Secret transmission of intelligence

Abstract

1. In a signal privacy system, means to convert a gradually varying signal current wave into a stepped wave, means to superpose on said stepped wave a masking current varying in magnitude in a manner unrelated to the signal to produce a stepped summation current wave for transmission, and means to eliminate short portions of the summation wave at the points where the steps occur to further conceal the signal.

Description

The present invention relates to the transmission of intelligence with privacy or secrecy. The invention is concerned with the securing of a high degree of privacy or secrecy of transmission and is especially directed to speech transmission, although it is not limited to speech transmission and particularly is this true as to some of its aspects and features.

The specific disclosure of the invention contained in this application is based upon the vocoder, a system of transmission and reception of speech disclosed and claimed in H. W. Dudley's U.S. Pat. No. 2,151,091, dated Mar. 21, 1939 and the invention is in the nature of an improvement upon the systems disclosed in applications of R. C. Mathes Ser. No. 412,054, filed Sept. 24, 1941 and H. W. Dudley Ser. No. 423,437, filed Dec. 18, 1941.

In the vocoder, speech is analyzed by subdivision of the frequency band into narrow bands and by integration of the energy in each band, giving a number of slowly varying speech-defining currents in separate paths, each current requiring for its transmission a band no wider than about 25 cycles. These low frequency currents are simultaneously transmitted to the receiving point. Speech is synthesized under control of these low frequency speech-defining currents with the aid of a source of energy having a frequency distribution covering the essential speech range. This energy is subdivided in frequency into narrow bands and the energy in each band is passed through a circuit whose admittance is made a function of the corresponding low frequency speech-defining wave. The outputs of all of these circuits are combined. For a more complete understanding of the vocoder, reference may be had to the Dudley patent cited.

While the vocoder itself gives a certain degree of privacy of transmission, a much greater and entirely different order of secrecy of transmission can be realized by operating upon the low frequency signal-defining waves to render their reception and recognition difficult in absence of knowledge of the particular way in which these waves are operated upon.

One manner of disguising these speech-defining signals is to add to them a current varying with time in some manner unrelated to the signal to produce an unrecognizable wave for transmission. At the cooperating receiver, an identical current is, in effect, subtracted from the transmitted wave leaving as residue the signal current. Some means is provided at both stations for furnishing the identical waves used for masking the signal, one such means comprising records prepared in advance. It has been shown in the Mathes and Dudley applications referred to that the secrecy can be greatly increased by use of reentrant coding, by which the total range of variation of the signal-plus-masking wave is reduced to, for example, the range represented by the signal alone or to an even smaller range. For example, if the signal has a range of variation from 0 to S and the masking current has a range of variation from 0 to M, the summation wave would have a total range of 0 to M + S, or if M = S, a total range of 0 to 2S. In using reentrant coding, the summation wave is used for transmission so long as its instantaneous value is not in excess of S, but whenever the summation wave rises above the value S, there is subtracted from it by suitable means a fixed value sufficient to bring the resultant into the range 0 to S, which is the range transmitted. This results in greater secrecy since it effectively conceals knowledge of the signal that might otherwise be gained from particular values such as the value 2S, which value could only mean that M and S were both at their maximum value, assuming M = S. For a further understanding of reentrant coding and its advantages, reference can be had to the Mathes and Dudley applications cited.

The process of subtracting the fixed value from the summation wave when its instantaneous magnitude exceeds the maximum value of the signal, that is, S, can be carried out in different ways. In the Mathes disclosure such subtraction is made on a frequency basis while in the Dudley application disclosure the subtraction is made on an amplitude basis. When the subtraction involves some type of switching or sudden transition, transients are likely to occur which might reveal the instants at which reentry is used, thus giving possible useful information to an outsider interested in cracking the coded message.

One object of the invention is to prevent indications from occurring in the transmitted waves of the points at which reentry takes place.

A more general object of the invention is to improve upon prior secrecy systems by use of a number of features operating together to give more positive and reliable operation.

Features of the invention comprise improved exciter and timing circuits, improved reentry circuits, a curbing circuit and other features to be indicated in the detailed description to follow of a complete one way secret telephone transmission system embodying the improvements constituting the present invention.

IN THE DRAWINGS

FIGS. 1, 2 and 3, when placed end to end with FIG. 1 at the left and with FIG. 4 immediately below FIG. 1, as shown in key FIG. 6, show a complete one-way system to the extent necessary for a full understanding of the inventive features claimed herein;

FIG. 5 shows voltage wave forms to be referred to in the description;

FIG. 7 shows an alternative reentry and curbing circuit using a stepper, and

FIG. 8 shows time relations applicable to the alternative circuit shown in FIG. 7.

The left-hand portion of FIG. 1 shows in block diagram the analyzer part of a vocoder while the right-hand portion of FIG. 3 gives a similar showing of the synthesizer part of a vocoder. The circuits and apparatus intervening between these portions are for coding, transmitting and decoding the vocoder channel currents so as to render the transmission secret. The secrecy is obtained by adding to the vocoder currents before transmission masking waves obtained from a suitable record, indicated at 10 in FIG. 1, and the decoding is accomplished by supplying identical masking waves at the receiver supplied, for example, from a suitable record indicated at 20 in FIG. 3. The recorded material is assumed to be known only to the communicating parties.

The single record 10 or 20 has recorded on it as many separate codes or masking waves as there are channels to be transmitted, for example, eleven if we assume ten speech spectrum channels and one fundamental pitch channel. A greater or lesser number may actually be used depending on requirements.

