Technical Note

Transcription

1 1R PV /,f_jl * s - y Technical Nte A. S. Griffiths Measurements f ELF Nise Prcessing 2 September 1975 Prepared fr the Department f the Navy under Electrnic Systems Divisin Cntract F C-02 by Lincln Labratry MASSACUSETTS INSTITUTE F TECNLGY LEXINGTN, MASSACUSEI M Apprved fr public release; distributin unlimited. {\W\l<^

2 The wrk reprted in this dcument was perfrmed at Lincln Labratry, a center fr research perated by Massachusetts Institute f Technlgy. The wrk was spnsred by the Department f the Navy under Air Frce Cntract F C-02. This reprt may be reprduced t satisfy needs f U.S. Gvernment agencies. This technical reprt has been reviewed and is apprved fr publicatin. FR TE MMANDER E^fgene C. Raabe, Lt. Cl., USAF Chief, ESD Lincln Labratry Prject ffice

3 MASSACUSETTS INSTITUTE F TECNLGY LINLN LABRATRY MEASUREMENTS F ELF NISE PRCESSING A. S. GRIFFITS Cnsultant Grup 66 TECNICAL NTE SEPTEMBER 1975 Apprved fr public release; distributin unlimited. LEXINGTN MASSACUSETTS

4 ABSTRACT This reprt describes nn-linear prcessing experiments with ELF nise recrded in Nrway, Saipan, and Greece. Effective nise levels, an imprtant measure f perfrmance fr a nn-linear receiver, were cmputed using sine-wave signals in the z and z bands. Althugh results were nt ptimum due t fixed (as ppsed t adaptive) clip levels, ntch filter effects and lcal sine-wave interference, they d shw diurnal variatins and d prvide a significant increase in the available data in this frequency range. 111

5 SUMMARY This reprts describes nn-linear prcessing experiments with ELF nise recrded in Nrway, Saipan, and Greece. The purpse f the experiments was t measure "effective nise levels," an imprtant measure f the perfrmance f a nn-linear prcessing receiver. The experiments used simulated sine wave signals in the z and z bands. The Nrway data (apprximately 6 hurs) was recrded during six perids f varius length frm January 1969 t September The Saipan and Greece data (apprximately 3 and 2 hurs, respectively) was recrded in May 1972, cincident with ne f the Nrway recrding perids. The results were btained using a fixed (as ppsed t adaptive) clipper and may therefre be subptimum. Emphasis was placed n examining highest effective nise levels which determine the minimum required signal energy received fr an ELF cmmunicatins system which must perate under wrst case cnditins. The highest levels (-130 db ) were bserved in the z band in Nrway; hwever, significant degradatin in the simulated receiver may be attributed t the prximity f the 50 z ntch filter and lcal interference. The highest Nrway levels in the z band were abut -137 db. While the data base fr these experiments was t small t measure seasnal variatins, diurnal The effective nise level is defined in Ref. 9 as where N..(dB ) = 10 lg iri (2 2 T/SNR) eft 10 s = injected signal strength (amp/m) = reference -field usually taken as 1 amp/meter T = matched filter integratin time (sec) 2 SNR = (mean /variance) at the matched filter utput.

6 variatins f 4-10 db were bserved (with the higher levels during the day). The highest levels in Saipan were abut -136 and -137 db, respectively, in the z and z bands. The highest levels bserved in Greece in the z band were abut -134 db. VI

7 Measurements f ELF Nise Prcessing I. INTRDUCTIN These experiments were part f Lincln Labratry's analysis and design f lng range ELF cmmunicatins spnsred by the Department f the Navy. Lng range cmmunicatin in the ELF band is pssible due t the relatively lw attenuatin rate f the atmsphere. Similarly the relatively lw attenuatin rate in sea water allws penetratin t submerged submarine antennas. The dminant surce f nise in this band is attributed t radiatin induced by lightning. Lcal nise characteristics are affected nt nly by nearby strms, which prduce large spikes, but als distant strms which generate backgrund nise with ccasinal spikes. Since there is a wrldwide variatin in thunderstrm activity there is a crrespnding variatin in ELF nise characteristics. An imprtant aspect f ELF nise is its distinctly nn-gaussian character. Wavefrms recrded thrughut the wrld display frequent spikes. Analysis f amplitude distributins similarly shws significant departures frm nrmal 2 distributins. While linear receivers are ptimum in a gaussian envirnment, mre sphisticated techniques are required t ptimize receiver perfrmance in a nn-gaussian envirnment. ' ' ' ' ' wever, extensive simulatins with digital recrdings f ELF nise indicated that a relatively simple clipper prvides significant imprvements cmpared t linear prcessing and that mre 9 sphisticated techniques which were examined yielded n further imprvement. Initial digital recrdings were wideband, had a wide dynamic range, and Antenna nise is als a factr

8 were relatively shrt (30 mins/tape). A series f analg recrdings were made fr prpagatin measurements ' ' ' and als t prvide a data base fr further receiver simulatins. In estimating receiver perfrmance under wrst case cnditins a very large data base is required. While these analg recrd- ings are still nt sufficient t base precise predictins f wrst case per- frmance, they represent a majr increase ver the digital recrdings. II. PRCEDURE Cncurrent with prpagatin measurements, ELF nise with an injected sine wave signal was recrded fr analysis f effective nise levels. During ne 15-day perid in May 1972 simultaneus recrdings were made in Nrway, Greece and Saipan. In additin, similar recrdings were made in Nrway during ctber- Nvember 1971, January 1972, September 1972 and July The recrding per- ids ranged frm 5 t 24 days; typically tw tapes each 6 t 12 hurs lng were recrded each day and subsequently analyzed. The injected sine wave, which was used fr calibratin f the prpagatin tests, was varied in frequency frm 42 t 48 z and frm 72 t 78 z. The in- jected signal level was designed t be high enugh t prvide reasnable mea- sures f signal-t-nise ratis ver perids cmparable t system message * lengths and t be lw enugh t reflect receiver perfrmance. In the higher frequency band the injected signal was apprximately -140 db Q and in the lwer band it was apprximately -135 db, * With nn-linear prcessing, the signal-t-nise rati is apprximately prprtinal t the signal strength fr small signals.

9 As shwn by Fig. 1 three channels f signa] plus nise were clipped and recrded in the field. The results described in subsequent sectins are based primarily n data frm tape recrder channel 2, which wes prprtinal t the time derivative f the -field ver the z range. The clip level n this channel was set at apprximately -123 db at 75 z in bth the Saipan and Nrway equipment and at apprximately -129 db at 75 z in the Greece equipment. Wavefrms n this channel shw definite clipping, but particularly in Saipan and Nrway the clip levels may have been set t high. As shwn by Fig. 2, three channels were prcessed in parallel in the labratry. The injected signal and the reference signal used fr matched filtering were recrded in the field and played back thrugh identical filters. Fluctuatins in tape recrder speeds thus intrduced identical phase shifts in bth signal and reference. Using a hybrid f analg equipment and a small cmputer (Varian 620/L), effective nise levels were calculated and displayed. Since the sftware was relatively straightfrward (invlving n digital filters) the prcessing culd be dne at 32 times real time. The lngest tapes f 12 hurs culd thus be prcessed in abut 22 minutes each, excluding setup time. In all the simulatins, message lengths f 13 minutes were used. The matched filter integratin time was typically 1-2 secnds. The effective nise levels (and simulatin parameters) were written n magnetic tape fr subsequent pltting and analysis. Results were als printed n the teletype cnsle, pltted n the CRT and cpies n a hard-cpy device while the analg tapes were rewund. By bandpass filtering the data befre sampling, a sample rate less than the Nyquist rate was pssible withut aliasing. In fact, the higher band ( z after the x32 playback speedup factr) was centered 480 z abve the sample rate f 1920 z while the lwer band ( after playback) was centered 480 z belw the sample rate.

