JOURNAL OF RESEARCH of the National Bureau of Standards-D. Propagation Vol. 66D, No. 3, Ma y-June 1962 Observations of Radio Phase Characteristics on a High- Auroral Path

1. W. Koch and W. M. Beery

Contribution from Central Laboratory, National Bureau of Standards, Boulder, Colo.

(Received June 22, 1961; revised November 7, 1961)

Experimental observations of phase perturbations to continuous wave and pulse signals, over short time periods, have been carried out on a high-frequency a uroral path. Statistics on phase perturbations occurring in time intervals of one t o t went y milliseconds were obtained for t he continuous wave signals. P ulse-t o-pulse phase st ability measurements were made, using one-millisecond pulses with a pulse repetition rate of 250 pulses per second. The phase of corresponding parts of successive pulses were compared continuously, and then t he integrated values of phase differences during one-millisecond pulse periods were deter­ mined. Comparisons of statistics of phase perturbations for continuous wave signals a nd one-milliseco nd pulse signals ind ic at e no significant differences for approximately four­ millisecond sampli ng intervals and comparable speeds on t his auroral path. 1. Introduction fairly long auroral path. Fast "flutter " type fading has been observed for many years on radio paths A high-frequency radio ignal, propagated by the crossing the auroral zone; phase per turbations of over long distances, usually arrives at radio may be expected to be quite severe the receiver via several different propagation paths under such conditions. and with different time delays for the various paths. Interferen ce between the multipath signal com­ 2. Experimental Facilities ponents, along with movements of ionospheric layers Observations wero carried out over a path from and irregularities, results in phase and amplitude Barrow, Alaska to Boulder, Colorado during D e­ per turbations of the received signal. E ven when cember 1959, and the first half of 1960 at one component is much stronger than the others, of 9.9475 M c/s, 14.688 M c/s, and 19.247 M c/ . The rapid movement of the ionosphere may still cause great circle distance is 4,470 km. Figure 1 shows phase per turbations of the received signal as the phase path-length changes. - APPROXIMATE ZONE OF MAXIMUM AURORAL ACTIVITY Bramley [1951] and Voelcker [1960] have consid­ ered the theoretical phase per turbations to be expected over short time periods with R ayleigh III fading signals. Observations of changes in phase path lengths for high-frequency radio waves have been obtained by several workers, notably Findlay [1947], [1951 ], and Price and [1957]; however, most of the observing equipment was limited to time resolutions of the order of a second or greater. Lutz, Losee, and Ladd [1959] have suggested that the amount of multipath falling within a received signal element, and hence the phase per turbations, may be reduced by a technique of transmitting short pulses separated in time by intervals greater than the longest multipath delay times ; they sug­ gested I-msec pulses at a repetition rate of 250 pulses per second. A phase-keyed communication system, using this separated pulse technique, would then compare the phase continuously between corresponding parts of successive pulses, the inte­ grated value of the phase comparisons determining the signalling state. 140' 130' The purpose of the observations reported in this 12 0' 11 0' paper is to provide some statistical information on phase perturbations occurring within a few milli­ FIGURE 1. Geographic locations of and receiver seconds for continuous wave and pulse signals on a stations. 291 the geographic location of the path relative to the is much stronger than others, the pulse-to-pulse undistmbed amoral Zone. A discription of th e anel con tinuous wave phase stabili ties would be the transmitting and receiving equipment is given in same; likewise, if flll of the important modes con­ appendix A. The phase stability recording and tribute energy elming a received pulse period, the analysis equipment is discussed in appendix B. pulse-to-pulse phase stability is not likely to be much better than continuous wave phase stability. 3. Phase Stability Characteristics If pulses arc used which are short enough to isolate the vflrious multipath components, and then phase Regardless of the number of multipath components comparisons made for successive pulses of each present in samples of the received wave, if the of the most stable modes, an improvement over ionosphere is essen tially fixed in characteristics and continuous wave phase stability should result dm-ing in space over the sampling interval time, there will periods when fl number of multipath components be no phase pertmbation from one sample to the are present. next. However, movements of irregularities and of t he regular layers in the ionosphere, as well as 3.1. Theoretical Distribution of Phase Varitaions changes in along the path, will cause changes in phase path lengths of the various received In the case of Rayleigh distributed faeling of a components and in the interference pattern between continuous wave carrier envelope, and with the the components. Since the ionosphere is constantly assumption of a normal-law power spectrum of the in motion, and especially so in the amoral regions, fading carrier, one may derive theoretical probability it is logical to assume that phase pertmbations will distributions of carrier phase pertm-bations for increase with the number of multipath components various length sampling in tervals and fading speeds present in the samples. Therefore, one might con­ from the work of Bramley [1951] and Voelcker clude that the phase stability would be better for [1960]. The expression short pulses than for continuous wave signals, with phase sampling intervals equivalent to the pulse P (IL1

obtained by integrating the expression given by Bramley [1951] for the probability density function of phase variations, is plotted on figure' 2; R, the envelope of the normalized autocorrelation function

