CA1182595A - Privacy communication system employing time/frequency transformation - Google Patents
Privacy communication system employing time/frequency transformationInfo
- Publication number
- CA1182595A CA1182595A CA000414741A CA414741A CA1182595A CA 1182595 A CA1182595 A CA 1182595A CA 000414741 A CA000414741 A CA 000414741A CA 414741 A CA414741 A CA 414741A CA 1182595 A CA1182595 A CA 1182595A
- Authority
- CA
- Canada
- Prior art keywords
- segments
- signal
- time
- system defined
- memory
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04K—SECRET COMMUNICATION; JAMMING OF COMMUNICATION
- H04K1/00—Secret communication
Abstract
Abstract of the Disclosure A privacy communication system digitizes a voice signal and divides the signal into different frequency bands or time segments and shifts the bands or segments in frequency and/or time under control of a continually changing pseudo-random key word to develop an encrypted transmitted signal having the same time/bandwidth product as the voice signal.
Description
5~
Back~round of the Invention This invention relates to a privacy communication system. It relates more particularly ~o an audio or voice scrambler in which a communication is rendered 5 unin~elligible so that its conten~ is unavailable to third parties, the communication being capable of being unscrambled after reception by an authorized person to recover the original voioe content.
Descri tion of the Prior Art P __ Privacy systems are in widespread use ~or renderlng audio signals, particularly voice signals, unintelligible for transmission over an exposed transmission link such as a telephone line ~o as to maintain ~he trans~ission priva~e, specifically to avoid reconstruction of the 15 voice content by unauthorized listeners. In such systems, the voice signals are typically encoded at a transmitting si~e using an encoding technique that involves scrambling or displacing the audio signals in the frequency domain, time domain or both. At the 20 receiving site, the scrambled signals are decoded by, in effect, reversing the encoding procedure to recover the original audio signals. Ideally, in any system o this type, the encoding technique used should rnake it extremely difficult for unauthorized listeners to decode 25 or "break" an intercepted scrambled signal, yet still permit recovery of the ~ransmitted informa~ion at the receiving site with good intelligibility and recognition by authorized listeners.
One proposed technique used to scramble voice 30 information, disclosed in U.S. Paten~s 3,921,151 and 3,970,790, for example, is to divide the ~ransmi~ted signal into segments of equal time duration. The individual ~ignal elemen~s are applied to a ~emory fot temporary storage. The stored segment~ are then read out 35 from memory in a random pattern as determined by a signal from a pseudo-random key code generator so as to scramble the order of the segments. They are transmitted in this scrambled order and momentarily stored at the receiving end, where the process is ef~ectively reversed to 5 descramble the signal elements. The former patent adds an additional element of randomness to the communication by time~reversing selected signal segments according to the output from the pseudo-random key code generator.
Attempts have been made to increase the security of 10 privacy systems by dividing the audio signal to be transmitted in frequency as well as in time to further scramble the transmitted signal. Arrangements such as this are disclosed in U.S. patents 4,149,035 and 4,221,931. Basically, the audio signal is divided into 15 two or more different frequency bands with the information in each band being digitized. Then each band is partitioned into a number of segments of equal time duration and these segments are transposed both within the same band and with segments in the other bands in 20 accordance with a code developed by a pseudo-random key generator. The signals are then converted back to analog form and combined for transmission to the receiving site, ~hich contains similar equipment ~or reversing the above procedure to recover the original audio siynal. The 25 system in the last-mentioned patent also timewise reverses certain of the signal segments in accordance with a second key code developed by the key generator to obtain an additional element of randomness in the scrambled signal. Even so, however, it may still be 30 possible for unauthori~ed listeners to decode or break an intercepted scrambled signal such as by analyzing the cadence and detec~ing recurring phenomena in the transmitted scrambled signal using present-day high~speed computers.
~ ~ ~Z~5 Hl4-015 Summary of the Invention It is therefore an object of the present invention to provide a privacy communication system having the ability to transmit scrambled audio signals with 5 increased security.
It is a further object of the invention to provide a system such as this in which the random nature of the scrambling of the audio signal is materially increased to reduce the periodicity of the transmitted signal 10 sequences.
Yet another object of the invention is to provide a secure communication system whose scrambled signals can be transmitted over a channel having no greater bandwidth than that required for the original audio signal.
A further object is to provide such a system which produces good quality in the received "clear" audio signal.
Still another object of the invention is to provide a secure communication system which is relatively simple 20 in construction, yet quite efficient and reliable in operation.
Other objects will, in part, be obvious and will, in part, appear hereinafter.
The invention accordingly comprises the features of 25 construction, combination of elements and arrangement of parts which will be exemplified in the Eollowing detailed description, and the scope of the invention will be indicated in the claims.
The present system has partic~lar applicability to 30 achieving high level communications security for audio bands in the speech frequency spectrum or channel extending from around 200 Hz to about 3000 Hz. To accomplish this, the sys~em divides ~he analog audio input signal into two or more frequency bands, preferably 35 of equal bandwidth. In general, one or more of the bands 5~
are transposed so that all the bands end up in the same low-frequency range. In the usual case, this is accomplished by separately transposiny the bands above the lowest frequency band to the frequency range o~ the 5 low ban~. ~he frequenc1es o~ one or mvLe ~
may also bP inver~ed. A~ this point, the system has produced a set of "frequency segments", all of which are now in the same low-frequency band, but which represent information originally contained in differen~ frequency lO bands. The signals in the respective frequency segments are then individually digitized and stored in a memory for digital processing. Such processing divides the frequency segments as a group into a set of successive "time blocks". Each of the frequency segments is thus divided into a succession of time segments, with each time block containing a number of these frequency-time segments equal to the number of frequency bands into which the original signal was divided.
Each such segment is read out of memory at a rate 20 which is n times the rate that it was written into memory, where n is equal to the number oE frequency-time segments in each block. This compresses each segment in time by a factor of l/n and expands it in frequency by a factor of n so that each segment now occupies l/n times 25 the time interval and has n times the bandwidth that it did prior to processing That is, it has a bandwidth equal to the entire bandwidth of the original input signal. Accordingly, each segment retains its original time-band~idth productO
For example, if the original audio channel has been divided into two frequency bands, each segment would be expanded in frequency by a factor of 2 and compressed in time by a factor of 2. Thus, when ail the segments derived from an input signal have been retrieved from 35 memory, they have collectively ~he same bandwidth and duration as the input signalD If the input signal were divided into five bands, the expansion and compression factor would be 5.
The time-compressed digi~ized segments are read out of memory under the control of a long, nonlinear pseudo-5 random key code developed by a key generator. The codedetermines the length of each time block and the order in which the segments within the block are retrieved.
Furthermore, each frequency-time segment may be reversed or not reversed in time as it is read out of memory 10 according to the key code. The digital signal thus obtained is converted to analog form for transmission to the reception site.
This scrambling process is repeated on successive increments of the audio signal being transmitted, the 15 audio signal being (1) divided into a plurality of frequency bands, (2) further segmented in time, and (3) compressed in time and expanded in frequency, with such time-compressed, digitally represented segments being reordered in time under control of the key generator, 20 returned to analog form and transmitted. Furthermore, the key code for encoding each successive audio signal block is developed pseudo-randomly so as to provide continual change in the encryption permutation appliecl to the input signal.
A receiving unit at the receiving site detects the transmitted analog signal and converts it to digital form. Then the above described scrambling process is carried out in reverse under control of a synchronized pseudo-random key generator and the recovered or clear 30 signal is reconverted from digital to analog ~orm so that it is intelligible to the authorized listenerO
Thus the compression time-wise and concomitant expansion in the frequency spectrum of selected frequency-time segments derived rom di~erent frequency 35 bands of the original audio signal, accompanied by the random time reversal and reordering of those segments in time, yields a transmitted signal which is extremely difficult to decode or break. Yet when the reverse algorithm is applied to the incoming signal at the receiving site, there results a received audio signal which contains substantially all of the intelligence of its original counterpart. Furthermore, since the transmitted signal has the same time-bandwidth product as the original audio signal, the transmitted signal requires no wider bandwidth than is required for the original audio signal.
To sum-marize, according to a first broad aspect of the present invention, there is provided a privacy communication system comprising A. means for splitting a voice signal into a selected number of frequency bands; B. means for generating a pseudo-random key word; C. means for dividing each said band into segments of different time duration in accordance with the pseudo-random key word; D. means for compressing said segments in time and expanding them in frequency by a factor equal to the number of frequency bands into which the voice signal was divided; and E. means for trans-mitting said time-compressed segments, the -transmitted signal having substantially the same time-bandwidth product as -the voice signal.
According to a second broad aspect of the present inven-tion, there is provided a privacy communication system comprising A. means for dividing a voice signal into signal segments; B. means for expanding said segments in time and compressing them in fre-quency; C. means for generating a pseudo-random key work; D. means for shifting said segments in frequency and in -time in accordance with the pseudo-random key word so as to stack said segments so that they span a frequency band which is approximately the same as the bandwidth of the voice signal; and E. means for transmitting the ~.
6a stacked segments, said transmitted signal having substantially the same time-bandwidth product as the voice signal.
