GEOPHYSICAL EFFECTS STUDY (REDACTED VERSION)

• Document Type: CREST
• Collection: STARGATE
• Document Number (FOIA) /ESDN (CREST): CIA-RDP96-00788R001800190001-3
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 53
• Document Creation Date: November 4, 2016
• Document Release Date: June 24, 1998
• Sequence Number: 1
• Publication Date: July 1, 1984
• Content Type: REPORT

Interim Report
Covering the Period 15 November 1983 to 15 July 1984
GEOPHYSICAL EFFECTS STUDY (U)
Ap ove or Release'2000/08/08 : Cl -RDP96-00788R001 00190001-3
SRI Project 6600
ESU 83-147
Copy No. .......".....
This document consists of 54 pages.
ved For Release 2000/08/08 : CIIA-RDP96-007888001800190001-3
333 Ravenswood Avenue ? Menlo Park, California 94025 ? U.S.A.
[phone] ? Cable: SRI INTL MPK ? TWX: 910-373-2046
 Approved For Release ~QQQ/ 0 OO788ROO1800190001-3
VIV~..L~~
LIST OF ILLUSTRATIONS . . . . . . . . . . . . . . . . . . . . . V
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . V
I OBJECTIVE . . . . . . . . . . . . . . . . . . . . . . . . . 1
II INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . 3
III PRESENT STATUS . . . . . . . . . . . . . . . . . . . . . . 5
A. General . . . . . . . . . . . . . . . . . . . . . . . 5
B. Literature Search . . . . . . . . . . . . . . . . . . 7
C. ELF Measurements . . . . . . . . . . . . . . . . . . . 7
1. Introduction . . . . . . . . . . . . . . . . . . 7
2. Los Altos Site . . . . . . . . . . . . . . . . . 8
3. ELF Data Acquisition Systems . . . . . . . . . . 8
a. Basic System Design . . . . . . . . . . . . 8
b. FFT Program . . . . . . . . . . . . . . . . 10
c. System Electronics . . . . . . . . . . . . . 10
d. ELF Electronics/Software Subtasks
(Status) . . . . . . . . . . . . . . . . . . 10
e. ELF System Calibration . . . . . . . . . . . 11
D. Satellite Downlink Geophysical Data-Acquisition
System . . . . . . . . . . . . . . . . . . . . . . . . 12
E. Geophysical Data/RV Performance Correlation
Analysis . . . . . . . . . . . . . . . . . . . . . . . 14
IV SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . 17
APPENDICES
A FAST FOURIER TRANSFORM ROUTINE FOR ELF DATA 19
B ULF TREE POTENTIALS AND GEOMAGNETIC PULSATIONS . . . . 31
C FORMAT FOR SATELLITE BROADCAST OF
SPACE ENVIRONMENT SERVICES . . . . . . . . . . . . . . 37
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Approved For Release $15 O 0788ROO18OO19OOO1-3
CONTENTS (U)
 Approved For Release 2QQQ(QCQ8A~IR.9 00788R001800190001-3
THIS PAGE INTENTIONALLY LEFT BLANK
iv
Approved For Release UNIN6-00788R001800190001-3
 Approved For ReleasE 2QQ, e?/q$ Sct ffff -00788RO01800190001-3
ILLUSTRATIONS (U)
1 ELF Data-Acquisition System . . . . . . . . . . . . . . . . . 9
2 Real-Time Geophysical Data Acquisition Via Westar IV
Downlink . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3 Real-Time Geophysical Data-Acquisition System . . . . . . . . 15
4 Geophysical/Performance Data-Analysis System . . . . . . . . . 16
TABLES (U)
1 Geophysical Data Bases . . . . . . . . . . . . . . . . . . . . 6
Approved For Release tp0 FtES00788ROO1800190001-3
 Approved For Release V em0w- ~E PQf rj-00788 R001800190001-3
Approved For ReleaseMAS5WIED6-00788R001800190001-3
 Approved For Release2000/08/08 : CIA-RDP96-007,68R001800190001-3
I OBJECTIVE (U)
The objective of this effort is to investigate the
possible effects of ambient geophysical/low-frequency electromagnetic
factors on remote viewing (RV)* performance
(U) RV (remote viewing) is the acquisition and description, by mental
means, of information blocked from ordinary perception by distance
or shielding.
Approved For Rele e,2000/08/08 : CIA-RDP96-0~788R001800190001-3
 Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
 Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
II INTRODUCTION (U)
SRI International is tasked 4
,to investigate a potential correlation between remote viewing
(RV) performance and ambient geophysical/extremely-low-frequency electro-
magnetic (ELF) activity. The possibility of such correlation is indicated,
for example, by studies showing psychophysiological effectsl'a* and
behavioral changes3#4 associated with ELF electro-magnetic fields. The
geophysical variables of interest include such factors as ELF intensity/
fluctuations, ionospheric conditions,
and solar'emissions (e.g., X rays and
answered in.this program are
? Do geophysical/performance
measurement of the ambient
be used as an indicator of
geomagnetic indices, sunspot number,
solar flares). The questions to be
correlations exist such that
geophysical variables could
expected performance?