100 ~-~-~--~-~----;,-----.---~ of the fading carrier, is given by the expression,

I- L>- o 3.2. Observed Phase Perturbations of CW Carriers W , (fJ ' ...... (fJ 40P ",,- u (fJ en 1\\'\ -""""' 0 C '"w "'~ ~ n !;;: u 2.0 x en \ 0 ° ~ w a: • "-0 "0,,- :::l z \ -,'" "'. 1'---0 1"-0 c- 1.0 , :!;.s", '- ...... 0 a: ", '" 0.8 s·c w I " , § \ 0 ,,~ Q. " ~ ,,~ ~S w " '" (j) ~ 0.4 ", r-.... w'" -...... = w 0.1 " "- ,. 0.08 lL ;: 0 ~ 1\ W '" 0.04 "ION CU RVE IS FOR A 200 SECOND SA MPL E 0.00 4

0.002 0.001 ;----0 1;;;:0--::40:----;60::----:8:::-0 ---;1;;;00--:';11:-0 -~1 4 0

DEGREES PHASE PER T URBATION 0.00 I o 20 40 60 BO 100 120 140 DEGRE ES PHASE PERT URBATI ON F I G URE 4. T ypical phase stabiltty oj 19.247 A1c/s CW stgnals betwee n sttccess'ive sam pling lJoints at 4,5 msec i ntervals FlOURIJ 3. Distribt,tions oj CW cania phase pertUl' bation fo r various fading m tes . between various length sampling intervals. Sempling period 200 sec IF 800 cis (-G db) 3.3. Observed Pulse-to-Pulse Phase Perturbations Media ll carric r·to·noise ratio 33 db A vcragc fading rate 7 cis 19.247 ;\le/s, April 25, 19GO; 1<156 M S'r Observations of pulse-Lo-pul e phase sLabiliLy were carried ouL from June 20, 1960 through July 2; 1960. perturbations for different Jading rates, all at a [nterference from other sLations was especially sampling interval length of 4.5 msec. Phase sta­ troublesome during this period, a nd much of t he bility was, in general, a function oJ the fading rate, data h ad to be discarded fo1' that )'ea O il . Reliable as migh t be expected. However, for 1 mscc sampling detection of rapid propagation-ind uced phase pertur­ intervals, figure 5, th ere app ears to b e very little b ations of a fading signal makes necessary n, l'ela­ change of phase perturbations as the observed fading Li vely lo w level of interfer ence and noise. The daLa rate increases above 2 c/s. The Rayleigh-fading presented h ere were obtained during late afternoon model discussed in 3.1 compares closely with the and early evening bours wh en interference was not exper imen tal distributions only in a limited number present at the 14.688 a nd 19.247 Mc/s frequencies. of cases and does not explain the observations for No reliable data were obtained aL t he 9.9475 M c/s 1 msec sampling intervals. The statistical model of frequency. a relatively constant vector plus a R ayleigh distrib­ uted vector for fast fading conditions suggested as a Q. Oscillograms of Pulses possible explanation for reduced depth of fading at these times [Koch and P etrie, 1961] might also b e Some o scill og~ ' am s of typi cal pulse waveforms, at used here to explain the relatively small change of the .10 k c/s reCeiver outputs, ar e shown on fi gw'e 6. phase perturbations for very short sampling in ter­ Oscillograms (a) through (g) were obtained at the > vals as tbe fading r ate increases. For the same 19.247 M c/s frequency from 1 48 Lo 1950 hours NI ST fading rate there appears to b e no significant differ­ on June 28, 1960. The rapidly cha nging muitipath ence in the distribution of phase p er turbation at th e conditions are evident fo r th is period; at Limes there various frequencies observed. It should be noLed, is little evidence of multipaL h components, a nd at however, that th e probability of occurrenee of other time multipath co mponenLs are obvious specific Jadin$ rates is dependent upon the observed withill a nd following the main pulses. Since each frequency [h .och and P etrie, 1961] and, therefore, of these oscillograms represent a single sweep of the the expected phase perturbations do exhibit some oscilloscope, th e ch angin g conditions from one pulse frequency dependenee. to the n ext may be observed. It would appear that, 293 100.0 Bo.O