10 In additin t the nise prcessing, spectral analysis f mst tapes was perfrmed using the same equipment. A Varian 620i cmputer was prgrammed using FFT subrutines t prvide the spectrum averaged ver the tape, nise pwer variatins as a functin f time, and als t plt a segment f the time wavefrm. Using a x32 playback factr this prcessing tk abut 22 minutes fr a 12-hur tape. Figure 3 illustrates the spectral analysis prcessing prcedure. The A/D cnverter was driven by a 250-z clck derived frm the recrded clck signal f 10 z. Fr a given channel, a buffer f 512 samples was filled (requiring 64 millisecnds f real time crrespnding t secnds f tape time). The pwer spectrum was cmputed using an FFT, buffered and cmbined with previus spectra fr subsequent averaging (requiring an estimated 1.1 secnds f real time r 35.2 secnds f tape time). Pwer estimates were btained by summing spectral values ver selected bands (e.g., z and z). This prcess was then repeated fr the next tw data channels. At the end f the input tape, the averaged spectra and the tempral pwer samples were cnverted t db, and written n an utput scratch tape fr subsequent pltting. Als written were tw secnds f wavefrms. As shwn in Table 1, the analysis band was z with 0.49 z frequency reslutin. Spectral samples were taken at abut tw-minute intervals n each channel, yielding 12 hurs x 30 samples per hur = 360 samples per channel.

11 TABLE 1 SPECTRAL ANALYSIS PARAMETERS Variables Sample Rate FFT Length Frequency Reslutin Analyzing Band Degrees f Freedm Prcessing Time per FFT N. f Channels Time Between FFT's n Input Tape 250 z 512 samples secnds z z 720 «35 secnds 3 ss2 minutes With x32 Playback 80 z 512 samples 64 msec z 0-40 z 720 «il. 1 secnds 3 ~3.3 secnds Results Table 2 summarizes the data derived frm Nrway, Saipan and Greece. Recrdings were made in Nrway n five ccasins. The highest levels were bserved in the z range in the Fall f 1971 and the summer f In bth Nrway and Greece there was nticeable pwer line interference at 50 z, which may have degraded the results particularly in the lwer band. At each site, in bth bands significant diurnal variatins were bserved. Fllwing is a summary f each set f measurements.

12 TABLE 2 SUMMARY F EFFECTIVE NISE LEVELS (db ) urs f z z Recrding Recrding data Site Perid Analyzed igh Medium Lw igh Median Lw Nrway ctber-nvember Nrway January Nrway May Nrway September Nrway July Saipan May Greece May Nrway: ctber Nvember 1971 ver a 22-day perid, 44 analg tape recrdings were made. Each day tw 8-hur tapes were recrded frm abut 20 t 04 and frm 10 t 18 GMT. During the first 12 days the injected signal was in the 72 t 78 z range. During the last 10 days the injected signal was in the 42 t 48 z range. Three channels, using the tw different clip levels and tw different frnt-end filters, were recrded in additin t the narrw-band channel, time cde, calibratin signal and reference signal. Frm a sample f 12 tapes with signals at 42 and 46 z representing abut 1 hurs f data ver a 6-day perid effective nise levels were derived and are summarized by Table 3. Sinusidal interference was bserved at 16 z and at 42 z. This interference, particularly at 16 z, was unusually high and quite

13 pssibly increased effective nise levels. Despite the pssible degradatin due t the ntch filter, the 46 z data was abut 2 db lwer than the 42 z data presumably due t the interference near 42 z. Because f pssible degradatin effects, this data may nt be representative. Analysis f these degradatin effects is presented in Appendix A. TABLE 3 EFFECTIVE NISE LEVELS MEASURED AT 42 AND 46 z FRM NRWAY, FALL 1971 Tape N. Start Date * Day/Night T.,., Effective Nise Level (db ) Injected Signal Frequency (z) Lw Median igh 33 ct. 28 Night Day Night Day Night Day Night 40 Nv. 1 Day 41 1 Night 42 2 Day 43 2 Night 44 3 Day Recrding times were 10 t 18 GMT during the day and 20 t 04 GMT at night.

14 Figure 4 is a plt f effective nise levels bserved in the z band ver the 12-day perid frm ctber 12 t ctber 24, Figures 5 thrugh 9 shw the same data n an expanded scale. As shwn by these figures, tw 7-hur segements were prcessed fr each day. ver this perid (representing 48 hurs f data) the median was abut -145 db and the extremes were and db. Table 4 shws a frequency distributin f the effective nise levels fr each 7-hur segment (32-13-minute messages) as well fr the entire perid. The highest levels were cncentrated ver brief perids. The afternns f ctber 13 and 15 prduced 26 f the 27 bservatins ver -140 db. Diurnal variatins are apparent; variatins greater than 10 db were bserved n ctber 13 and 15. Figure 9 shws the average effective nise level variatin as a functin f time f day. The peak at abut 13 GMT is abut 5 db higher than the minimum at abut 03. These diurnal variatins culd nt be related t changing levels in lcal man-made interference; spectral measurements ver a 7-hur perid n ctber 15 (see fr example, Figs. 10 and 11) remained vir- tually cnstant while the effective levels drpped 7 db frm abut nn t 16 GMT. Nrway: January 1972 ver a 6-day perid in January 1972, abut 54 hurs f ELF data with an injected 76 z signal were recrded and subsequently prcessed. The effective nise levels ranging frm -143 t -152 db, shwn by Fig. 12 are cnsiderably lwer than ther Nrway results. Cnsequently, they d nt directly affect minimum transmitter pwer requirements, but their deviatin frm previus bservatins suggests seasnal fluctuatins which shuld be cnsidered in 8