Brief Description of the Drawings For a fuller understanding of the nature and objects of the invention, reference should be had -to the following detailed description, taken in connection with the accompanying drawings, in which:
Fig. 1 is a functional block diagram of one end of the privacy communication system made in accordance with this invention;
Fig. 2 is a detailed block diagram of the scrambling circuitry used in the system of Fig. l; and Figure 3 is a diagrammatic view illustrating the operation of the Fig. 1 system.
Description of the Preferred Embodiment _ Refer now to Fig. 1, which illustrates the transmitting end of a privacy communication system which accomplishes time/
frequency transformation in accordance with this invention. The system, shown generally at 10, is interposed be-tween an audio signal source 12, a microphone for example, and a transmitter 14.
The audio signal from source 12 is applied to a wave shaping network 16 for filtering out signal components which may lie outside the bandwidth of the transmission channel. For example, when speech is to be transmitted, the H1~-015 network 16 may include filter elements which attenuate at frequencies below about 300 Hz and above about 2500 Hz.
After passage through the network 1~, the audio signal is applied to bandpass filters 15 and 20, which 5 divide the signal into two bands. In the illustrated system, filter 18 passes only those frequency components of the audio signal from about 300 Hz to 1150 Hz.
The filter 20 passes only those frequency components from about 1550 up to 2500 Hz. The audio signal is thus 10 split by the system 10 into a high frequency Band A and a low frequency Band B. A gap is thus provided between the two passbands to form a guard band so that effective filtering of the two passbands takes place. This improves the quality of the transmitted signal without 15 appreciably degrading its information content.
The high frequency band signal passed by the filter 20 is frequency shifted approximately to the lower band. It is also inverted in frequency so that the frequency distribution of the inverted form of that 20 higher frequency band resembles that of the lower frequency band. The resultant signal has an amplitude distribution that is more regular than that of a normal speech signal. Therefore it is more di~ficult for an unauthorized listener to extract cadence information from 25 a transmission that may facilitate "breaking" the transmission.
To effect the frequency translation and inversion, the output of filter 20 is applied to a balanced modulator 22 where it is modulated wi~h a square wave 30 derived from a system clock 24. For voice signals o~ the type herein involved, the frequency of the square wave is selected to be approximately 2700 Hz. The output of modulator 22 is applied to a low-pass filter 28 which passes only the lower sideband portion of that signal 35 which is a replica of signal component in ~and Al but extending in frequency from 1150 down to 300 ~iz.
The signals from filters 18 and 28 are both applied to an analog gate 32. Gate 32 is gated by a signal from a system clock so that the gate alternatel~ passes the voltages ~rom the filters 18 and 28. In the present 5 instance, a sampling rate of 3.90625 KHz (hereinafter rounded off to 3.9 KHz) is employed. In other words, the clock signal applied to gate 32 is about 3.9 KHz. Thus, there appears at the output of gate 32 a string of vol~age samples ~hich are selected alternately from 10 Band A and Band B. These samples are applied to an analog-to digital converter 36 which converts each successive sample into a serial 8-bit binary representation. The digital output from converter 36 is then applied by way of a serial-to-parallel converter 38 15 to a conventional digital random access memory 46.
Typically, memory has 409S locations, each location containing a signal sample.
When a secure transmission commences, the system generates a series of WRITE signals. At the first such 20 signal, a WRITE address is applied to the memory 46 to address the first location in memory 46 so that the first 8-bit signal sample derived ~rom fre~uency Band A
is loaded into that section location. At the next WRITE
signal, the next WRITE address is applied to memory ~6 25 and the first sample from Band B is written into the next memory 46 location. The third WRITE signal causes the next sample from Band A to be stored into the third memory ~6 location, and so on~
During writing, then, the successively addressed 30 locations in memory 46 receive samples alternately ~rom Band A and Band B, corresponding to the successively sampled voltages appearing at the OUtpl~t o~ gate 32.
Since gate 32 samples each band at a 3.9 KHz rate, e.g.
every 256 microseconds, an 8-bit word represen~ing a 35 voltage sample ~rom either Band A or Band B is loaded ~14-015 into memory 46 a~ a rate of 7.8 KHz or every 128 microseconds.
In accordance with the illustrated system, after sufficient data has been written into memory 46, the 5 system initiates a READ routine and generates a series of READ signals. ~pon the occurrence of each such signal, an 8-bit word is read out of memory 46. The WRITE and READ signals alternate so that data is retrieved from memory concurrently with the storage of new signal data.
The successive digitized signal samples read out of memory 46 are applied via a parallel-to~serial converter 56 to a digital-to-analog converter 58.
Converter 58 converts the successive samples to successive voltages that are applied to a lowpass 15 filter 6~. The output of the filter 62 is a scrambled audio signal that is fed to the transmitter 1~ for transmission to a remote receiver.
The signal scrambling accomplished by the illustrated embodiment of the invention takes place in 20 the R~AD routines that retrieve the stored signal samples from the memory 46. During these routines, signals derived from frequency Bands A and B are retrieved from memory 46 in a pseudo-random fashion with respect to (a) the lengths or time durations of the segments (b) the 25 order of the segments, i.e. from Band A and then Band B, or vice versa, and (c) their direction, i.e. forward with time or reversed~
More specifically, retrieval from memory is divided into blocks, each of which, in the present example, 30 contains two equal length segments re~resen~ing the Band A and Band B components o the original audio signal. Since the blocks can have different lengths, it is obvious that while the segments in a given block are of equal length, the segments in different blocks can 35 have different lengths.
Since the samples in each segment are read out of memory 46 in succession (whereas they were loaded into the memor~ alternately with samples from the other frequency band), they come out for transmission at a rate 5 which is twice the read-in rate, i.e~ 7.8 KHz vice 3.9 KHz. This compresses each segment in time by a factor of two and doubles its frequency. However, the tirne-bandwidth product is the same as that of the original audio signal component. This is ill~strated 10 diagrammatically in FIG. 3. As shown there, the original audio signal is divided into two bands, i.e. Band A and Band B. Band A is divided into sample time-segments Al, A2, A3, ...AN which may be of different lengths. sand B
is likewise divided into time segments Bl, B2, B3~ ...BN.
15 During the READ routines, corresponding segments from each band are compressed in time and expanded in freq~ency and comprise successive blocks, i.e. segments Al, and Bl, form Block l; segments A2 and B2 comprise Block 2, and so on. Those blocks are then transmitted.
20 Each block and the segments it comprises can have different lengths as shown. Also the two segments forming each block can be read out for transmission in either order and in the forward or reverse direction timewise as shown by the arrows in FIG. 3. Thus, for 25 example, in Block 1, the segment Al, is transmitted in the forward direction, followed by segment 81 the reverse direction. In Block 2, segment B2 is transmitted before segment A2 and both segments are transmitted in the forward direction. Also, as is apparent frorn FIG. 3, 30 Block 2 is shorter than Block 1. The remaining blocks comprising the audio signal are formed and transmitted in a similar ~ashion. FIG. 3 shows that, in all cases, each segment has the same time-bandwidth product after scrambling as it did be~ore scrambling.
In the present system, the segment read~out order and direction as well as the length of each segment are under the control of a random number key code generator 66. More particularly, before each block of samples is stored in memory 46 for transmission, a new five-bit pseudo-random number from the generator 66 is 5 applied to various circuit elements to determine the manner in which ~he segments in that block are to be arranged. The first bit of the key number determines the order in which the two segments comprising that block are to be read out. If the first bit has one value, say, 10 ZERO, then the Band A segment is read out before the Band B segment in that block, i.e., see Block 1 in FIG. 3. On the other hand, if the first bit has a value ONE, then the Band B segment is read out before the ~and A seg~ent, viz. Block 2 in FIG. 3.
The second and third bits in the pseudo-random number produced by generator 66 determine the block length. With two bits, four different block lengths can be selected. In the present system, the four block lengths are 1024, 1280, 1536 and 1792 signal samples.
Bits 4 and 5 of the pseudo-random number determine the direction in which each of the segments in the block will be read out of memory ~6, i.e. in the forward or reverse direction. That is; if bit 4 is a ZER~, the first segment of the block is read out forwardly as shown 25 in Block 1 in FIG. 3; if it is a ONE, that segment is read out in reverse as depicted in Block 3. Similarly, the second segment is read out forwaldly or in reverse depending upon whether bit 5 is a ZERO or a ONE.
~s is customary in secure communications systems of 30 this general type, an initial synchronization signal is transmitted in order to synchronize the section of the system at the transmitter with the comparable section at the receiver. Also, at that time, the system is initialized to reset the various components o~ the 35 system. Then the WRITE routine commences as described above so that a succession of Band R and Band B signal samples are loaded alternately into successive addresses in memory 46.
FIG. 2 depicts in block form a circuit that operates in accordance with the flow chart of FIGo 2 during signal 5 transmission. The clock 24 of F~G. 1 continually provides alternate READ and WRITE signals that condition the circuitry for alternately writing signal samples into the memory 46 and retrieving them from the memory. The clock, which is of conventional design, also provides, 10 during the read and write intervals, a series of phase pulses Pl, P2, etc. A control logic unit 68 provides the control signals for the various other units in FIG. 2.
For example, it provides the "load" signals for the counters and registers, as well as other signals 15 described below. It comprises a conventional assemblage of flip-flops and gating circuits that pass various timing pulses from the clock 24 and other signals whose timing is derived in part from the clock. The details of these circuits will not add to an understanding of the 20 invention and they are omitted from this description for the sake of clarity.