? If so, can optimum performance windows be identified?
(U) The structure of the program that will address the above issues
consists of
- SRI (Menlo Park, California location)
- Time Research Institute (Los Altos, California
field station).
? Real-time geophysical data acquisition via NOAA
(National Oceanic and Atmospheric Administration)
Westar IV satellite downlink.
? Computer correlation studies of RV performance versus
variables of interest.
(U) References are listed at the end of this report.
3
Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
 Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
 Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
UNCLASSIFIED
III PRESENT STATUS (U)
A. (U) General
(U) In order to accomplish the goals set out in Section II, the
program has been designed as a joint effort between SRI International and
Time Research Institute of Los Altos, California, with SRI as the prime
contractor. Time Research Institute is a research organization that
specializes in temporal analysis of geophysical variables and their po-
tential correlation with phenomena of interest, such as weather patterns,
earthquakes, and so forth.
(U) With regard to the present effort, Time Research Institute is
responsible for establishing the appropriate hardware and software systems
for collecting and analyzing data concerning environmental conditions and
their correlation with RV performance. The purpose of the correlation
study is to determine whether RV performance is enhanced or degraded by
measurable changes occurring in the geophysical (including solar-
terrestrial) environments. The specific data bases under consideration
in this effort are given in Table 1.
(U) Should correlations between geophysical variables and RV
performance be found, the application potential of the effort is twofold:
(1) Time periods in which enhanced RV performance might be
expected could be identified, resulting in increased
quality and accuracy of information obtained through
such channels; similarly, time periods in which degraded
RV performance might be expected could be avoided. Thus,
optimum performance windows would be identified.
(2) An increased understanding of the types of environmental
changes that correlate with RV performance could provide
clues as to the mechanisms involved in RV functioning.
Such knowledge would lead to more focussed research on
factors that could enhance RV performance, and would
also provide information critical to the development of
defensive countermeasures against RV.
5
Approved For Release 2JIYGt8ASJS FffJ~0788R001800190001-3
 Approved For Release V nff%?t"bOO788R0O18OO19OOO1-3
(U) GEOPHYSICAL DATA BASES
? Solar-Terrestrial
- Geomagnetic
Ground-measured indices Ap, K, etc.
Satellite-measured intensities
- Solar flux (MHz)
15,400 1,415
8,800 606
4,995 410
2,800 245
2,695
- Sunspot number
- Solar flares
- Interplanetary magnetic field
- Solar wind (Pioneer XII)
- Protons
- Cosmic ray indices (neutron monitor)
? Ionospheric Measurements
- Sudden ionospheric disturbances (SIDS)
- Auroral electrojet
- Radio propagation quality indices
? ULF/ELF
- 30 frequencies (from 1 to 30 Hz)
(U) The tasks listed in Section II (literature search, real-time
ELF measurements, real-time geophysical data acquisition via satellite
downlink, and correlation studies of RV performance versus geophysical
variables of interest) have been prioritized with the goal of producing
the longest possible ELF data base during the period of this contract.
Therefore, while all of the tasks are being pursued in parallel, ELF-
related tasks have been the focus of attention to date.
Approved For Release (0/(k . O-OO788ROO1800190001-3
 Approved For Release 2o00/08/08 S CIA- I F I E D 07888001800190001-3
B. (U) Literature Search
(U) The purpose of the literature search is to acquire and
integrate information from recent work done in the field of biological
response to ELF and geophysical conditions. Approximately 30 keywords
are in use as input to a computerized literature search. Forty manu-
scripts have been obtained to date and are in the process of being
reviewed, plus additional sources of literature have been identified and
will be retrieved as priorities permit.
C. (U) ELF Measurements
1. (U) Introduction
(U) Although the ELF frequency range (3 to 300 Hz) has been
studied in some detail, many unknowns remain. For example, although it
is known that ELF frequencies generated by geophysical means (e.g., elec-
trical storm activity) tend to distribute themselves globally, little
information is available on the variation of the ELF environment from
location to location. Therefore, local variations may exist that are
caused by both manmade sources, and by the geological structure of the
area. In the San Francisco Bay Area, manmade sources that generate ELF
on a local scale include motors, telephone lines, power lines, and
electrical subways [Bay Area Rapid Transit (BART)], and it needs to be
determined whether the emission from such sources constitutes a signif-
icant contribution to the omnipresent global ELF field.
(U) In order to address the above issue, two ELF monitoring
stations are being set up--one at SRI Menlo Park (in the RV Laboratory),
the other at the Time Research Institute field station, 17 km distant.
