Correlation data communications system.

Field of the invention

0001The present invention relates to digital data communications systems and, more particularly, to such systems which employ correlation detection to determine the identity of an incoming encoded digital code word.

Background of the invention

0002The use of correlation techniques in a data communications system decoding system to identify an incoming digital code word is well known. In a correlation decoding system a correlation detector is provided which matches an incoming digital code word on a bit-for-bit basis with a plurality of locally generated reference code words. When a substantial match with one of the reference code words is found, the correlator indicates this fact by an output signal, the amplitude of which represents the degree of bit-for-bit matching, and the reference code word with which the incoming digital code word is then being compared is identified as the received code word.

0003If a communication system employing a correlation detector in the decoding system uses Single Side Band (SSB) modulation for transmission and reception, Doppler shifts in the medium and carrier reinsertion errors in the receiver will shift each frequency in the demodulated output signal by some translation error which is the same for all output frequencies at the demodulator at any given instant, but which can vary with time due to changes in relative velocities of the transmitter and receiver and to variations in frequency standards in the transmitter and receiver. For SSB receivers and transmitters in the high frequency (HF) band (2 to 30 MHz), frequency translation errors in the demodulator output of up to + 100 Hz are common.

0004When a frequency translation error occurs, it has been found that the output signal of the correlation detector rotates in phase at the translation error frequency providing the envelope of the output signal of the correlation detector with an amplitude which is sinusoidal, varying at the translation frequency. With very low (or even zero) translation errors, if the phase of the correlation detector output is near a sinusoidal null of the amplitude variation, the correlation detector output will be low or zero and will remain so until the phase rotates away from the null. Such a low output of the correlation detector may cause the decoding system to fail to properly recognize a match between an incoming digital code word and one of the internally generated reference code words.

0005It is also often desirable to retrofit a data communications system to an existing operational radio transmitter and receiver system to avoid the expense of designing an entirely new transmission and reception system for data transmission. If a data communications system is retrofit to High-Frequency(HF) radio transmitters and receivers designed for Single Side Band (SSB) voice transmissions, a problem exists in that the limited audio band pass (usually 300 to 2500 Hz) of the transmitters and receivers prohibits use of wide-band spread- spectrum techniques commonly employed at UHF and microwave frequencies to improve code transmission reliability. At the same time the many forms of signal degradation associated with HF transmitters and receivers, such as fading, high interference, multipath propogation, etc. places demands on the data communications system which are not easily satisfied by simple frequency-shift keying or other conventional digital transmission techniques. To successfuly permit utilization of in place High-Frequency SS_B radio transmitters and receivers for data communications, the data communications system must: <ul id="ul0001" list-style="none"><li>a. provide a unique wave form in space which is highly unlikely to be duplicated by probable sources of interference or to be modified by such interference to produce undetected output errors;</li><li>b. provide a high degree of error detection and correction;</li><li>c. provide the ability to accept the waveform to which a receiver is synchronized while rejecting interference (either from other transmitters of from multipath propagation of the desired transmission) from similar waveforms not precisely in synchronism;</li><li>d. permit as high a communications data rate as possible within the transmitter/receiver bandpass limitations;</li><li>e. be implemented through addition of external encoder and decoder circuitry without modification of existing transmitters and/or receivers and without access to transmitter carrier frequency control circuits to achieve phase or frequency modulation nor to receiver RF or IF circuitry as would be required for demodulation of frequency or phase modulation or conventional correction of SSB translation errors; and</li><li>f. provide for synchronization of the demodulator to received signals and compensation for frequency translation errors over the same range of signal-to-interference ratios as required for signal processing after synchronization is achieved.</li></ul>

Summary of the invention

0006The present invention has been devised to eliminate the frequency translation problem in data communications systems employing correlation detection techniques in the decoder and to provide a reliable and accurate data communications system which can be easily retrofit to existing High Frequency SSB radio transmitters and receivers.