The memory 46 is addressed by a write counter 70 for the write operations. For read operations, it is addressed by a concatenation of the read counter 72 and a ~5 single bit from the logic unit 68 as described below.
The contents of the counters 70 and 72 are applied to the address port of the memory 46 by way of a multiplexer 74 under control of the READ and WRITE signals. That is, when the WRITE signal is asserted, the multiplexer 30 connects the write counter 70 to the address port and when the WRITE signal is not asserted, i.e. when the READ
signal is asserted, the multiplexer 74 connects the read counter 72 and the single bit from logic unit 68 ~o the memory ~6 address port. The memory 46 is connected to a 35 data bus 73 to which the serial-to-parallel converter 38 and the parallel-to-serial converter 58 are also ~z~
connected. The memory and the converter 58 are co~pled to the bus 73 by the WRITE signal. The READ signal and a P pulse cause the converter 58 to receive data from the bus 73 during the read operations of the memory. The 5 WRITE signal is also used as a read/write control signal for the memory 46. Each memory operation is triggered by a Pl pulse from the clock 42.
In accordance with the algorithm described above, a FIFO register ~L receives the two bits from generator 66 10 indicating the block length for readout operations; a second FIFO register RS receives random number bits 1, 4 and 5 that determine segment sequence and reversal of the readout operations for each block; and a third FIFO
register ~NB receives the four most significant bits of 15 the beginning address of each block being written into the memory ~6.
The binary representations of the four block lengths listed above are provided directly by the two block length bits in the register BL. Speci~ically, the latter 20 bits are used as bits 8 and 9 in the block length number, a ONE is inserted as bit 10 and the lower order bits are all ZEROS. For a memory capacity of 4096 locations, 12 bits (i.e. bits 0-11) are needed for the memor~ address function. The beginning address of each block will 25 contain ZEROS for bits 0-7. Accordingly, generation of beginning addresses of blocks and other operations relating to beginning addresses involve only the four most significant bits, i.e. bits 8-11. Therefore, these operations, as well as storage of the beginning 30 addresses, can be performed by 4-bit circuitry.
The circuit depicted in FIG. 2 operates as follows.
On synchronization, the various registers are cleared, the first pseudo-random number from the generator 66 is loaded into the registers BL and RS and, the block length 35 is loaded in~o a block counter 75. The BNB FIFO register contains the beginning address of the first block, i.e.
~ 015 - 0000. In response to clock pulses the write counter 70 provides a succession of memory addresses for storage of the signal samples from the converter 38. However, the read operations are inhibited by the logic unit 68, e.g.
5 by preventing "load" pulses from reaching the counter 56.
The block counter 75 counts down in response to the pulses that advance the write counter 70. It thus reaches ZERO when the first block of samples has been stored in the memory 46. It thereupon emits an output 10 signal that causes the next random number to be loaded into the registers BL and RS, with the new block length from the generator 66 also being loaded into the counter 75. The same output signal also causes the count in write counter 70 to be loaded into the BNB register.
15 This address is the beginning address of the second block.
This operation continues indefinitely, with a succession of pseudo-random numbers being loaded into the registers BL and RS and block-beginning addresses being 20 loaded into the BNB register. The contents of these registers are used in the memory retrieval operations, which will now be described.
More specifically, the address in the BNB register, with the most significant bit inverted, is applied to a 25 comparator 76. The bit inversion effectively advances the address by 2048. Thus, with the BNB register initially cleared, the address 2048 is applied to the comparator 760 The other input to the comparator is the content of the write counter 70~
Accordingly, when 2048 signal samples have been stored in the memory 46, the counter 70 advances to a count of 2048 and the comparator 76 emits an output signal to the logic unit 68. The logic unit thereupon initiates the read operationsO
If bit 4 of the pseudo-random number contained in output stages of the BL and RS registers is a ZERO, ~ 915 indicating forward retrieval of the first segment in the block, ~he beginning address of the block is loaded into the read counter 72. Specifically, the counter 72 receives the output of an adder 78, which sums the 5 address in the output stage of the BNB register wi~h the output of a multiplexer 80. If random number bit 4 is a ZERO, an UP/DOWN signal from the logic unit 68 causes the multiplexer 80 to select zero as its input and the adder 78 thus passes the block beginning address in the 10 BNB register to the counter 12.
If random number bit 4 is a ONE, the last address in the block is loaded into the read counter 7~ for retrieval in the reverse direction. The state of the UP/DOWN signal causes the multiplexer 80 to select the 15 block ;ength from the BL register. However, the sum of the block length and the block beginning address, as provided by straight addition in the adder 78, is the beginning address of the next block. This can be converted to the last address in the present block in 20 either of two ways. One of these is to load the sum into the read counter 72 and then decrement the counter before the counter beyins addressing the memory 46. The second arrangement, accomplished by additional circuitry (not shown), is to subtract the block length increment (256 in 25 the present example) from the block length, the block beginning address or the sum o~ the two, and preset all the lower order bits in the counter 72 to ones.
The UP/DOWN signal from the logic unit 68 also controls the direction in which the read address 30 counter 72 counts. That is, if random number bit 4 is a ZERO, the UP/DO~N signal causes the counter to count up and if that bit is a ONE, it causes it to count down, thereby providing the required forward or reverse retrieval from the memory 46.
As mentioned above, the least significant bi~ o~ the memory address for retrieval operations is provided by
Back~round of the Invention This invention relates to a privacy communication system. It relates more particularly ~o an audio or voice scrambler in which a communication is rendered 5 unin~elligible so that its conten~ is unavailable to third parties, the communication being capable of being unscrambled after reception by an authorized person to recover the original voioe content.
Descri tion of the Prior Art P __ Privacy systems are in widespread use ~or renderlng audio signals, particularly voice signals, unintelligible for transmission over an exposed transmission link such as a telephone line ~o as to maintain ~he trans~ission priva~e, specifically to avoid reconstruction of the 15 voice content by unauthorized listeners. In such systems, the voice signals are typically encoded at a transmitting si~e using an encoding technique that involves scrambling or displacing the audio signals in the frequency domain, time domain or both. At the 20 receiving site, the scrambled signals are decoded by, in effect, reversing the encoding procedure to recover the original audio signals. Ideally, in any system o this type, the encoding technique used should rnake it extremely difficult for unauthorized listeners to decode 25 or "break" an intercepted scrambled signal, yet still permit recovery of the ~ransmitted informa~ion at the receiving site with good intelligibility and recognition by authorized listeners.
One proposed technique used to scramble voice 30 information, disclosed in U.S. Paten~s 3,921,151 and 3,970,790, for example, is to divide the ~ransmi~ted signal into segments of equal time duration. The individual ~ignal elemen~s are applied to a ~emory fot temporary storage. The stored segment~ are then read out 35 from memory in a random pattern as determined by a signal from a pseudo-random key code generator so as to scramble the order of the segments. They are transmitted in this scrambled order and momentarily stored at the receiving end, where the process is ef~ectively reversed to 5 descramble the signal elements. The former patent adds an additional element of randomness to the communication by time~reversing selected signal segments according to the output from the pseudo-random key code generator.
Attempts have been made to increase the security of 10 privacy systems by dividing the audio signal to be transmitted in frequency as well as in time to further scramble the transmitted signal. Arrangements such as this are disclosed in U.S. patents 4,149,035 and 4,221,931. Basically, the audio signal is divided into 15 two or more different frequency bands with the information in each band being digitized. Then each band is partitioned into a number of segments of equal time duration and these segments are transposed both within the same band and with segments in the other bands in 20 accordance with a code developed by a pseudo-random key generator. The signals are then converted back to analog form and combined for transmission to the receiving site, ~hich contains similar equipment ~or reversing the above procedure to recover the original audio siynal. The 25 system in the last-mentioned patent also timewise reverses certain of the signal segments in accordance with a second key code developed by the key generator to obtain an additional element of randomness in the scrambled signal. Even so, however, it may still be 30 possible for unauthori~ed listeners to decode or break an intercepted scrambled signal such as by analyzing the cadence and detec~ing recurring phenomena in the transmitted scrambled signal using present-day high~speed computers.
~ ~ ~Z~5 Hl4-015 Summary of the Invention It is therefore an object of the present invention to provide a privacy communication system having the ability to transmit scrambled audio signals with 5 increased security.
It is a further object of the invention to provide a system such as this in which the random nature of the scrambling of the audio signal is materially increased to reduce the periodicity of the transmitted signal 10 sequences.
Yet another object of the invention is to provide a secure communication system whose scrambled signals can be transmitted over a channel having no greater bandwidth than that required for the original audio signal.
A further object is to provide such a system which produces good quality in the received "clear" audio signal.
Still another object of the invention is to provide a secure communication system which is relatively simple 20 in construction, yet quite efficient and reliable in operation.
Other objects will, in part, be obvious and will, in part, appear hereinafter.
The invention accordingly comprises the features of 25 construction, combination of elements and arrangement of parts which will be exemplified in the Eollowing detailed description, and the scope of the invention will be indicated in the claims.