It is anticipated that the SRI environment may be a "noisy" one due to
the large amount of electronics known to be in the area. Data from the
two sites, taking the field site as a reference, are to be compared in
order to begin to differentiate the naturally-occurring ELF from the
manmade noise occurring at the location where RV is being carried out.
Approved For ReleasEUN0./ASlf"-00788 R001800190001-3
 Approved For Release Ufft!VA. !95 J Jq E9E)OO788ROO18OO19OOO1-3
2. (U) Los Altos Site
(U) Since May 1982, Time Research Institute has been operating
an ELF monitoring site in Los Altos, collecting data twice daily for the
purpose of correlating ELF disturbances with various phenomena of interest.
In this period, analysis techniques were developed that are directly
applicable to the present task.
(U) One of the first tasks was the upgrading of the Los Altos
ELF monitoring site to provide coverage during power interrupts. Research
was performed to determine the best power-interrupt system, including
generators. The selection criteria chosen for the generator sought to
optimize power output, cost effectiveness, and reputation for reliability,
with the consideration that this system might serve as almodel for addi-
tional sites in the client community. On this basis, a 3500-W Kubota
generator was bought and installed at the site. A PTI Industries "Data-
shield" device was also purchased and installed, for use in conjunction
with the generator (which must be started manually). This device powers
the required electronics for a period of twenty minutes on its own while
awaiting generator startup. Furthermore, an automatic alarm telephone
dialing system dials up as many as four individuals should a power failure
occur while no one is at the site. The two devices working together have
protected the system on numerous occasions from power interruptions--
including an 8-hr outage planned by Pacific Gas and Electric Company.
Thus, since installation of the power-interrupt equipment, there has been
no loss of data collection/storage.
3. (U) ELF Data Acquisition Systems
a. (U) Basic System Design
(U) With the requirement that two ELF monitoring sites
be implemented for the program--one at SRI and one at Time Research
Institute--it was decided that the two systems would be made identical.
In this way, differences between the two systems would be minimized, thus
reducing the opportunity for artifactual differences between the two sys-
tem outputs.
Approved For Release 2LDKbA S,iF-b--O788ROO18OO19OOO1-3
 Approved For Release 2ffWQ8,i ft1f pf t?0788R001800190001-3
(U) Figure 1 is a block diagram of the basic ELF data
acquisition system. The ELF signal is collected by an antenna, amplified,
and then digitized by an analog-to-digital (A/D) converter so that the
signal can be input into an IBM PC Jr. computer for the purpose of analy-
sis by a fast-fourier-transform (FFT) program on at least an hourly basis.
The recorded data are then transferred by floppy diskette to an IBM XT
computer for further handling.
ELF
INPUT
DATA
A/D
CONVERTER
COMPUTER
IBM PC JR.
fHOURLY
DATA
MAKES SIGNAL READABLE BY
COMPUTER
DETERMINES INTENSITY OF SIGNAL
FREQUENCY COMPONENTS
FLOPPY DISKETTES TRANSPORTED
TO IBM XT COMPUTER
(U) As indicated in the above system description, an
integral part of the data acquisition system is computerized record
keeping, using IBM systems--both the IBM PC Jr. and the IBM XT. The
software is written and compiled on the XT, used as a master system, and
then run on the PC Jr. (the PC Jr. is not itself large enough to run a
compiler, nor are there compilers written for it). Beyond this, however,
in spite of the much publicized "compatibility" between the various
Approved For Release 2UNJLAtSIORD0788R001800190001-3
 Approved For Release WL. VAglfffff tf 0788R001800190001-3
IBM PC systems, some development time had to be expended to ensure that
programs compiled on the XT (that were concerned with communication with
external devices) could be run on the PC Jr.--because they handle the
addressing of communication ports differently.
b. (U) FFT Program
(U) A consultant was hired from the Radidscience Labora-
tory at Stanford University to critique the ELF system at Time Research
Institute as it stood at program start. As part of that critique, he
recommended changes in the FFT software to increase its accuracy and
running speed. As a result, new FFT software was written in compiled
BASIC and then debugged. The program description is given in Appendix A.
This task was completed in mid-April.
c. (U) System Electronics
(U) The prototype system of Figure 1 was assembled and
installed for our purposes at the Los Altos site. Early tests indicated
that modification of the existing electronics was required because
(1) the system was sensitive to ground-loop problems, thus the preamplifier
had to be redesigned to include an isolation amplifier, and (2) when the
new system was installed in mid-June, it was found that a slowly-varying
dc level was superimposed on the incoming ELF signal, resulting in ex-
cursions that exceeded the limits of the A/D converter at amplification
levels required for good signal analysis. Therefore, new circuitry was
designed to eliminate the dc problem. With these changes:, the ELF detec-
tion system is scheduled for completion of testing, debugging, and
calibration at the Los Altos site in August, before its sister system
is installed at SRI.
d. (U) ELF Electronics/Software Subtasks (Status)
(U) The status of the ELF electronics/software subtasks
Approved For Release UN 3 q FIED.00788R001800190001-3
 Approved For Release 20UJA(0d.A&&tFJE D788RO01800190001-3
(U)
? Subject to the requirement that the basic microprocessor
units to be used in the program would (1) be able to com-
municate with an IBM XT, (2) have at least 64K of memory,
(3) possess diskette storage capability, and (4) be cost
effective, the IBM PC Jr. was selected from among the
various alternatives, and two units were purchased.