15 w rj i s 1 N 5. z I fa en E3 5 ~J 1-1 Z w > U w E fa w fa 33 ca -a > rj QJ D SB 0) > TB301 9ET- L l- 8EI- - 6CT- n- I*7l- 2<7l- EVT- *"' 5<7l- 9n- MX- sn- en rm r rm r r rm r 1-1 >9 m <r 1-1 in <r N rs N r-^ ~ m MJ- in N in N in U U r^ I** N N p^ U <-< N u p*. N N U U u u U rm N <r u U <t r^ <r 3 N N p* N N N N N U N i- d N N N r N N -< Mt N N i-, U -* U u r-~ d P-* N Mf N r N N U0 N N ~? N N N N u -ar^ N U0-3- N N r-. P~ b CQ Pi t/} U z UJ - 6<7T- gi- 0 a <u ^ 3 N D - 33 QJ N~-- h fa 1-. <u r V. n) v r«- c i 1 i rm m N <r 1 r N h. m 1 <r U u N N 1 1/1 h* r- N W0 r^ rn 1 N \ N r- r N 1 v r"» N 1 r r-~ N <r r 1 N UI r rm S r* r^ 1 1 K r^ 1 eh r^ N h«. 1 rm 1 NC r^ - 1 N r-~ <r ~j r 4-1 B»* 6-! a

16 estimating system perfrmance. As shwn by Fig. 12, five recrdings each 8-12 hurs were analyzed. The first tw (January 3 and 4, shwn in detail by Figs. 13 and 14) were recrded during the day. n bth days there was a rise f abut 4 db frm late mrning t mid-afternn. The last three (January 5-8, shwn in detail by Figs. 15 ) 16, and 17) were recrded frm late evening (22 GMT) t the next mrning (abut 09 GMT). All three f these recrdings shwed a 4-6 db drp frm late evening t very early mrning fllwed by a rise t mid-mrning. Althugh these are nt cntinuus 24-hur recrdings, they d suggest diurnal variatins f 6-8 db with a minimum in very early mrning and a maximum in the afternn r evening. Spectral analysis was perfrmed t nte prminent man-made interference and t determine whether variatins in effective nise level culd be related t variatins in spectra. The "sft" limiter channel which had very little clipping (see Fig. 18) was used fr spectral analysis while the "hard limiter" channel (Fig. 19) was used t cmpute effective nise levels. Cmparisn f spectra taken at 30-minute intervals ver a 12-hur perid shwed very little variatin. Fr example, Figs. 20 and 21 are spectra taken at 11 GMT and 19 GMT n January 4. ver this perid the effective nise level rse 2 db. When these spectra (and thers) are superimpsed nearly identical levels f sine wave interference are seen at abut 16, 42, 63, 84 and 105 z. The cnsistency f this interference is mst clearly shwn by a spectrum averaged ver several hurs using the hard limiter channels, fr example, Fig. 22; these lines and a smaller ne at 125 z are evident frm this spectrum which als shws the signal energy at 76 z. 10

17 Nrway: May 1972 ver a 15-day perid in May 1972, simultaneus ELF recrdings were made in Saipan, Greece and Nrway. As was bserved frm the Fall 1971 recrdings, the lwer band data frm Nrway and Greece was apparently degraded by lcal interference and by the effects f the 50 z ntch filter. Analysis f these effects is presented in Appendix B. Figure 23 shws the frequency distributin fr the higher band data recrded nearly cntinuusly frm May 1 t May 10. Table 5 shws the distributin f effective nise levels fr each f the tapes. The dd numbered tapes were recrded frm 0630 t 18 GMT and the even numbered tapes frm 1830 t 06. Cnsistent with ther bservatins, diurnal variatins f abut 8 db were apparent. Figures 24, 25, and 26 shw the effective nise levels n the three nisiest days. With the exceptin f Fig. 24 which shwed an unexpected peak arund midnight (pssibly due t lcal interference), the highest levels were bserved in the late afternn (dd numbered tapes). Nrway: July 1973 Frm July 15 t July 18 recrdings were made with injected signals at 76 z and 46 z. As shwn by Fig. 27, the 76 z levels were significantly lwer than 46 z levels except during a brief perid n July 16 when there was very strng lcal interference. As with the Fall 1971 and May 1972 data, the high levels at 46 z may be attributed t the prximity f the 50 z ntch filter. Saipan: May 1972 In May 1972 simultaneus recrdings were made in Saipan, Nrway (previusly described) and Greece. There were recrded and subsequently analyzed abut 215 hurs in the z band and abut 110 hurs in the z band. 1 1

18 6VT- m *71- p* r^ s s w rj 5 Z 2: w > t M Pi W 8 I pa -a aj > ai n 0 i: 0) > 0) 14 km w L*?\- 9*71- SVT- m- T<7T- ^T- 6et- un «m m m ^-\ s <r s vt cn -cr in r-~ -cr m s S 0 m S m CN -cr cn <r s CN cn s r^ -cr CN CTs s -cr <r 0 S r-) CTS -cr -cr as CTs CN cn <f s as m u-i s CN 3 CTS r^ cn p» S CTs u-i s r- -cr CN U0 S s S in c Csl U S S s s -3- S r-~ ( r-^ m s r -3-8ET- ci d w rj LZX- rj S 9ET- >s 0 3 cu ^-s 3 N cr x s s s r- s r-s r» r r r-- r^ r> r~ r*» r~ hs r-~ r-~ r^ r"s a) >s i-j cd w as Q ^ ai a. id M 1 i ~cr cn cn <r -cr m L S cn S s r"» cn <r m s CTs r-~ CTs 0 S CN n) u 0 s-s sj 12

19 Figures 28, 29, and 30 are typical spectra frm the three data channels. In cntrast t the Nrway and Greece equipment, the ntch filter was centered at 60 z, althugh n interference at this frequency is apparent. Interference at abut z was bserved n all spectra. The signal energy visible n these figures at 72 z is the injected signal. The 60 z ntch appears deeper and the sine waves slightly less prminent in Fig. 28 due t the higher clip level n this channel. The difference in clip levels is shwn by typical wavefrms, Figs. 31, 32, and 33. Energy arund 7 z is apparent frm the spectra and frm the wavefrm shwn by Fig. 31. Effective nise levels frm channel 2, prprtinal t d/dt, are sum- marized by Figs. 34 and 35 and by Tables 6 and 7. In the high band less than ne percent f the bservatins exceeded db and nne exceeded -137 db. In the lw band less than tw percent exceeded -139 db and nne exceeded -138 db. As seen frm Table 6, May 4 (Tape 37), May 7 (Tape 3013), and May 10 (Tape 3019) standut as the nisiest days in the z band. Figures 36, 37, and 38 shw the effective nise level variatins n thse three days. In each case the peak is arund GMT, fllwed by a gradual decline, reaching a minimum arund GMT. As shwn by Table 7 the highest levels in the lw frequency band were n May 11 and 12. Figures 39 t 42 shw the fur highest level days. The daily minima were typically arund GMT (like the high-frequency band). The maxima were nt sharp and were nt cnsistently at the same time. 13

