US2293501A - Method of and means for reducing the effects of multipath phenomenon - Google Patents

Description

Aug. 18, 1942. y c. w. HANSELL METHOD .OF AND MEANS FOR REDUCING THE EFFECTS OF MULTIPTH PHENOME NON Filed March l5, 1940 TOR. W HANSELL ATTORNEY.

UNITED STATES Patente-d Aug. 18, 1942 Clarence W. Hansell, Port Jefferson, N. Y., assignor to Radio Corporation of America, a corporation of Delaware Application March 15, 1940, Serial No. 324,051

PATENT OFFICE Claims.

This invention relates to a method of and means for reducing the effects of the multipath phenomenon upon radio signals.

It is known in the radio art that short wave radio transmission is accompanied by the ardifferent directions.

transmitter.

currents.

the receiver.

Zero modulation frequency.

(Cl. Z50-8) vention, the power received over different space circuit paths will change frequency at different times and in a succession which is in inverse order with respect to the effective path lengths. During the transition period at the receiver,

This peak phase 5 rival of signal power at the receiver over more power of both transmitter frequencies will be than one path. Several explanations have been present simultaneously and give rise to two re- `given for these multiple paths, among which are: sultant currents at the two frequencies. When First, that the waves which reach the receiver these two currents are passed through the limiter have been refracted and reflected between the they result in an output current from the limiter ionosphere and the earth a number of times, and which is at the frequency of the stronger input second, that there exists refraction from different current component and this output current is ionized layers or regions in the ionosphere. phase modulated by the weaker current, at fre- There have also been many instances in long disquencies corresponding to the difference between tance communication of the transmission of the the frequencies of the input currents and mulsame signal simultaneously around the world in tiples of the difference frequency, and by a peak A common result of the angle whose sine is the ratio of the weaker curmultipath phenomenon between transmitter and rent to the stronger current. receiver is that the useful signal modulations ardeviation is always less than plus and minus 90. riving at the receiver are badly distorted and do 2o The unbalance between the two components of not faithfully reproduce the modulations of the current may be expected to change abruptly with arrival of the frequency change over some The present invention has for its primary cbone of the multiple space paths. Therefore, genject to reduce the effects of the multipath pheerally, the carrier current in the limiter output nomenon which produces distortion at the rewill be keyed from one frequency to another alceiver, and this I accomplish by employing what most as if the signal had traveled over only the is known as spacing wave or frequency shift keyone path, the strongest ray of the multipath preing, which requires the transmission of two reladominating in producing the useful modulation tively close frequencies, one frequency being output. The presence of phase modulation of known as the marking wave while the other frelimiter output current during the transition pequency is known as the spacing wave, and reriod will not be of much consequence if phase ceiving the radiated waves with a frequency variation due to the useful signal adds up with modulation type of receiver designed to respond time to a much higher value than the peak plus to modulation frequencies down to the lowest reand minus phase shifts due to the modulation. quired to reproduce the signals, which may re- From another point of view, the phase variations quire response down to zero frequency, that is, are, of course, the cause of, or a result of, and down to direct current or steady state output equivalent to frequency variations. The phase modulation causes a small equivalent frequency According to the preferred embodiment of the modulation of the limiter output current but the invention, I employ a spacing wave or frequency useful signal produces a relatively large freshift keyed transmitter in which the frequency quency shift which lasts for a much longer time shift between marking and spacing frequencies than one cycle of the frequency modulation. is greater than the modulation band width re- Also, the undesired detected phase modulation quired to nearly enough reproduce the signal at beats between the two frequency components For receiving the signal from the may be eliminated by means of a low pass filter spacing wave keyed transmitter, I employ a freif all component frequencies of the beats lie quency modulation receiver wherein the keyed above the frequencies of the useful modulation. wave is received through an amplitude limiter Consequently, for least distortion, the frequency followed by a frequency modulation detector shift employed for spacing wave keying should which is broad enough to utilize all significant be somewhat greater than the modulation band frequencies and designed to respond down to required by the signal and greater than the highest required modulation frequency in order If, now at the transmitter of the invention, the that the beats may be eliminated. carrier frequency is changed suddenly from one The foregoing statements regarding Suppres- 5.5