? Preamplifiers, low bandpass filters, and amplifiers have
been designed, and one complete prototype system has been
assembled. Amplifiers and filters have been assembled
for the second system to be installed at SRI.
? Design specifications have been completed for the ELF
antenna. Assembly of this antenna is pending the results
of a calibration task (described in Section III.D.3.e).
? A survey of A/D converters that would be compatible with
the other system components was completed, and the
selected units were purchased.
? Communication between the IBM PC Jr. and the A/D con-
verter has been established, enabling the computer to
read the incoming signal.
? Software has been developed and debugged that: (1) reads
the communications RS 232 port of the PC Jr. input from
the A/D converter, (2) performs FFT analysis of the signal,
and then (3) writes half-hourly averages of 19 different
frequencies (from 1 to 29 Hz) to a computer diskette.
Further software refinements will continue to be made, such
as determining daily maximum values for each frequency.
This software has been implemented and debugged. The sys-
tem is ready to begin data acquisition upon implementation
of the dc-level-controlling hardware. Minor enhancements
of the software will continue.
? The Los Altos system is now in operation and is being
tested. It is anticipated that the SRI system will be
implemented in August.
e. (U) ELF System Calibration
(U) System calibration is proceeding. The Stanford
University consultant mentioned earlier is an expert in the areas of ELF
and VLF measurement, antenna design, and spectral analysis. His cali-
bration program is being carried out at the present Los Altos site using
specialized, sophisticated instruments from the Stanford Radioscience
Laboratory.
R,4rl ACZ~&Crl
Approved For Release 20/US/08'C1 W- 0788ROO1800190001-3
 Approved For Release iMtO/Q%3l .t 0788ROO18OO19OOO1-3
(U) As part of the calibration process, certain power-
line and power-supply "noise" sources were identified that could have
produced artifacts in the historical and ongoing data. Therefore, in-
creased electronic-filtering measures are being incorporated into the
system, including software changes to recalibrate past data already in
storage.
(U) Finally, as part of the calibration process, three
systems are to be monitored simultaneously: (1) the original unmodified
Los Altos system on which the historical ELF data have been acquired;
(2) a new system (described in this report) using an ELF wire-coil
antenna designed by and on loan from the Radioscience Laboratory at
Stanford University; (3) the new system, but with a bioantenna (oak tree)
as an antenna, which is a procedure recommended by Stanford (see Nature
reprint, Appendix B). The results from the calibration tests will be
used to calibrate the historical data base, and to fix the final antenna
design.
D. (U) Satellite Downlink Geophysical Data-Acquisition:System
I
(U) A near real-time satellite downlink system for solar-terrestrial
data has recently become available from NOAA (National Oceanic and Atmo-
spheric Administration). With this unit, it is possible to provide
immediate feedback and/or analysis in conjunction with RVsessions.
(Normally, there are long delays in procuring solar-terrestrial data;
without the downlink, delays of 10 days to 6 months are standard.) The
downlink system provides for accumulation of a detailed data base directly
on computer diskettes. (See Appendix C for an item-by-item description.)
(U) A satellite controller and a dish antenna for the downlink
system were ordered and installed at the Los Altos site early in the
project (see Figure 2). At the time of this writing, specifications for
data-acquisition software for the IBM PC Jr. have been completed, and
first-order software has been written that captures the data to computer
diskettes. Because of the large volume of data transmitted each minute,
a double-density, double-sided diskette fills in about 21 hours. Disk-
Approved For Releas6U ' I/A S Sl4rFtM-OO788ROO18OO19OOO1-3
 Approved For Release 2&fWt8Aftff ftt?0788RO01 800190001-3
CONTROLLER
F IBM PC JR,
DATA
LDISKETTE
IBM XT
COMPUTER
BURROUGHS B6900
COMPUTER
FIGURE 2 (U) REAL-TIME GEOPHYSICAL DATA
ACQUISITION VIA WESTAR IV
DOWNLINK
(U)
ettes have been changed on an almost daily basis since early April in
order to begin to acquire a long-term data base.
(U) To increase the amount of data that can be stored on a
diskette, it is necessary to separate numerical data of interest. The
necessary software is now being written that will be able to identify
the various data types as they are transmitted, so that only the data of
interest will be transferred to an appropriate data file on the diskette.
Statistical analysis can then be done on the data in this form.