20 6<7T- m ^ s (N as in m cn r*. r^ -! < > 02 W 55 M W g ^^ 1 33 M «a (XI w t/) > i J > 01 s -) 0) m <C > 0 d z w c/> 0) s pa > w 4-1 J 01 < m U < rj w rj c 2 8<7T- M- 9*7T- snvnevim- "[«?!- VT- 6CT- 8ET- Ltl- s as CN m - cn as r S as s 1--. s en s S IN m -3- sr cn CN CN CN i- r <r r-. en m cn m CN S as in r- s CN r- as S m i- - ^-1 d m in en vt r-- CTv Cl <r p~ CN in r^!- en CN an r^ CN -.-r CN m in m CN cn <r IN un CN CN S CN i CN as m 0s) s -* <r as t-t S -3- CN S as as cn i i- CN r--. m cn a- St S m s ejs as as s S «3 «3- cn as d!- cn -3- -a- -3- V w 9ET- >s y a <U /, 3 N VD s S s <N CN CN CN CN CN CN CN <r a r~ l> r~- r-. r~- r-~ r-~ r^ r^ r-^ 1-^ r- r* r^ r^ r^. r- r-~ r-^ IX QJ ^ M b /-^ 01 >> 4-t. CNI CN cn C1 s* st in m S S r*. r-~ CTi S - a ~ t-i SsS 01 i m <r m S r^ as CN en <r m S r~ as <«a CN 4J Fl «en 3 en Cl en en en cs-i en en en en m cn m cn en en cn cn in! 14

21 w I I «en M 1 M in en en i > 1-1 B S 'a 6*1-8<7T- Ln- 9*T- I" SVT- 33 a) > a) J ai t Q Z > W VVT- en- Z*T- VT- 6 T- 8CT- ZCT- - c m S 1-1 S > m r-l m sr S S s m S S r~ -* c in i-i \ s s as 1-1 m -I c s r-l S T c S m 0> S -* S S d S r-l S S m c eg r-l Date Frequency Tape (May) (z) -» -» S S S st l i en sf * r-l <t m -* r-l s a- <t «* -* U0 S ITl 4-1 8^5 s 0 15

22 Greece: May 1972 Like the Nrway data, the data in the lw frequency band frm Greece was crrupted by interference and the effects f the 50 z ntch. Typical spectra (Fig. 43) shwed interference arund 30, 35, and 1 z. Pwer line interference was apparent in the 50 z ntch. T determine the effects f sinusidal interference, experiments with different integratin times were dne. Effective nise levels attributed t randm nise are nt a functin f integratin time due t the nrmalizatin t a 1 z bandwidth. The effect f nearby sine wave interference, hwever, is a functin f the match filter spectral windw and hence the integratin time. The effect f interference is shwn by Fig. 44 which shws that 2 db lwer effective nise levels were btained by lengthening the integratin time and hence reducing the matched filter bandwidth. Analysis f the higher frequency data frm Greece was less affected by the interference and ntch filtering. Results are summarized in Fig. 45, which shws abur ne percent f the bservatins exceeding -135 db. Table 8 shws that mst f the higher levels were cncentrated n three tapes recrded n May 3, 7, and 10, which are als shwn by Figs. 46, 47, and 48. As in Saipan, diurnal variatins f 10 db r mre are apparent with peaks arund GMT and minima arund GMT. 16

23 6<7T*!- ZVI- 9- m m N CI eg <t in m m v en N N m N. <N <t N m ^-i r-< en en r- U1 -I I PN N ft N v N N N -* -* N & W Z M g - \ as X 03 1 ^-^ en aj Q 0) > > en Z 0) 0) < as > Pi 0) w > -i PQ U W CJ 0) rj N 14 h» U 3 N w W W w s t > u rj w t M as w > U w E E w S<7l- «7VT- C<7T- T7T- VT- 6EI- 8ET- ET- 9CT- 5ET- VCT- r~ en <n m N N en <t r~ r-1 in en en N N i-l r» m en N in n <t N -* N in N r~^ m m eg -< m en v en en en fil <r in en ^ r~ r^ (N m <f r-\ en en N vd IN. en m en i- i- N en rg r^ r-i <t m <r N sf <r -* m i-l <N <T I N en en -i -3- r-l V en N en N m -T rg vd v -Cf N Nt in N r~- r^ Nt N r- m r- en m m N ei N -9 ^ <n r~ \ r~ <n r», en >N d 0),-N 3 N N S vd N tn IN a> * a r» r- r^ r^ r^ r~ r~ Is. r~ r^ r^ h* r- i~» r- r-~ r~ 1**. r~ h* <D vj (X N» en m en N p~ N -a- N -3- N N l"n N en ^-t i en en <r «r in in N N r^ r^ N N Q «^ 1-1 *c I en <t in vd r~ (Ts en <J- in v r~ N (I) a a p i-l ^-t u Q <f 0 3 <~M IN ^1 esi CNI tn]»4 17

24 ACKNWLEDGMENT Principal cntributrs t the wrk reprted here, in additin t the authr, were Dale McNeill wh prvided much f the sftware, gden Nackney wh develped the hardware, and Mark Saklad wh did much f the prgramming. 18

25 LP ANTENNA CLIPPER CANNEL 2 BANDPASS FILTER NTC FILTERS 3-dB CTAVE FILTER CLIPPER CANNEL 1 CLIPPER CANNEL 5 TAPE RERDER ATTENUATR SIGNAL GENERATR CANNEL 4 TIME DE CANNEL 7 GENERATR Fig. 1. Simplified blck diagram f recrding equipment Fig. 2. Simplified blck diagram f nise prcessr. 19

26 ANALG TAPE RERDER 8 kz CLCK -4 LPF f = 40 z 3 BUFFERS LPF f«40 z LPF f =40 z MPUTE AND STRE PWER UTPUT TAPE A/D BUFFER 512 SAMPLES FFT AND SQUARE RUNNING SPECTRUM AVERAGE FINALIZATIN CRT INTERRUPT WAVEFRM BUFFERS BUFFERS Fig. 3. Simplified flw chart f spectral analysis ll-«-1i!l 76 z-. 74 z-. 76 z 72 z 78 z 76 z 1 -,38 IB > U 1 1 l 1 z U > U UJ Ik I l -144 <4 - * it ~ u * -I50l l 1 1 / 11 1 I III I 1 1 ' 1 IJ l«m 16 (7 IB «CTBER Fig. 4. Effective nise levels at z recrded in Nrway, ctber 13-24, t 20

27 11»-1- UTU -74 z- > UJ z Fig. 5. Effective nise levels frm Nrway, ctber 13-15, j - 76 z- V Fig. 6. Effective nise levels frm Nrway, ctber 16-18,

28 -134 ll-6-tlt ~b -138 X m UJ -140 > UJ UJ > -144 UJ " UC «0 V J* j /* h p V ' i irf ' \ < V 1 Fig. 7. Effective nise levels frm Nrway, ctber 19-21, 1961, -78 i- -76 i UJ -140 > 2 Fig. 8. Effective nise levels frm Nrway, ctber 22-24,

29 -134 II-8-ISI?; -138 Z s I -( (42 z ul 1- V \ > N -IS ( TIME (GMT) Fig. 9. Average diurnal variatin in effective nise level recrded in Nrway, Fall 1971, z. 50 ( j S 1 UJ h 20 ift-1 i IK i uin /V*i f r (^ K! l 1 ( FREQUENCY (i) Fig. 10. ELF spectrum recrded in Nrway, GMT, ctber 15,