value to another then, at the receiver of the insion of signal elongations may not be strictly true in all cases because, although a frequency change arriving over one path may swing the balance of current, arrival of a frequency change over another longer path may sometimes swing the balance back again momentarily. Consequently, when there are a considerable number of paths carrying currents of about the same order of magnitude we cannot be sure that any one path current will completely predominate. In such cases, multipath effects maybe greatly reduced but will not be completely eliminated. In order to obtain the best results and to reduce the possibility of frequently changing relative strengths of currents received over several paths which may cause first one and then another path current to predominate, I propose to add to the effectiveness of the present invention by employing means for, reducing the number of multiple paths, by employing sharply directive antennae arranged to transmit and receive at low angles. Tests have shown that short radio Waves transmitted over great distances show increasing time delay with increasing angles to the horizon and that the increase in time delay with increase in angle is roughly proportional to the square of the angle above horizontal. I also prefer to reduce the number of probable multiple paths by employing the highest radio frequencies which are capable of being used reliably for communication between transmitter and receiver at any partics ular time. Experience With long distance short Wave circuits has shown that there is always an upper limit to the frequencies which can reach the receiver with useful strength, at any particular time, and that the number of signicant transmission paths decreases successively down to zero as the frequency is increased to and beyond the limit.

An important' advantage of the present invention lies in the fact that it makes possible a reduction in the width of frequency bands occupied by transmitters as now designed and operated throughout the world. In radio telegraph, printer and facsimile transmission it has been an almost universal practice to key the transmitter power on and off in accordance with the signals. A typical on-off keying system is described in my United States Patent No. 1,808,226. This practice Was originally adopted partly because it was the only method applicable to early long wave,

mechanically driven alternators and partly because it was believed to provide a maximum message speed for a given frequency band.

On-off keying of Vacuum tube radio telegraph transmitters has never been entirely satisfactory because of signal wave form distortions, because of extreme electrical stresses in tubes at the beginning of each marking period and because of difficulties from interference between transmitters. Interference of the key click type has been experienced too frequently. At the beginning of each marking period the tubes Work into short circuited output impedances which build up to a normal steady state value as circulating currents are established in the high frequency circuits. During this period momentary parasitic oscillations may exist which give rise to interference. Avoidance of these parasitics requires a care and thoroughness in transmitter design, initial adjustment and operation which long experience has demonstrated cannot be relied upon in commercial service. It is also almost inevitable that vacuum tube amplier power supply and biasing potentials change when power is thrown on to transmit a signal. It takes time for the potentials to settle down to steady state values and during this time potential variations cause undesired variations in the amplitude, phase or frequency of the signaling current. When the radio frequency excitation to the usual transmitter is interrupted by the keying, at the end of a marking period, power remains in the circuits and dies out without benefit of the frequency determining excitation current. The time required for the circuits to stop oscillating may be greatly increased if regeneration is present. This may result in click interference to reception of other transmitters unless the circuits have been carefully adjusted and maintained to hold correct frequency Without benefit of the frequency determining excitation.

Gradual application and removal of excitation to remove key click interference is not very effective because of limiting used in successive stages. It may be made more effective by keying as near as possible to the final output stage but at the expense of possible difficulty due to idle wave radiation. Very little idle wave radiation can be tolerated because of limiting frequency used in the receivers. The only nearly perfect technical solution to thet on-o keying problem is believed to be that which was provided by absorber keying in early British short wave transmitters. This solution has not been generally accepted in the art because it required substantially double the tube complement and twice the input power consumption needed for more generally adopted types of keying arrangement.

The present invention recognizes that We may reduce the keying problem if we employ spacing wave or frequency shift keying instead of on-oif keying. Then there will be no transients in the power circuits, tubes will not be subjected to instantaneous overloading, the output will never be free from frequency guidance and by means of one transmitter we may obtain the equivalent of two transmitters handling the same message.