(U) Other data bases are continuing to be maintained for this
project by Time Research Institute. Files of 2800-MHz solar flux, the
planetary magnetic activity index (Ap), the Anchorage magnetic index,
Approved For ReleaseUWE" TRJ*&00788 R001800190001-3
 Approved For Release ft
and the Stanford mean solar magnetic field are updated on a weekly or
monthly basis as the data are available. ELF data from the "old" system
are recorded twice daily; ELF data from the "new" system are being
recorded at the following intervals: half-hourly averages (48 times a day),
and two sets of half-daily averages (twice a day) at 00:00 and 12:00 UT,
and at 04:30 and 16:30 UT. Whole-day averages are also being recorded.
(U) To summarize the status of the satellite-downlink data-
acquisition system:
? The downlink-geophysical system is in place and in operation
at the Los Altos site.
? Partial data acquisition from the downlink is in ,place.
? Software for the final data-acquisition system will be
completed and implemented in August.
E. (U) Geophysical Data/RV Performance Correlation Analysis
(U) Data for RV sessions are to be analyzed for statistical cor-
relation with respect to the battery of geophysical data sets listed in
Table 1, and those acquired by Time Research Institute via measurement
(Figure 3). Recording of RV and geophysical data is now in progress.
The overall system for data acquisition and analysis of RV performance/
geophysical data is depicted in Figure 4. When enough data have been
collected toward the end of the contract period, analysis will be performed.
The tasks described in earlier sections are in preparation for this task,
and therefore have received the bulk of the effort. Certain subtasks in
the analysis task, which require a longer leadtime, have, however, already
been completed in preparation for the analysis.
(U) The primary statistical program that will be used to scan the
data for possible relationships is called EPOCH ANALYSIS. This program
reads two files simultaneously. The first file is an event file, the
second a data file. The program first reads an event, then scans the
data temporally backward and forward in time around the event. This
information is stored, a second event is read, and so forth. When all
Approved For Release 2U C8A *F ED0788R001800190001-3
 Approved For Release 20 J ff f ftQ788R001800190001-3
rz:~~
DATA
DISKETTE
DATA
DISKETTE
DATA
DISKETTE
(U)
the events and surrounding data have been read, a printout is created
that lists appropriate cross-correlation statistics between event and
data elements.
(U) Preliminary scans of data generated during an approximate
100-site series with one remote viewer have been carried out. The
session quality was graded on a scale of 0 to 3and correlations
between solar magnetic field and solar sunspot number were investigated.
Some correlation between RV performance and solar sunspot number was
found, which, if substantiated by further data, would indicate the
possibility that performance might improve immediately after a peak in
the sunspot number, and would deteriorate just before the sunspot number
peaks in its 27-to-29-day cycle. This result is based on data points
that are too small in number to be taken seriously at the point, however,
and is mentioned only to give an example of the types of correlations
that will be sought out and examined during the course of study.
Approved For Release W/bA SISMMD0788R001800190001-3
 Approved For ReleaseLWejQk SFj. ffb-00788 R001800190001-3
PERFORM-
ANCE
DATA
ELF
DATA
GEO-
PHYSICAL
DATA
DATA BASE
MANAGEMENT SYSTEM
BURROUGHS B6900 COMPUTER
ANALYSIS
REQUIRING LARGE M4INFRAME
(e.g., MULTIVARIATE ANALYSIS)
EPOCH
ANALYSIS
GRAPHS/
STATISTICS
STATISTICAL
ANALYSIS
GRAPHS/
STATISTICS
STATISTICS
AND
GRAPHS
MANUAL
ANALYSIS
Approved For Release UNI?8lA.SWFEIED00788R001800190001-3
 Approved For Release 2% 1(Q$/,TA CW- s I F I E D 7888001800190001-3
IV SUMMARY (U)
(U) Approximately 80 percent of the project's data-acquisition
tasks have been completed, and all of the equipment and hardware have
been purchased and delivered.
(U) System calibration should be completed in the near future,
and ELF and downlink-data acquisition will have begun in their final
formats.
(U) At the above point, the focus of effort will turn to analysis
of past and present geophysical and ELF data, soon to be followed by the
initiation of correlation studies of these data against RV performance.