30 50 IS-6-1G7?4 m J LALAJ 1 - * J *1% V M M Vd f i v Id f (/> V 1/ V \k *Y*W FREQUENCY (z) Fig. 11. ELF spectrum recrded in Nrway, GMT, ctber 15, [ ia- 6-16T251 UJ -146 > z JAN 3 JAN 4 JAN S JAN 6 JAN 7 JAN 8 Fig. 12. Effective nise levels at 76 z recrded in Nrway January 2,

31 1 ll-j-ltt!s I m -144 J UJ -146 > UJ _1 UJ! g X V / AAJ ^, A/v ' vv \A -^ TIME (GMT) Fig. 13. Effective nise levels at 76 Kz recrded in Nrway, January 3, IB-6-16T2T -142 I J -146 > UJ -J!/> -14t z UJ > -1*0 UJ li- ft TIME (GMT1 Fig. 14. Effective nise levels at 76 z recrded in Nrway, January 4,

32 116-B- 16T2») > UJ _J -146 ifl -14«Vv V \r\ V 22 2* (0 TIME (GMT) Fig. 15. Effective nise levels at 76 z recrded in Nrway, January 5-6, u-t-«?ij] Y \ AA v \ \J f^a J /wv TIME (GMT) Fig. 16. Effective nise levels at 76 z recrded in Nrway, January 6-7,

33 -1* l«t50 UJ > ^K/^J -ts« * TIME IGMT) Fig. 17. Effective nise levels at 76 z recrded in Nrway, January 7-8, SENDS Fig. 18. Typical wavefrm frm "sft limiter channel, Nrway, January

34 Fig. 19. Typical wavefrm^ trn "hard" limiter channel, Nrway, January t FREQUENCY (i) Fig. 20. ELF spectrum recrded in Nrway, GMT, January 4,

35 ll-6-(8?j v A 1 \j* / r w v k k, II FREQUENCY (i) Fig. 21. ELF spectrum recrded in Nrway, GMT, January 4, >'35 3 IT FREQUENCY (j) Fig. 22. ELF spectrum frm "hard" limiter channel recrded in Nrway, GMT, January 4,

36 23.3 l» s <r 2.4 db ^J I (34 Fig. 23. Distributin f effective nise levels fr Nrway data in the z band (10 days in May 1972). db. ls w^ > -142 \ W V ^Y TIME (GMT) Fig. 24. Effective nise levels at 76 z recrded in Nrway, May 2-3, 1972 (tapes 13-14). 30

37 ] ta-e-tens v^ > -M2 V i.»/l A V r JtfSM I \ V T (8 TIME (GMT) Fig. 25. Effective nise levels at 72 z recrded in Nrway, May 7-8, 1972 (tapes ) [ll-t-iitjs Z fi -J -111 > > -142 UJ b. vv /** r V\A. V \AV^ W T t TIME (GMT) 0, - t± IAL Fig. 26. Effective nise levels at 78 z recrded in Nrway, May 10-11, 1972 (tapes ). 31

38 11B-S-1C7401 STRNG LCAL INTERFERENCE W I J 1 If i " I > -140 I I ^ / I V! V ; JULY 15 JULY 16 JULY 17 JULY 18 +» 46I- Fig. 27. Effective nise levels recrded in Nrway, July U-6-K 74l p \ 50 FREQUENCY (j) Fig. 28. ELF spectrum recrded in Saipan, GMT, May 3, 1972 (tape 35, channel 1, 3 db/ctave, sft limiter) 32

39 50 $ 30 ^nb» 3 cr - u S> M 1 k/ } ^ f JLr ^ \ ll-(-1tt42 S FREQUENCY (i) Fig. 29. ELF spectrum recrded in Saipan, GMT, May 3, 1972 (tape 35, channel 2, d/dt, hard limiter). i i-6-ie M3 *\ 1 F *V Vv A ^ r a 20 \ V j ^*-*>»~V <~TT s FREQUENCY (j> Fig. 30. ELF spectrum recrded in Saipan, GMT, May 3, 1972 (tape 35, channel 5, 3 db/ctave, hard limiter) 33

40 Fig. 31. Typical wavefrm recrded in Saipan, May 3, 1972 (tape 35, channel 1, 3 db/ctave, sft limiter). Fig. 32. Typical wavefrm recrded in Saipan, May 3, 1972 (tape 35, channel 2, d/dt, hard limiter). 34

41 Fig. 33. Typical wavefrm recrded in Saipan, May 3, 1972 (tape 35, channel 5, 3 db/ctave, hard limiter). 29 ia ] db i L i M db0 Fig. 34. Distributin f effective nise levels fr Saipan data in the z band (10 days in May 1972). 35

42 u-s KMl 25 <r 15 db I i db Fig. 35. Distributin f effective nise levels fr Saipan data in the z band (5 days in May 1972) (38 ~ -1*0 i -* n z > A* A MM /' ^ v \,/v IB-S-1S749 ] It. UJ -130 \ \ Vj " WV «TIME (GMT) Fig. 36. Effective nise levels at 72 z recrded in Saipan, May 4-5, 1972 (tapes 37-38). 36

43 ll-t-ltj50-1*0 > 111 -I s - ««^M/ WAJ A \ \ A, / Y ^ V 18 TIME (GMT) 06 Fig. 37. Effective nise levels at 72 z recrded in Saipan, May 7-8, 1972 (tapes ). ffi > Id J u z \l u - u \i Ul b A >. \ w\ v hm % ll-(-l(7st J -152 (B TIME (GMT) Fig. 38. Effective nise levels at 78 z recrded in Saipan, May 10-11, 1972 (tapes ). 37

44 «-« X m > w W -1A lavi UJ u Id Ik JKA\ A V W^ pry \ A, ^ V*/ / -14» 1830 TIME (GMT) Fig. 39. Effective nise levels at 46 z recrded in Saipan, May 11-12, 1972 (tapes ). U-«-16T55 ~b -US X CD > -142 X V ^ V w A/V \ v V y TIME (GMT) Fig. 40. Effective nise levels at 42 z recrded in Saipan, May 12-13, 1972 (tapes ). 38

45 -132 (l-«-is75» -134 " "<36 X B > w g " 4 A i- u 111 ii. t "'44 V A/ \ yk\/ k V V / ^ l ; -I4» 0» TIME (GMT) Fig. 41. Effective nise levels at 48 z recrded in Saipan, May 13-14, 1972 (tapes ). (I-I-«IS5 -'38 > -142 \ f*\ vv M V V / /V/ s Vw TIME (GMT) Fig. 42. Effective nise levels at 42 z recrded in Saipan, May 14-15, 1972 (tapes ). 39

46 ,1 jn i V Vsi ^J 1 w * [l8-6-l FREQUENCY (i) Fig. 43. ELF spectrum recrded in Greece, May 3, 1972 (tape 2025) TIME (GMT) Fig. 44. Effective nise levels at 42 z recrded in Greece, May 15, 1972 (tape 2028). 40