Preparatory to adding the code currents or masking waves to the vocoder channel currents, each vocoder channel current passes through a stepper to change the current from a gradually varying current to one changing in abrupt steps. Each code current is similarly converted to a stepped wave, the steps in both the message and code currents being similarly timed so as to coincide with each other. Thus, the summation current is also a stepped wave in which each step level is equal to the sum of the message and code at the corresponding instant. This summation wave is then put through a reentry circuit to bring its total amplitude range down to the range represented by the signal alone, signal referring, of course, to the vocoder channel current. This is the range actually transmitted. The output is also inverted in order to facilitate removing the key at the receiving end. The channels are multiplexed together on different carrier frequencies for transmission over a common medium.

At the receiver, a reverse operation takes place, the key supplied from the record 20 being first combined with the received secret message currents (after both message and key currents have been put through steppers) and reentry is made to restore the vocoder signals to their true form. Since an inversion was made before transmission an inversion is necessary also at the receiver to restore the signals to proper form. With this brief over-all description as introduction, a more detailed description of the various parts of the system will now be given.

A speech input, such as a microphone, is shown at 11 leading through equalizer 12 to a branch point. Branch 13 is the fundamental pitch channel and leads through band-pass filter 14, rectifier 15, frequency measuring circuit 16 and low-pass filter 17 for deriving a direct current whose magnitude varies from instant to instant in accordance with variations in fundamental pitch of the speaker's voice. The rest of the analyzer comprises a group of channels, of which there may be ten, by way of example, each consisting of a band filter 18, rectifier 19 and low-pass filter 21. Filters 18 have different pass bands for subdividing the speech band into relatively narrow bands.

It is desirable at this point to introduce a rather large amplification of all of these channel currents and since they comprise frequencies including direct current (zero frequency) and extending to about 25 cycles, magnetic amplifiers can advantageously be used. For this purpose a 2-kilocycle source of waves, shown at 23, is used with its output carefully regulated by suitable means to a constant value and thoroughly filtered to remove harmonics. This wave is put through windings on magnetic cores with other windings leading to the output circuit 24, the windings being arranged to be balanced when there is no signal input so that none of the 2-kilocycle wave or its harmonics then gets into the output circuit 24. A signal winding 25 included in the output circuit from filter 21 surrounds both cores and unbalances the circuit in proportion to amplitude of the input signal to let through a corresponding amount of the double frequency (4-kilocycle) which is the principal component. The usual choke coil and series resistor are shown at 26 for keeping harmonics out of the 2-kilocycle supply circuit. This type of modulator is of itself known in the art and is shown as a preferred form, although other known amplifiers might be used instead (for a fuller disclosure of magnetic amplifier see Burton U.S. Pat. No. 2,164,383, July 4, 1939). It is not necessary to rectify the waves in circuit 24 since the varying amplitude 4-kilocycle wave is well suited as input to the message stepper. It will be clear that one source of 2-kilocycle wave supplies all of the vocoder channels and that each channel includes a similar amplifier.

The message stepper comprises, specifically, five gasfilled tubes 27, 28, 29, 30 and 31 having grids connected to graduated points along potentiometer resistances 32 bridged across circuit 24. By means of an exciting and bias control circuit to be described at a later point, the grids and plates of these tubes have applied to them rectangular voltages of the type shown in FIG. 5, upper two wave forms. The common grid lead 33 has applied to it a voltage which is 4 volts negative with respect to the cathodes for 2 milliseconds and is then 150 volts negative with respect to the cathodes for 18 milliseconds. The common cathode lead 34 has a voltage of 150 volts negative with respect to ground applied to it all the time except for 2-millisecond interruptions which are for the purpose of restoring the stepper tubes to normal approximately every 20 milliseconds for a brief instant. As seen from FIG. 5 the interruption of the plate supply comes just before the grids are driven in the positive direction. The plates are driven positive with the grids so as to allow the stepper tubes to be triggered provided there is a signal present in circuit 24, and the number of tubes that are triggered depends upon the peak amplitude of the signal in the circuit 24 in the 2-millisecond interval when the grid is driven in the positive direction. Immediately thereafter the grids are driven highly negative relative to the cathodes so that the signal in circuit 24 no longer has any control over the tube discharge. In this way the stepper tubes sample the signal current for an interval and if the signal has sufficient amplitude one or more tubes break down depending on the signal peak amplitude in that interval. The tubes broken down remain so until their plate supply voltage is interrupted after 18 milliseconds. They then restore momentarily and if the signal amplitude has changed in the meantime, a greater or lesser number of tubes break down upon the next 2-millisecond exposure. Assuming, by way of example, that the 4-kilocycle voltage range in circuit 24 is from 2 volts for no signal input to 50 volts for maximum signal input, the potentiometers 32 would be set so that for a signal input greater than 2 volts peak value but less than 10 volts, no tube breaks down; for a signal greater than 10 volts but less than 18 volts, tube 27 breaks down; for signal greater than 18 volts but less than 26 volts, tubes 27 and 28 break down, etc. All tubes break down when the signal exceeds 42 volts. The currents in the various tubes of the stepper flow through individual plate resistors 36 and return to the plate supply source through common resistor 35 and ground. The current through 35 is, therefore, the sum of the currents through the tubes that are broken down at any one time and the voltage developed across resistor 35 is proportional to the signal amplitude at the sampling times and remains constant between samplings times. Further, the voltage developed across 35 is made proportional to the signal by properly proportioning the resistances 36 in series with plates of the stepper tubes.

