-J CM<3 a z p" U\ CO IP. rC, CM r—
1
f—
1
c c c c c c c c c c c c co 2T UJ -1 in o r—
)
c C c c c c c c c c c c c c c c cK <r _•> r-t ON ,—
1
c c C c c c c c c c c c c c c c c ca. co o c m c c c c c c c (
;
c c c c c c c c c c cUi c o c c c c c c c c c c c c c c c c c c c
^ o a c o c c c c c c c c c c c c c c c c c c c—* Ll
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
Q co OLU or r<" 1—
1
1—
1
<r c CM CM <!- vO <j- <}- CM c~ cv CM CM CO m CM c CM-T •« UJ C CO r- r—
<
CM c c C2 c c c c. c c c c v~ c c c cor M I
s- vO <j- r-< c c c c c c c c r c c c c; c c c_ co _) c vO r-
<
c c C" c c c c c c f. c r c c c c c cLl_ < Ji" c CO c c c c c c c c c: c c c c c c c c c c
12" o c o c c c c c c r c c c c c c c c c c c c
U_i a or c o c o c c c c c c c c c c o c c c c c cCK •— LL
G. to•*
1 1 1 1 1 1 1 1 1 ! 1 1 1 1 1 1 1
UJ • (X in <r CO r—
1
h- vO <\ c a r^ CM <J C" CM CM CM c CO r—
1
CM• O' to cc h- r~ CO r- ro 1—1 -1 1—
1
c c c c c c e c c c c cW> —
\
a r- i—
i
r\i r\i c c c c c c c c c c c c c e c c cU a l C c cf c c c. c c. ^ c c c c c c c c c c c cu c vO c c c c c c c c c c c c c c c c c c co Ll in c c o o c c c c c c c c c t c c; c c c c cor O c o o c c c c r c c c c c c c c c c c c ca
>u.
O 1 1 1 1 1 I 1 1 1 i i 1 1 1 1 I 1 1
"5 (—
< »—
i
l0 c o o <t CM CM rvj <!- c LI vj" CM c CM CM CM «—
1
CO r-H (\. CMor oo Q. cc >.o ^0 cc CO C c c c c c c c c c c c c c c ciD ~z. 'j_f r CM CM CM c C c c r c ( c c c c o c c c c co u 1— r (X' r- o c e c c c c < c c c c c c c c c cor o 00 c c—
1
c c c c c c c c (. c c c c c c c c c cCL c c c c c c c c c c c c c c c c c c c c c
_J 2T c c c c c c c c c c c c c c c c c c c c co <I •—
r- ori— tr;
1 i i I 1 i 1 1 i 1 1 1 1 1 1 1
< v_; to oo OO CO CM r—
I
<r a <t <r vC in <)• <J <t CM CM CO CM 1— CMH- LU O r-
i
r- r—
1
vD C' Cvj r—
1
1— c c c c c c c o o c C_ c c<r a or C co <r <r I—
1
c t c: c c c c c c c c c c c c cCs 00 u c <r <i o c r* c c c c c c c c c c c c c c c
< c c cm c c c c c c c c c c c c c c c c c ch- or c c o c c c c c c c c c c c c c c c C: c c3 Li- o o o c c c c c c c c. c c c c c c c c c c cQ. 's- <Z 6 LU 1 1 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1
•—
i
a or
105
I/)
i—
l
WO
r-co<j-voccvOiNj<t-cco cm i-h cm • \C r- \D r-i c<H o r-i en sA o r- c; r*-
o co lp c c,s co ^o in voc <f c m" h co C' rci o c o o cm o r-t r-i
o c c • c c o o o c c o o o o o o coco o
c> ^J vO vO o CO <c CO sj lp oo r-co CM I
s- CM r—t c r- o r-i LP CO 00r- CO r- CO CO CvJ c r- co c o CO<t lO m C rvj CM i-H IP. c <J~ ^o C\J
c C\J <* LP m CI c <f r\j CM r- Hc rH o o o o c c O o o CSJ
<oroocra
illCO O CO C if'
Ovl I
s- r-l <t LP
I I I l I I
CO <f i-H CO r—I O I
s-
vo cm co co cm o coIs- CO vO vO I
s- I
s- ^H
vO O vD 00 •-
ct iT\ c
.
un •-
co c c C i-
o o c c c
r-iI
s-Is-
<o CO •—
c o cO O C
c cC C o
co co s i—
<) CI f\J -v M I
s- Lf
c cc c
I
C". ir. O C «-h O- vO coO vD i-h <r lT' c o o^o co in ix oo ^o r— —
i
O CO C C\j rH CO CO o^o oo ir r-- I
s- oo r-i vO
O O r-i C O O '"H Oo o o o c c o o
nu_i
orou_
I
UJCCa
1/5
U.I
Uoora
<or
oa
o
<
a.
l/) CDU.i o_j ^:o. os; u<r uj</) Lf)
I
o oo ccm ur- h-4
co 5
• oo
Ll.' CO
—< u_
o ot0
ll. aO LlI
i—
o
<LU lOor oor or
c uu <
I
O QI- <Z> Ui<. or
I I I I I I I I I I I I
4 vD CO 00 CC CO Is- LP v.0 -T i-H CO O CO CM <f C" vO CM vO <T
c", Is- i-H a-i-H I
s- vO c lp a: r—
i
cm <r •—
'
co c lo a. -4- 1—1 ok0 C lO vO C C\' lP 00 uO CO in I
s- CM co O C' (\ A N O COLTi vO C CO O Lf> <KMMvOlf\OO^Hir IT. rH O vO LPCO C C l-H in C oo C r-l LA IP CO s
' O <] r-l c CO. C vO <fi-H o C C C C O C C C O C O C• O i-h c c c c Cr-H O O 0"0 OO O C o o o o o c c O O C O O
I I I I I I
co <f c c ir co I
s- r- -* lp. <r <f cm -j-I
s- 1-h r- —i <t ^o m
I
s- C> vO O- I
s- Is- C^ vO C I
s- vO n0 O rH LT' c^ 0- <—I LO O C")
co O if' c- o- r r- c vO ^ co co rvi <)• c. 4 v o o^ rH ccv c cr 1- c -
! i—
1
<t <r co c-v r*-s
I
s- a c ic <- <j a> cm
c vO c c <+ vr lc r- ip .-<<}- c r-- c u c o lc ir\ <j vf'
r- O C C r- rH C O C Or-tOOOrHCrHC C rH OrH O C C C C C C C C OOC C O C C C C C c
I I I I I I I
<fv0OCOr- CO -t —1CNJ rH rH vO CM CM <t O <t O IP. CO <t"
O". CM CO CI Cl LP O LP. CM —1 <f oi >0 1^- LP CO C - CM PJ O rH
m co c <t-^j-ooaccisO<t"r— voc r-^cpi— cMiPrHCO r-l C C C P. LP — C vO j- ClflrlC O" C" I
s- LT" CO ovO ri C if. C 1—
1v£) co \D CO I
s- vO C PJ C rH 1-^ O- IP LP LP.