0007Accordingly, one object of the invention is the provision of a digital data communications system wherein the effects of frequency translation errors induced during transmission or reception of a digital code word are eliminated at the output of the decoding system correlation detector thereby improving the overall reliability of the decoding operation. This object is achieved by providing a feedback system which detects the effects of a frequency translation error at the output of the correlation detector and applies a frequency correction signal component to received digital code words which offsets the frequency translation error at the input to the correlation detector. This object can alternatively be achieved by providing a minimum shift keying (MSK) encoding apparatus in the encoding system of the invention which generates any differential MSK code which is differentially processed by decoding structures in the decoding system to reduce the effects of frequency translation errors.

0008An additional object of the invention is the provision of a data communications system which permits use of unmodified currently operational High-Frequency (HF) radio transmitters and receivers designed for single side band (SSB) voice transmissions to provide high-quality digital data communications and which satisfies the . transmission and reception requirements noted earlier. This object of the invention is achieved through a data communications system, employing an encoding and decoding system, which uses: (1) an encoding system which generates pseudorandom binary code words as start-stop phases of a repeating pseudorandom code for transmission on which data to be transmitted is impressed by cyclic code shift keying (CCSK), (2) a decoding system which employs correlation detection of received baseband pseudorandom binary code words, and (3) feedback from the correlation detector output through a feedback control loop which eliminates the above-noted frequency translation errors from the received pseudorandom binary code words and permits the decoding system output to be an undistorted representation of the data provided as input to the encoding system. In lieu of the feedback control loop, the encoding system may encode the pseudorandom binary code words using a differential code MSK technique with the MSK coded pseudorandom binary code words being differentially processed in the decoding system to eliminate frequency translation errors.

0009A further object of the invention is the minimization of distortion of the pseudorandom binary code words caused by passage through the limited bandpass of existing receivers and transmitters by selection of a code for which truncation of the code's frequency spectrum below 300 Hz and above 2500 Hz does not significantly degrade signal-to-interference rations, nor unacceptably degrade the correlation function of the codeword.

0010A further object of the invention is the provision of a method for CCSK encoding of pseudorandom binary code words which permits decoding by simple and economical circuitry without the need for microprocessors and look up tables, or other more complex methods conventionally used for CCSK decoding.

0011An additional object of the invention is the provision of a unique encoding system for generating pseudorandom binary code words having an overall D.C. balance (equal number of ones and zeros) which may be used to advantage in a data communications system.

0012An additional object of the invention is the provision of a unique encoding system for generating minimum shift keying (MSK) signals from applied binary code words.

0013These and other objects and advantages of the invention will be readily understood from the following detailed description of the invention which is presented in connection with the accompanying drawings.

Brief description of the drawings

0014<ul id="ul0002" list-style="none"><li>Fig. 1 illustrates a unique digital encoding system which may be used in the data communications system of the invention;</li><li>Fig. 2 illustrates a decoding system which may be used in the data communications system of invention;</li><li>Fig. 3 illustrates a modification to the encoding system shown in Fig. 1;</li><li>Fig. 4a and 4b respectively illustrate the envelope of the output of a correlation detector used in the decoding system, with and without aberrations caused by a frequency translation error;</li><li>Fig. 5 illustrates a representative peak finding circuit which can be used in the Fig. 2 decoding system;</li><li>Fig. 6 illustrates another modification to the encoding system shown in Fig. 1;</li><li>Fig. 7 illustrates a modification to the decoding system illustrated in Fig. 2;</li><li>Fig. 8 illustrates a modification to the encoding system illustrated in Fig. 1;</li><li>Fig. 9 illustrates a modification to the decoding system illustrated in Fig. 2; and</li><li>Figs. 10a, lOb, 10c, 10d and 10e are waveforms which illustrate operation of the encoding and decoding systems of Figs. 8 and 9.</li></ul>

Detailed description of the invention

0015The data communications system of the invention is particularly designed to be incorporated into existing transmitting and receiving systems which have a voice data channel and for the purpose of exemplary description the invention will be described for use in a SSB (single side band) High Frequency voice communications system. However, it should be appreciated that the data communications system of the invention can be used in other transmission/reception systems as well, and that the encoding system of the invention may be used with or without the decoding system of the invention and vice versa.