The secret key recorded on record 10 may be prepared in any suitable manner, for example in the manner disclosed in Newby-Vaughan application Ser. No. 456,356 filed Aug. 27, 1942, now U.S. Pat. No. 3,373,245. The key for any one channel should ideally consist of a succession of current values occurring in random order in successive time intervals equal to the time intervals used for the stepper, that is, intervals of 20 milliseconds duration. All of these keys (eleven in the example assumed) can be modulated on individual carrier frequency waves and recorded, as in multiplex carrier transmission, on the record 10. (A duplicate recording is made on record 20.) The keys for the various channels can be readily separated by means of filters 40,40, etc. and suitable amplifiers 42, etc. may be used in each key channel to bring the key waves to the same amplitude range as the signals.

The key stepper is entirely similar to the message stepper and, therefore, requires no separate description. As the key varies in amplitude in random fashion it is sampled by the key stepper and a corresponding number of tubes break down (or none if the key is of less value than step No. 1), producing a voltage drop across resistor 45 proportional to the key amplitude at a given time.

The key is added to the message by the resistance bridge consisting of resistors 46, 47, and 48, resistances 46 and 47 being large compared to 35, 45, or 48 which are about equal. The summation current flows through resistor 48 to ground. The voltage developed across 48 (negative with respect to ground) and representing the summation of the message and key at any instant is applied to the grid 66 of duplex tube 50 for transmission in a manner to be described.

A second resistance bridge consisting 51 and 52 is connected across resistors 35 and 45 and its mid-point connects through resistance 53 to the control grid of pentode 54, this grid having connection through resistors 55 to positive pole of 150-volt battery 56. Resistance 55 is of the order of a megohm and is much larger than 51, 52, or 53, these each in turn being, say, three times larger than 46 or 47. The net result is that tube 54 is positively biased into its saturation region normally, and all of the first five negative voltage steps applied from resistors 35 and 45 are insufficient to reduce its space current below the saturation value but the sixth step swings the grid from positive to negative through its entire characteristic to the cut-off point. Higher voltages in the same direction are unable to reduce the space current further since it is already reduced to zero by the sixth step. Tube 57, on the contrary, has its grid normally biased to cut-off by negative voltage from battery 61. The actual bias on the grid of tube 57 is determined by the constant negative voltage from 61 and the variable voltage applied from the plate of tube 54. When the space current of tube 54 is reduced to zero its plate voltage assumes its highest positive value and this throws the bias on the grid of tube 57 so far positive as to cause saturation current to flow through the latter tube. For all voltage steps in excess of step 6 applied to the input of tube 54, therefore, saturation current of constant value flows in tube 57 and through resistor 65 to ground. The direction is such as to make the ungrounded terminal of 65 positive. This potential is, therefore, subtracted from that existing across resistor 48 as a result of direct application thereto from resistors 35 and 45. When current flows in resistor 65, therefore, the voltage that is applied to the grid 66 of duplex tube 50 is the difference between the voltages existing across 48 and 65. Grid 66 of tube 50 is connected to the center point of the bridge consisting of 48, 65 and the two equal resistors 67 and 68, these latter being larger than either of the resistors 48, 65 which are equal to each other.

In this way, reentry is accomplished. To recapitulate briefly, if no tube in either stepper is broken down, the voltage applied to grid 66 of tube 50 is zero, corresponding to step zero. If one stepper tube is broken down, a small negative voltage is applied to grid 66, corresponding to step 1. If five stepper tubes are broken down, substantially five times as great a negative voltage is applied to grid 66 corresponding to step 5. If six stepper tubes are broken down, a negative voltage of step zero value is applied to grid 66, this resulting from six steps of negative voltage across resistor 48 and the constant reentry six units of positive voltage across resistor 65. If all ten stepper tubes are broken down, a negative voltage of four-step value is applied to grid 66.

The great advantage of stepping the message and key waves and using reentry for obtaining secrecy has been pointed out in the Mathes application. It can be graphically illustrated by the following table in which the message is assumed to have any one of six values including zero and the key can have any one of the same six values. The figures in the table are the values after reentry (sum minus 6) where reentry is used, and represent the

Message Values O 1 2 3 4 5 ______________________________________ Key 0 0 1 2 3 4 5 Values 1 1 2 3 4 5 0 2 2 3 4 5 0 1 3 3 4 5 0 1 2 4 4 5 0 1 2 3 5 5 0 1 2 3 4 ______________________________________

coded signal. Any given message value can become any other message value after coding so that no one value, such as 0 or 5, constitutes a clue, since these can represent all values.

It has been observed that when the tubes 54 and 57 operate as described to produce reentry, transient effects may occur which show up in an oscillogram as peaks or spires of current at the edges of the current steps. While these can be attenuated by filters to a large extent, their presence in any detectable degree is undesired since a close study of the wave form might reveal the instants when reentry is made. Since the secret wave produced consists of flat-topped pulses, it is only necessary to transmit the central part to convey all the useful information contained in the wave and there is no loss of informational content by suppressing the beginnings and endings of the impulses. By suppressing these portions, all clue as to reentry times is effectively removed. This is accomplished by the curbing circuit now to be described.