a c c o c <-- o c c c o e'e c cm c: cm c c c oc c c c. c c c c C o c: c c2 CO C C C C C C C
I I I
C CO r- vf -1" C vO CO I
s- -H PJ CM r- 0" 0O CO Is- CO ip, C CO
C CO CO LP C CO CO ^0 vO C OCMCMOOOCMr-O- <f C Mc cm *o <( vf) cc h m co a —< p- lp cm r- <r ix
; cm r—. a <f
c~ ai ex 'i 00 vc 1—
i
rH 00 co \0 co lp lp lp i-hI
s- vo i—' r-- i-h cC CO, PJ co 1— <f- C. I
s- CM vf CO CM <f r- •- <r I
s- CC i—
I
o oCC'CCr-,CrHCC C- C- CCOr-iCt-CrHrHCO O C C C C O C C C OOOO O C C C C o c
I I I I
106
13
tO
to
i-h <t c- i—i -o co .-h —• r- -t [•- r~- «X5 ex oo vO Lf> in c rn
C c\i r-- m vf) <f \0 h i> vo vO wp- tr r- >-
< m m .-n -t r- c> r- r-N (i i—
i
CO <j ^o oo c i*~ F-- c mrx ex txj rv ex ex rvj (' <vi r\i rx
r-i vO u- cr ro m h coo rvi .-, o-c co m vOr- in <h rvj i—
i
r-» i—
i
or c O C C O O C
C c o o o c c c c c cc c o o c o c o o o o o o c c o c c
c o co o c
->_ «
< >-
K u15 "Z.
O • ll '
(V —
)
C_ 10 ~'.
UJ UJ^3 _J «
1—
1
a U_•»~
a < oUJ i0 £V"".'
U-i
K o MO oU. in
1 Is- C
UJ ro o:or. u_a •
UJ •'
(/) »
• .—
.
lT>
to o a.UJ ;•:' <^
uo u_ inCXL oa u_
>- O'. t-< I—
H
IT)
a 10 a.
O -i UJ
O UJ i—
a: o lOQ.
_J .7"
o <r >—
*
f- cr:
(— to< u lO
h- UJ O<r n_ or
Q (.0
<(— (V
Z> UJ ca 2: <l;s O U'i—* a o.:
c; rH <f->• rr oo c .; in r- c <f rv c; -d cc c <fr m a
I
s- or- r-i ex rv r\i > cr rvj r\j oo '-n ac Lf c m ^o vo <f <r cr-i rr u". ^ lp cO p- vfl vO r> oo co o- <r i—
'
a- c CO Is- <f
fXI ,_| C (Vy ,p, v-
J-- Q r^. (^ (-,, ^ cy, y. y- ^ fXJ f^ ,_, r_, ,_,
rvj r\j r-i ex rvj rv rvj co rvj r\j ^^ —
:
r-i c C O c C O O Co c o c o c o c ^~ o o o c c c o o o o c cc c c c c c o c c c o o o o c c c c c: o o
a" vr C if. c •!' m I ro o^ co c m oo ur cc ct if >o (^ <tc if- rvj r\j i— in m >o r- a r-* vo cm •4- m k> h^o •* cm<t c C I
s- cr r\i t-i >- r\j ry- co <f O- m vO cor--- co mr-i CH if. in 0> 00 -'C'i ifi i—: a c^ cr: ir C^ ro rv: h r-n r-^
n r— r\. f\j rvj rv r^' rvj rvj rvj rv; rH r-s c C c c c c C Cc c c r c r~ o c o c c c o o c c c c c c cO C C C C C 7. c o c o o c c c c o c c o o
r- c if. in in t-< o vJD C c"\ .1 rr c C <^ vC a <f ro mo- h o L! r\' rv r< if C -t <f vO \T- -t ^0 COvC> C O O M O --4 CO vO CO r- c f -J O o if
rv c C rv vo if O r- a' <j i— r- v* <i' k <r r rv.: rv o o c\ n rvj cv, n rv r- c c c
4 <r O ^0 inC n •—I rH r-l
C C C' c co c c c c c co c e c <:- c c
o c o c c:. c c c c c o e oc c o c c c c c c o o o o
p- <J- ro ir -.o ro <<"C c — '.' r\j <t r~- ^r cv ir <—i if oci—i r- if. cr vO iT' rvj rvj r-n ir> r~ O o" m o^ rH O to c lca c"i
ir vo >n r- in r\j rvj m <.o rvj co u~ <}• •-' c r- ir h ir, inr—i c r-H Ci <t P- oo no rjv r~ rv r— <t
rv; rvj r\ ! rvi rvi r-i rx; r\j nj rvj rvj rvj r-i
C C "" c c c c c o c o cc c" c <". c c c c a c c c c
O- vO cc rf r>. r\i h r-i
c c c c c c c oC C; C C C C C Cc c o c c o c c
r-i oc a in co m r— o c m f\: r\ r\ r- m cr . h c: r*~ co cmit. rv ixi r>ocjvcy>r*-r-iovvorvjc cr- r- ^o <p cc r- r~ vo
s a c<~. r-: rvj rvj <t r— >- <J- •—
i
c° cc c" c- cr, <j r- C ^o >3
O c ><"> c^ ro ifi r- co rjv o- m rvj c<~ C vO ir> co rv <xi <—
t
•—
<
C rv rvi rvj rvj nj r\j rvj rvj rvj rvj r\i rH i-h c C C C C O Cc o c c c o o c c o o o e re c o o c o cCCCCCCOC '. c c: c c c c c c c c c o
107
P
CO
co <o cm <j- cm <f m r-n ,—, oo r^~ lp n c\j <r —
i
in <r p- r- .nco O co r-- <*•. c <t lt ? rH '0 ix, ..r O C ro CM n- i/ a) nooo co r—
i
m <t c\i if oo :o en i>i r\j -\i <f p- .i - p ^r o ,n in
o <r p- p- p- o~> ^o c -.o co rg vo m r— >o r-i p- m r-n <r p-
00 CO —
I
00 v£> [*- vO vO f^- CO <t iP .-H rH C i! p- r-n vO f-*
r- o <f r- -o vD r~ -J co i\i o ^ o o o ro r- <)- co '-0 coO O O O O O C C. O © •*• G r-i o c: o c © o o o
co< LUcm _Ju a.
o ^r
OS <a. cO
y O'-' O
in•
Q CM co
U CM Q• r
(V • OO Li I uu. CO UJ
1
>—i <-0
LU o 1
cc z oQ- r£
CO uZ> 1—1
• _J 3^CO Q.Li o(^. _J oo < CO
Cd 2a. 6 u_
i—
<
o'7 CO
< toCC Ll aO O UJ
o t—
oca o
co
o r— •—
1
i— <r_i CO
< LU <s>
i— cc Q<r CC aQ o u
u <j
h- 1
D o cCL h- <.
z D Ll:
t—
1
< or
I I i I I I I I I I
<j- ro co vo o p .+ o —
i
r-- .-.*- o r (x; -r r\j -? p- in -h oin r- m co O if oo .o •—
•
-t in in t—
i
co r-n oo <t oo O vO or*- vO co cp -t p" o m ^n • o >o :- cm rH <m & m a>co i—
i
vo <—
•
m r-i vO p- O -1" "" oo oo o '•*P •--
'
rr > •-! o co-d- m co cm vO P- m3 cm so -p r> m -+ in '.<" <f co o co ^j <j-
© o vc m oo u o> cm r~ c cm o co en c if r~ co vo c1 - mrHQCOOOCCOOr-iC O O i-» O O C; C G O
I I I 1 I I I I
vO o^ cm oo i—i rf> a- cm •> p*- i—i rH in .''\t rvi h vo ('- m <!