The present system has partic~lar applicability to 30 achieving high level communications security for audio bands in the speech frequency spectrum or channel extending from around 200 Hz to about 3000 Hz. To accomplish this, the sys~em divides ~he analog audio input signal into two or more frequency bands, preferably 35 of equal bandwidth. In general, one or more of the bands 5~
are transposed so that all the bands end up in the same low-frequency range. In the usual case, this is accomplished by separately transposiny the bands above the lowest frequency band to the frequency range o~ the 5 low ban~. ~he frequenc1es o~ one or mvLe ~
may also bP inver~ed. A~ this point, the system has produced a set of "frequency segments", all of which are now in the same low-frequency band, but which represent information originally contained in differen~ frequency lO bands. The signals in the respective frequency segments are then individually digitized and stored in a memory for digital processing. Such processing divides the frequency segments as a group into a set of successive "time blocks". Each of the frequency segments is thus divided into a succession of time segments, with each time block containing a number of these frequency-time segments equal to the number of frequency bands into which the original signal was divided.
Each such segment is read out of memory at a rate 20 which is n times the rate that it was written into memory, where n is equal to the number oE frequency-time segments in each block. This compresses each segment in time by a factor of l/n and expands it in frequency by a factor of n so that each segment now occupies l/n times 25 the time interval and has n times the bandwidth that it did prior to processing That is, it has a bandwidth equal to the entire bandwidth of the original input signal. Accordingly, each segment retains its original time-band~idth productO
For example, if the original audio channel has been divided into two frequency bands, each segment would be expanded in frequency by a factor of 2 and compressed in time by a factor of 2. Thus, when ail the segments derived from an input signal have been retrieved from 35 memory, they have collectively ~he same bandwidth and duration as the input signalD If the input signal were divided into five bands, the expansion and compression factor would be 5.
The time-compressed digi~ized segments are read out of memory under the control of a long, nonlinear pseudo-5 random key code developed by a key generator. The codedetermines the length of each time block and the order in which the segments within the block are retrieved.
Furthermore, each frequency-time segment may be reversed or not reversed in time as it is read out of memory 10 according to the key code. The digital signal thus obtained is converted to analog form for transmission to the reception site.
This scrambling process is repeated on successive increments of the audio signal being transmitted, the 15 audio signal being (1) divided into a plurality of frequency bands, (2) further segmented in time, and (3) compressed in time and expanded in frequency, with such time-compressed, digitally represented segments being reordered in time under control of the key generator, 20 returned to analog form and transmitted. Furthermore, the key code for encoding each successive audio signal block is developed pseudo-randomly so as to provide continual change in the encryption permutation appliecl to the input signal.
A receiving unit at the receiving site detects the transmitted analog signal and converts it to digital form. Then the above described scrambling process is carried out in reverse under control of a synchronized pseudo-random key generator and the recovered or clear 30 signal is reconverted from digital to analog ~orm so that it is intelligible to the authorized listenerO
Thus the compression time-wise and concomitant expansion in the frequency spectrum of selected frequency-time segments derived rom di~erent frequency 35 bands of the original audio signal, accompanied by the random time reversal and reordering of those segments in time, yields a transmitted signal which is extremely difficult to decode or break. Yet when the reverse algorithm is applied to the incoming signal at the receiving site, there results a received audio signal which contains substantially all of the intelligence of its original counterpart. Furthermore, since the transmitted signal has the same time-bandwidth product as the original audio signal, the transmitted signal requires no wider bandwidth than is required for the original audio signal.
To sum-marize, according to a first broad aspect of the present invention, there is provided a privacy communication system comprising A. means for splitting a voice signal into a selected number of frequency bands; B. means for generating a pseudo-random key word; C. means for dividing each said band into segments of different time duration in accordance with the pseudo-random key word; D. means for compressing said segments in time and expanding them in frequency by a factor equal to the number of frequency bands into which the voice signal was divided; and E. means for trans-mitting said time-compressed segments, the -transmitted signal having substantially the same time-bandwidth product as -the voice signal.
According to a second broad aspect of the present inven-tion, there is provided a privacy communication system comprising A. means for dividing a voice signal into signal segments; B. means for expanding said segments in time and compressing them in fre-quency; C. means for generating a pseudo-random key work; D. means for shifting said segments in frequency and in -time in accordance with the pseudo-random key word so as to stack said segments so that they span a frequency band which is approximately the same as the bandwidth of the voice signal; and E. means for transmitting the ~.
6a stacked segments, said transmitted signal having substantially the same time-bandwidth product as the voice signal.
Brief Description of the Drawings For a fuller understanding of the nature and objects of the invention, reference should be had -to the following detailed description, taken in connection with the accompanying drawings, in which:
Fig. 1 is a functional block diagram of one end of the privacy communication system made in accordance with this invention;
Fig. 2 is a detailed block diagram of the scrambling circuitry used in the system of Fig. l; and Figure 3 is a diagrammatic view illustrating the operation of the Fig. 1 system.
Description of the Preferred Embodiment _ Refer now to Fig. 1, which illustrates the transmitting end of a privacy communication system which accomplishes time/
frequency transformation in accordance with this invention. The system, shown generally at 10, is interposed be-tween an audio signal source 12, a microphone for example, and a transmitter 14.
The audio signal from source 12 is applied to a wave shaping network 16 for filtering out signal components which may lie outside the bandwidth of the transmission channel. For example, when speech is to be transmitted, the H1~-015 network 16 may include filter elements which attenuate at frequencies below about 300 Hz and above about 2500 Hz.
After passage through the network 1~, the audio signal is applied to bandpass filters 15 and 20, which 5 divide the signal into two bands. In the illustrated system, filter 18 passes only those frequency components of the audio signal from about 300 Hz to 1150 Hz.
The filter 20 passes only those frequency components from about 1550 up to 2500 Hz. The audio signal is thus 10 split by the system 10 into a high frequency Band A and a low frequency Band B. A gap is thus provided between the two passbands to form a guard band so that effective filtering of the two passbands takes place. This improves the quality of the transmitted signal without 15 appreciably degrading its information content.
The high frequency band signal passed by the filter 20 is frequency shifted approximately to the lower band. It is also inverted in frequency so that the frequency distribution of the inverted form of that 20 higher frequency band resembles that of the lower frequency band. The resultant signal has an amplitude distribution that is more regular than that of a normal speech signal. Therefore it is more di~ficult for an unauthorized listener to extract cadence information from 25 a transmission that may facilitate "breaking" the transmission.
To effect the frequency translation and inversion, the output of filter 20 is applied to a balanced modulator 22 where it is modulated wi~h a square wave 30 derived from a system clock 24. For voice signals o~ the type herein involved, the frequency of the square wave is selected to be approximately 2700 Hz. The output of modulator 22 is applied to a low-pass filter 28 which passes only the lower sideband portion of that signal 35 which is a replica of signal component in ~and Al but extending in frequency from 1150 down to 300 ~iz.
The signals from filters 18 and 28 are both applied to an analog gate 32. Gate 32 is gated by a signal from a system clock so that the gate alternatel~ passes the voltages ~rom the filters 18 and 28. In the present 5 instance, a sampling rate of 3.90625 KHz (hereinafter rounded off to 3.9 KHz) is employed. In other words, the clock signal applied to gate 32 is about 3.9 KHz. Thus, there appears at the output of gate 32 a string of vol~age samples ~hich are selected alternately from 10 Band A and Band B. These samples are applied to an analog-to digital converter 36 which converts each successive sample into a serial 8-bit binary representation. The digital output from converter 36 is then applied by way of a serial-to-parallel converter 38 15 to a conventional digital random access memory 46.
Typically, memory has 409S locations, each location containing a signal sample.
When a secure transmission commences, the system generates a series of WRITE signals. At the first such 20 signal, a WRITE address is applied to the memory 46 to address the first location in memory 46 so that the first 8-bit signal sample derived ~rom fre~uency Band A
is loaded into that section location. At the next WRITE
signal, the next WRITE address is applied to memory ~6 25 and the first sample from Band B is written into the next memory 46 location. The third WRITE signal causes the next sample from Band A to be stored into the third memory ~6 location, and so on~
During writing, then, the successively addressed 30 locations in memory 46 receive samples alternately ~rom Band A and Band B, corresponding to the successively sampled voltages appearing at the OUtpl~t o~ gate 32.
Since gate 32 samples each band at a 3.9 KHz rate, e.g.
every 256 microseconds, an 8-bit word represen~ing a 35 voltage sample ~rom either Band A or Band B is loaded ~14-015 into memory 46 a~ a rate of 7.8 KHz or every 128 microseconds.
In accordance with the illustrated system, after sufficient data has been written into memory 46, the 5 system initiates a READ routine and generates a series of READ signals. ~pon the occurrence of each such signal, an 8-bit word is read out of memory 46. The WRITE and READ signals alternate so that data is retrieved from memory concurrently with the storage of new signal data.
The successive digitized signal samples read out of memory 46 are applied via a parallel-to~serial converter 56 to a digital-to-analog converter 58.
Converter 58 converts the successive samples to successive voltages that are applied to a lowpass 15 filter 6~. The output of the filter 62 is a scrambled audio signal that is fed to the transmitter 1~ for transmission to a remote receiver.