Referring to FIG. 5, lowest diagram, the curbing voltage consists of a normally positive voltage for 14 milliseconds at constant value and falling to zero for 6 milliseconds. The 6-millisecond interval begins just before the restoring period of the steppers and lasts until after the next step has been taken. It is in this interval that the reentry takes place if at all. The curbing wave of the type shown in FIG. 5 is generated in the common exciting and timing part of the system to be described, and is applied to the lead 71 leading, for the channel shown in detail, to the lower end of resistor 72 (and to similar resistances in all the other reentry and curbing circuits). Resistance 72 may have a value of 70,000 ohms, for example. The opposite terminal of this resistor is supplied from source 61 and by means of a slider 73 on resistor 72 suitable normal bias voltages can be taken off and applied to both grids 66 and 70 to bring them to the proper points on their operating characteristics. This is the condition existing for 14 milliseconds out of each 20 milliseconds. During the 6 milliseconds when the 150 volts positive voltage is removed from lead 71, the grids 66 and 70 are swung negative beyond cut-off by source 61 and tube 50 is blocked during each such 6-millisecond interval. This in effect eliminates the beginning and end of each pulse and confines transmission to the central 14-millisecond portion.

Reference will now be made to the exciter and impulser circuits shown in FIG. 4. The exciter circuit is shown as comprising two banks of three tubes each, these being shown as pentodes 75, 76, and 77 in the upper bank and 78, 79, and 80 in the lower bank. These tubes have their control grids driven from a source of 50-cycle current 81 of constant frequency through phase shifting circuits. This current is amplified at 82 and applied to the two banks of tubes through transformers 83 and 84. When the key is used, the 50-cycle per second comes from the record to synchronize the impulses with the key.

Considering first the tubes 75 and 78, a negative half cycle of the 50-cycle input wave swings the grids of both tubes beyond cut-off. At some point in the cycle the grids are driven in the positive direction a sufficient amount to transmit current and the current rises quickly to full value. Due to the phase shifting circuits 85 and 86 the 50-cycle wave on the grid of tube 75 leads that applied to the grid of tube 78 by 2 milliseconds. Tube 75 first transmits current through a path which can be traced from its plate over lead 95, resistor 96, lead 97, potentiometer resistor 98 to ground at 99, to ground at 100 (adjacent the exciter and at the right) upper part of potentiometer resistor 88 in the output of rectifier 87, and the minus 400-volt cathode lead 92. This current in flowing through resistor 96 throws the grid of tube 101 negative beyond cut-off and shuts off the flow of current from rectifier 102 to output potentiometer 103 the effect of which will be further discussed presently.

At a time 2 milliseconds later, tube 78 sends current from its plate through resistors 90 and 91, lead 92, lower part of resistor 88 to cathode lead 93. The current flow through resistors 90 and 91 cuts off the tube 75 by blocking its grid, and this tube remains blocked until the grid again swings toward positive when the process repeats itself.

The pair of tubes 76 and 79 operate in analogous manner to send current from the plate of tube 76 over lead 104 and resistor 105 to ground 99 for 2 milliseconds to throw the grids of the parallel bank of tubes 106 beyond cut-off for 2 milliseconds and cut off the supply of rectified current to potentiometer 98.

The pair of tubes 77 and 80 operate in analogous manner to send current from other plate of tube 77 over lead 107, resistors 108 and 109 to ground, thereby placing a blocking potential on the grid of the tube 110 in the reentry impulser for a period of 6 milliseconds to cut off the supply of rectified current to potentiometer 111.

By means of the phase shifters similar to 85 and 86 in the inputs to the exciter tubes the relative timing can be readily obtained to correspond to that indicated in FIG. 5 for the three voltage supplies.

In order to be able to supply sufficient power for all of the channels from common apparatus and to regulate closely the voltages supplied, a number of voltage regulators are used, as disclosed in FIG. 4. These are essentially similar. For example, in the cathode impulser, the output of the rectifier 115 is filtered and the resulting direct current is put through potentiometer resistor 98 in series with the bank of tubes 106 which are inserted in the positive lead to ground 99. Pentode tube 116 is connected across the line with its cathode connected to the negative side and its plate connected through resistor 105 to the positive side. Its control grid is connected in series with negative battery 117 to a point in resistor 98 so that a suitable fractional part of the output voltage can be applied to the control grid. The regulation is based on the assumption that the voltage of battery 117 remains sufficiently constant. This battery may have some convenient voltage such as 90 volts. Just enough opposing voltage is tapped off from resistor 98 to bring the control grid of tube 116 to some suitable control point on the tube characteristic. This establishes a normal value of current through resistor 105 which fixes the normal regulating voltage on the grids of tubes 106. Variations in voltage across resistor 98 swing the grid of tube 116 above or below its normal value; these variations are amplified and applied to the grids of tubes 106 in such sense as to oppose the assumed variations in output voltage by varying the drop of potential across tubes 106. In this way the circuit regulates itself to a constant output voltage of 150 volts. This is applied over conductor 34, as previously described, between ground and the cathodes of all of the steppers (FIG. 1). When the voltage of the grids of tubes 106 is thrown negative by current received over lead 104 the tubes 106 change their function from regulating to blocking for the 2-millisecond interval.

The tubes 110 and 120 of the reentry impulser operate in similar manner to maintain constant voltage across the terminals of resistor 111 so as to place 150 volts positive to ground on lead 71, except when the output current is interrupted for 6 milliseconds by current received over lead 107. (For convenience of drawing the power source 63 is shown at several different points but may be a single power supply such as house lighting circuit or other supply.)