fM vO O m O i— ^ <t" cm O -+ ir\ ro a- <r <f n O a^ o.cm m vO co in ac c in vO cm fM
i- 00 d C^ r~-
m cm vO r>- * f*-- c oo r- c\i c in >n >o n <r <t <f c m coh f\| h m vO CM <| in i—
I
r~- (Ml CM co. ro LP, CJ1 -n CO O C\J r—
CM vO 0> O C <—
i
C C 00 —< -H iO r— vf) r> r- r- O * H (0O O O O h O ho O O r-i c O © O © © pH C r-i O
I I I I I I I I I I
cnr\jaooiO\oc><t"0 fTx C7> C CM co ,-( a-, if vQ ex1 c-" r—
vO r-- vO CO ^H CM vO i—; P- 00 f- ;T rH C- >.r; oA r- <*" O O' i—
i
O <f 'O CM C\J IT'' f— O cr P- i.'r < O -t LP C r- vO T'
1 —
<
If J'
o- vo c <?• r- c i— m f"- cr r- co .—
i
a. c r- r-< a o <f
O cc <r vc o- vf r- <f o'i cm a' c oo -—
<
a o' vr n oc c in
r- cm i—i oo i—i r-H on co oo in oo -o m r^ in c vo .-.h i-h r-i oc o r-i o h c r c c c o C o o c i-i c r-i c- •-< o
I I l l l I l
r-co in o i—i nj a- oo in on —i cm h <o oj c- o c? r^- u
r-<T(M<l"lD>OH0 © <r O «.0 cm O i—ir- C m h <f h
r-H oo o o- co r- oo <t co c;> oo m 'O r-i r-i m vo i- —
<
o^ o>
CT- CO iO —I C C <"• r r-H d' f 1 O CM <t ff in 4- CO CO Cco ^o c> co cm lp ro >o <f oo cr. nj ">
i.o, c in in c ~ r*- c>
<i- o a-' o r-i n i- <j- oo oo -o '- f 'o c\' o- o o c "• o
<f O O C W C O C DOC C C O C O C *-i c* < ©I I I I
o c r-H c c c a -.1
i I II IIvo ;jo '.tn oj o >o "- m o> m c r-n
o r-H co c —i o r- c "• cnj -t ^ r-i <t m -j- o ct> o- r-- o->
O co rn <t r- p- m in <f t> en <f 'O r-i <(• r-H c oc C r-i oO in a) P c 4 O I
s- --H oo r\) in -t <( oo m co ^o c- r- cm
O co r-i o-n co cm cc co r- o -t in o vo c co co. o vo o r-
o r-t co ri o^ <r in lp -X) o m p o o r-n p- c co cm cc. <r6 o c c c c c c o c o c 3 r-H c. o e c ceo
i i I I I I I I I I
108
• rv vO <f IX. r—
(
r^ **j' c X r^ l—l rv r. <r rv Cl c- CO r-l <f CO
c •Is- vO CO c 1—1 lO a? o CO rv rv OJ LO. C vO CO CO (X! vO c I
s-1—
1
tO o c (VI to 1—
1
i—
i
lO o- r—
1
vO -t CM C rv c If, CX rv CO LO vO10 Ijj r- c 00 r- CO t—
i
>—
i
MD LO ,<i^H O^l r—
*
a: lO Cl cv: rv r—
1
r—
1
c"
_l •—
<
LO 1—
1
r-l OvJ C\l m CM rv rv rv i—, i—
1
c c c c c c c c1 4 Q_ o c c c o c c c C c: O c. c c c c c c c c elO
<c o c. c c c o c c c C o o c c c c c c c c
*~lO •
< VQ' C u f\l ^o r—
1
<f r- 1 rv Is- ' rv •."• oo . >o .—
1
c<" 00 o rH <rIs-
O c ^z ,—
I
lO, <f r- I
s- tO J- r<"i r\j rvi rv LO, Ll c I
s- a c rv CO Is- o
o lO UJ o <r (X) r- cr Ol stf <t ^o O rH CO o LT 1—1 I
s- r—
,
^-t- r- lO rOcc C\J
—
1
I
s- io vO v^' V 0"'
I
s- r- rv ;xj fl <J- CVI r^_ lO r CO OJ r-H r-( 1—1
a. CM a i—
i
rH r-l 1—
1
cv r-l rv rv CM OJ CXi r-i I—
1
c c c c c: c c cUJ O •—
1
c c c c C C C O C o C c c c c c o c o^r •
Lul
to
cr.
Ll.
o c c c o c c c o c c o C_ c c c c c o c c •
o t—
4
CJu o or CO u" o vO f—
1
c (J0 o vO fj> <* <r CO vO OJ I
s- I—
1
CC vO o^ CC''.' S' LU v0 CO C". lO 0"^ rv a.) m c O V5 O r- 00 lC CO I—* vO 1—1 cr.
<Y. is] —
1
^ I
s- vO '-O;" >n CO C
'/ rv C vO c <f vC vO OJ CO I
s- vO coo tO C" o co 00 CXI ^-1 tOi to rv OJ rv vO 1—
1
r- lO Ci CO rv 1—
1
r—1 r-l
Ll _.; l-H C7- r—
1
r- cv rv rv rv o: rv rv r-l c—
1
c C c c C" c c c1 .J O c C o c c c O C. c C
' C c c c c c c c c clul Cl Cv
' c C o c o c O C O o o o t
"
c c c c c o c c(V Ll
a< c
~r • --]- C\! CT <f to ^ r-H lO, m lO CO c^ oo <r <l CO Is- CO CO CO o
• a tO f\i <J- <r <. <f v£> f\l oo o CNj O CO r>< Osj vC vO rv o OJ J- 1—1
10 >—
i
a I
s - t—1 vO Is- ^0 (X, v0 a-> f
—
<r f' r> I
s- <f Ol CO r~ c CO vO <t-
Li 1 lO o pH 0" c vf r- r\ to o <} tn o" Is- C o. vf' C<" O (X 1—
1
r—
1
I—
I
t; (M O' r^ t- o. rv r\ o> o o. rv r— i—
•
C c c c c c c co u_ LO O C c. o c c. C o o o . o o o c c o c o c cm o c o c c o c C o o c Ci o c c c C" c o o c ca
>-Ll_
ojr h-<r •—• tO <t to f\l lO CO <r r
.—* <r Ol 1—
1
cr vO c ^;J- c vO rf, vO C" 00
oi to Q. o r-- lO c lO C: 6 Is- vf) lO I
s- r- (T' d 0> lO Is- CO o o 00
ID ?; LU lO o c (X <fr f—
i
i—
*
00 r- r~j -•t cc l-H c <+ <f c Is- 0- I
s- co
O a. 1 h- (Nl a i-H cr 00 c<N to o tO v0 •D I
s- i—
I
a vC <+ c CX! 1—
1
l—l 1—1
or a 10 (XI r-H OJ —
i
I-t rv rv OJ rvi rv CVJ r-H r-l c c c c: c o c cCL o c O c r- C c c C" C Q o c c c c c c o c c
_i ^* c c o c r c c c O c c c c C c c c c o c co < «—
<
1—I— to
< u cO a* r- CXi Is-
Is-
Is- C '.