40.0 ~ ~ \~~ I 20.0 (a) 19.247 MCIS, 1848 MST, JUNE 18,1960; (e) 19.247 MC/S,1937 MST, JUNE 18,1960; l'\, I 10.0 \\ PU LSE FAD ING RATE -I CIS PULSE FAD ING RA TE-9 CIS B.O

4.0 ~ i 1 o~ .'. . t . f ~ \~r'~ g 2.0 ~~ "";'~I~t,, ~ ' ~1rf""1 : "'''l,il''''i''''''~ , : . \ ~\ .' . 'j . '" ~ (b) 19.147 MCIS, 1853MST, JUNE 18, 1960; (f) 19.147 MeIS, 1950MST, JUNE 18,1960; ~ t\ PULS E FAD ING RATE - I CIS PUL SE FAD ING RATE - 10 CIS ~ ~ " I ~ ~ . ~ 0.2 \ ~ < w 0.1 ,. O.O B \ ;::

~ o 0.04 \ ~\

~ '"« \ \~\ (e) 19 .147 MCIS , 1909 MST, JUNE 18,1960; (g) 19.247 MClS, 1950 MST, JUNE 28, 1960; ! 0.02 PULSE FADING RATE - 2 CIS PULSE FADING RATE -10 CIS 0.01 0.008

0.00 4 0.002 ~I 0.00 I o 10 20 30 40 50 60 70 (d) 19.247 MC/S, 1923 MST , JUNE28, 1960; (h) 14.688 MC/S, 2048 MST , JU NE 27 , 1960; DEGREES PHASE PERTURBATION PULSE FAD ING RATE - I CIS PULSE FAD ING RATE - 3 CIS ) FIGURE 5. Phase p e1" tU1'baHons between 1 msec sampling points f or various average fading rates (CW carrier) . NOTE: ALL OSC ILLOGRAMS EXCEPT (g) HAVE SWEEP- TIME Sampling period 200 sec OF ONE MILLISECOND PER MAJOR SCALE DIVISION; If bandwidth 800 cIs (-6 db) Median carricr-to-l1oise ratio 22 to 37 db OSCILLOGR AM ( g) HAS SWEEP TIME OF TEN MILLISECONDS 14.688 Mcls, December 1959; midday PER MAJOR SCALE DIVI SION . at times, multipath effects are quite variable for FIGURE 6. S ingle-sweep oscillograms of recewed 1 msec pulses. Note: All oscillograms except (g) h ave sweep time of 1 mscc per major scale successive pulses, and hence pulse-to-pulse phase division; pertlU'bations would be quite significant part of Oscillogram (g) has swceptlme of 10 illsec per major scalc division. the time. observations indicated that at certain times when a b . Distribution of Pulse-to-Pulse Phase Perturbations number of discrete multipath components were Distributions of phase perturbations of the re­ observable on the ionograms (all within one milli­ ceived pulse signals were obtained for data samples second delay), the phase perturbations were less of 200 sec in length. Sample results of the 19.247 than at other times when the received sweep-fre­ Mc/s observations of pulse-to-pulse phase stability for quency pulses were simply smeared out to 200 !-,sec good signal-to-noise ratios are given in figure 7 for in width. Thus it would appear that the most signifi­ ~ several different pulse fading rates. The median cant short interval phase perturbations on this path I pulse signal-to-noise ratios were 16 to 24 db for these are a function of very rapid moving ionospheric observations. The results at 14.688 Mc/s showed irregularities. This observation also indicates that no significant difference for comparable fading rates. the occurrence of many multipath components does Sweep-frequency oblique-incidence ionograms, ob­ not itself produce phase perturbation. tained by Tveten [1960] on the Barrow-Boulder path near the times when pulse phase stability observations 3.4. Comparison of Continuous Wave and Pulse­ were made indicated that at all times multipath to-Pulse Phase Stability components arrived with a delay of about 100 !-,sec with respect to the first received component, and From the discussion in the previous section, one only occasionally did significant multipath com­ surmises that most of the multipath components ponents arrive with a delay that exceeded 1 msec at causing large phase perturbations over short time either of t he two frequencies used for phase observa­ intervals were well within a delay of 1 msec with tions. Transmitted pulses of the sweep-frequency respect to the first arriving component. It might, equipment were 100 !-,sec in width. Careful examina­ therefore, be expected that the pulse-to-pulse sta­ tion of the sweep-frequency records and the phase bility with 1-msec pulses, and continuous wave 294 Although the use of separated pulses for phase­ I00 ~:::Z;:--'---;::,;;:c""", .""0 ~5M"'S"' T.""J"'UL'='y 0:=,• ""'6~0"" ; P"'U;=;LS 7. "",A"'o,'7"NG::=;R""ATo;.=_"O<"',"c51/, .~~ (2) 1953 M S T, J UN E 2 B,1960PULSE FADI NG RAT E-leis keying applications would appeal' to have some (3) 191 4 MS T, J UNE 28,t960 PuLSE FADING RA TE- 3c/ s merit as compared with continuous waye phase­ \ (4 ) 1950 MS T, JU NE 28,I 960PULSE f ADIN G RATE - IDe/s ~o a {51195 4 M ST JUNE 2 8, l96 0 PULSE FAD ING RAT E - lOc/s keyed systems, there does not seem to be much • advantage to the technique, as presently eyolved, o '"• for an aurorally-disturbed propagation medium . The rapidly changing ionospheric characteristics cause seyere phase perturbations for a small pel'­ cen tage of the tim e on the path inyestigation for both separ ated pulses and continuous wave signals. Phase-keying systems r cquiring 4-msec phase sta­ bility do not appeal' to be practical on such a path. The use of pulses much shorter than a millisecond to separate the mod es, togethcr with phase informa­ o~ tion from only the most stablc modes, may provid e o an improvement, although this has not been in­ O.I I:------\--~---~:---A_--\---l vestigated. "'. Appendix A . Transmitting and Receiving ~. Equipment 0.0 1 b------:o'r------'::",."".=----Lj The transmitting station was at the camp of the Artic R esearch Labor aLory, Barrow, Alaska. Thrce o (3) kilowatt transmittcrs wcrc used with prccise EACH DI ST RI BUT ION CU RV E IS \ FDA A 200 SE CO ND SAM PLE carrier frequ ency con trol and frequcncy stability to o bcttcr than one par t in 108 per day. A quar tcl'­