The rectifier 121 and regulator 122 may be of the same types as already described including a shunt pentode, opposing battery and series triode, to regulate to constant voltage of 150 volts across points 123, 124. Two additional tubes 125 and 126 operate to provide a constant voltage of 4 volts across resistor 127, when tube 101 is cut off, these tubes operating in similar manner to those already described. When, therefore, the current in resistor 103 via tube 101 is reduced to zero for 2 milliseconds, the upper end of resistor 127 has the same potential as the cathode lead 34 (also lead 97) while the lower terminal of resistor 127 is then 4 volts negative relative to the cathode lead 34 due to regulated voltage set up by current via tube 126. The grid bias lead 33 connects to this last point and thus the grids of the steppers during the 2-millisecond exposure period are biased 4 volts negative to their cathodes. In addition, during this interval of 2 millisecond, the alternating current impedance between leads 34 and 33 is kept low to avoid pick-up difficulties. In the 18 -millisecond interval between exposure times, the stepper grids are thrown 150 volts negative to their cathodes. When the tube 101 delivers current to resistor 103 and develops 150 volts across it, this negative voltage saturates tube 125 to thus block tube 126 and remove the 4 volt supply. Tubes 101 and 130 regulate to provide 150 volts across resistor 103 except when the regulator tube 101 is disabled for the 2-millisecond interval.

It was described above how the reentry and curbing circuits including resistors 48, 65, 67, 68 and tubes 54 and 57 impress voltage steps on the grid 66 of tube 50 corresponding to the coded signal to be transmitted. It will be understood that similar apparatus operating in similar manner is used in each of the eleven channels, as indicated in FIG. 1. The eleven separately coded channel waves may be sent to the distant receiving station in any suitable manner as by ordinary multiplex transmission but the method illustrated comprises frequency modulation of individual carrier waves by the several coded channel currents. This method was disclosed in the Mathes application but the apparatus of the present disclosure is specifically different and will now be described.

The duplex tube 50 feeds its output to a special inductance coil 140 comprising a number of separate windings mounted on a magnetic core such as permalloy. This coil forms the tuning inductance of oscillator 141, the tuning capacity being 142. The duplex tube 50 must amplify currents of low frequency including zero and it is necessary to balance the circuit carefully against battery voltage variations, heater variations and other low frequency variables. For this reason the tube 50 has signals applied only to grid 66, the grid 70 and its cooperating anode being for balance and inverting purposes. Since increasing steps from the steppers throw grid 66 more negative an inversion of the signal takes place in tube 50, resistor 53 being set to give full unbalance output at step 0 and balanced output through both primary coils at step 5. The output coil has two windings, as shown, one in each plate circuit, and the potentiometer 143 between these coils aids in balancing the circuit against the variables mentioned. The coil 140 is operated at high saturating flux to minimize hysteresis and give as nearly as possible linear variation of inductance with amplitude of impressed signal. The signal should swing over only a relatively small portion of the total characteristic. Amplified signal currents in the winding 144 vary the saturation and therefore the effective tuning inductance, thus modulating the frequency of oscillations generated by tube 141. Amplitude modulation is minimized by gas discharge tube 145 which breaks down on wave peaks of greater than a specified amplitude, and these excess peaks are dissipated in the series resistance 146. The oscillator shown is of the type disclosed in L. A. Meacham U.S. Pat. No. 2,163,403, June 20, 1939 including as bridge arms the tuned circuit consisting of coil 140 and condenser 142, and resistance arms comprising 148, 149, and 150. Gas tube 145 and resistor 146 take the place of the lamp type limiter of the Meacham patent. Thermistor 151 is connected across arm 149 to compensate for ambient temperature variations. The output frequency modulated oscillations are taken off through transformer 152 and applied to outgoing line 154 through band filter 153. The frequency modulators for the other channels feed out through similar filters 153 to outgoing line 154. The mean frequency of the oscillators in the different channels may be 170 cycles apart beginning with, say, 765 cycles for the lowest frequency, by way of example. Each band filter 153 would then have a pass-band 100 cycles wide with the mean frequency of the particular channel at one edge of the filter band and the frequency may be modulated to a depth of 100 cycles from mean frequency so that only those modulated wave components lying to one side of the mean frequency of the oscillator are transmitted. The total transmitted band in this case is within 3 kilocycles and the transmission can be carried out over an ordinary telephone circuit, carrier channel or radio channel.

The secretly transmitted wave after traversing the transmission line or channel 154 to the distant receiving station is separated into its components channel bands by filters 155 which are similar to filters 153, and these individual frequency modulated waves are applied to frequency modulation detectors 156, each consisting of an amplifier-limiter, and a tuned circuit (slope circuit) 157 slightly off-tuned from the wave to be received so that the frequency modulated wave is converted to an amplitude modulated wave. This latter is detected in diode 158 to give a 0 to 25-cycle output wave which is impressed on the grid of amplifier 160. Diode 158 is biased from negative battery 159 through resistance paths that permit the plate to be negative toward the cathode to just rectify the weakest input amplitude modulated wave. Amplifier 160 is cathode-coupled to low-pass filter 161 in a fully degenerative stabilized feedback circuit to stabilize the amplifier so as to give satisfactory direct current operation.