Approved For Releas4 N / S SlPRt -00788R001800190001-3
 Approved For Release UQQ/9?x,(0$ SQ,P~-007888001800190001-3
THIS PAGE INTENTIONALLY LEFT BLANK
18
Approved For Release 20U?JeLA.%f"eD0788R001800190001-3
 Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
UNCLASSIFIED
FAST FOURIER TRANSFORM ROUTINE FOR ELF DATA (U)
Approved For Release 20QCyfft%.A1g-?ff788R001800190001-3
 Approved For Release 2000/08/08 : CIA-RDP96-00788ROO1800190001-3
UNCLASSIFIED
THIS PAGE INTENTIONALLY LEFT BLANK
20
Approved For Release eL/(knMt"6-OO788ROO1800190001-3
urim
 Approved For Release 2twtk"fftb 0788ROO1800190001-3
FAST FOURIER ROUTINE FOR ELF DATA (U)
************************************************************************
PROGRAM DESCRIPTION
************************************************************************
;THIS PROGRAM USES THE FAST FOURIER TRANSFORM (FFT) ALGORITHM TO
;CALCULATE THE SPECTRAL MAGNITUDES OF AN ARRAY OF CLOSE-PACKED REAL INPUT
;DATA POINTS. THE PROGRAM IS ORGANIZED AS FOLLOWS:
1. AT THE START, DATA POINTS ARE STORED IN THE REAL ARRAY X(1),
WHERE 1 RUNS FROM 0 TO NX-1. NX MUST BE AN INTEGRAL POWER OF 2,
OF THE FORM NX - 21"MX (IE, MX - LOG2(NX)). SUCCESSIVE ELEMENTS
OF X REPRESENT SUCCESSIVE SAMPLES OF AN INPUT SIGNAL, SAMPLED AT
REGULAR INTERVALS OF TIME DT.
2. THE REAL ARRAY X(i) IS TREATED FOR THE FFT AS A COMPLEX ARRAY
OF NX/2 PAIRS OF REAL AND IMAGINARY ELEMENTS. THAT IS, THE REAL
ELEMENTS OF THE ARRAY ARE THE EVEN-NUMBERED INPUT SAMPLES, AND
THE IMAGINARY ELEMENTS OF THE ARRAY ARE THE ODD-NUMBERED INPUT
SAMPLES. AFTER THE FFT IS CALCULATED, AN ADDITIONAL STEP IS USED
TO EXTRACT THE SPECTRUM OF THE REAL INPUT DATA. TRANSFORMING
CLOSE-PACKED DATA IN THIS WAY, EVEN THOUGH IT REQUIRES AN
ADDITIONAL STEP FOR THE REAL TRANSFORM EXTRACTION, IS FASTER THAN
TRANSFORMING A COMPLEX ARRAY OF NX REAL-IMAGINARY PAIRS (2*NX
ELEMENTS), WHERE ALL OF THE INITIAL IMAGINARY VALUES ARE ZERO.
3. THE FFT IS CALCULATED AS FOLLOWS:
B. THE FFT IS CALCULATED USING DECIMATION IN TIME, WITH
ANGLE ARGUMENTS IN EACH SUB-DFT APPEARING IN NATURAL
(IE, INCREASING) ORDER.
C. AFTER THE LAST PASS THROUGH THE FFT ALGORITHM, THE
ARRAY X CONTAINS SPECTRAL VALUES IN NORMAL ORDER, WITH
EACH EVEN POINT A REAL VALUE AND EACH SUBSEQUENT ODD
POINT THE CORRESPONDING IMAGINARY VALUE.
4. FOLLOWING THE FFT THE REAL TRANSFORM IS EXTRACTED. NOTE THAT
IF THE SAMPLING INTERVAL (THE TIME BETWEEN SUCCESSIVE SAMPLES) IS
DT, THEN THE TOTAL SIGNAL INTERVAL PROCESSED IS NX*DT (THE FFT
ASSUMES THAT THE MISSING RIGHT END POINT IS THE SAME AS THE FIRST
POINT). THE FFT GENERATES A SPECTRUM CONTAINING VALUES AT
INCREMENTS OF DF IN FREQUENCY, WHERE DF - 1/(NX*DT). IF WE WERE
TO TRANSFORM AN ARRAY OF NX COMPLEX DATA POINTS (WHOSE IMAGINARY
VALUES WERE ZERO, SINCE WE ARE CONCERNED WITH A REAL SIGNAL) WE
WOULD GENERATE NX COMPLEX SPECTRAL POINTS RANGING IN FREQUENCY
FROM .0' TO (NX-1)*DF Hz. HOWEVER, THE POINTS FROM (NX/2+1)*DF Hz
TO (NX-1)*DF Hz ARE MERELY THE COMPLEX CONJUGATES OF THE LOWER
POINTS, AND CONTAIN NO ADDITIONAL INFORMATION. THIS IS BECAUSE
THE NYQUIST RATE, OR THE HIGHEST UNALIASED FREQUENCY THAT CAN BE
SAMPLED, IS 1/2*NX*DF. WHEN WE TRANSFORM NX REAL CLOSE-PACKED
POINTS AND THEN EXTRACT THE REAL TRANSFORM, WE GET ONLY THE LOWER
NX/2 COMPLEX SPECTRAL VALUES, WHICH ARE ALL THAT ARE NEEDED.