47 v 3.2 db J«-134 Fig. 45. Distributin f effective nise levels fr Greece data in the z band (10 days in May 1972) l-*-ttTM -134 " -IS* 81 -««$ -142 u UJ k (A M \ m v,, Ml >^A \ jr V VM -we TIME (GMT) Fig. 46. Effective nise levels at 72 z recrded in Greece, May 3-4, 1872 (tapes 25-26). 41

48 t I U u > -142 \A V 18 TIME (GMT) \ \/ Fig. 47. Effective nise levels at 72 z recrded in Greece, May 7-8, 1972 (tapes ). « t T CD T> 111 > Ui -J -138 UJ -140 u g UJ > -142 K tf B -144 n, j/wi y V i/i/ \ A / w 18 TIME (GMT) A V^ V Fig. 48. Effective nise levels at 72 z recrded in Greece, May 10-11, 1972 (tapes ). 42

49 REFERENCES 1. A. L. Bernstein et al., "Lng Range Cmmunicatins at Extremely Lw Frequencies," Prc. IEEE 62, (March 1974). 2. J. E. Evans, "Preliminary Analysis f ELF Nise," Technical Nte , Lincln Labratry, M.I.T. (26 March 1969), DDC AD F. artley, " Wideband Technique fr Imprving FSK Receptin in Atmspheric Nise," Prc. Nat. Electrnics Cnf. 24, (1968). 4. D. L. Snyder, "ptimal Binary Detectin in Knwn Signals in a Nn-Gaussian Nise Resembling VLF Atmspheric Nise," WESN Tech. Papers, pt. 4, WESN, Ls Angeles, Calif., Sessin 6, Paper 4 (1968). 5. V. R. Algazi and R. M. Lerner, "Binary Detectin in White Nn-Gaussian Nise," Lincln Labratry, M.I.T. (Unpublished Grup 65 reprt) Ye. Antnv, "ptimum Detectin f Signals in Nn-Gaussian Nise," Radi Eng. Electrn. Phys., 1_2, (1967). 7. S. S. Rappaprt and L. Kurz, "An ptimal Nnlinear Detectr fr Digital Data Transmissin Thrugh Nn-Gaussian Channels," IEEE Trans. Cmmun. Tech. M-14, (1966). 8. J. E. Evans, "Prbability Density Functin Estimatin (with Applicatins t Receiver Design fr Receptin in Nn-Gaussian Nise)," Technical Nte , Lincln Labratry, M.I.T. (29 August 1969), DDC AD A. S. Griffiths, "ELF Nise Prcessing," Technical Reprt 490, Lincln Labratry, M.I.T. (13 January 1972), DDC AD J. E. Evans, D. K. Wlllim, and J. R. Brwn, "Descriptin f the Lincln Labratry Wideband ELF Nise Recrding Systems," Technical Nte , Lincln Labratry, M.I.T. (14 April 1972), DDC AD D. P. White and D. K. Willim, "ELF Prpagatin Study (Phase II - Fall 1971)," Technical Nte , Lincln Labratry, M.I.T. (15 February 1972), DDC AD D. P. White, "ELF Prpagatin Measurements (Phase III - Fall 1971)." Technical Nte , Lincln Labratry, M.I.T. (31 July 1972), DDC AD D. P. White, "ELF Prpagatin Measurements at Nrway, Saipan, and Greece (Phase IV - Spring 1972)," Technical Nte , Lincln Labratry, M.I.T. (7 December 1972), DDC AD

50 APPENDIX A Analysis f Interference Effects n z Data Recrded in the Fall f 1971 in Nrway Examinatin f the Nrway data recrded in the Fall f 1971 indicated in- termittent man-made interference which prbably significantly increased the effective nise levels in the z band. igh levels f 16 z interference were bserved n several ccasins. Figures A-l and A-2 illustrate spectra taken frm these Fall 1971 tapes, using 1/2 z reslutin averaging 2 blcks ver three-hur perids using tape recrder channel 1, the "sft limiter" channel. As shwn by Fig. A-3, the clip level is exceeded a very small fractin f time n this channel. n all tapes analyzed n the preliminary spectral analysis there was significant interference at abut 16 z, db abve the backgrund. There was als interference at abut 42 z (2 t 10 db abve the backgrund) at abut 63 z (1-7 db abve the backgrund) and at abut 84 z (2-8 db abve the backgrund). These spectra als shw the relatively wide ntch at 50 z withut any visible 50 z interference. Figure A-4 shws a plt f a typical wavefrm frm the "hard limiter" channel after bandpass filtering. Figures A-5 thrugh A-10 are effective nise level plts in the z range frm this channel. As cnsistently bserved in the Nrway data, diurnal variatins were apparent with a definite upward trend in the late afternn; in five f the six afternns examined there was a rise f 3-4 db frm GMT. ver a 24-hur perid there was a 3-6 db variatin. Figures A-7 and A-8 are effective nise levels measured at 46 z, while 44

51 Figs. A-5, A-6, A-9, and A-10 are 42 z data. The ntch centered at 50 z culd be expected t degrade the 46 z results. Interference near 42 z culd degrade the effective nise levels at this frequency. There were n dramatic differences between the 42 z and 46 z data, althugh the 46 z tapes had slightly lwer levels (by 2 db n the average). Figure A-7 shws a rapid drp in effective nise level frm a very high level, abut -132 db t a lw value f abut -140 db ver a 4- hur perid in the middle f the night. Because such a drp had never been seen n previus Nrway recrdings, particularly at night, the high levels were attributed t lcal interference. T cnfirm this assumptin further spectral analysis was cnducted n this Tape (37) and n Tapes 34 and 40 which shwed shrt-term variatins. Figures A-ll and A-12 are spectra taken at 2030 and 2130 GMT f Tape 37 when the effective nise level was rapidly drpping, as shwn by Fig. A-7. These spectra shw unusually high levels f 16 z interference, 20 db abve the backgrund cmpared t previus bservatins when it was db abve the backgrund. N ther spectral lines are utstanding; Fig. A-ll shws sme energy at abut 46 z which was the signal frequency. Bth Figs. A-ll and A-12 indicate energy at 42 z. Frm the early drp in effective nise level, ne might expect nticeably mre interference n the earlier spectrum. wever, Fig. A-9 indicates that the drp may nt have been mntnic. Starting at 2030 GMT, successive spectra every 30 minutes shwed the level at 16 z t vary: 6, 5, 7, -2, 3, -2, 0, 0, 0, -2 db with respect t ne digital unit. Figures A-13 and A-14 are later spectra in this tape when the effective nise level was lw. The 16 z is still prminent, but abut 5 db lwer than the earlier 45