The signal scrambling accomplished by the illustrated embodiment of the invention takes place in 20 the R~AD routines that retrieve the stored signal samples from the memory 46. During these routines, signals derived from frequency Bands A and B are retrieved from memory 46 in a pseudo-random fashion with respect to (a) the lengths or time durations of the segments (b) the 25 order of the segments, i.e. from Band A and then Band B, or vice versa, and (c) their direction, i.e. forward with time or reversed~
More specifically, retrieval from memory is divided into blocks, each of which, in the present example, 30 contains two equal length segments re~resen~ing the Band A and Band B components o the original audio signal. Since the blocks can have different lengths, it is obvious that while the segments in a given block are of equal length, the segments in different blocks can 35 have different lengths.
Since the samples in each segment are read out of memory 46 in succession (whereas they were loaded into the memor~ alternately with samples from the other frequency band), they come out for transmission at a rate 5 which is twice the read-in rate, i.e~ 7.8 KHz vice 3.9 KHz. This compresses each segment in time by a factor of two and doubles its frequency. However, the tirne-bandwidth product is the same as that of the original audio signal component. This is ill~strated 10 diagrammatically in FIG. 3. As shown there, the original audio signal is divided into two bands, i.e. Band A and Band B. Band A is divided into sample time-segments Al, A2, A3, ...AN which may be of different lengths. sand B
is likewise divided into time segments Bl, B2, B3~ ...BN.
15 During the READ routines, corresponding segments from each band are compressed in time and expanded in freq~ency and comprise successive blocks, i.e. segments Al, and Bl, form Block l; segments A2 and B2 comprise Block 2, and so on. Those blocks are then transmitted.
20 Each block and the segments it comprises can have different lengths as shown. Also the two segments forming each block can be read out for transmission in either order and in the forward or reverse direction timewise as shown by the arrows in FIG. 3. Thus, for 25 example, in Block 1, the segment Al, is transmitted in the forward direction, followed by segment 81 the reverse direction. In Block 2, segment B2 is transmitted before segment A2 and both segments are transmitted in the forward direction. Also, as is apparent frorn FIG. 3, 30 Block 2 is shorter than Block 1. The remaining blocks comprising the audio signal are formed and transmitted in a similar ~ashion. FIG. 3 shows that, in all cases, each segment has the same time-bandwidth product after scrambling as it did be~ore scrambling.
In the present system, the segment read~out order and direction as well as the length of each segment are under the control of a random number key code generator 66. More particularly, before each block of samples is stored in memory 46 for transmission, a new five-bit pseudo-random number from the generator 66 is 5 applied to various circuit elements to determine the manner in which ~he segments in that block are to be arranged. The first bit of the key number determines the order in which the two segments comprising that block are to be read out. If the first bit has one value, say, 10 ZERO, then the Band A segment is read out before the Band B segment in that block, i.e., see Block 1 in FIG. 3. On the other hand, if the first bit has a value ONE, then the Band B segment is read out before the ~and A seg~ent, viz. Block 2 in FIG. 3.
The second and third bits in the pseudo-random number produced by generator 66 determine the block length. With two bits, four different block lengths can be selected. In the present system, the four block lengths are 1024, 1280, 1536 and 1792 signal samples.
Bits 4 and 5 of the pseudo-random number determine the direction in which each of the segments in the block will be read out of memory ~6, i.e. in the forward or reverse direction. That is; if bit 4 is a ZER~, the first segment of the block is read out forwardly as shown 25 in Block 1 in FIG. 3; if it is a ONE, that segment is read out in reverse as depicted in Block 3. Similarly, the second segment is read out forwaldly or in reverse depending upon whether bit 5 is a ZERO or a ONE.
~s is customary in secure communications systems of 30 this general type, an initial synchronization signal is transmitted in order to synchronize the section of the system at the transmitter with the comparable section at the receiver. Also, at that time, the system is initialized to reset the various components o~ the 35 system. Then the WRITE routine commences as described above so that a succession of Band R and Band B signal samples are loaded alternately into successive addresses in memory 46.
FIG. 2 depicts in block form a circuit that operates in accordance with the flow chart of FIGo 2 during signal 5 transmission. The clock 24 of F~G. 1 continually provides alternate READ and WRITE signals that condition the circuitry for alternately writing signal samples into the memory 46 and retrieving them from the memory. The clock, which is of conventional design, also provides, 10 during the read and write intervals, a series of phase pulses Pl, P2, etc. A control logic unit 68 provides the control signals for the various other units in FIG. 2.
For example, it provides the "load" signals for the counters and registers, as well as other signals 15 described below. It comprises a conventional assemblage of flip-flops and gating circuits that pass various timing pulses from the clock 24 and other signals whose timing is derived in part from the clock. The details of these circuits will not add to an understanding of the 20 invention and they are omitted from this description for the sake of clarity.
The memory 46 is addressed by a write counter 70 for the write operations. For read operations, it is addressed by a concatenation of the read counter 72 and a ~5 single bit from the logic unit 68 as described below.
The contents of the counters 70 and 72 are applied to the address port of the memory 46 by way of a multiplexer 74 under control of the READ and WRITE signals. That is, when the WRITE signal is asserted, the multiplexer 30 connects the write counter 70 to the address port and when the WRITE signal is not asserted, i.e. when the READ
signal is asserted, the multiplexer 74 connects the read counter 72 and the single bit from logic unit 68 ~o the memory ~6 address port. The memory 46 is connected to a 35 data bus 73 to which the serial-to-parallel converter 38 and the parallel-to-serial converter 58 are also ~z~
connected. The memory and the converter 58 are co~pled to the bus 73 by the WRITE signal. The READ signal and a P pulse cause the converter 58 to receive data from the bus 73 during the read operations of the memory. The 5 WRITE signal is also used as a read/write control signal for the memory 46. Each memory operation is triggered by a Pl pulse from the clock 42.
In accordance with the algorithm described above, a FIFO register ~L receives the two bits from generator 66 10 indicating the block length for readout operations; a second FIFO register RS receives random number bits 1, 4 and 5 that determine segment sequence and reversal of the readout operations for each block; and a third FIFO
register ~NB receives the four most significant bits of 15 the beginning address of each block being written into the memory ~6.
The binary representations of the four block lengths listed above are provided directly by the two block length bits in the register BL. Speci~ically, the latter 20 bits are used as bits 8 and 9 in the block length number, a ONE is inserted as bit 10 and the lower order bits are all ZEROS. For a memory capacity of 4096 locations, 12 bits (i.e. bits 0-11) are needed for the memor~ address function. The beginning address of each block will 25 contain ZEROS for bits 0-7. Accordingly, generation of beginning addresses of blocks and other operations relating to beginning addresses involve only the four most significant bits, i.e. bits 8-11. Therefore, these operations, as well as storage of the beginning 30 addresses, can be performed by 4-bit circuitry.
The circuit depicted in FIG. 2 operates as follows.
On synchronization, the various registers are cleared, the first pseudo-random number from the generator 66 is loaded into the registers BL and RS and, the block length 35 is loaded in~o a block counter 75. The BNB FIFO register contains the beginning address of the first block, i.e.
~ 015 - 0000. In response to clock pulses the write counter 70 provides a succession of memory addresses for storage of the signal samples from the converter 38. However, the read operations are inhibited by the logic unit 68, e.g.
5 by preventing "load" pulses from reaching the counter 56.
The block counter 75 counts down in response to the pulses that advance the write counter 70. It thus reaches ZERO when the first block of samples has been stored in the memory 46. It thereupon emits an output 10 signal that causes the next random number to be loaded into the registers BL and RS, with the new block length from the generator 66 also being loaded into the counter 75. The same output signal also causes the count in write counter 70 to be loaded into the BNB register.
15 This address is the beginning address of the second block.
This operation continues indefinitely, with a succession of pseudo-random numbers being loaded into the registers BL and RS and block-beginning addresses being 20 loaded into the BNB register. The contents of these registers are used in the memory retrieval operations, which will now be described.
More specifically, the address in the BNB register, with the most significant bit inverted, is applied to a 25 comparator 76. The bit inversion effectively advances the address by 2048. Thus, with the BNB register initially cleared, the address 2048 is applied to the comparator 760 The other input to the comparator is the content of the write counter 70~
Accordingly, when 2048 signal samples have been stored in the memory 46, the counter 70 advances to a count of 2048 and the comparator 76 emits an output signal to the logic unit 68. The logic unit thereupon initiates the read operationsO
If bit 4 of the pseudo-random number contained in output stages of the BL and RS registers is a ZERO, ~ 915 indicating forward retrieval of the first segment in the block, ~he beginning address of the block is loaded into the read counter 72. Specifically, the counter 72 receives the output of an adder 78, which sums the 5 address in the output stage of the BNB register wi~h the output of a multiplexer 80. If random number bit 4 is a ZERO, an UP/DOWN signal from the logic unit 68 causes the multiplexer 80 to select zero as its input and the adder 78 thus passes the block beginning address in the 10 BNB register to the counter 12.