I
s- on vO o CO c o LO I
s- lO o~ CO c COr- UJ O i—
i
Is- to o- i—
.
v£) co cf. a to lO vO lO. a. tO rv a 00 .—
<
Lf\ r—
1
< (X a: <r <—• <f c Is-
C1
: 00 <t C" <t o r—
1
sr-l <J cr cr or a- a <f
G L/) u t—
i
rr r- r—
i
tOi rv <t- o> v0 lO iO» CO i—H c: f
—
<f o^ rv rH r-l •—
1
< ,
—
r—
1
i—
i
<X) 1—
1
CVJ rv rv rv rv rv l—
1
i—
1
—I c c C"_ c r" c c(- a: o o o c c C c C c C Q o c c c c c c c c; cD •UJ o o o o c c C c c o o o o c- c c c C c c c cCL "-T <t
~z 6 UJH-l a K
109
3LUT.
<f OlflN ONi^ C O vf <f ^ N m vC u> r-i oo mr—
i c go og vo f\i r^ -—i?> u r-or-o-cxivOcor-Hcr-r-t
md er so oo >-h en c :r\ r-- c\ co eo o co vO r—i r- vo >-t r- co
oa"r-iogr-Hvor-1^- oc cc ifi <f o h n m o vo co cor-- a: ao co a -
, cm m o <f- c cr- I
s- -J- lo <f- m <± ro no r-n i—I
s- G c\j <r r-< no m cm o c <f en c\) c I
s- c <f <t lt cm Oi—i CX i—< C\J CM i-H CM <—
I
CM C. CO O CO C CM r-H CM r-H C\J CM CMo COu LLi
1 i 1 1 1 1 1 1 1 1
u _l
< a >—i <t CO en CO vO O 1—
1
en sO CO sO "C" en Is- en CO CM CO r-H 00
2T lf\ v0 O CO I-H o <t ON -4- o ^O r-H CM r-H r-H <f- <t On vO o nO_J < on CO CO CM CO m a- o CM <}- in CO, vO vO CO I
s- m O en CO r-H
< c0 I—
1
Lf, o- <t en CO CO '» CC; CC o a- vO CXI vO vO I
s- CO vO o r*-
> CM IC LT, o r-H O I
s- o m I
s- CO CX' c> CO O I
s- i—
i
co <f t> CMo O lO C On <f o*. I
s-: o vO 4 c. o o r- CM CO CO <! CO C7n O co
x_ o • CM CO r—
1
r-H CM <-i CO o CM c. <l- o CM r-H CO —
1
CM CM i—1 CO r-H
UJ en
QL CMCM
CO
o1 1 1 1 1 1 i 1 1 1 1
LU z: nO r— <t i-H r- CM i—
i
vT- CO a 0> CO Is- o CO i-H CO O O O <f
c0 • o i—
I
r- CO. p- r—
1
I
s-Is- en <f vO o O o vO <f CO, •^j I
s- CO CO i—
(
>—
1
LU u o no CO o vO O o 00 o 00 cri i—i r-- a; o <r en vO 00 CO r-H
o to LU r- : I—
1
CO I
s- r-H Is- CO I
s- vO r-H <f c< CM os o vO <f I
s- 00 r-2T HH cO CM <)" c CO s o I
s- o o er ef> I
s-I
s- vO en CM en o <f CO Is-
o 1 CO On CO o f—
1
<J CM o en vO CO c J- r-H O CM ^O CM en co o• z Q i-h t—
•
CM c> CO o co o CM c CO r~\ fM CM! CO CM CM CO r-l CO r-H
CO O'CO CO u 1 1 1 1 1 1 1 1 1 1 1
tu 3 t—
1
u _J ^r*I
s- r- cf\ r\i en Is- C r- o c <r I
s- O r— o- Is-
I
s- o <j- CO (XI
o a O CP i-H CO nO 00 CI <t- I
s- CM a~: v0 Is- vi- CM sO ef^ o> CO vO C^:
cy. C <r c* Is-
i—
i
c C: CO r^ c- CM •tf vj- n Is- c I
s- CO r-l I
s- <fD. _j c CO <r t— CM f- G- vO s
- f^- v0 en <!s
er O C- vO CO CC' 00 co< CO I
s- r- <r o Ch «o ^H CM en o <r vO 00 CO CO r—t 00 m O <jv CO
21 2 m CC CO CO CO CO i—
1
o 00 er <) CO o CXi <J" CM I
s- <f vO o o< o Ij_ i—
i
c CO o CO o CO f—
1
CM r—
1
CM i—
i
f—
1
CM i-H CO r-H co O CO Gcv r—
4
OO c0
I 1 1 1 1 1 1 1 1 1
o c0Qi U_ a CO CO c r—
1
CM o CM o r-H lO Ox vO CM o co ^o c m <r co OQ. o UJ CO CM 00 vt CO I
s- O r-H CO o CM s r-H <f <5 en vO O r^ o s-
h- CM vO 0" r-H er Is- O CM (MJ t—
1
<)- CO CO <— O a- r»- i-H 00 r-H e>
5! 77- c0 I
s- CM r-H r—
1
r-H co CM ^H 'O c CO <f CO r-H cc o O NO s- On CO
O Q CXi CM v0 <)" CO r-H i—
1
o co vO O c 0" 0^- o a> m o r-H CM 4OT 1—
I
2T <f CX O 1—1 4- O o- en h- en s- I—
1
lO \0 CM o •X) a- o- -i ou_
<_J
t—
4
m C CO c CO i—
1
CM t—
I
CM CM r-H CM i-H txj C Csl o CM O co r-H
< cO 1 1 1 1 1 1 1 1 1 1
I— UJ e0
< or: O o vO vO CO CM o CO o r
—
CM en 00 O en o o r-H 4 vO CO s-
Q or tr o r—
1
cc c CO s o CO in CM m c. en <f e'- <r CO nO vO CO oo u o 1—
1
o- cr 4 CO o CM o> <r CO vO r-l c: en r-H s CO o <r <]-
I— u < c 1—
1
O CM CO o uO o o CO r-H <f o- ef-. r-H CO CM en O CM CO
id 1 o a r- o a) CM o CO r—
1
o m o co en o CO I
s- r-H nO 00 ^0
a o o c CT <t no CM CM vO 00 o 1—
1
I
s- CO O <f- 1—
1
CO •J- en O r-H <t
i— r— < c e CM o CO i—
I
i-H i-H CM CO l—l (Xi r-H co o CM c (XJ r-H CO r-H
D ~> UJ
O < Qi 1—
1
! i1 1 1 1 1 1 1 1 1
110
QLU 00 O CM CO CMX KT, «X) o CO r- o co c a^ CM x > <J- c o -cf CO CO o> ro COto LU. —
i
00 c r—
1
iM vO r-l m c c . r-H CM ; vD I
s- vO < sO I
s- vO-I _J c <r r- 4 .—
i
4 I
s- >: M. Is- c I 'M .—
i
'. vO I
s- -4 ro 1—
1
_l a CM vD r\j t—
1
m CM s- o- N • c<~ uo Pi :*-
,
CM t-H o c: c o c cQ. c f\J o c o o c r C c O c C o c o ( c c c c5 <". c C c c c c c c c c C' c c C - c c c c c c c6 w •
u >-
u C U< O ."J or o 1—
1
I—
1
sj" c v0 CO v," <J r—
I
o (<" vO co CO O c CVI <r ain Li i
ro, rv CM o in r- .;•
J c\ p I
s- m I
s- -f: o rv ' <r .r f-r
_j cm z< <J vD 'AJ vO ,-\ o Q« !-H - f\l ' :—I
s- IXN. c Is- c I
s-
<£ rv o rv c<~ CO in —
i
1—
1
vC 1— 3 <t •i c lT> <r .—iIs- CO 1 P
-! o CM
> UJ CSJ cc> H ro in r-J v0 I." c\ f V f- CO CM r—
I
r—
i
o c c c c co • m C r- c t , o c o c c c o o c c o c c c c c. o2; uJ U_ c o o ( o c o c. c c c o c. c c o c o r_- c eLU 10[V >— o
o :rr
LU Z U ; co o CO CO m <!- LT --r <Jr Lf <r .—
I
1—
1
p-i CC . : Is- vO vO <f
CO N CO ro o <t rH en r~- rr r- c~ CO <t ^+ s CO Is- . T
. r—
i
vO -j r-•—i 10 CM .—
i
CM <r in o h- r- <J r- i—
i
c CVj o m m m Is- CV CO co
o **">" vC <mj C C CO c lT rv ir I
s- CM r <!- r- c> LT <f c\ Cn'
2 Zl G P* o -t rO CO ro <t -T r% pf in rvi r-i ^-H i- i r c c c c C"
a. c< C sO o O O O r C C c o O o o c C c C' o c (
• Li. c c c o C c C' c c c c C c c. c v c c c c ci/) _J
10LUU O t/i •d r- in rv r—
i
in sO 1—
1
ir 1
1
r\ m vC' (Ml in a- I's
<f nj
o — .a r-. o n > P- CO <r 01 c r CM -H I
s- .>. c CM <M o~ vO 1-
Q to l_ vD a. c C0 cc if h r- fv v^ 0' c r-l. 1
1 i—
i
<£i vC c >.c.