By means of the invention the bands 0f side frequencies of the two keyed transmitted Waves may overlap into the same frequency bands and, as an ultimate limit, the two signaling wave bands of frequencies may occupy the same band. Therefore, as an ultimate limit the frequency modulation receiver of the invention permits spacing wave keyed signals to occupy no more than the same total frequency band that is required for ori-off keyed signals on a single wave telegraph transmitter.

With the use of my special frequency modulation type of telegraph receiver, I can obtain substantially twice the signaling speed, for any given band with, and greater signal to noise ratios than with the two receiver or equivalent combination employed in known types of spacing wave keying systems. By means of the present invention, it is possible to carry out frequency shift keying in the same total frequency band as is required for on-off keying without sacrificing signal to noise ratio, if we apply band width determining circuits in the output connections of transmitters and then employ relatively large frequency shifts within the output circuit pass band. Then each signaling wave will be transmitted by a form of asymmetric side band transmission and will be capable of `a rate of response corresponding to substantially the whole pass band.

I have determined, both by theory and experiment, that, when two points are joined by carrier current circuits of any chosen frequency band width, then by means of the present invention, signals may be transmitted by frequency shift keying of a carrier current `at substantially the same speed and quality as can be obtained when the signals are transmitted by keying the carrier power on and off. Thus, I have shown that, fundamentally, frequency shift keying, or frequency modulation requires no greater frequency band than on-off keying, or amplitude modulation.

By reducing multipath distortions and the effects of noise, the frequency shift keying system of the present invention will provide increased maximum traffic speeds and improved reliability.

The following is a more detailed description of means for practicing the invention accompanied by a drawing, wherein Fig. l illustrates schematically a transmitter which can be used with the present invention, Fig. 2 illustrates a special frequency modulation type of receiver which can be used with the transmitter of Fig. l to achieve the results of the invention, and Fig. 3 illustrates the response characteristics of the frequency modulation detector in the receiver.

Referring to Fig. l, there is shown a transmitting system adapted for use with frequency shift keying. This system includes an amplifier and rectifier I which provides a rectangular wave direct current output and is controlled by a suitable keyed audio tone arriving over line 2. Putting it in other words, apparatus I provides a direct current output which is turned on and off or which changes from one fixed value to another depending upon whether or not there exists carrier current energy in line 2. The output from rectifier l controls the frequency shift transmitter 3 which functions to send out a marking frequency and a spacing frequency over directive antenna 4. Between the transmitter and the antenna there may be inserted a narrow band pass filter to limit the width of the band of frequencies transmitted so that interference to reception of other transmitters may be minimized. This lter may be omitted without adverse effects upon the signaling in those cases where interference between transmitters is not a problem, which will usually be the case at the present time.

As an illustration of the frequencies to be transmitted, the carrier marking frequency may be, say, 20,000,000 cycles per second `and the carrier spacing frequency may be 19,999,400 cycles per second, thus providing a frequency shift of 600 cycles, while the highest required modulating frequency for these marking and spacing frequencies may be, say, 500 cycles per second at a telegraph keying speed corresponding to 100 dots per second. If desired, the transmitter may have a greater band width than that indicated but a telegraph speed up to 100 dots per second would be typical of present day speeds on long distance short wave transmitters and 600 cycles frequency shift would be near the minimum practical value.

superheterodyne receiver 6 which ampliies the 75 radio signals and beats the incoming signal with the output of a suitable local heterodyne oscill-ator I4 in a detector to produce energy of intermediate frequency which is then passed on to the frequency selective amplitude limiter 'I. From the amplitude limiter I the limited signal energy of intermediate frequency is passed through back-to-back frequency modulation detectors 8 and 9 (sometimes called discriminator or off-tuning circuits) which are broad enough in operating frequency range to include all of the significant transmitted frequencies, and designed to respond down to Zero modulation frequency. These back-to-.back detectors include a pair of off-tuned circuits and rectifiers associated therewith, the purpose thereof being to change the frequency modulation to amplitude modulation which is a substantially true reproduction of the original frequency change keying, or frequency modulation. The off-tuned resonance curves for detectors B and 9 are illustrated in Fig. 3 of the drawing. Curve A represents the change of output current of detector 8 of Fig. 2, when the transmitter frequency is changed, and curve B represents the change of output of detector 9, when the discriminator circuits of the detectors are tuned respectively to frequencies F1 and F2, below and above the band of useful signaling frequencies delivered to the frequency modulation detectors. The two transmitter signaling frequencies, `after heterodyning, may be as represented by frequencies C1 and C2 in Fig. 3. The outputs from detectors 8 and 9 are combined with opposing polarities and passed through a low pass filter system lil, the latter having a uniform response band less than the frequency changes at the transmitter and serving to eliminate undesired phase modulation beats which lie above the frequencies of the useful mcdulation. The output from the low pass filter system i0 may then be utilized in any desired manner such as by controlling a tone keyer I I to transmit suitable signals over a line I2 to a central station where the signals can be recorded.