Following the low-press filters 161 are magnetic amplifiers 162 which may be like the magnetic amplifiers 22 of FIG. 1. They are supplied with 2-kilocycle energizing current from source 163 through purifying filter 164 as in FIG. 1.

The message steppers and key steppers may be identical with those of FIG. 1. The reentry and curbing circuits may be identical with those of FIG. 2, but in this case the output of the reentry and curbing circuit feeds into the low pass filter 170 in each channel instead of into a frequency modulator as in FIG. 2. The cathode impulser, grid impulser, reentry impulser and exciter circuit may be the same as those of FIG. 4.

The action of the transmitting and receiving circuits in coding and decoding will be made clear from considering a specific case. Suppose a signal element after the transmitting message stepper has an amplitude corresponding to step 2, or for convenience, two arbitrary units, and that the key has an amplitude of three units. These are added to give a five-unit pulse which after curbing is transmitted. In order to be able to use dupliciate apparatus for coding and decoding, however, the reverse (in the telegraph sense) of the five-unit pulse (5--5) is actually used for transmission, that is, zero unit. This goes through the transmission channel and is recovered at the distant receiver as zero unit. The key (three units) is added making three units, which is reversed by subtraction from five units, giving two units, which is the original signal. In this case no reentry was involved since the five-unit summation pulse was of insufficient amplitude to reduce the space current of tube 54 in the reentry circuit to zero. To take another example, let the signal again be two units but the key five units making a coded pulse of seven units. This operates the reentry circuit to give a coded pulse of one unit which is sent as (5-1) or four units. This is received and has added to it the key (5) making nine units which after reentry becomes three units, and this is reversed to yield the original signal of two units.

The reversal can take place at various points in the transmission. For example, besides the method described using tube 50, it can be effected by reversing the polarity of winding of the frequency modulating input coil or it can be done in a direct current circuit by subtracting the pulse from a constant voltage. At the receiver it can be done by changing the slope circuit tuning from one side to the other of the wave frequency, or by subtracting the direct current pulses from a constant voltage. With whatever inversion method that is used, the inverting means at the transmitter and at the receiver must be properly coordinated.

It is necessary to drive the records 10 and 20 in close synchronism with each other and this may be done by suitable means known in the art such as are used in synchronizing television transmitter and receiver distributors, printing telegraph distributors or other synchronized apparatus. In some instances, constant frequency oscillators may be used at the separate stations, such as crystal controlled oscillators like that of the Meacham patent, driving synchronous motors for the turntables. The showing of the disc record is illustrative only, since various other types of records including photographic and magnetic could be used.

In some cases it may be found feasible to omit the message steppers and key steppers at the receiver, thus simplifying to that extent the apparatus used. Even if the steps are not well defined out to the edges of the "flat" portions, the edge effects are eliminated by the curbing circuit. On the other hand, if the steppers are used, it may be feasible to omit the curbing circuits at the receiver since the effect of the spires of current at the edges of the steps, or other transients, would not lessen the secrecy of the system but would rather appear as distortion which might be tolerated in some cases.

Following on from the output of the receiving reentry and curbing circuits, each channel leads to a low pass filter 170. The uppermost channel, which is the fundamental pitch channel, leads to switching amplifier 171 and tone source 172 and the noise source 173 is connected to the input of the amplifier 171. In the absence of a voiced sound, that is, on using vocal cord vibrations, the tone source 172 is biased in inoperative condition, while the switching amplifier 171 is in its transmitting condition and transmits continuous energy spectrum waves from the noise source to the supply circuit 174. Voiced sounds produce current in the pitch channel which disables the switching amplifier and removes the blocking bias on the tone source 172 to permit fundamental and harmonic wave energy to flow into the output supply circuit 174, and the pitch of the fundamental wave is determined as a function of the amplitude of signal in the pitch channel.

The remaining channels lead to variable gain amplifiers 175 and the channel currents unblock these amplifiers in proportion to the strength of signal in the particular channel. Energy in the supply circuit 174 is distributed to amplifiers 175 through selective filters 176 having pass ranges similar to those of the filters 18 of FIG. 1, or other suitable pass ranges. Output filters 177 have similar pass ranges and lead in common to the receiver 178 which may be a telephone receiver or loudspeaker. Equalizer 179 may have a characteristic to compensate for that of equalizer 12 of FIG. 1, or otherwise equalize transmission.

The portion of the system to the right of filters 170 is, as stated, the receiving end of a vocoder and is in accordance with the disclosure of Dudley's patent which may be consulted for a more detailed understanding.

In FIG. 7 there is shown an alternative circuit which may be used in place of the reentry and curbing circuit of FIG. 2. Beginning at the left, this figure shows a message stepper at 200 and a key stepper at 201 which may be the message and key steppers shown in detail in FIG. 1. At the lower right of FIG. 7 is shown a frequency modulated oscillator 203 which may be the frequency modulated oscillator shown in detail in FIG. 2 with its specially constructed modulating transformer 140. The circuit elements shown between the steppers 200 and 201 and the oscillator 203 of FIG. 7 comprise the alternative circuit referred to and will presently be described.