52 spectra. While the variatins in 16 z interference d nt clsely track the variatins in effective nise level, there is sufficient crrelatin t suspect that the effective nise level variatin n this tape is nt due t natural effects. Examinatin f sample wavefrms, Figs. A-15, A-16, and A-17 indicate that the 16 z interference dminates the backgrund but that its amplitude fluctuates. n ctber 29 and Nvember 1, (Tapes 34 and 40) there was a 3-4 db rise in effective nise level during the afternn (Figs. A-5 and A-8). Spectral analysis f ctber 29 data shwed a fairly cnsistent decrease in 16 z interference as the effective nise level increased. Fr example, Fig. A-18 shws a level f -1 db at 01 GMT when the effective nise level was -136 db while Eig. A-19 shws a level f -7 db at 14 GMT when the effective nise level had risen t db. Sample wavefrms, Figs. A-20, A-21, and A-22 again indicated that the 16 z interference level was varying althugh the different scales exaggerate this effect. Like the ctber 29 results, analysis f Nvember 1 data did nt shw dramatic crrelatin between interference levels and effective nise levels. While there was a fairly steady rise in effective nise level frm abut 11 GMT, -138 db t abut 1630 GMT, -135 db during this perid, the nly nticeable interference was 16 z (Fig. A-22, fr example) which fluctuated frm -3 t -7 db with a general dwnward trend. In summary the nighttime variatins in effective nise level may be due t man-made interference at 16 z while the afternn variatins as previusly bserved are prbably due t variatins in the ELF backgrund. 46

53 50 [tt-l-k)k «0 S 30 GJ 8; 2 (0 J/ \ A 1 K \ **«* / \^f t v\»a> \ 1 r \ v ) 50 FREQUENCY (z) Fig. A-l. ELF spectrum recrded in Nrway, GMT, Nvember 2, 1971 (tape 41). 50 «-«-1«T«j] U «* A Vi A..»fc f j \ /^ ta "W«An \/ rf> Iv V / 50 FREQUENCY (i) Fig. A-2. ELF spectrum recrded in Nrway, GMT, Nvember 2, 1971 (tape 42). 47

54 l8-6-!6764 J L J I I L 0.5 TIME (sec) Fig. A-3. Typical wavefrm recrded in Nrway, Fall 1971 (tape 41, channel 1, 3 db/ctave, sft limiter). Fig. A-4. Typical wavefrm recrded in Nrway, Fall 1971 (tape 41, channel 5, 3 db/ctave, hard limiter). 48

55 [ ' X 5 UJ > UJ UJ tn -*M I A ft M,i.A V h 1 z W V A S.IM u UJ b. u. UJ 1 / u r V »> t4 TIME 1GMT) Fig. A-5. Effective nise levels at 42 z recrded in Nrway, ctber 28-29, 1971 (tapes 33-34). ] V li i/r Kii, wv / - - -» 1 TIME (GMT) Fig. A-6. Effective nise levels at 42 z recrded in Nrway, ctber 29-30, 1971 (tapes 35-36). 49

56 l l u -I i UJ > -136 f- I \ \j^ Vt v /I \ « TIME (GMT) (i A Fig. A-7. Effective nise levels at 46 z recrded in Nrway, ctber 30-31, 1971 (tapes 37-38). ]«-6-l 769! d*«!>.116 b i "v rwv \fj* \ji Tl \ 10 TIME (GMT) Fig. A-8. Effective nise levels at 46 z recrded in Nrway, ctber 31-Nvember 1, 1971 (tapes 39-40). 50

57 4» 1 l» V V i\ vw ^ n TIME (GMT) l«-6-l6770 Fig. A-9. Effective nise levels at 42 z recrded in Nrway, Nvember 1-2, 1971 (tapes 41-42) ] AW i«\. fi f y u \ ii TIME (GMT) Fig. A-10. Effective nise levels at 42 z recrded in Nrway, Nvember 2-3, 1971 (tapes 43-44). 51

58 1 t8-6-l Y / h I (ta tau V,A»» ^ V 1 I i. 1 FREQUENCY (i) Fig. A-11. ELF spectrum recrded in Nrway, GMT, ctber 30, i B-6-16' j P \l Vw fu W\ i * r V \ 1 FREQUENCY (z) V/ i. -JL. Fig. A-12. ELF spectrum recrded in Nrway, GMT, ctber 30,

59 1 FREQUENCY (z) Fig. A-13. ELF spectrum recrded in Nrway, GMT, ctber 30, «g 30 a 20 FREQUENCY (jl Fig. A-14. ELF spectrum recrded in Nrway, GMT, ctber 30, 1971 (tape 37). 53

60 Fig. A-15. Typical wavefrm recrded in Nrway, 2230 GMT, ctber 30, 1971 (tape 37). Fig. A-16. Typical wavefrm recrded in Nrway, 2230 GMT, ctber 30, 1971 (tape 37). 54

61 Fig. A-17. Typical wavefrm recrded in Nrway, 2330 GMT, ctber 30, 1971 (tape 37) AV1 \, Jfk (i i v if Tnv V v * V!' FREQUENCY (z) Fig. A-18. ELF spectrum recrded in Nrway, GMT, ctber 29, 1971 (tape 34). 55

62 l8-6-l67b0 1 c I- u in 1 1 USt nfrs V U>, 1 M \, *A *J \ 1 FREQUENCY (z) \ 1 Fig. A-19. ELF spectrum recrded in Nrway, GMT, ctber 29, 1971 (tape 34) I llik h IL ii i i ili LI Ii i i i i i i i i i TIME (sec) Fig. A-20. Wavefrm recrded in Nrway, 13 GMT, ctber 29, 1971 (tape 34). 56

63 Fig. A-21. Wavefrm recrded in Nrway, 14 GMT, ctber 29, 1971 (tape 34) TIME (wc> J L Fig. A-22. Wavefrm recrded in Nrway, 15 GMT, ctber 29, 1971 (tape 34). 57

64 l8-6-!6784 m 30 1 I If 1 M M ^v» h Tj V l t FREQUENCY 1z). JU^A Fig. A-23. ELF spectrum recrded in Nrway, GMT, Nvember 1, 1971 (tape 40). 58

65 APPENDIX B Analysis f z Data Recrded in Nrway and Greece, May 1972 Frm May 11 t May 15, the injected signal was varied frm 42 t 48 z. Labratry calculatin f effective nise levels yielded results which were unexpectedly high. Subsuquent analysis indicated that the high levels may be attributed t degradatin due t the ntch filter and due t lcal interference, Figure B-l is a spectrum f Nrway Tape 1025, which had an injected signal at 48 z (nt apparent in plt) and which had very, very high effective nise levels (-136 t -131 db ). Five features are apparent frm this plt: (1) Extrardinary interference ccurred arund 42 z and varius harmnics. (2) The 50 z ntch is brad enugh t seriusly degrade a 48 z signal. (3) N 50 z is apparent althugh ne might expect t see a line in the ntch. (4) N 150 z is apparent, althugh interference is very prminent arund 165 z. (5) Unlike Fall 1971 data, 16 z interference is nt apparent. Figure B-2 is a spectrum f Nrway Tape 1028 which had a 42 z signal and cmparatively lw effective nise levels (-139 t -135 db ). In cntrast t Fig. B-l, the interference at 42 z and harmnics is virtually absent the energy at 42 z is prbably signal. Interference at 16 z is apparent. The 50 z ntch appears deeper than frm Tape 1025, suggesting the first tape had clippings and assciated spectral smearing. Still n 50 z r 150 z is ap- parent but there is a suggestin f interference at 165 z. The ntch filter which attenuates the 50 z pwer lines by abut 50 db als attenuated the 5 0