a o- rv CO c a- v0 in r I- vC <j Is- r-- I
s- ^0 Is -
1/ p p rvU_ in <f •—I m r—
1
cvj CO v-». r . I
s- ni c .-"1 H c C c c •'•
t
7>" c c r—
1
C a o C o C r c C c: o o C" c c c c c c
< Ll c C o c o C c <T c c c ^~c c c .V c C. c c
Q' >- Oo t—
c — lOQ' to a. o r- o> m o r-l <j r I
s-( I
s- 00 <r. ^o Is
, r-l CM 0> s- c <j"
Ql Z LU s<fr co s c CO i— j < r I
s- <f t--^ I h CM nj vO vC c >—
1
LU H- i.." P~ ,
-^ r\ 1
—
<t ^ i-
c . i .;« I—
1
r CO vO -7 r-^ CO r—
i
Is-
*"" O lT' ,-H s- cl- Lf rr-
i
.' v£ -^j Cf CO <r c ')•-Is- '.0 M? 00 rvj If cv
6 vf; CO lO rv. r—
I
CO cj c <!' lT <JTj CV: .—
»
1
—
i—
i
c c c. c c cor _J .7" c c c C o c - C c C ^~ c c C.1 c c c o c c cu. < —
Of.
c c c C c e c c C c c c c C; c c c c c~ c c
< \~ W:
h- U LTi
<r Lll O Is- >0 4" vO CM <|-
i I
s- <r c 11-! ro 00 CO m ro <r CM c CC vOQ a. a i—
i
m C\i <—
1
rvi CO o r< I
s - vj" LP, I
s- C^ <f sIs- ro I
s- ro pp I
s-
10 u ro CO 00 CO c v0 m CM vT m I—
<
vO r-i Is- .—
i
fM vO o r> If. CXI
1— <3 v£> c^ CM r-H O C CO r- CM CM ro c <!' i—
i
Ch ir. CO CO m p^
3 ly: CO en I—
1
<t O 4 o- I
s- <r 4- vO rvi CM CV CM C c c c c c0_ Uj a O o c o c O c c C o C C c o O CJ' c. c c: c cI— ~< <!' o o o o c c c. o c c o c O o c o c c e o C"
oC LUo_ cr
111
X Scale- 16 msec/inch. Y Scale- 0.5 units/inch.Figure I4- Pure signal auto-correlation over 3750
sat pies. •Estimated s/i\i ratio, plus 55 DB«
112
rvTO
toTOto.
R
•
-
10csto
cs05
ncsTO
CM
TO
i
i
P3>
.'••*
:
ii.
,._..— ,_,^ JJ_ _^_>l
—TOIS
*
REQ
812 a 10 ;83 dm -802
X Scale- 0.02 units/inch. Y Scale- 100 cps/inch,'Igure 5- ?iire signal power spectral density over
373*0 samples. 125 estimates 5 cps apart.
-Q06
X Scale- .001 units/inch. Y Scale- 100 cps/inch.
Figure 6- Power spectral density of noise, 3750 samples,
125 estimates 5 cps apart.
!\
to.1 C^
1
1
i
!
i
!
! 1
I(®Iro
f i-O
/ to/ ®
*/ 3^y^ <S>
J TO
s~"~^
roi T~ to
.
1 TO
•<cjr C\J
TO—
-
TO
r-l
TOW
~z?
TO
1 [ c^ TO
3FQ
-336 -<d#I -333 -332 -331
X Scale- .001 units/inch. Y Scale- 100 cps/ir.ch
Figure 7- Power spectral density of noise, l87p0 samples,125 estimates 5 ops apart.
-Q3S
X Scale- .001 units/inch. Y Scale- 100 cps/inch.Figure 8- Power spectral density of noise, 37500 samples,
125 estimates 5 cps apart.
116
- ±2/Z*&£^T/QA/ w,*> & t//J
X Scale- lj.0 msec./inch. Y Scale- 0,5 units/inch.Figure 9- Auto-correlation of ODB. signal plus noise.
37£0 samples .
117
rftE3
OCT.:
X Scale- 0.02 units/inch. Y Scale- 100 cos/inch.Figure 10- Power spectral density of ODB. signal after
noise removal. Alpha calculated on RBAR(T).3750 samples .
118
a
10t—
i
i
X-Scale- I4.0 msec. /inch, Y Scale- .5 units/inch.Figure 11- ODB. signal after noise removed. Alpha
determined on RBAR(T). 3750 samples .
119
</'«"
X Scale- .002 units/inch. Y Scale- 100 cps/inch.Figure 12- -10DB. P sd. before noise removal.125 estimates, 5 cps apart, over 3750 samples.
120
-a
oI
II
2
2
G _
-086 •805 -884 -803 -882
X Scale- .010 units/inch. Y Scale- 100 cps/inch.Figure 13- -10DB. p Sd. after noise removed. 3750 samples.
Alpha determined on RBAR(T).
121
X Scale- .01 units/inch. Y Scal«- inn ~~„ /* uPigura IK- -10DB
. p.d.W noise^ov"? l^O^lesAlpha determined on RBAR(T).sampj.es.
122
N -^
&
-012 -010 -003 -006 -004 -002
. X Scale- .02 units/inch. Y Scale- 100 cps/inch.Figure 15- -10DB. psd. after noise removal. 18750 samples.
Alpha determined on RBAR(T).
r-(
«—
I
BEST
ERR-015 028
X Scale- l\0 msec. /inch. Y Scale- 0.5 units/inch.Figure 16- -10DB. signal after noise removal. l87!?0 samples.
Alpha determined on RBAR(T).