Approved For Release 1?'LA5SIFC D-00788R001800190001-3
 Approved For Release 0nffA -_ q,fPf -00788 R001800190001-3
51
5. NEXT, THE SPECTRUM IS CONVOLVED WITH A SHORT WINDOW FUNCTION.
THE REAL AND IMAGINARY (EVEN AND ODD) POINTS ARE CONVOLVED
SEPARATELY, SINCE THE WINDOW FUNCTION IS A SEQUENCE OF REAL
NUMBERS. WINDOWING IS NECESSARY IF THE SIGNAL FILTERS WE ARE
EFFECTIVELY SYNTHESIZING ARE TO HAVE A USEFUL SHAPE. WITHOUT
WINDOWING WE WOULD FIND THAT EACH SPECTRAL FILTER WOULD HAVE A
NARROW PASSBAND BUT SIGNIFICANT SIDELOBE RESPONSES. THAT IS, THE
MAGNITUDE OF A SPECTRAL LINE X(1) WOULD DEPEND NOT ONLY ON SIGNAL
COMPONENTS NEAR 1*DF IN FREQUENCY, BUT ALSO SIGNIFICANTLY ON
COMPONENTS AT OTHER FREQUENCIES AS WELL. WINDOWING BROADENS THE
SHAPE OF THE PASSBAND OF EACH SPECTRAL FILTER BUT DECREASES THE
SIDELOBE RESPONSES. THE AMOUNT OF BROADENING AND SUPPRESSION OF
SIDELOBES DEPENDS ON THE WINDOW ORDER, OR THE LENGTH OF THE
WINDOW FUNCTION WHICH IS CONVOLVED WITH THE RAW SPECTRUM.
6. FINALLY, THE MAGNITUDE OF THE SIGNAL AT EACH SPECTRAL
FREQUENCY IS CALCULATED AS THE ROOT SUM OF THE SQUARE'S OF THE
REAL AND IMAGINARY SPECTRAL COMPONENTS (IE, THE VECTOR MAGNITUDE
OF EACH COMPLEX SPECTRAL POINT) AND SCALED TO MAKE THE PROCESSING
GAIN INDEPENDENT OF THE SIZE OF THE BLOCK LENGTH NX. WE COULD
ALSO CALCULATE THE PHASE OF THE SPECTRAL POINTS BUT THIS
INFORMATION ISN'T TOO VALUABLE FOR OUR USE. NOTE THAT THE
HIGHEST MEANINGFUL SPECTRAL FREQUENCY DEPENDS ON THE ANTI-
ALIASING LOW-PASS FILTER THAT IS USED WHEN THE SIGNAL; IS SAMPLED.
THERE IS NOT MUCH MEANING TO SPECTRAL POINTS ABOVE THE FILTER
CUTOFF FREQUENCY, WHERE SIGNALS ARE ATTENUATED AND FREQUENCY
ALIASING BECOMES A PROBLEM. THUS, IN THIS PROGRAM WE DO NOT
CALCULATE SPECTRAL COMPONENTS ABOVE ABOUT 30 Hz (THE NOMINAL
CUTOFF FREQUENCY OF THE LOW-PASS FILTER).
;THIS PROGRAM IS WRITTEN IN BASIC FOR THE IBM PC COMPUTER. HOWEVER, IT
;IS EASILY ADAPTED TO OTHER MACHINES. FOUR THINGS TO WATCH OUT FOR WHEN
;TRANSFERRING THE PROGRAM TO ANOTHER VERSION OF BASIC ARE:
1. ALL ARRAYS START WITH INDEX 0. THAT IS, THE ARRAY X(NX) RUNS
FROM X(O) TO X(NX-1). ARRAY REFERENCES WILL NEED TO BE CHANGED
IF THE PROGRAM IS TO BE USED ON MACHINES WHERE THE FIRST ELEMENT
OF ARRAYS HAS INDEX 1.
2. ALL VARIABLES WHOSE NAMES START WITH THE LETTERS "In THROUGH
"N" ARE IMPLICITLY INTEGERS (DEFINT I-N STATEMENT). THESE
VARIABLES MAY NEED TO BE RENAMED. IT IS IMPORTANT THAT LOOP
COUNTERS AND ARRAY INDEX VARIABLES BE INTEGERS BECAUSE INTEGER
ARITHMETIC (WHERE AVAILABLE) IS FASTER THAN FLOATING-POINT
ARITHMETIC.
3. THIS PROGRAM USES THE INTEGER DIVISION OPERATOR "\". THIS
MAY BE CHANGED TO "/" IN OTHER VERSIONS OF BASIC.
4. THIS PROGRAM USES LOGICAL OPERATORS ("NOT", "AND" "OR") ON
INTEGER VARIABLES (NOT LOGICAL VARIABLES) IN THE BIT-REVERSAL
ROUTINE. IF THESE OPERATORS ARE NOT AVAILABLE THE ALTERNATE
VERSION OF STATEMENTS 3000-3190 USING ONLY INTEGER ARITHMETIC
MUST BE SUBSTITUTED.