0016One exemplary encoding system which may be used in the invention is illustrated in Fig. 1. In this system source digital code works having k bits are transmitted as digital code words having 2^k bits, 2^k_l bits being generated by a pseudorandom code generator 15, and an additional single bit being added to the output of the pseudorandom code generator to form an even number of bits for transmission. Thus, for example, if a five bit digital source code word is input to the encoding system, the output thereof would be 32 (2⁵) chips (bits) in length.

0017Referring to Figure 1, a data source 11 supplying a serial stream of digital code is clocked by the output of a frequency divider 29 which applies a clocking signal at a predetermined bit rate f_b. The clocking signal is also fed as a clocking input to a data register 13 which receives at a serial input thereto the output from data source 11. Data register 13 is of the type which may receive data into the various register positions either by the serial input connected to the output of the data source 11 or by parallel input which is connected to the output of a data modifier 37, which is described more fully below. The data register contains a number of bit positions corresponding to the number of bits from data. source 11 to be encoded on an output pseudorandom binary code word, e.g. 5 for the example given above. After the selected number of data bits from data source 11 have been loaded into data register 13 its contents are transferred in parallel to a pseudorandom code generator 15 by an ENTER signal (word rate clock signal fw) applied to code generator 15 by frequency divider 27. Pseudorandom code generator 15 is a well known device which generates at an output, taken, for example, at register stage 5, a pseudorandom binary code word which is 2^k-1 bits long where k is the number of register stages (which correspond to the k bits of a digital code word to be encoded). The number of possible pseudorandom binary code words which can be generated by code generator 15 is 2^k-1 and each is a different start-stop phase of a repeating 2^k-1 bit pseudorandom code pattern. The code generator 15 is essentially a shift register which continually cycles an input code word thereto through the register with register stages 3 and 5 being-summed by an exclusive OR circuit 3 and fed back as a serial input to the code generator 15. A clock signal supplied to the clock terminal of the pseudorandom code generator 15 advances the serial data through the code generator in cyclic fashion. The clock signal is supplied by an inverter 21 which receives a clocking signal through AND gate 7 from the output of frequency divider 31 which supplies a clocing signal at chipping rate fc. Thus, the serial data stream at the output of the pseudorandom code generator 15 (bit position 5) has a chipping rate of f . The first five bits of the pseudorandom binary code word at the output of code generator 15 correspond to the five bits originally loaded into code generator 15.

0018After generating a pseudorandom binary code word corresponding to an input digital code word, pseudorandom code generator 15 is reloaded with a new incoming digital code word by the output of frequency divider 27 which applies a word rate clocking signal f to the ENTER input of pseudorandom code generator 15.

0019As discussed above, a pseudorandom coded output signal having 2^k bits is generated by the encoding system. However, a characteristic of pseudorandom code generator 15 is that 2k-1 bits are generated. Accordingly, an additional bit is generated and added to the output of pseudorandom code generator 15. To accomplish this, an AND gate 33 is connected to the output stages of code generator 15 to determine when a predetermined code condition exists therein. For example, AND gate 33 may be arranged to detect the state "11110". Upon detection of this state, an output signal from AND gate 33 causes a one-shot flip flop 39 to generate a pulse signal which inhibits the clocking signal output of AND gate 13 so that no clocking signal is applied at the clock input to pseudorandom code generator 15 for the time duration of the one-shot flip-flop 39. The pulse from the one-shot flip-flop is sufficient to prevent one clock pulse from the output of frequency divider 31 from being applied through an AND 19 and inverter 21 to the clocking input of pseudorandom code generator 15. Accordingly, whatever bit code ("0" for the code "11110") is then in register stage 5 of the pseudorandom code generator 15 is held for an additional clock signal. When the output pulse from one-shot flip-flop 39 disappears AND gate 19 is reenabled and clock signals from the output of the divider 31 are again supplied to the clock input of the code generator. Thus, for each n-bit digital code word inputted to code generator 15 from the data register 13, a 2k digital code word is generated at the output thereof.

0020The output of the pseudorandom code generator 15 is supplied to a transmitter 41 which may be a conventional RF transmitter using any suitable type of modulation. As noted, a SSB HF transmitter is particularly suitable.