If random number bit 4 is a ONE, the last address in the block is loaded into the read counter 7~ for retrieval in the reverse direction. The state of the UP/DOWN signal causes the multiplexer 80 to select the 15 block ;ength from the BL register. However, the sum of the block length and the block beginning address, as provided by straight addition in the adder 78, is the beginning address of the next block. This can be converted to the last address in the present block in 20 either of two ways. One of these is to load the sum into the read counter 72 and then decrement the counter before the counter beyins addressing the memory 46. The second arrangement, accomplished by additional circuitry (not shown), is to subtract the block length increment (256 in 25 the present example) from the block length, the block beginning address or the sum o~ the two, and preset all the lower order bits in the counter 72 to ones.
The UP/DOWN signal from the logic unit 68 also controls the direction in which the read address 30 counter 72 counts. That is, if random number bit 4 is a ZERO, the UP/DO~N signal causes the counter to count up and if that bit is a ONE, it causes it to count down, thereby providing the required forward or reverse retrieval from the memory 46.
As mentioned above, the least significant bi~ o~ the memory address for retrieval operations is provided by
2~i~5 1~
the logic unit 68. This bit is derived from bit 1 of the random number, as contained in the RS register~ and specifically if this bit is a ZERO, the first segment of the block being retrieved from memory will be the A
5 segment, which is contained in even memory addresses.
Accordingly the least significant bit (LSB) provided by the logic unit 68 is a Z~RO. Conversely, i~ random number bit 1 is a ONE, the least significant bit of the memory address will be a ONE during retrieval of the 10 first segment, thereby ~roviding retrieval of the B
segment, which resides in the odd numbered addresses.
Also, at this time, a read counter 83 is loaded with one-half the block length.
The read operation begins immediately and runs in 15 synchronism with the write operation. Each time the read address counter 72 is incremented or decremented to provide a new memory address, it changes the address by a count of 2, thereby providing only even or odd addresses, in accordance with random number bit 1. At the same 20 time, the read counter 83 is decremented. When the read counter reaches the end of the segment A or B being retrieved from the memory 46, the write counter 83 will have counted down to zero. The res~lting signal from the counter 83 is applied to the control logic unit 68 which 25 responds by setting up the read address counter 72 to retrieve the second segment in the block.
Specifically, the logic unit generates a new UP/DOWN
signal corresponding to bit 5 of the random number and accordingly causes the read address counter 72 to be 30 loaded with the beginning address or the last address of the block, depending on whether bit 5 indicates that the second segment of the block is to be re~rieved in the forward or reverse direction. The logic unit 68 also inverts the least significant bit of the memor~ address 35 so that if the segment contained in the even num~ered addresses of the block was the first segment retrieved, ~ ~ ~Z55~
H1~-015 the segment contained in the odd numbered addresses will now be retrieved, and vice versa. The retrieval operation then continues so as to retrieve from the memory 46 ~he second segment in the block.
When retrieval of the second segment has been completed, the write operation ~ill have addressed 20~8 locations beyond the end of the block from which data is being retrieved and the comparator 76 will therefore emit 10 another output signal. This siqnal causes the logic unit to shift the next pseudo-random number to the output stages o~ the BL and RS registers and shift the beginning address of the next block to be retrieved to the output stage of the BNB register. The logic unit 68 then 15 operates as described above to initiate retrieval of the first segment of the next block.
This sequence continues as long as the secure transmission persists. That is, the system writes sample data into memory 46 alternately from Band A and Band 20 and then retrieves that data out of the memory and transmits it with a system delay corresponding to 2048 samples. The samples are retrieved from the memory in successive data blocks having one of four selected lengths as determined by the key word produced by 25 generator 66 prior to transmission of each new block.
Furthermore, the same key word determines the order in which the Band A and Band B data segments comprising that block are transmitted and whether each such segment is read out and transmitted in the forward or reverse 30 direction. Resultantly, the transmission ~rom system 10 is reasonably secure. Yet the transmit~ed scrambled signal can be recovered easily at the receiving location by a comparable system operating in synchronism with generator 66.
It will be apparent from the foregoing that a larger number of bits in the key word will permit a larger 1~
selection of possible block lengths. It should also be appreciated that the audio signal can be divided into more than two ~requency bands. Furthermore, the segments in adjacent frequency bands can have coincident-in time 5 boundaries, as described herein, or the different bands may be asynchronously divided into segments that do not coincide in time. Still fur~her, the signal samples can be scrambled as thel are being read into memory section ~2 rather than during readout as specifically 10 described herein.
It is also quite feasible to reverse the order of the processes that are carried out to encrypt the audio signal to be transmitted. For example, the audio signal can be partitioned into segments of different time 15 duration before it is divided into different ~requency bands rather than the reverse as specifically described above. In other words, a~ter the entire frequency spectrum of the audio signal is divided into segments of different time duration, each such segment can be 20 digitized and stored in memory. The stored in~ormation may then be read out of memory at a slower rate equal to the reciprocal of the rate at which it was read into the memory. This effectively expands each signal segments in time and compresses it in the frequency domain.
On read-out the segments are permuted according to a pseudo-random key code from the key generator 66 and converted to analog form with sr without time reversal.
After such conversion, some of the segments are shifted ~ to different frequency ~ands thereby to permit the 30 segments to be stacked frequency-wise so as to fill the entire audio frequency spectrum and time domain of the audio signal being transmitted. This composite signal and each segment thereof also has the same time-bandwidth product as the original audio signal and each segment 35 thereof. The composite signal may then be transmitted to a receiving unit which applies a reverse algorithm to the received signal to recover the intelligence in the transmission.
Also, to further increase the difficulty of decoding the transmission, ~he signal sequences in the different 5 frequency bands may not be stacked into columns of synchronous time-wise segments. Instead, the sequences in adjacent bands may be shifted in time to provide an asynchronous transmission of the signal sequences from the different frequency bands.
By the same token, different commutations and permutations of the aforementioned signal processing steps may be carried out on the signal to be transmitted.
Thus the original analog audio signal may be time-divided into full spectrum segments of different time duration as 15 described above. Then, some of these full spectrum segments can be frequency divided into partial spectrum signal sequences. The analog information may then be digitized and stored in memory. The full spectrum segments and the partial spectrum sequences, up~n 20 readout~ are time-wise reversed and permutated according to the pseudo-random key code from the key generator.
Also during readout some of the segments and/or sequences can be expanded or compressed in the time domain, thereby to respectively compress or expand their bandwidth so ~5 that each segment retains its original bandwidth-time product.
It will thus be seen that the objects set forth above, among those made apparent from the preceding description, are efficiently attained, and, since certain 30 chanses may be made in the akove construction without departing from the scope of the invention7 it is intended that all matter contained in the above description or shown in the accompanying drawings be interpreted as illustrative and not in a limiting sense.
Z5~
It is also to be unders~ood that the following claims are intended to cover all of the generic and specific features of ~he invention herein describedO
the logic unit 68. This bit is derived from bit 1 of the random number, as contained in the RS register~ and specifically if this bit is a ZERO, the first segment of the block being retrieved from memory will be the A
5 segment, which is contained in even memory addresses.
Accordingly the least significant bit (LSB) provided by the logic unit 68 is a Z~RO. Conversely, i~ random number bit 1 is a ONE, the least significant bit of the memory address will be a ONE during retrieval of the 10 first segment, thereby ~roviding retrieval of the B
segment, which resides in the odd numbered addresses.
Also, at this time, a read counter 83 is loaded with one-half the block length.
The read operation begins immediately and runs in 15 synchronism with the write operation. Each time the read address counter 72 is incremented or decremented to provide a new memory address, it changes the address by a count of 2, thereby providing only even or odd addresses, in accordance with random number bit 1. At the same 20 time, the read counter 83 is decremented. When the read counter reaches the end of the segment A or B being retrieved from the memory 46, the write counter 83 will have counted down to zero. The res~lting signal from the counter 83 is applied to the control logic unit 68 which 25 responds by setting up the read address counter 72 to retrieve the second segment in the block.
Specifically, the logic unit generates a new UP/DOWN
signal corresponding to bit 5 of the random number and accordingly causes the read address counter 72 to be 30 loaded with the beginning address or the last address of the block, depending on whether bit 5 indicates that the second segment of the block is to be re~rieved in the forward or reverse direction. The logic unit 68 also inverts the least significant bit of the memor~ address 35 so that if the segment contained in the even num~ered addresses of the block was the first segment retrieved, ~ ~ ~Z55~
H1~-015 the segment contained in the odd numbered addresses will now be retrieved, and vice versa. The retrieval operation then continues so as to retrieve from the memory 46 ~he second segment in the block.
When retrieval of the second segment has been completed, the write operation ~ill have addressed 20~8 locations beyond the end of the block from which data is being retrieved and the comparator 76 will therefore emit 10 another output signal. This siqnal causes the logic unit to shift the next pseudo-random number to the output stages o~ the BL and RS registers and shift the beginning address of the next block to be retrieved to the output stage of the BNB register. The logic unit 68 then 15 operates as described above to initiate retrieval of the first segment of the next block.
This sequence continues as long as the secure transmission persists. That is, the system writes sample data into memory 46 alternately from Band A and Band 20 and then retrieves that data out of the memory and transmits it with a system delay corresponding to 2048 samples. The samples are retrieved from the memory in successive data blocks having one of four selected lengths as determined by the key word produced by 25 generator 66 prior to transmission of each new block.