Approved For Release U NARAA~4 FMW-00788 R001800190001-3
 Approved For Release 2000/08/08 : CIA-RDP96-00788R001800190001-3
UNCLASSIFIED
wxwwxxxw*wwxxxwwxwx**w*w*wx**xw*w*xx*w*wx*w*xwx******xw*wx*xw***********
w
CHOICE OF SAMPLING TIME AND BLOCK SIZE w
w
xwwxww******wx*x****************w*w*****wxwxxxxwxxww*w**********xx******
;THE SAMPLING TIME DT AND THE BLOCK SIZE NX TOGETHER DETERMINE THE
;NUMBER OF SPECTRAL POINTS CALCULATED AND THEIR SPACING IN FREQUENCY.
;THE CHOICE OF SAMPLING TIME ALSO AFFECTS THE FREQUENCIES AT WHICH POWER-
;LINE HARMONIC INTERFERENCE THAT IS PASSED BY THE ANTI-ALIASING LOW-PASS
;FILTER WILL APPEAR IN THE SPECTRUM. THE FOLLOWING DISCUSSION ASSUMES
;THAT THE 36 Hz LOW-PASS FILTER WILL BE USED TO ATTENUATE HIGH FREQUENCY
;SIGNAL COMPONENTS BEFORE SAMPLING.
OF THE INPUT SIGNAL THAT IS SAMPLED CONTAINED ONLY FREQUENCIES BELOW
06 Hz IT WOULD BE SUFFICIENT TO SAMPLE AT THE NYQUIST RATE OF 69 SAMPLES
;PER SECOND. HOWEVER, SINCE THE FILTER DOES NOT HAVE INFINITE
;ATTENUATION ABOVE 30 Hz IT IS NECESSARY TO SAMPLE AT A SOMEWHAT HIGHER
;RATE, AND THEN DISCARD THOSE SPECTRAL POINTS REPRESENTING SIGNALS ABOVE
;36 Hz.
1. ONE CONSIDERATION IN CHOOSING THE SAMPLING RATE IS THE ACTUAL
RATE AVAILABLE WITH A GIVEN A/D CONVERTER. WITH THE CMC BUSSTER
D16 A/D CONVERTER, SAMPLING TIMES CAN BE AS SMALL AS ,0..0'Z05 s.
HOWEVER, IF MORE THAN ONE SIGNAL IS TO BE DIGITIZED AT ONE TIME,
THE MINIMUM SAMPLING TIME IS 0.21 s PER CHANNEL, OR A SAMPLING
RATE OF 166 SAMPLES/SECOND. THIS SEEMS LIKE A REASONABLE CHOICE.
2. A SECOND CONSIDERATION IS THE EFFECT OF THE SAMPLING RATE ON
THE FREQUENCIES OF ALIASED POWER LINE HARMONICS. ONLY HARMONICS
AT 6Z AND 128 Hz ARE LIKELY TO BE A PROBLEM. WITH DT ? 6.61 S,
THE NYQUIST RATE IS 56 Hz, SO THE SPECTRUM WILL CONTAIN POINTS AT
FREQUENCIES FROM 6 TO 50 Hz. ANY BPI Hz SIGNAL THAT IS DIGITIZED
WILL APPEAR IN THE SPECTRUM AT 40 Hz, WHICH POINT WILL BE THROWN
OUT, SO 66 Hz INTERFERENCE WON'T BE A PROBLEM. HOWEVER,
INTERFERENCE AT 126' Hz WILL APPEAR AT 20 Hz IN THE SPECTRUM, AND
THIS FACT MUST BE KEPT IN MIND WHEN ANALYZING THE DATA. (HIGHER
HARMONICS APPEAR AS: 180 AT 20 Hz, 240 AT 40 Hz, 300 AT 0 Hz,
366 AT 40 Hz, AND SO ON.)
;GIVEN A SAMPLING TIME
;FOR DIFFERENT CHOICES
DT ? 0.01 s, WE CAN CALCULATE THE SPECTRAL SPACING
OF BLOCK SIZE NX, AND WE FIND THE FOLLOWING:
NUMBER OF POINTS C= 36 Hz
64
128
0.7813
Hz
38
256
0.3906
Hz
76
512
0.1953
Hz
153
1924
0.6977
Hz
3Z7
2648
6.6488
Hz
614
23
Approved For Release UNWAS QP6-00788R001800190001-3
 Approved For Release U2000/08/0 N CL $ SC-16-00788R001800190001-3
*
WINDOWING
;*
;AFTER THE SPECTRUM OF THE INPUT SIGNAL HAS BEEN FOUND, IT IS CONVOLVED
;WITH A SHORT WINDOW SEQUENCE TO IMPROVE THE SHAPE OF THE SYNTHESIZED
;FILTERS. THOUGH THE SPECTRAL VALUES ARE COMPLEX, THE WINDOW SEQUENCE IS
;A SEQUENCE OF REAL NUMBERS, SO THE CONVOLUTION IS DONE SEPARATELY FOR
;THE REAL AND IMAGINARY SEQUENCES IN THE SPECTRUM. THE CONVOLUTION IS OF
;THE FORM
X(j)