0021The word rate f_w, bit rate f_b, and chip rate f_c clocking signals are respectively provided by dividers 27, 29 and 31 which are run off oscillator 23 operating at a frequency of 10 f . Divide-by-ten divider 27 and divide-by-two divider 29 receive the output of oscillator 23 through a divide-by-thirty-two frequency divider 25, while divide-by-ten frequency divider 31 directly receives the output of oscillator 23.

0022As well known, when a pseudorandom code generator is loaded with an all zero code word, e.g. "00000" for a 5 bit system, its output will be a stream of all zeros and will lock in this state. To eliminate the possibility of an "all zero" code word being loaded into the code generator 15, data modifier 37 continuously monitors the output of data register 13 and whenever a prohibited digital code word is entered therein, for example, all zeros, the data modifier 37 substitutes a different code word into the data register 13. The data modifier 37 receives the word rate f output from divider 27 and checks every digital code word in data register 13 just before its being loaded into pseudorandom code generator 15 so the forbidden code never is loaded in the code generator. When a forbidden code word is detected the data modifier 37 substitutes a different digital code word into register 13 and supplies a gating signal to data register 13 which allows the data register to accept, by parallel input, the output of data modifier 37.

0023Data modifier 37 may also be used to prevent any other incoming digital code words from being fed to random code generator 15. For example, the digital code word "11111" can be converted to the sequence "01111" by data modifier 37 in order to simplify decoder circuitry (or if the pseudorandom code generator is using inverted logic wherein "11111" would be the prohibited sequence).

0024It should be noted that the addition of a chip (added bit) to a conventional maximal length code word of length 2^k-1 to produce a 2^k-chip word is not required for CCSK encoding and, in fact, degrades to some extent, the desirable correlation characteristics of the 2k-1 chip binary code word produced by code generator 15. However, the added chip permits one to attain an additional pseudorandom binary code word for encoding a forbidden n-bit data word (e.g. all zero). Another way to encode a "forbidden" all-zero (or all-one for code generators with inverted logic) data sequence is to generate in the encoding system any selected one of the other 2^k_l sequences, but with the code inverted (all zeros changed to ones and all ones to zeros). The decoding system correlation detector can sense this inversion and substitute the forbidden sequence for the inverted word while, at the same time, inhibiting circuits (such as the detranslation feedback loop described below) which would be adversely affected by the inversion. This alternative encoding technique is illustrated in Fig. 6.

0025As shown in Fig. 6, a data modifier 111 monitors the parallel output from data register 13 for a prohibited code word and substitutes a permitted code word into register 13 as described earlier. In addition, the data modifier generates a control signal which enables a gate 113 and disables a gate 115 through inverter 119. Gate 113 receives at another input an inverted version of the output of code generator 15, while gate 115 receives at another input the non-inverted output of code generator 15. In this arrangement when data modifier 111 detects the presence of a prohibited bit pattern in data register 13 it loads a different, permitted code word therein and supplies a control signal on line 121 which disables gate 115 and enables gate 113 to thereby allow the output of code generator 15 to be inverted by inverter 117 prior to its being fed to transmitter 41. The inverted pseudorandom binary code word can then be detected in a system decoder and converted to the proper bit patterns.

0026In the process of generating the 2k bit pseudorandom binary code word which is applied to transmitter 41, the encoding system is arranged so that the added bit, which occurs by holding the code generator 15 in a predetermined output position for one clock pulse by AND cate 33 and one-shot flip-flop 39, can be a bit value which obtains a balance in the overall number of ones and zeros in the 2^k bit encoded output. In the arrangement shown, this balancing occurs by the addition of a "0" to the 31 (2^k-1) bits generated by the output of the pseudorandom code generator 15 when operating normally. This additional "0" is thus generated by sensing a bit pattern in the code generator 15 which has "0" in the bit 5 stage. One such convenient bit pattern is "11110" as described above. However, it should be clear that the additional zero could be generated at other points in the output of the pseudorandom code generator 15 to provide an added "0" to balance the overall number of ones and zeros in the transmitted pseudorandom binary code word. Moreover, if pseudorandom code generator 15 produces one more "0"'s than "I"'s in its output (as would occur, for example, with inverted logic), then the added bit would be a "1".