Another :arrangement of back-to-back detectors, given .by way of example, which is known in the art and can be used is described in Usselman Patent No. 1,794,932, dated March 31, 1931, to which reference is made.

The transmitting and receiving antennae 4 and 5, respectively, are preferably of the sharply directive type and designed to function at low angles; further, the highest practical radio frequencies are preferably employed, al1 to reduce r the number of multiple paths.

In the operation of the system of the invention there will be two :predominant resultant currents at the receiver during the transition period, one at the marking frequency and one at the spacing frequency. One or the other of these frequency components will determine the center frequency in the limiter output, depending upon which is stronger. The weaker component can only temporarily phase modulate the stronger component `current and these modulations, since they are at frequencies higher than the frequencies required to reproduce the useful signal satisfactorily, may be eliminated =by the low pass filter in the output `circuits of the receiver. The balance between the two component currents may be expected to change suddenly with signal arrival over some one of the multipaths, so that nearly always it will appear that only one path is carrying the signal, and distortion is therefore reduced toward the condition which exists for single path transmission, and signal timing or length variations will be greatly reduced. Qualifications on these conditions of operation have previously been discussed.

It will be noted that my system provides greatly improved signals in telegraph, printer and similar .communications circuits whenever multipath distortions are present. I obtain this improvement by substituting frequency shift keying for power on-off keying; by employing frequency shift greater than the frequency range required by signals of acceptable qualit by employing an amplitude limiting frequency modulation type of receiver ldesigned to respond down to the lowest frequencies required by the signaling; and by limiting the receiver output frequency band to a width less than the transmitter frequency shift produced by the signals.

Although I have not shown the means for doing it in Fig. 2, I prefer to employ a receiver equipped for automatic tuning, This may be done by any one of a number of means known in the art but superior results will be obtained if there is used the automatic tunin-g system which I have devised especially for use with frequency shift keying, `and which is described in my copending application Serial No. 395,556 filed May 28, 1941.

Also, improved results will be obtained if two or more receivers are employed, with outp-uts combined, if these receivers are connected with antennas spaced apart or with antennas designed to receive Waves of different polarization. A preferred form of frequency modulation diversity receiving system is one in which the outputs from the several receivers into their combined output circuit is more or less proportional to their respective inputs. In general, receivers with strongest input will have best signal to noise ratio in their outputs and consequently should be made to provide a larger portion of the useful output in order to provide a good signal to noise ratio. A preferred form of frequency modulation diversity receiver system is described in my copending United States application Serial No. 326,129, filed March 27, 1940.