124
10
i
S+N
X Scale- 16 msec/inch. Y Scale- 0.5 units/inch,Figure 17- -10DB. auto-correlation before noise
removal . 22£00 samples*
125
&
882l -^V-**^ '^~A'
r.-jC'^ W5 ^ eie884 886
X Scale- 16 msec/inch. Y Scale- 0.5 units/inch,Figure 18- Estimated auto-correlation of noise in -10DB.
signal plus noise input, averaged over37500 samples.
126
f «EQ
X Scale-
Figure 19-,002 units/inch. Y Scale- 100 cps/inch.-10DB. power spectral density beforenoise removal. 22500 samples, 125 estimates,5 ops apart.
127
CS>
\n
C3I
1—\
03CMCS
i
X ScaleFigure 20 -
16 msec/inch. Y Scale - .5units/inch.-10 DB. SNE Signal afeer noise removal infrequency domain. 22500 samples.
128
rv
•REQ
1
^=T
£
-616 -893 -666 -664 -86Q
X ScaleFigure 21 -
.02 units/inch. Y Scale - 100 cps/inch.Power spectral density of -10 DB. SNRafter noise removal in frequency domain.22500 samples.
129
APPENDIX C
DETECTION AND LOCATION OF SIGNALS IN NOISE
In Appendix A and B methods were developed for digital analysis of
analogue signals.
The data utilized in the above appendices was in a sense artificial, in
as much as it was generated in the laboratory.
To test the proceedures developed using a live signal, where signal
characteristics and noise statistics are generally not known, a 200 cps
sinusoidal signal source was placed in an unknown noise field produced by
patrons of a cafeteria.
The signal source was set up at one end of the .dining room, and its'
volumn raised until complaints were voiced. The source level was then
lowered until the complaints ceased. This level, which we shall refer to as
the db level was measured to be 38 milliwatts at. the voice coil of the loud-
speaker. At this level the sound was occasionally audible at various loca-
tions within the room, but was not detectable aurally at the microphones.
A stereo recorder was set up at the far end of the room, using two micro-
phones symmetrically placed about the longtitudinal center line of the room.
A number of three minute recordings were made of signal plus noise, re-
duced signal plus noise (-9. 3 db), and noise alone.
The noise level of the crowd varied very considerably over the hour and
was punctuated by an occasional dropping plate. The period from 12:30 to
12:40 was selected for further study as the number of diners was essentially
130
constant over this period. The details are listed in Table I„
By cross -correlating the left and right channel information it is possible
to locate the source and to estimate its frequency.
.
The bearing of a source, with respect to the center-line may be deter-
mined by measurement of the difference in arrival times of sound waves at
the receiving sensors.
Assume as in Figure 1, a distant sinusoidal source, at.amangle 8 mea-
sured from the center-line or dead-ahead position, and two sensors symmet-
rically placed about the center-line spaced d units apart.
Assuming plane wave propagation, with no mulUpath effects, and in the
absence of noise, the difference in travel time or phase of ray 2 is a direct
measure of the angle 0.
2-
Ret)
K T,
/IN\
\i r 1 1 >
X
\yFigure 1. (a) The basic DF Problem, (b) Cross -Correlation of Mi, Hn
i rT/2
.
Forming RM , M (T) = lim T -*•<*> ~ J (cos w t)(cos w (t-r))at1 4 i _T/2
(1)
131
= B |T/W
?RM1M2< T) = M. J i [cos
(Wo^ * Wo^ + CW dt
2TT _ ^/Wq * u
o
w f **°r 1
RM1M2 (T) " ° + cos (W T -a) (3)
The first peak in the correlation curve occurs when
Wo Tj = V or T,
= JL. = A sin>9 (4)
Subsequent peaks in RjvilM2 (1") occur when
W f* - y = N (2TT ) ,N = 0, 1, fe, 3 ...
Peaks obtained for n\0 represent ambiguities in the bearing angle. The
_ d_
P" cfirst ambiguity occurs when T D = — sin &8orA0 = 9a -6= sin
-1 E
= sin-l C^TT_
dW
The first peak at 1"! is of interest. In this case using room measure-
ments and an assumed velocity of propogation of 1129 ft/second, Ttwas
calculated to be 1. 5506 msec, for an angle of 10. 4°. For small angles we
may assume the time and the associated angle are linearly related. Under
this assumption the arrival time would change very nearly 150 U. sec. per
degree.
The available processing system however could- sample the record at
best every 200 M. sec. , resulting in a bearing accuracy of about 1. 33°. To
improve the bearing accuracy, the data was slowed by a factor of four,
grequency shifting all components down by a factor of four. This has the
132
effect of increasing the arrival time by a factor of four resulting in a bearing
accuracy of 1. 33/4° or . 333°.
The mechanics of the speed reduction are shown in Table 2.
This is more bearing accuracy than is justifiable but since a reduction
in data speed of at least two was necessary to digitize, a 5 Kc analogue re-
cord, the extra factor of two was used to increase bearing accuracy.
The data speed reduction meant that the original. 200 cps signal would
appear as 50 cps with the first peak of the cross -correlation curve at T =
6. 02 msec. The data was sampled at intervals of 205 JLLsec. , and hence
the signal, if present, should occur between 29 and 30 shifts of the correla-
tion function argument.
2. Data Analysis
Twenty blocks of data were digitized for each of three cases; signal to
noise ratio db; -10 db; and noise alone.
Due to difficulties experienced with the clocking system, only the first
block of a twenty block run was clocked, with clocking of subsequent blocks
internally. Due to the variations in the time necessary for the 163 tape unit
to recover from a writing operation, a cumulative error in clocking arose:
Hence only the first few blocks are time synchronized to an acceptable degree.
Fortunately this was long enough for sucessful signal detection.
The data was normalized in the computer by the computed estimator of the
standard deveation, and the cross -correlation of the normalized data formed.
The time of integration was fixed due to the finite block length. Assum-
133
ing a process is stationary and ergodic, (at least short-time stationary) the
effective integration time may be extended by averaging in time, ie
N
N=l* R
iab(r) " -£ % f a (t)fb(t-r) (5)
and R^bCf) = — i> R- ab (1 )
K i=l
- K NRiab(T) = ± £ 0- £ f a CT) f h(t-T))K ^ N N=l
NK= j^ £ f-.(-r) i b (t -T) (6)
The running average cross -correlation function of the db signal plus
noise was formed over the first five blocks and is shown in Figures 3 through
5 as a "hatched-in" curve.
3. Results
Experiments with averaging more than five curves were not successful,
due to the previously mentioned cumulative timing error.
The cross -correlation of db signal plus noise was abserved to be es-
sentially sinusoidal with the first peak occuring between 29 and 30 shifts
exactly as predicted.
The frequency of the sinusoid however was observed to be 150 cps, or
three times the expected signal frequency, indicating detection was being
made on the third harmonic of the source!
No firm explanation for this is offered. However it is noted that the
134
oscillator produced 0. 1% third hormonic, and the 20 W. push-pull amplifier
used would be expected to produce considerable cross -over distorion when
operated at such low power levels. In addition the tape re-recording process
tended to favour the upper registers.
A power spectral density was made of the db signal plus noise, and
of noise alone, using a 6. 125 cps filter bandwidth. The normalized results,
shown in Figure 2, reveal the pronounced 150 cps component in the signal
recording, with the general shape of the remaining portion being similar to
the noise power spectral density.
The programming mecessary is included as Program SIMSIG in Appendix
B.
The results obtained with the signal at -9.3 db were inconclusive and
are not included.
4. Conclusion
Provided accurate time synchronization between right and left channels
can be maintained during the digitizing process, cross -correlation of the two
channels provides both bearing and frequency information.