0027The balancing of the overall number of ones and zeros in the generated 2^k bit pseudorandom binary code word has the advantage of providing a DC balance to the overall signal which makes for a simpler design of decoding circuitry. Table I illustrates an exemplary pseudorandom coding (2^k) obtained at the output of code generator 15 for the various states of a 5-bit input data word. The asterisks denote zeros added in the manner described above. <tables id="tabl0001" num="0001"><img file="EP0102918A2_D0001.tif" /></tables>

0028As evident form the above Table, each of the pseudorandom binary code words generated at the output of code generator 15 represents a different start-stop phase of an endlessly repeating pseudorandom n-bit pattern, where n=2k-1 and k and n are integers.

0029Allthough the encoding system of Fig. 1 generates a preferred 2^k bit pseudorandom code from a k-bit data code word, it can also be used to generate a 2^k-1 bit pseudorandom binary code (Table I without the added zeros) by the elimination of inverter 35, gate 33, one shot flip-flop 39, gate 19 and inverter 21 and directly connecting the output of divider 31 to the clock input of code generator 15. These changes will enable code generator 15 to operate as a conventional pseudorandom code generator and there will be no addition of a bit to the output of code generator 15 as described earlier. The data modifier circuit shown in Fig. 6 will work with the encoding system configured as just described to produce a 2^k_l pseudorandom code output in which each of the binary code words generated at the output of code generator 15 represents a different start-stop phase of a repeating n-bit pattern where n=2^k-1.

0030The data communications system of the invention employs correlation detection in the decoding system to reproduce the input data words to register 13 from the transmitted and received pseudorandom binary code.

0031A characteristic of decoding systems employing correlation detection for identifying incoming digital data, such as the pseudorandom binary code generated by the encoding system of Fig. 1, is that the pulsed output of the correlation detector, which occurs whenever a substantial match is found between an incoming pseudorandom binary code word and one of a plurality of locally generated reference pseudorandom binary code words, rotates in phase at the frequency of any frequency translation error in single side band transmission systems with which the data communications system may be used. Such errors can be caused by Doppler shifts in the transmitting medium and carrier reinsertion errors in the receiver demodulator. This results in the correlation detector output signal envelope having a sinusoidal amplitude component impressed thereon due to the frequency translation error (Fig. 4a). The translation error shifts each frequency of the demodulator output by some translation error which is the same for all demodulated output frequencies at any given instant, but which varies with time due to changes in relative velocities of the transmitter and receiver and to variations in frequency standards in the transmitter and receiver. As noted earlier, for SSB transmitters and receivers in the high frequency (HF) band (2-30 Mhz) translation errors of up to ± 100 hz are common.

0032The frequency translation error, which occurs as amplitude and phase variations in the output waveform of the correlation detector, may result in a correlation detector output signal being "missed" in the decoding system because it may occur at a point near a sinusoidal null in the output envelope of the correlation detector or because it has an inverted polarity. To eliminate this problem, the decoding system of the invention, as illustrated in Fig. 2, includes a detranslation circuit which eliminates the sinusoidal variations in the output of the correlation detector.

0033Before describing Fig. 2 in detail, it should be pointed out that the decoding system shown therein correlates incoming pseudorandom binary code words with similar locally generated reference pseudorandom binary code words. However, it should be understood that the decoding system of the invention can be used with any type of digital coding and that pseudorandom coding is not at all necessary. However, for the purpose of simplifying subsequent description, the pseudorandom binary code words as generated by the encoding system of Fig. 1 will assumed as being received at the input to the decoding system of Fig. 2.