The principles of the present invention and transmitters and receivers similar to those shown in Figs. 1 and 2, respectively, except for electrical constants and adjustments, can be used to reduce multipath effects in television or facsimile transmission. by using frequency modulation. Since no secondary path component of received current, after amplitude limiting at the receiver, can produce a steady state change in frequency of a larger current received over a main path, therefore in television by means of frequency modulation, no object in a transmitted image which requires steady state conditions to be reached for reproducing it, can appear as a secondary image because the secondary path currents can not cause steady state shifts in frequency at the receiver corresponding to the steady state amplitude shifts which they can produce. To illustrate the principles, assume that in a frequency modulated television system we have secondary path waves arriving at the receiver five microseconds later than primary path waves and with an amplitude of Assume also that we have a fairly large dark object on a grey background in a transmitted image and that it takes twenty microseconds for the scanning spot to pass over the object. Transmission of the image of the object may be assumed to cause a rectangular wave form of transmitter frequency change from 20% to 80% and back again. Let 100% frequency change be two megacycles. During the first five microseconds after arrival of the frequency change we will have the direct and main path currents arriving with frequencies differing by 1.2 megacycles and beating together to give a peak phase variation in the output of the receiver amplitude limiter of plus and minus arctangent 0.5=26.5 degrees or 0.46 radian. This corresponds to a peak frequency deviation of 0.55 megacycle at the rate of 1.2 megacycles. For the next fifteen microseconds the modulation is absent and we have simply the new frequency. Then, for another five microseconds, we are back to the old frequency again but we have again the 0.55 megacycle frequency modulation at the rate of 1.2 megacycles, for the second ve microseconds, then the old frequency Without modulation. It will be noted that the center or midfrequency input to the detector follows the transmitter frequency modulation very closely and that the secondary path current can only produce a rapid modulation above and below the correct value.

The distortion in the case of frequency modulation in a television system would not be observed at all if the modulation frequency could be made too high to pass through the video frequency circuits, or if the modulation were at a greater rate than the rate of passing over the image elements. This at once suggests a means for substantially eliminating effect of multipath distortion in frequency modulation television.

If we should transmit television (likewise facsimile in transoceanic circuits) by the constant frequency variable length dot method, by switching the transmitter frequency back and forth between two fixed values, and by varying the percentage time spent at each frequency in accordance with light in the image transmitted, then the beat between currents received over different paths might always be too high to affect the apparent quality of the reproduced image.

For 441 line television with an aspect ratio of 4/3 and a frame frequency of 30 per second, the

- scanning spot will pass over the picture at a rate of 7,780,000 or say 8,000,000 picture elements per second. In a CFVD (constant frequency-variable dot) frequency modulated television system the minimum separation between the two spacing wave frequencies to average out the effect of multipath beating within the time of one picture element would then be 8,000,000 cycles. With this or greater frequency spacing the beating effect shown in the frequency modulated signals would be invisible at normal viewing distances and substantially no effect of multipath reception should be apparent.

If we assume that the CFVD dot rate is 8,000,000 dots per second and that sufficient fidelity of reproduction will be obtained by transmitting up to third harmonics of the dot frequency, then the total radio band width required will be 48 megacycles. Detail corresponding to the fifth harmonic would require megacycles and so on.

Even if the CFVD method is not employed, frequency shift television and facsimile modulation should make possible the elimination of multiple images and reduce the effect of multipath to short transient distortions or beats which are less serious than the distortions and multiple images encountered in using amplitude modulation.

Large percentage light variations in the images would produce modulations of kinescope currents in the presence of multipath currents in the receiver at a high rate but these would not be very objectionable because of their fine structure and because of lack of distortion of average kinescope current. Small percentage light variations would produce a much greater percentage distortion in kinescope currents but these distortions are not very apparent to the eye, which responds only to percentage variations. That is, light modulations which can be badly distorted due to multipath effects are also weak and relatively unimportant.

The invention also finds application in facsimile systems, particularly over long distance short wave radio circuits where multipath distortions have now been a serious diiculty, and have limited operating speeds, for sixteen years.

Assuming, for example, that tape facsimile transmission at 30 revolutions per second requires signal response at frequencies up to 900 cycles per second, we might employ spacing wave frequency change of say 1200 cycles. Then, when, due to multipath, currents at both frequencies are arriving at the receiver simultaneously, they will beat together to give phase modulation at 1200 cycles, above the signaling frequency band, and their effect may be eliminated after detection. 'Ihe apparent carrier or center frequency, after the limiter in the frequency modulation receiver, will be at one of the two frequencies which, at any particular time, has the greatest vector sum in the input to the limiter. The balance between amplitudes of currents at the two frequencies may be expected to swing more abruptly, to produce a change in center frequency, than would be the change in amplitude of an on-off keyed signal under like multipath conditions. The signal elongation and smearing will therefore be reduced.