The process of averaging cross -correlation curves must be done with
care, ensuring accurate time synchronization and continuing the averaging
process only as long as the signal characteristics remain stable.
135
ANALOGUETIME
DATA DIGITALSTATE RUN
12:30-12:33 "O" DB. 0001 0021 0001 0024 (R)
TO0001 0021 0024 0024
0001 0032 0001 0024 (L)
TO0001 0032 0024 0024
12:35-12:33 "-10"DB. J
12:41-12:44 NOISE C
Digitized on Reserve Tape 18
Sampling Interval 205 Ms.
0002 0021 0001
TO0024 (R)
0002 0021 0024 0024
0002 0032 0001
TO0024 (L)
0002 0032 0024 0024
0003 0021 0001TO
0024 (R)
0003 0021 0024 0024
0003 0032 0001TO
0024 (L)
0003 0032 0024 0024
TABLE I: Details of Data Digitized.
STEP PLAYBACK BAND-W. RECORD BAND-W. SIGNAL
Concertone 50 cps Ampex FR-100 - 1. 1 Kc. 200
7.5 IPS 10 Kc. cps
Ampex - 650 cps Concertone 20 - 650 100
7.5 IPS 7.5 IPS cps cps
Concertone 10 - 325 cps Ampex 10 - 325 50
3.75 IPS 7.5 IPS cps cps
TABLE II: Frequency Reduction Process.
136
of\J
.-H
•
o 1—
1
UJ '!<
O —
-
i
—
QLu <rsl CK
>—i
1 co*—
O LlUJ Z'c »—
.
LOID r- ~-
Lli r— X.— C: on Is-
"***
.
—
•
l0 UJ 1
v0v0
LO — CO u X oLU to Li.' —
«
LU H- <J3 OLU LU LO Lli to »-. lC + 0^ — Q LU vOCC LU ll dc 00 2 !— < iX> IT ro X < Q vOO DC LU o lu UJ lV r- — o LU CO t— OC o X •
UJ O --.' LU Ixl ^ X O • 1- < C CO r- < r) LU 1— c s
Q UJ 13 c 0i i—
i
Lfl LT> r~ Ll c U ' £V (D o CO '0 IU Q r- UJ e? o *• Q Ll e Q r~ 2" Q r^ K' o Li X a s_ a X .X) G Ui < -~
<£ ^ O <c LU >- r
—
• < 'r \ Si Ll C <r I
s- < >|c a o o o. o to• r- < X £-< ^r
Is- o r- Q r- <3' O t— — i—
1
r- Q _j 1 <L "* o < u2f LU LU I— o Q_ _J u> r- ~* *^^ -O 1— ,; Ll
iV. — < <c 1— <r LlI o_- UJ
e. r— X — UJ »— I — _J — i—
i
-— LU —
»
r> V. u_ (Y- v0 "-^ o rsl t_ LP. LL. K 1 o ' a Q r- L0!— <c I— r- rvj I
s - c: f-~. <^ r- o'^ r— Lj r- f\i Q ;1-'
i
—
vU u < 1• r~-
;-- X o •1 i V. UJ 1—
I
r- u <t U . vO 2 or o o • < 1— • o o o KD X _lLli o X c X o T_" c "T" ^ T" C "7" o r-1 I
s- CO r- ~r r-H <t r- • CO c 1— 1— o o • • Ll! r- _lcc _J lT\ —
i
^o rH c" l—l IT t—
i
CM <—
i
O »—
i
x li'\ r—
1
< 1— « h- Ll 1 C rH Is- <; OQ (^ o o Q ^- >—
1
CVJ
DC t—i U r—\ u_ 1—
1
U. rH U_ »—
!
IL i—
I
i . '—1 l! II II UD <^. iNj '."!II ii il il II II O" CO M u. S! 1—
!
1
< ^y — w — •
—
— »— — — ~-* — -— —~ — il h- c o II II II II V_7 X :~j LO o LO II rH L0 21 rH
ro <f t— 1— h" 1— r— r— h- I— h- r- r— i— 1 o LU <z i.l . i
—
r> ID c U h- <t LU Ijj UJ o II II1—
.
z:•v ££ <C < <£ •:. <£ < < < < <£ < <c <r X (Y. Q LU < Ll i u L_J h- ll; a Q Q UJ e> O. v/: H r-
U. o > 21•
*-; 1
'
I ;
.
.> "> ;r s ^. :>~, a: 1— (— 1— M q: C (C I ll; X X ^L o o UI < n 2!co c ::c Q£ £V or. (V or o- r2 O: 0' 0^ a: : 2: LL. L L 2 1— r~ o a.
—
^
r- r— H- < :r o L>. Z) -D t—
i
<r cr O O C o O C c c O o 3 6 O Lu < X X <-:. u Li. X _j < r- 1
—
K- 1— <r >— K—* < <L DCrH
c£30
G. u_ Ll ij_ U_ Li <'- — Ll. Ll U. Ll Ll Ll Lj i— H~ '- 1— ISJ iV.1
or <r r- CO .-o m GO O u cO r- 1— a.
o i—i r\i i< s*d' LC'. ^ r- a, a- a r-i l\4
o _? O o O ,; , ,f c O o r-H rH 1—i
—
)
i—
i
rH >—
i
—
i
r-H i—
1
—
>
i
—
i—
i
>—
"
r—
1
i—
1
r—
f
u
137
c: cu c C u Oo LL! ii d LUI— /*\ c X dUJ I— o a — z>
I Ll X i 00 <(— N or < h-•^ n «- #* ** #v
c p-i (M co <r LP vO r- 00 Ui O Ho c O c: O c c c o O —l i—
l
o: cr a: or cr d: c: c c o; o: a: ir z 20.0.0.0-0.0 oaaao.o_o.LLj'Lu
138
uoMpu-Tca Resourt
© — 10. 4il
% -4
$ — 83.^154c* = S£>. 173*
t = 93 -72>S6
At s i.SSot wsec
VetoCixy or Soowo TAwtNi
to og iu<» '/*ec
M»CPoPHOM£ CtfZ *1lCfc0PrtC*J£ Ch ••
FiGufcS 2. ! G-eOHCTfrv OP trte LuwCrt- Room pAoftten
139
X Scale- 2.05 msec/inch. Y Scale- 0.5 units/inch,
Figure 3- ODB. signal cross-correlation over 3950samples.
140
rflu
X Scale- 2.05 msec/inch. Y Scale- 0.5 units/inch.
Fig-are ]>.- ODB. signal cross-correlation over 110i>u
samples.
141
liflR
roi
X Scale- 2.05 msec/inch. Y Scale- 0.5 units/inch.Figure 5- ODB. signal cross-correlation over 19750
samples.
142
X' Scale- 10.25 msec/inch. Y Scale- 0.1 volts/inch,
Fig-are 6- ODii. signal auto-correlation over 11850samples.
143
X Scale- 10.25 msec/inch. Y Scale- 0.5 volts/inch.Figure 7- Noise auto-correlation over 11850 samples.
144
..
.
^ J
I:
J
9
i
oV
^s«
\T\
o
8I/)
figure 6- Relative power spectral densities of noisealone, ,and signal .plus noise, Normalizedto 100/* .