The message stepper 200 and key stepper 201 are supplied with timed impulses from the cathode impulser 204 and grid impulser 205, these being controlled in turn from exciter 206 driven from 50-cycle standard frequency source 81, these equipments all being the same as those disclosed in FIG. 4, except that in this case the exciter 206 does not have the third pair of tubes used in FIG. 4 to supply a timing pulse to the reentry impulser, which in this case is not used. The control impulses for the steppers of all channels are supplied over leads 33 and 34 and the timing pulses for the impulsers 204 and 205 are supplied over leads 95 and 104 as previously described in connection with FIG. 4. The stepped message currents in resistor 35 and the stepped key currents in resistor 45 are added in the resistance bridges 46,47 and 51,52 as in the case of FIG. 2, the center point of the first pair of resistors leading through magnetic amplifier 207 and the center point of the second pair of resistors leading to the grid of reentry circuit tube 208. Magnetic amplifier 207 is like magnetic amplifier 22 of FIG. 1, and the signal input coil is connected in series with 4-kilocycle filter coil 209. The output currents from the magnetic amplifier 207 are applied to the output stepper consisting of five gas-filled tubes 210 to 214 having their grids connected to graduated points on potential dividing resistors 215 connected in bridge of output branch 216 of the amplifier 207. The arrangement is similar to that of the message and key steppers in that tubes 210, etc. fire in different numbers depending upon the maximum amplitude of the voltage existing, at the instants of exposure, across circuit 216. The current in the common output resistor 218 is, therefore, proportional to the sum of the currents flowing in the plate circuits of the tubes 210, etc. that are conducting at one time and the plate resistors 217 of the individual tubes are proportioned to make the current in resistor 218 proportional to the maximum voltage existing across circuit 216 at the sampling times. A fractioanl part of this output current is taken through adjustable resistance 221 and is sent into winding 222 of frequency modulating coil 140, this winding corresponding to the winding 144 of FIG. 2, so that from this point on the action of modulating the frequency of oscillator 141 and transmitting the modulated waves through filter 153 to line 154 is the same as in the previously described case.

As noted in the description of the system disclosed in earlier figures, it is desirable to invert the signals before transmission. This is done in the present circuit by providing battery 225 which applies a positive potential to the input terminal of the amplifier 207 of such value that when the output voltage from the message and key steppers is zero, all five of the tubes 210 to 214 are broken down and maximum value of modulating current is impressed on the frequency modulation circuit. Since the output voltage from the message and key steppers applied to lead 226 is negative it subtracts from the constantly applied battery voltage and when the total output from the steppers has a value equal to five steps the voltage in the output of amplifier 207 is reduced to a point where none of the tubes 210 to 214 fires.

The reentry circuit comprises tubes 208 and 227, the cathode of the latter being connected to lead 226 and eventually to ground through the amplifier 207. Assuming increasing output of summation message and key currents, the first five steps reduce the normally positive grid voltage of tube 208 but on step 5 the tube still continues to transmit saturation current which, flowing through output resistances to which the grid of tube 227 is connected, is sufficient to keep tube 227 from passing current. On the sixth step, tube 208 is cut off allowing tube 227 to transmit saturation current which is adjusted to such value as again to cause all five stepper tubes 210 to 214 to fire. Steps 7, 8, 9, and 10 result in the firing of four, three, two and one of the stepper tubes respectively. In this way reentry is accomplished. It will be observed that the tubes 208 and 227 operate in the same manner as the tubes 54 and 57.

The time relations involved in the operation of the FIG. 7 circuit are given in FIG. 8. The upper two curves apply to the message and key steppers and are the same as the upper two curves of FIG. 5. The output stepper has its exposure time set somewhere near the middle of the 18-millisecond output pulse coming from the message and key steppers. This allows time for transients to subside. This timing relation is shown by the grid bias supply curve of the output stepper, the third curve of FIG. 8. The output stepper tubes are restored by reduction of their plate-to-cathode voltage to zero. This occurs at a time about 6 milliseconds before they are to be fired on the next succeeding impulse. The plate voltage is held at zero for 4 milliseconds while the grids are at negative blocking potential and the plate voltage is again thrown positive just before the grids are unblocked to expose the stepper tubes. This results in an output pulse duration of about 14 milliseconds with a 6 millisecond spacing. These time relationships are shown in the lower two curves of FIG. 8.

Referring again to FIG. 7, an exciter circuit 230, grid impulser 231 and cathode impulser 232 are shown for the output stepper. Exciter 230 may be similar to the exciter circuit shown on FIG. 4 (but in this case only the first two pairs of tubes are needed in the exciter, since there is no curbing control lead 107 required). The grid impulser may be the same as that of FIG. 4 (involving tubes 101, 130, 125 and 126) and the cathode impulser may be the same as that shown on FIG. 4. Since the timing on the cathode impulser, however, calls for current interruption in the impulser of 4 milliseconds instead of 2 milliseconds, the phase shifting circuits in the exciter must be adjusted to send a 4-millisecond control impulse over the lead 104 in place of a 2-millisecond control impulse as was assumed in the case of FIG. 4. Since the operation of exciter 230 and impulsers 231 and 232 is delayed in time with respect to the exciter and grid and cathode impulsers of FIG. 4 by an amount indicated in the curves of FIG. 8, a phase shifter 233 is inserted between the source 81 of 50-cycle current and the exciter 230 to introduce the requisite time delay.

Since with the circuit of FIG. 7 the sampling of the message and key stepper output current occurs in the middle of the pulse period, the beginnings and ends of the pulses are eliminated and spires of current or transients caused, for example, by reentry operation are suppressed. This output stepper therefore, also acts to curb the impulses before transmission.