0.001 '-----'2'-0 ----'4'-0 -----'6-0 ----'80--....1.- ----'-12-0 ---'140 wavc ycr tical an tcnna and r adio wire grounclplane 0 10 0 were used with each transmitter. DEGR EES PHASE PERTURBAT ION F or the con tinuous wave observations, the tr ans­ mitters wcre operated without . For the F I GU RE 7. T ypical pulse-to-pulse phase stability at 19.247 NIcis for various f ading Tates. pulsc-to-pulsc phase obsel'Ya tio;n s, the r adio fre­ quency drivc for the transmi tters was gated (( on" and (( off" with a prccisc pulse gcnera tor. The" on" phase stability for sampling intervals equivalen t to periods wcr e 1 mscc in lcngth, with a pulse r epetition tbe pulse repeLition interval, would not be appreci­ frequency of 250 cis, pl'oyiding an (( off " timc of 3 ably different. Comparison of fi gure 7 with fi gure msec bctwccn thc end of one pulse and the start of 4 does, in fact, indicate that for approxima tely 4- the next pulse. msec sampling in tervals and comparable fading The receiying station was locatcd at the NBS speeds, pulse-Lo-pulse and continuous waye phase T able M esa field site n eal' Boulder, Colo. R eceiying perturba tions arc not significan tly difJ'erent ,on this equipment consisted of high quality communication path . typc receiYers wi th external ll ctcrodyn e oscillators. 4 . Conclusions The oscillator stability was within one par t in 108 pCI' day. Automatic gain control (AGC) of the Tbe statistics derived from the phase stability receivers was deactiyatcd chuing phase stability obseryation show an expected correlation with f::tdin g observation periods. The r eceiyer carrier frequen­ rate for the 4.5 msec sampling interval, but are cies wer e translated down to 2.5 kc/s and 10 k c/s for relatively un correlated with fading rate changes continuous wave and pulse-to-pulse observations, nboye 2 cis for the 1 msec sampling interval on this respectiyely. Receiving antennas consisted of half­ auroral path. The observed phase perturbations horizontal dipoles at a heigh t of one are not adequa tely described by the R ayleigh­ wavelength above the ground. fading model a large part of the time. The proba­ bility of occurrence of large phase perturbation are Appendix B. Phase Stability Recording and less for the experimental distributions that predicted Analysis Equipment from the R ayleigh-fading model. The experimental distributions approach the distribution that would For the continuous wave phase stability obserya­ be ob tained if the received signal consisted of the tions the receiver 2.5 k c/s output was connected to a sum of a specular componen t and randomly varying phase detector for comparison with a precise £re­ componen ts. The theoretical distribution derived quency standard. The output voltage of t he phase from the Rayleigh fading model can be used to detector was recorded on frequency modulated bound the actual phase pcrturbation distributions, channel of a tape recorder. Phase per tw'bations that is, the probability of occurrence of a spccific between various length sampling intervals, ranging phase per turbation should not exceed the value from 1 to 20 msec, were analyzed from the I' produced given by the R aylcigh-fading model. phase detector recordings. At the beginning of each 295 sampling interval the voltage output of the p hase The experimental study of phase per turbations, detector was clamped to zero r eference; at the end of reported herein, was sponsored by the U . S. Air j the sampling in ter val the change of the phase de­ F orce Ii'\Trigh t Ail' D evelopm ent Division, Wright­ tector output voltage (representing the phase pcr­ Patterson Air F orce B ase, Ohio . T he assistance of "1 tmb ation between the beginning and end of t he the Arctic R esearch L aboratory , Office of Naval interval) activated gate circuits which h ad thrcshold R esearch , P oint B arrow, Alaska, is gratefully levels equal to or below th is voltage, allowing ftCknowledged for providing facilities and housing for associated decade coun tel'S to register one count. t ransmitter operation . Special t hanks are due T he percentage of time that the phase per tmbation L . H. T veten of NBS for providing sweep-frequency exceeded the phase angle associated with each ionograms, and M . N esenbergs of NBS fo r integrat­ threshold level is therefore r egistered on the r espec­ ing the theoretical probability density function of tive decade counter. Calibration of the phase detec­ phase variations. Othe NBS personnel contributing tor was performed before each r ecording, using an substantially to the project include W . B . H arding, external oscillator and phase shifter . G. E . \i'\Tasson, G. G. Ax, and J. L . IVorkrnan F or the pulse-to-pulse phase obser vations t he (B arrow, Alaska). received signals, translated to 10 kc/s, were recorded on m agnetic tape. The signals from the m agnetic 5. References tape reproducer were fed to t he inputs of two ch annels, one of the two channels incorporating a Bramley, E. N., Diversity effects in spaced-aeri al reception of 4-msec m agnetostrictive delay line. The pulse ionospheric waves, P roc. I nst. Elec. Engrs. Part III 98, m odulated 10 kc/s ou tputs of the two channels were No. 51, 19- 25 (J an. 1951). F indlay, J . 'vV., Measurements of changes of t he phase-paths of compared in a phase detector. Continuous relative radio waves in the ionosphere, N ature 159, No. 4028, 58- 59 phase information of the carrier for two consecutive (J an. 11 , 1947). pulses, over t he 1 msec pulse periods, was provided Fi ndlay, J . W ., T he p hase and group paths of radio waves b y the phase detector. T he instan taneous phase ret urned from region E of t he ionosphere, J ATP 1 No. 5/6, information was integrated for 1 msec periods, and 353- 366 (195 1). K och, J. 'vV., a nd H . E. P etrie, Fading characteri stics obser ved t he resultan t integrated volt age was propor tional to on a high-frequency auroral radi o path (to be published, t he integrated phase differences for the t wo consecu­ 1961 ) . tive pulses. The integrated output of the phase Lutz, S. G., F . A. Losee, and A. ·W. Ladd, Pulse phase-change d etector was sampled at the end of each 1 m sec signalling in the p resence of ionosp heri c multi path distor­ t ion, IRE T rans. on Comm. Systems CS- 7, No.2, 102- 110 integrating period by amplitude discriminators for a (June 1959). set of decade counters. T hus, the distribution of P rice, R ., a nd P . E . Green, Jr., M easurement of ionosphcri c ) phase per turbations between consecutive pulses was path-phase fo r oblique incidence, Natu re 179, No. 455, "1 obtained from the counter indications for an observ­ 372- 373 (F eb. 16, 1957). T veten, L. H., p rivate co mmunication (1960). ing period. Calibration of the phase detector and Voelcker, H. B., Phase-shift keying in fading channels, Proc. associated circuitry was accomplished b efore each I nst. Elec. Engrs. 107 B, No. 31, 31- 38 (J an. 1960). analysis period with the aid of an auxiliary oscillator, adjusted precisely to the recorded carrier frequency, and a calibrated phase sllifter in one of the channels. (P aper 66 D3- 197)