145
INITIAL DISTRIBUTION LIST
No. Copies
1. Defense Documentation Center 20Cameron Station
Alexandria, Virginia 22314
2. Library 2
U. S. Naval Postgraduate SchoolMonterey, California
3 Commanding Officer 1
Canadian Defence Liaison Staff
2450 Mass. Ave. N.W.Washington, D.C. 20008
4. Dr. Glen A. Myers (Thesis Advisor) 1
Department of Electronics Engineering
U. S. Naval Postgraduate School
Monterey, California
5. Dr. L. V. Schmidt 1
Department of Aeronautical Engineering
U. S. Naval Postgraduate School
Monterey, California
6. LT Donald Leichtweiss, USN 1
840 Lottie Ave.
Pacific Grove, California
7. LTNeilA. Barrett, R.C.N. 1
1038 Broncho Road (
Pebble Beach, California
8. Mr. W. H. Littrell 1
Code 3150U. S. Naval Electronics Laboratory
San Diego, California
9. Dr. H. A. Titus 1
Department of Electrical Engineering
U. S. Naval Postgraduate School
Monterey, California
146
UNCLASSIFIEDSecurity Classification
DOCUMENT CONTROL DATA - R4D(Security claaailication of title, body ol abstract and Indexing annotation must bt* entered whan the overall report le elaaeilied)
I. ORIGINATING ACTIVITY (Corporate author)
U. S. Naval Postgraduate School
Monterey, California
2a. REPORT SECURITY CLASSIFICATION
UNCLASSIFIED26 GROUP
NOT applicable3. REPORT TITLE
Extracting Analogue Signals from Noise Using a Digital Computer
4. DESCRIPTIVE NOTES (Type ot report and Ineluaiva dates)
Master's Thesis in Engineering Electronics5 AUTHORfSJ (Lett name, tint name. Initial)
Barrett, Neil A.
6. REPORT DATE 7a. TOTAL NO. OF PACES
144
7b. NO. OF REFS
8a. CONTRACT OR GRANT NO.
b. PROJECT NO.
9a. ORIGINATOR'S REPORT NUMBERfSJ
9b. OTHER REPORT no(S) (Any other numbere that may be aaalgnedthia report)
10. AVAILABILITY/LIMITATION NOTICES
of this report from DDC.This document* has been -;*r<^
jTC!^* and &hle; its dis„rib
rd for public
iion is unlimited
11. SUPPLEMENTARY NOTES 12. SPONSORING MILITARY ACTIVITY
Department of National Defence CanadianDefence Forces Headquarters, Ottawa,Ontario, Canada.
13. ABSTRACT
It is frequently convenient in data processing to convert analogue to digital
data for computer assimilation. .A convenient method of such conversion has beendeveloped and used in the study of correlation detection of the audio signals cor-
rupted by noise.
?•
A method to use apriori knowledge of the corrupting noise to increase proces-
sing gain has been studied. In the case of detection of a sinusoid in noise, anadditional gain over convertial auto-correlation of up to 14. 5 db has been achieved.
Finally, a signal source located in an unknown random noise field was detectedclassified and located in relative bearing by the cross -correlation of the signals
received from two spatially separated sensors.
DD .MS. 1473 UNCLASSIFIED
147Security Classification
UNCLASSIFIEDSecurity Classification
14.KEY WORDS
Analogue to Digital
Correlation DetectionNoise Removal
LINK A
ROLE WTLINK B
ROLELINK C
ROLE
INSTRUCTIONS1. ORIGINATING ACTIVITY: Enter the name and addressof the contractor, subcontractor, grantee, Department of De-fense activity or other organization (corporate author) issuingthe report.
2a. REPORT SECURITY CLASSIFICATION: Enter the over-all security classification of the report. Indicate whether"Restricted Data" is included. Marking is to be in accord-ance with appropriate security regulations.
26. GROUP: Automatic downgrading is specified in DoD Di-rective 5200. 10 and Armed Force* Industrial Manual. Enterthe group number. Also, when applicable, show that optionalmarkings have been used for Group 3 and Group 4 as author-ized.
3. REPORT TITLE: Enter the complete report title in all
capital letters. Titles in all cases should be unclassified.If n meaningful title cannot be selected without classifica-tion, show title classification in all capitals in parenthesisimmediately following the title.
4. DESCRIPTIVE NOTES If appropriate, enter tha type ofreport, e.g., interim, progress, summary, annual, or final.
Give the inclusive dates when a specific reporting period iscovered.
5. AUTHOR(S): Enter the name(s) of authors) as shown onor in the report. Enter last name, first name, middle initial.
If military, show rank and branch of service. The name ofthe principal author is an absolute minimum requirement.
6. REPORT DATL: Enter the date of the report as day,month, year, or month, year. If more than one date appearson the report, use date of publication.
7a. TOTAL NUMBER OF PAGES: The total page countshould follow normal pagination procedures, Le. , enter thenumber of pages containing information.
76. NUMBER OF REFERENCES: Enter the total number ofreferences cited in the report.
8a. CONTRACT OR GRANT NUMBER: If appropriate, enterthe applicable number of the contract or grant under whichthe report was written.
86, 8c, & 8d. PROJECT NUMBER: Enter the appropriatemilitary department identification, such as project number,subproject number, system numbers, task number, etc
9a. ORIGINATOR'S REPORT NUMBER(S): Enter the offi-
cial report number by which the document will be identifiedand controlled by the originating activity. This number mustbe unique to this report.
96. OTHER REPORT NUMBER(S): If the report has beenassigned any other report numbers (either by the originatoror by the sponsor), also enter this number(s).
10. AVAILABILITY/LIMITATION NOTICES: Enter any lim-
itations on further dissemination of the report, other than those
Imposed by security classification, using standard statementssuch as:
(1)
(2)
(3)
(4)
(5)
"Qualified requesters may obtain copies of thisreport from DDC""Foreign announcement and dissemination of thisreport by DDC is not authorized.
"
"U. S. Government agencies may obtain copies ofthis report directly from DDC Other qualified DDCusers shall request through *
-0»
"U. S. military agencies may obtain copies of thisreport directly from DDC Other qualified user*shall request through
"All distribution of this report is controlled. Qual-ified DDC users shall request through
If the report has been furnished to the Office of TechnicalServices, Department of Commerce, for sale to the public, indi-cate this fact and enter the price, if known.
1L SUPPLEMENTARY NOTES: Use for additional explana-tory notes.
11 SPONSORING MILITARY ACTIVITY: Eater the name ofthe departmental project office or laboratory sponsoring (pay-ing for) the research and development. Include address.
13. ABSTRACT: Enter an abstract giving a brief and factualsummary of the document indicative of the report, even thoughit may also appear elsewhere in the body of the technical re-port. If additional apace is required, a continuation sheet shallbe attached.
It ia highly desirable that die abstract of classified reportsbe unclassified. Each paragraph of the abstract shall end withan indication of the military security classification of the in-formation in the paragraph, represented as (TS). (S), (C), or (U).
There is no limitation on the length of the abstract. How-ever, the suggested length is from 150 to 225 words.
14. KEY WORDS: Key words are technically meaningful termsor short phrases that characterise a report and may be used asindex entries for cataloging the report. Key words must beselected so that no security classification is required. Identi-fiers, such as equipment model designation, trade name, militaryproject code name, geographic location, may be nsed aa keywords but will be followed by an indication of technical con-text. The assignment of links, roles, and weights is optional.
DPJ FORMVJ 1 JAN 94 1473 (BACK) mssSaWLusn
148 Security Classification