In the case of a large number of multiple paths, a single signal pulse with relatively long spaces between pulses and which is shorter than about half the delay time between shortest and longest path might sometimes not register at all in the output of the receiver. This might be caused by the fact that the early current components to change frequency at the receiver would be changing back again before the later current components had changed. Then they balance between currents at the two frequencies at the receiver might not be thrown over by the signal.

The result would tend to be a signal shortening by about half the delay time. The shortest and longest important paths between the transmitter and the receiver are those particular paths in which the signals are strong enough compared to other multipath components to have a substantial effect on the reproduction of signals. 'I'he remedy for the above trouble would be to provide an adjustable signal elongation at the transmitter to compensate for shortening at the receiver due to the use of the frequency change modulation system in the presence of bad multipath conditions, by means of which the marking pulses are lengthened relative to the spaces.

Means for providing an adjustable elongation of rectangular wave keying have been described in my United States Patents Nos. 2,185,192 and 1,808,220, which describe how the relative marking and spacing time intervals can be varied.

Another remedy would be to apply adjustable slope frequency selectivity before the limiter in the receiver, to aect the balance between currents at the two frequencies. In practice this may often be accomplished by means of a small detuning of the receiver achieved through adjustment of the frequency of heterodyne oscillator I4 of Fig. 2.

Although I have described my invention in its application to long distance radio circuits, I realize that it is also applicable to many other kinds of communication circuits including both radio and wire line electrical circuits, sound and supersonic mechanical wave signaling systems, or to any other system where multiple paths or multiple reflections of waves may `occur on the way from transmitter to receiver.

What is claimed is:

1. In a frequency change radio signaling system where the transmitted signal characters are short compared with the difference in time of signal travel between shortest and longest significant multiple paths extending between the transmitter and the receiver and also short compared with the spaces between signal characters, a carrier current telegraph communication system comprising, in combination, a transmitter having means to shift the frequency of a carrier current between two fixed values to produce marking and spacing waves, means at said transmitter for differentially modifying the relative lengths of marking and spacing time prior to transmission of said waves, to cause the marking wave to be lengthened for an interval of time approximately half the delay time between the shortest and lon-gest important multiple paths, and a frequency modulation receiver responsive to said marking and spacing waves of different frequencies.

2. In a frequency change radio signaling system where the transmitted signal characters are short compared with the difference in time of signal travel between shortest and longest signicant multiple paths extending between the transmitter and the receiver and also short compared with the spaces between signal characters, in a carrier wave telegraph signaling system, in combination, a transmitter having means for changing its frequency of operation between two xed values, a frequency shift receiver having means to controllably and dierentially modify the relative strengths of -currents at said two values of frequency, said receiver having means coupled thereto for substantially removing amplitude variations of the combined currents, and a frequency shift detector coupled to said last means.

3. In a system for reducing the effects of multipath distortion in radio communication wherein there is an overlap of the significant received signals, a transmitter for sending out marking and spacing frequencies of fixed values which differ from each other by a predetermined amount, and a frequency modulation receiver having an amplitude limiter and also selective circuits for passing only a range of frequencies smaller than said predetermined amount but sufficient to produce readable signals.

4. The method of reducing the effects of multiple signal paths in radio communication between a transmitter and a receiver, which comprises alternately changing the signaling frequency at the transmitter between two frequencies whose difference is -greater than the frequency range required at the receiver to reproduce signals of acceptable quality, receiving the transmitted signals, limiting the amplitude of the received signals, and reproducing the received signals only in a range of frequencies less than the frequency difference between the two transmitted frequencies but suicient to produce readable signals.

5. A method of reducing the effects of multiple signal paths in radio communication which comprises, at the transmitter, alternately changing the signaling current frequency between two fixed values and, at the receiver limiting the amplitude of the signal carrier currents, converting the frequency change into differential change of amplitude of two carrier currents in conversion circuits broad enough to include all signaling frequency components, converting the two carrier currents into varying direct current signal currents, differentially combining the last signal currents, passing the combined currents through a low pass filter having a uniform response pass band less than the frequency changes at the transmitter, and utilizing the resulting current obtained from said lter.

CLARENCE W. HANSELL.

Images