66 injected signal: 10 db at 48 z, 5 db at 46 z and 3 db at 42 z. In a linear system with a narrwband signal, this signal attenuatin wuld have little effect since the nise in the matched filter band wuld be similarly attenuated. In a nn-linear system, hwever, ut-f-band nise is flded int the signal band. With nn-gaussian nise, it is difficult t analyze quantitatively the fld-ver effects. In rder t determine if the high effective nise levels were due t sine wave interference near the signal frequency, simulatin experiments, varying the matched filter integratin times (ranging frm 1 t 16 secnds) were perfrmed n tw tapes. If degradatin had ccurred due t sine wave interference clse t the signal frequency (nt capture effects), lnger integratin times (i.e., narrw matched filter bandwidths) wuld reduce this effect. The tw Nrway tapes shwed n nticeable imprvement with increased integratin times. 60

67 1 l8-6-!6785 s => IT J V ^ / [ 1 K >w 1 FREQUENCY (i) Fig. B-l. ELF spectrum recrded in Nrway, GMT, May 13, 1972 (tape 1025). t FREQUENCY (z) L Fig. B-2. ELF spectrum recrded in Nrway, GMT, May 14, 1972 (tape 1028). 61

68 External Distributin List Chief f Naval peratins (0P-941P) Department f the Navy Washingtn, D.C Chief f Naval peratins Dr. R. E. Cnley (P-94) Department f the Navy Washingtn, D.C Chief f Naval Research (Cde 418) Attn: Dr. T. P. Quinn 8 Nrth Quincy Street Arlingtn, VA Cmputer Sciences Crp. Systems Divisin Attn: Mr. D. Blumberg 6565 Arlingtn Blvd. Falls Church, VA Directr Defense Cmmunicatins Agency Cde 960 Washingtn, D.C (10 cpies) IIT Research Institute Attn: Mr. A. Valentin, Div. E 10 W. 35th Street Chicag, Illinis Naval Civil Engineering Labratry Attn: Mr. J. R. Allgd Prt ueneme, CA Naval Electrnics Labratry Center Attn: Mr. R. 0. Eastman 271 Catalina Blvd. San Dieg, CA Naval Electrnic Systems Cmmand Attn: PME-117, Captain W. C. Cbb Department f the Navy Washingtn, D.C Naval Electrnic Systems Cmmand Attn: PME-117T, Mr. J. E. DnCarls Department f the Navy Washingtn, D.C Naval Electrnic Systems Cmmand Attn: PME , Captain J. Gallway Department f the Navy Washingtn, D.C Naval Electrnic Systems Cmmand Attn: PME A, Dr. B. Kruger Department f the Navy Washingtn, D.C Naval Electrnic Systems Cmmand Attn: PME Department f the Navy Washingtn, D.C Navay Electrnic Systems Cmmand Attn: PME Department f the Navy Washingtn, D.C Naval Electrnic Systems Cmmand Attn: PME , Mr. Lery S. Wznak Department f the Navy Washingtn, D.C Naval Facilities Engineering Cmmand Attn: Mr. G. all (Cde 054B) Washingtn, D.C Naval Intelligence Supprt Center 4301 Suitland Rad Washingtn, D.C Naval Research Labratry Attn: Mr. R. LaFnde 4555 verlk Avenue, S.W. Washingtn, D.C

69 Naval Research Labratry Attn: Dr. Jhn M. Gdman, Cde verlk Avenue, S. W. Washingtn, D.C Naval Research Labratry Mr. Charles J. Stillings Cde 5437 Washingtn, D.C New Lndn Labratry (3 cpies) Naval Underwater Systems Center Attn: Mr. J. Merrill New Lndn, Cnn The Defense Dcumentatin Center Attn: DDC-TCA Camern Statin, Bldg. 5 Alexandria, VA Naval Telecmmunicatin Divisin Attn: Dr. N. McAllister, 0P941T Department f the Navy 4401 Mass Avenue Washingtn, D.C Naval Electrnic Systems Cmmand Attn: Mr. W. F. Larsn, PME Department f the Navy Washingtn, D.C Mr. Gerge L. Dwns, Manager (3 cpies) Strategic Systems, Electrnic Systems Grup GRE Sylvania 189 B Street Needham eights, Mass Mr. S. M. ergert, Attn: W31 Directr, Natinal Security Agency Ft. Gerge G. Meade, Maryland

70 UNCLASSIFIED SECURITY CLASSIFICATIN F TIS PAGE (When Data Entered) I. REPRT NUMBER ESD-TR REPRT DCUMENTATIN PAGE READ INSTRUCTINS BEFRE MPLETING FRM 2. GVT ACCESSIN N. 3. RECIPIENT'S CATALG NUMBER 4- TITLE (and Subtitle) 7. AUTRS Measurements f ELF Nise Prcessing Griffiths, Andrew S. 5. TYPE F REPRT & PERID VERED Technical Nte 6. PERFRMING RG. REPRT NUMBER Technical Nte NTRACT R GRANT NUMBERfW F C PERFRMING RGANIZATIN NAME AND ADDRESS Lincln Labratry, M.I.T. P.. Bx 73 Lexingtn, MA NTRLLING FFICE NAME AND ADDRESS Naval Electrnic Systems Cmmand Department f the Navy Washingtn, DC MNITRING AGENCY NAME 4 ADDRESS (if different frm Cntrlling ffice) Electrnic Systems Divisin anscm AFB Bedfrd, MA PRGRAM ELEMENT, PRJECT, TASK AREA & WRK UNIT NUMBERS Prgram Element N N Prject N. 2D REPRT DATE 2 September NUMBER F PAGES SECURITY CLASS, (f this reprt) Unclassified 15a. DECLASSIFICATIN DWNGRADING SCEDULE 16. DISTRIBUTIN STATEMENT (f this Reprt) Apprved fr public release; distributin unlimited. 17. DISTRIBUTIN STATEMENT (f the abstract entered in Blck 20, if different frm Reprt) IB. SUPPLEMENTARY NTES Nne 19. KEY WRDS (Cntinue n reverse side if necessary and identify by blck number) ELF nise Navy cmmunicatins '0. ABSTRACT (Cntinue n reverse side if necessary and identify by blck number) This reprt describes nn-linear prcessing experiments with ELF nise recrded in Nrway, Saipan, and Greece. Effective nise levels, an imprtant measure f perfrmance fr a nn-linear receiver, were cmputed using sine-wave signals in the z and z bands. Althugh results were nt ptimum due t fixed (as ppsed t adaptive) clip levels, ntch filter effects and lcal sine-wave interference, they d shw diurnal variatins and d prvide a significant increase in the available data in this frequency range. DD F0RM 1 JAN EDITIN F 1 NV 65 IS BSLETE UNCLASSIFIED SECURITY CLASSIFICATIN F TIS PAGE (When Data Entered)