Furthermore, the same key word determines the order in which the Band A and Band B data segments comprising that block are transmitted and whether each such segment is read out and transmitted in the forward or reverse 30 direction. Resultantly, the transmission ~rom system 10 is reasonably secure. Yet the transmit~ed scrambled signal can be recovered easily at the receiving location by a comparable system operating in synchronism with generator 66.
It will be apparent from the foregoing that a larger number of bits in the key word will permit a larger 1~
selection of possible block lengths. It should also be appreciated that the audio signal can be divided into more than two ~requency bands. Furthermore, the segments in adjacent frequency bands can have coincident-in time 5 boundaries, as described herein, or the different bands may be asynchronously divided into segments that do not coincide in time. Still fur~her, the signal samples can be scrambled as thel are being read into memory section ~2 rather than during readout as specifically 10 described herein.
It is also quite feasible to reverse the order of the processes that are carried out to encrypt the audio signal to be transmitted. For example, the audio signal can be partitioned into segments of different time 15 duration before it is divided into different ~requency bands rather than the reverse as specifically described above. In other words, a~ter the entire frequency spectrum of the audio signal is divided into segments of different time duration, each such segment can be 20 digitized and stored in memory. The stored in~ormation may then be read out of memory at a slower rate equal to the reciprocal of the rate at which it was read into the memory. This effectively expands each signal segments in time and compresses it in the frequency domain.
On read-out the segments are permuted according to a pseudo-random key code from the key generator 66 and converted to analog form with sr without time reversal.
After such conversion, some of the segments are shifted ~ to different frequency ~ands thereby to permit the 30 segments to be stacked frequency-wise so as to fill the entire audio frequency spectrum and time domain of the audio signal being transmitted. This composite signal and each segment thereof also has the same time-bandwidth product as the original audio signal and each segment 35 thereof. The composite signal may then be transmitted to a receiving unit which applies a reverse algorithm to the received signal to recover the intelligence in the transmission.
Also, to further increase the difficulty of decoding the transmission, ~he signal sequences in the different 5 frequency bands may not be stacked into columns of synchronous time-wise segments. Instead, the sequences in adjacent bands may be shifted in time to provide an asynchronous transmission of the signal sequences from the different frequency bands.
By the same token, different commutations and permutations of the aforementioned signal processing steps may be carried out on the signal to be transmitted.
Thus the original analog audio signal may be time-divided into full spectrum segments of different time duration as 15 described above. Then, some of these full spectrum segments can be frequency divided into partial spectrum signal sequences. The analog information may then be digitized and stored in memory. The full spectrum segments and the partial spectrum sequences, up~n 20 readout~ are time-wise reversed and permutated according to the pseudo-random key code from the key generator.
Also during readout some of the segments and/or sequences can be expanded or compressed in the time domain, thereby to respectively compress or expand their bandwidth so ~5 that each segment retains its original bandwidth-time product.
It will thus be seen that the objects set forth above, among those made apparent from the preceding description, are efficiently attained, and, since certain 30 chanses may be made in the akove construction without departing from the scope of the invention7 it is intended that all matter contained in the above description or shown in the accompanying drawings be interpreted as illustrative and not in a limiting sense.
Z5~
It is also to be unders~ood that the following claims are intended to cover all of the generic and specific features of ~he invention herein describedO
Claims (22)
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:
1. A privacy communication system comprising A. means for splitting a voice signal into a selected number of frequency bands; B. means for generating a pseudo-random key word; C. means for dividing each said band into segments of different time duration in accordance with the pseudo-random key word; D.
means for compressing said segments in time and expanding them in frequency by a factor equal to the number of frequency bands into which the voice signal was divided; and E. means for transmitting said time-compressed segments, the transmitted signal having substantially the same time-bandwidth product as the voice signal.
means for compressing said segments in time and expanding them in frequency by a factor equal to the number of frequency bands into which the voice signal was divided; and E. means for transmitting said time-compressed segments, the transmitted signal having substantially the same time-bandwidth product as the voice signal.
2. The system defined in claim 1 wherein the dividing means divides the signal segments in adjacent frequency bands so that they have coincident-in-time boundaries.
3. The system defined in claim 1 wherein the dividing means divides the signal segments in adjacent frequency bands so that they have non-coincident-in-time boundaries.
4. The system defined in claim 1 and further including means for inverting the signal segments from at least one frequency band prior to compressing them timewise.
5. The system defined in claim 1 and further including means for repeatedly changing the key word controlling the encryption of the voice signal.
6. The system defined in claim 1 wherein the splitting means split the voice signal into two said bands.
7. The system defined in claim 1 wherein the compressing means comprise memory; means for writing data from each band in succession into said memory at a selected rate; means for reading band data comprising successive bands from said memory at said rate so that the data from each band is read out at a rate which is a multiple of the number of bands.
8. A privacy communication system comprising A. means for dividing a voice signal into signal segments; B. means for expanding said segments in time and compressing them in frequency; C. means for generating a pseudo-random key word; D. means for shifting said segments in frequency and in time in accordance with the pseudo-random key word so as to stack said segments so that they span a frequency band which is approximately the same as the bandwidth of the voice signal; and E. means for transmitting the stacked segments, said transmitted signal having substantially the same time-bandwidth product as the voice signal.
9. The system defined in claim 8 wherein the shifting means shifts the segments so that the stacked segments have coincident-in-time boundaries.
10. The system defined in claim 8 wherein the shifting means shifts the segments so that the stacked segments have non-coincident-in-time boundaries.
11. The system defined in claim 8 and further including means responsive to the key word for reversing selected signal segments in time prior to shift-ing them in frequency.
12. The system defined in claim 8 wherein the expanding/compressing means includes a memory; means for writing the signal segments into said memory at a selected rate; and means for reading said signal segments out of said memory at a rate slower than the selected rate.
23. The system as defined in claim 22 and further including means for repeatedly changing the keyword controlling the encryption of the voice signal.
24. The system as defined in claim 22 wherein the order of transmission of said segments is determined in accordance with said keyword.
23. The system as defined in claim 22 and further including means for repeatedly changing the keyword controlling the encryption of the voice signal.
24. The system as defined in claim 22 wherein the order of transmission of said segments is determined in accordance with said keyword.
13. The system defined in claim 1 wherein the order of transmission of said segments is determined in accordance with said keyword.
14. The system defined in claim 13 and further including means for repeat-edly changing the keyword controlling the encryption of the voice signal.
15. The system defined in claim 1 wherein certain segments determined in accordance with said keyword are reversed in time.
16. The system defined in claim 15 and further including means for repeat-edly changing the keyword controlling the encryption of the voice signal.
17. The system defined in claim 15 wherein the order of transmission of said segments is determined in accordance with said keyword.
18. The system defined in claim 8 wherein the durations of said signal segments is determined in accordance with said keyword.
19. The system defined in claim 18 and further including means for repeat-edly changing the keyword controlling the encryption of the voice signal.
20. The system defined in claim 8 wherein the order of transmission of said segments is determined in accordance with said keyword.
21. The system defined in claim 20 and further including means for repeat-edly changing the keyword controlling the encryption of the voice signal.