0034Referring to Fig. 2, a High Frequency single sideband, receiver 43 receives and demodulates a transmitted signal containing a stream of pseudorandom binary code words. In normal operation, this signal would then be fed to a correlation detector circuit which would compare the arriving pseudorandom binary code words successively with each one of a plurality of locally generated reference pseudorandom code words, the correlator detector signaling a match when found. In the decoding system of Fig. 2, the incoming demodulated pseudorandom binary code words are fed to a balanced modulator 49 where the pseudorandom binary code words modulate a first local carrier at the output of a voltage control oscillator 73. The output from balanced modulator passes through a low pass filter 51 which removes a higher frequency product of the modulation process. The resulting single side band at the output of filter 51 is demodulated by using it to modulate a second local carrier whose frequency is fixed at the approximate center frequency of the voltage controlled oscillator 73 supplying the first local carrier. The second local carrier is supplied by oscillator 101. The upper sideband of the resulting sideband pair is removed by filter 54 leaving a lower side band which is the same as the received signal except that an additional translation error, relative to the received signal, equal to the difference in frequency between the first and second local carrier frequencies has been introduced. The output of low pass filter 54 is applied as an input to correlation detector 53. A feedback loop, described in further detail below, is used to adjust the voltage controlled oscillator 73 so that this translation error is equal and opposite to that present in the received signal and the single sideband resulting from the second modulation is the originally transmitted pseudorandom binary code without translation error.

0035The center frequency of the voltage controlled oscillator 73 is selected to facilitate removal of one of the modulation side bands by filter 51 and may be, as in the embodiment here discussed and illustrated in Fig. 2, the chipping frequency fc selected for the system. The control range of the voltage controlled oscillator 73 is made wide enough to permit the output frequency to be varied over a range equal to the expected range of the translation error.

0036Correlation detector 53 may be a standard integrated circuit device, for example, a Reticon R5403. The correlator contains (for a 32 chip code word) two 32- chip "bucket brigade" delay lines. Incoming pseudorandom binary code words are shifted through one of the delay lines by two phase clock signals Ø₁ and Ø₂. The other delay line serially receives the locally generated reference pseudorandom binary code words which are shifted in by clock signals ø₃ and ø₄. A multiplier circuit is provided between each cell of the delay line for the incoming pseudorandom binary code words and a like numbered cell for incoming pseudorandom reference binary code words. A strobe signal to the STROBE input of the correlation detector causes the signals in the delay line cells to be transferred to the multiplier circuit where they are multiplied and the resulting products summed. When the STROBE input is held in a predetermined logic "1" state the contents of the delay line cells will be repeatedly supplied to the multiplier circuit for multiplication. The products of the multiplier circuit are summed into the correlation detector output to provide a continuous summation of the products of each chip. When a reference and incoming pseudorandom code word are alike, all 32 products are positive and add to produce an output pulse 32 times as large in amplitude as a signal due to the multiplication of a single chip. When some chips are alike and some are unlike, the negative products formed by unlike chips are substracted from the sum of products of the like chips and provide an amplitude at the output of the correlator which is proportional to the degree of match found between the incoming pseudorandom binary code word and a locally generated reference pseudorandom binary code word.

0037The incoming digital code word is advanced through the correlator by the output of oscillator 101 at the chipping frequency f which is supplied to the first clock Ø₁ input to the correlator 53 and by an inverted signal applied through inverter 73 to the second clock Ø₂ input to correlator 53.

0038The reference code words, generated by reference code generator 83, are advanced through correlator detector 53 by two phase clock signals Ø₃ and Ø₄ at rate of 32 x f formed by a multiplier 81 which multiplies the output f of oscillator 101. Because of the 32 to 1 frequency ratio between the incoming pseudorandom binary code word and reference pseudorandom binary code words which are clocked into the correlator 53, all 32 possible reference codes are stepped through the reference delay line between successive clock cycles which control the shifting of the incoming digital code words through correlator detector 53.

0039Code generator 83 generates a continously repeating code identical to the basic code generated at the output of the encoding system (Fig. 1). Since the number of cells in the correlator detector is same as the number of chips in the code, the five chips in the reference code generator are always the first five chips of the word in the correlator. Thus, when a match is found between an arriving digital code word and a locally generated reference code word by correlator detector 53 the output thereof sets a transfer register 85 to the same five bits as exist in the reference code generator 83. This data, which is a replica of the k-bit digital code word originally loaded in data register 13, is then transferred to output buffer 87 at the start of the next word interval as determined by frequency divider 77, where it is serially read out at a bit rate determined by frequency multiplier 79 which multiplies by five the output of frequency divider 77.