For purposes of full disclosure of the improvements constituting the invention, a complete one-way system has been disclosed, and this has involved the inclusion of features which in and of themselves form no part of the invention. In this disclosure, values and other detailed matters have been given by way of illustrative example with no intention of the limiting the scope of the invention thereby. The features which constitute the present invention and which are believed to be patentably novel are particularly pointed out in the claims, which follow.

Claims

What is claimed is:

1. In a signal privacy system, means to convert a gradually varying signal current wave into a stepped wave, means to superpose on said stepped wave a masking current varying in magnitude in a manner unrelated to the signal to produce a stepped summation current wave for transmission, and means to eliminate short portions of the summation wave at the points where the steps occur to further conceal the signal.

2. In a signal privacy system, means to convert a gradually varying signal current wave into a stepped wave, means to add thereto a masking wave, means to subtract from the resultant wave a fixed amplitude whenever the resultant wave exceeds in amplitude the maximum amplitude of the signal wave, to provide a stepped wave for transmission, and means to eliminate short portions of said last wave prior to its transmission, such eliminated portions including the points at which the steps occur.

3. In speech privacy transmission, means to derive from speech waves a number of low frequency component current waves, means to convert said current waves to stepped waves, means to add key impulses to the stepped waves to disguise the latter, and means to suppress the portions of the resultant waves occurring at the beginnings and ends of the steps.

4. In speech privacy transmission, means to derive from speech message waves a number of low frequency component waves of gradually varying amplitude, means to convert said latter waves to waves whose amplitudes change suddenly in abrupt steps, means to add a masking current to the stepped waves to produce resultant stepped waves, and means to interrupt transmission of the resultant wave during the times that the amplitude is changing from one stepped value to the next.

5. In a secret transmission system for signals, trigger tubes and means to apply the signals thereto, a timing circuit for triggering said tubes on and off at timed intervals to transmit segments of the signal differing in amplitude from one segment to the next, means to add a masking current to said current segments, said masking current also consisting of current segments of substantially the same duration as the segments produced from the signals to produce a resultant current for transmission consisting of segments, and means to suppress the beginnings and ends of the segments of the resultant current prior to transmission.

6. In a secret transmission system for signals, a source of signal waves, a stepper circuit for producing from said signal waves output waves varying in abrupt steps, a source of key waves, means to combine said key waves with said output waves to produce stepped resultant waves, a reentry circuit for restricting the total amplitude range of variation of said resultant waves to a range of the same order as the maximum range of signal variation, and a curbing circuit for deleting portions of the resultant waves including the beginning and end of each step.

7. In a secret transmission system having a transmitter and receiver with sources for supplying duplicate key waves at both transmitter and receiver, a message coder at the transmitter comprising a message stepper, key stepper, reentry circuit and curbing circuit and at the receiver a message decoder comprising a message stepper, key stepper, reentry circuit and curbing circuit for recovering the original message.

8. A receiving circuit for secret coded message waves resulting from combination of message waves and a secret key wave; said receiving circuit comprising a message stepper, a source of duplicate key waves, a combining circuit for stepped message waves and key waves, and a circuit for suppressing portions of the combined waves during which transitions occur from an amplitide representing one step to an amplitude representing another step.

9. In a system of speech transmission including analysis of speech waves into low frequency defining waves at a transmitting point and synthesis of speech waves at a receiving point under control of said defining waves, duplicate key wave sources at both points, means to convert the defining waves each to a stepped wave before transmission, means to combine the key wave with the stepped wave to produce resultant waves, means to reduce peak portions of the resultant waves where necessary to confine their total amplitude range to the normal range of said stepped wave alone, and means to suppress transmission of the portions of said last waves within which transitions occur from one stepped value to another to conceal evidence of peak reductions.

10. In a secret speech transmission system, means to analyze speech message waves into low frequency defining waves, a stepper circuit for each defining wave comprising a respective group of gas discharge tubes, means to set said tubes to break down individually at different instantaneous amplitudes of the given defining waves, a timing circuit for subjecting said steppers periodically to control by the respective defining wave, a source of key waves for each defining wave, means to combine the stepped defining waves and key waves to produce resultant coded waves and a reentry circuit and a curbing for each coded wave.

11. In a privacy system, a source of message waves, a source of key waves, a message stepper and a key stepper each comprising a group of gas discharge tubes having stepped break-down characteristics from tube to tube, a timing circuit, means under control of said timing circuit for generating timed impulses and applying the same to said groups of gas discharge devices in common, said impulses during their occurrence exposing said steppers to the influence of said message waves and key waves respectively, and means for restoring said gas discharge tubes to normal at a given time after the end of each exposure period.

12. In a secret signaling system, means to convert message waves into waves varying in amplitude in abrupt steps, means to add thereto a secret key wave also varying in abrupt steps to produce summation currents varying in abrupt steps, means to reduce the amplitude range of said summation currents to substantially the amplitude range represented by the message waves, and means to transmit the central portion only of each summation current step.

13. In a secret signaling system, means at a transmitting point for automatically coding a message wave to render the transmission secret comprising a message stepper, a source of key waves and means to add the key waves to the stepped message waves to produce a stepped summation wave, means to reduce by a fixed quantity the amplitude of the summation wave having peak amplitudes in excess of a given maximum, and means to invert the resultant waves, a source of duplicate key waves at a receiving point, means to add the key waves from said latter source to the received waves to produce summation waves, means to reduce by a fixed quantity the amplitude of the last-mentioned summation waves having peak amplitudes in excess of a given maximum, and means to invert the resultant waves to recover the message waves.