22. The system defined in claim 8 wherein certain segments determined in accordance with said keyword are reversed in time.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US317,947 | 1981-11-04 | ||
| US06/317,947 US4433211A (en) | 1981-11-04 | 1981-11-04 | Privacy communication system employing time/frequency transformation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA1182595A true CA1182595A (en) | 1985-02-12 |
Family
ID=23235953
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA000414741A Expired CA1182595A (en) | 1981-11-04 | 1982-11-03 | Privacy communication system employing time/frequency transformation |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US4433211A (en) |
| EP (1) | EP0093159A4 (en) |
| CA (1) | CA1182595A (en) |
| WO (1) | WO1983001717A1 (en) |
Families Citing this family (39)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4965825A (en) | 1981-11-03 | 1990-10-23 | The Personalized Mass Media Corporation | Signal processing apparatus and methods |
| US7831204B1 (en) | 1981-11-03 | 2010-11-09 | Personalized Media Communications, Llc | Signal processing apparatus and methods |
| USRE47642E1 (en) | 1981-11-03 | 2019-10-08 | Personalized Media Communications LLC | Signal processing apparatus and methods |
| US4591673A (en) * | 1982-05-10 | 1986-05-27 | Lee Lin Shan | Frequency or time domain speech scrambling technique and system which does not require any frame synchronization |
| US4750205A (en) * | 1982-05-10 | 1988-06-07 | Lee Lin Shan | Frequency or time domain speech scrambling technique and system which does not require any frame synchronization |
| EP0095923A3 (en) * | 1982-06-02 | 1985-08-21 | THE PLESSEY COMPANY plc | Communications scrambling systems |
| US4551580A (en) * | 1982-11-22 | 1985-11-05 | At&T Bell Laboratories | Time-frequency scrambler |
| US4608456A (en) * | 1983-05-27 | 1986-08-26 | M/A-Com Linkabit, Inc. | Digital audio scrambling system with error conditioning |
| EP0138485B1 (en) * | 1983-09-29 | 1990-04-04 | Nippon Telegraph And Telephone Corporation | Radio reception system for a phase modulation signal |
| EP0139496B1 (en) * | 1983-09-30 | 1990-05-23 | Nippon Telegraph And Telephone Corporation | A radio transmission system for a phase modulation signal |
| US4716586A (en) * | 1983-12-07 | 1987-12-29 | American Microsystems, Inc. | State sequence dependent read only memory |
| US4742543A (en) * | 1983-12-22 | 1988-05-03 | Frederiksen Jeffrey E | Video transmission system |
| FR2561473A1 (en) * | 1984-03-13 | 1985-09-20 | Trt Telecom Radio Electr | CRYPTOPHONY SYSTEM FOR NARROW BANDWIDTH LINKS |
| AU580769B2 (en) * | 1984-05-05 | 1989-02-02 | British Encryption Technology Limited | Communications system |
| US4680791A (en) * | 1984-05-31 | 1987-07-14 | Nec Corporation | Digital video signal process apparatus for use in a video tape recorder |
| US4802220A (en) * | 1985-03-20 | 1989-01-31 | American Telephone And Telegraph Company, At&T Bell Laboratories | Method and apparatus for multi-channel communication security |
| US4763357A (en) * | 1985-04-18 | 1988-08-09 | Barr William S | Method and apparatus for providing secure electronic communications |
| AU6128686A (en) * | 1985-06-27 | 1987-01-30 | Codart Communications Inc. | Scrambling apparatus |
| SE458443B (en) * | 1985-07-03 | 1989-04-03 | Torbjoern Hahn | SYSTEM FOR STORAGE OF LIQUID OR GAS IN A SPACE IN MOUNTAIN |
| US4724541A (en) * | 1985-07-24 | 1988-02-09 | Mallick Brian C | Data-dependent binary encoder/decoder |
| GB2182229B (en) * | 1985-10-25 | 1989-10-04 | Racal Res Ltd | Speech scramblers |
| CA1288182C (en) * | 1987-06-02 | 1991-08-27 | Mitsuhiro Azuma | Secret speech equipment |
| US4827507A (en) * | 1987-06-19 | 1989-05-02 | Motorola, Inc. | Duplex analog scrambler |
| GB2207328A (en) * | 1987-07-20 | 1989-01-25 | British Broadcasting Corp | Scrambling of analogue electrical signals |
| JPS6424648A (en) * | 1987-07-21 | 1989-01-26 | Fujitsu Ltd | Privacy call equipment |
| US4914696A (en) * | 1988-08-15 | 1990-04-03 | Motorola, Inc. | Communications system with tandem scrambling devices |
| USRE38627E1 (en) | 1991-05-15 | 2004-10-19 | Interdigital Technology Corp. | High capacity spread spectrum channel |
| US5253296A (en) * | 1991-11-26 | 1993-10-12 | Communication Electronics | System for resisting interception of information |
| KR100716441B1 (en) * | 1999-03-03 | 2007-05-10 | 소니 가부시끼 가이샤 | Recording apparatus, recording method, reproducing apparatus, and reproducing method |
| US7543148B1 (en) * | 1999-07-13 | 2009-06-02 | Microsoft Corporation | Audio watermarking with covert channel and permutations |
| JP4319984B2 (en) * | 2002-09-02 | 2009-08-26 | セッテック インコーポレイテッド | Copying apparatus for copying recording medium, method thereof, and computer program thereof |
| US7099221B2 (en) | 2004-05-06 | 2006-08-29 | Micron Technology, Inc. | Memory controller method and system compensating for memory cell data losses |
| US20060010339A1 (en) * | 2004-06-24 | 2006-01-12 | Klein Dean A | Memory system and method having selective ECC during low power refresh |
| US7340668B2 (en) * | 2004-06-25 | 2008-03-04 | Micron Technology, Inc. | Low power cost-effective ECC memory system and method |
| US7116602B2 (en) * | 2004-07-15 | 2006-10-03 | Micron Technology, Inc. | Method and system for controlling refresh to avoid memory cell data losses |
| US7894289B2 (en) * | 2006-10-11 | 2011-02-22 | Micron Technology, Inc. | Memory system and method using partial ECC to achieve low power refresh and fast access to data |
| US7900120B2 (en) * | 2006-10-18 | 2011-03-01 | Micron Technology, Inc. | Memory system and method using ECC with flag bit to identify modified data |
| US20090169001A1 (en) * | 2007-12-28 | 2009-07-02 | Cisco Technology, Inc. | System and Method for Encryption and Secure Transmission of Compressed Media |
| US8878041B2 (en) * | 2009-05-27 | 2014-11-04 | Microsoft Corporation | Detecting beat information using a diverse set of correlations |
Family Cites Families (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE889311C (en) * | 1951-08-21 | 1953-09-10 | Hans W R Krueger | Procedure for secret telephony by means of any, individual, irregularly running encryption of the speech frequencies |
| CH518658A (en) * | 1970-07-07 | 1972-01-31 | Patelhold Patentverwaltungs Un | Process for encrypted message transmission by interchanging information elements over time |
| US3723878A (en) * | 1970-07-30 | 1973-03-27 | Technical Communications Corp | Voice privacy device |
| US3921151A (en) * | 1971-06-21 | 1975-11-18 | Patelhold Patentwerwertungs & | Apparatus for enciphering transmitted data by interchanging signal elements of the transmitted data without overlapping or omitting any elements within the transmitted signal train |
| US4020285A (en) * | 1972-09-29 | 1977-04-26 | Datotek, Inc. | Voice security method and system |
| US4068094A (en) * | 1973-02-13 | 1978-01-10 | Gretag Aktiengesellschaft | Method and apparatus for the scrambled transmission of spoken information via a telephony channel |
| DE2307441C1 (en) * | 1973-02-15 | 1975-05-07 | Licentia Gmbh | Method for obfuscating speech signals |
| CH558993A (en) * | 1973-03-19 | 1975-02-14 | Patelhold Patentverwertung | PROCEDURE AND DEVICE FOR ENCRYPTED MESSAGE TRANSMISSION. |
| CH580893A5 (en) * | 1973-07-02 | 1976-10-15 | Gretag Ag | |
| DE2359673C2 (en) * | 1973-11-30 | 1982-05-27 | Licentia Patent-Verwaltungs-Gmbh, 6000 Frankfurt | Method for disguising the information content of frequency bands |
| US4058677A (en) * | 1974-04-26 | 1977-11-15 | Lear Siegler, Inc. | Sound scrambling equipment |
| CH589390A5 (en) * | 1975-08-19 | 1977-06-30 | Patelhold Patentverwertung | |
| CH607506A5 (en) * | 1976-06-01 | 1978-12-29 | Europ Handelsges Anst | |
| FR2375011A1 (en) * | 1976-12-22 | 1978-07-21 | Commissariat Energie Atomique | BALANCED ARTICULATED MANIPULATOR |
| US4217469A (en) * | 1977-03-15 | 1980-08-12 | Emilio Martelli | Coding and decoding apparatus for the protection of communication secrecy |
| US4221931A (en) * | 1977-10-17 | 1980-09-09 | Harris Corporation | Time division multiplied speech scrambler |
-
1981
- 1981-11-04 US US06/317,947 patent/US4433211A/en not_active Expired - Lifetime
-
1982
- 1982-09-28 EP EP19830900089 patent/EP0093159A4/en not_active Withdrawn
- 1982-09-28 WO PCT/US1982/001384 patent/WO1983001717A1/en not_active Application Discontinuation
- 1982-11-03 CA CA000414741A patent/CA1182595A/en not_active Expired
Also Published As
| Publication number | Publication date |
|---|---|
| EP0093159A1 (en) | 1983-11-09 |
| US4433211A (en) | 1984-02-21 |
| WO1983001717A1 (en) | 1983-05-11 |
| EP0093159A4 (en) | 1984-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA1182595A (en) | Privacy communication system employing time/frequency transformation | |
| US4052565A (en) | Walsh function signal scrambler | |
| US4221931A (en) | Time division multiplied speech scrambler | |
| US4434323A (en) | Scrambler key code synchronizer | |
| US5253294A (en) | Secure transmission system | |
| US4893339A (en) | Secure communication system | |
| US5682425A (en) | Information signal transmission system | |
| US5682404A (en) | Method and apparatus for signal transmission and reception | |
| US5488661A (en) | Data communication system and method with data scrambling | |
| US4100374A (en) | Uniform permutation privacy system | |
| AU606550B2 (en) | Analogue scrambling system with dynamic band permutation | |
| US4392021A (en) | Secure facsimile transmission system using time-delay modulation | |
| US4107458A (en) | Cipher computer and cryptographic system | |
| US4443660A (en) | System and method for encrypting a voice signal | |
| EP0018784A1 (en) | Communication system with digital audio scrambler and unscrambler subsystems for transmission of audio intelligence through a television system | |
| US3518547A (en) | Digital communication system employing multiplex transmission of maximal length binary sequences | |
| US4649419A (en) | Pseudo-random binary sequency generator | |
| US4379205A (en) | Analog signal scrambling system | |
| EP0117276B1 (en) | Privacy communication apparatus | |
| US4893308A (en) | Method and apparatus for time companding a digital voice signal | |
| US4302628A (en) | Analog signal encrypting and decrypting system | |
| US4133977A (en) | Voice scrambler using syllabic masking | |
| US4283602A (en) | Cryptographically secure communication system | |
| US4058677A (en) | Sound scrambling equipment | |
| US4361729A (en) | Narrowband analog message privacy system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MKEC | Expiry (correction) | ||
| MKEX | Expiry |