0040At the initiation of a code transmission from the encoding system (Fig. 1) a preamble pseudorandom binary code word is repeatedly generated in a predetermined number of times by repeatedly supplying a predetermined k-bit data word to data register 13. The resulting pseudorandom code word which is repeatedly generated is used by the decoding system to synchronize with the encoding system. A data decoder 131 in the decoding system is connected to the output of reference code generator 83 and determines whenever the k-bit data word appears which is used in the encoding.system to generate the preamble pseudorandom binary code word. Upon detecting this signal decoder 131 supplies a pulse strobe signal, through OR gate 143 to the STROBE input of correlation detector 53 causing it to load the corresponding reference pseudorandom binary code word in the reference delay line cells of the correlation detector into the correlation detector multiplier circuit. This reference pseudorandom binary code word is held in the multiplier circuit until a new reference pseudorandom binary code word can be loaded which occurs upon the occurrence of another strobe signal. The held reference pseudorandom code word is repeatedly compared in correlation detector 53 with the incoming preamble pseudorandom code and when a predetermined number of "matches" is found, the decoding system is considered synchronized with the encoding system.

0041To determine synchronization a counter 151 is connected to the output of correlation detector 53. Counter 151 counts the "matches" found by the correlation detector and whenever a predetermined number of "matches" is found counter 151 sets flip-flop 145 which supplies a continous strobe signal through OR gate 143 to the STROBE input of correlation detector 53. This causes correlation detector 53 to continuously feed new reference code words as they appear from code generator 83 into the multiplier circuit for comparison with pseudorandom code words arriving from the encoding system. The strobe output of flip-flop 145 also enables a gate 141 which permits, when enabled, the output of correlation detector 53 (whether positive or negative) to be supplied to a transfer register 85. The output of correlation detector 53 is always supplied to peak finder 63 to enable the translation error feedback loop, described in greater detail below, to operate during synchronization. After synchronization is achieved, transfer register 85 is operated by the positive output only of correlation detector 53 to output on n-bit data word corresponding to an arriving pseudorandom binary code word.

0042The synchronization detecting circuits including counter 151 and flip-flop 145 are reset upon the cessation of a transmission from the encoding system and this is recognized by a missing pulse detector 153 which supplies a signal which resets counter 151 and flip-flop 145 after the output of correlation detector 53 has disappeared for a predetermined length of time.

0043The predetermined number of preamble pseudorandom code words which are transmitted are sufficient in number to allow adequate time for the frequency translation error feedback loop, described below, to eliminate the effect of frequency translation errors at the input (and correspondingly the output) of correlation detector 53.

0044If a data modifier 37 (Fig. 1) was used in the encoding system, a similar data modifier 161 is required in the decoding system to detect the arriving pseudorandom binary code for the prohibited k-bit code word and to load transfer register 85 with the proper k-bit code word to which the arriving pseudorandom code word corresponds. If data modifier 111 (Fig. 6) is used in the encoding system to generate an inverted pseudorandom binary code word for a prohibited k-bit digital code word, the data demodifier 161 of Fig. 2 would be replaced by the circuit illustrated in Fig. 7. In the Fig. 7 circuit, the correlation detector output (whether positive or negative) which appears at the output of gate 141 after synchronization is achieved (and the translation frequency error feedback loop is locked) is supplied to a negative pulse detector 137. A characteristic of correlation detector 53 is that if an "inverted" arriving pseudorandom code word is compared with its "non-inverted" counterpart, a negative output signal occurs, the amplitude of which is determined by the degree of "non-match" between the compared code words. Thus, negative pulse detector 137 will determine when an "inverted" pseudorandom code word arrives from the encoding system. The output of negative pulse detector 137 is supplied to an input of AND gate 139 which receives at its other input the output of gate 141. Gate 139 functions to inhibit the negative correlation detector output from gate 141 from being supplied to peak finder 63. Thus the peak finder 63 is not affected by negative outputs of correlation detector 53 after the feedback loop described below is locked. The output of negative pulse detector is also supplied to a pulse generator 135 which generates a CLEAR signal which is supplied to transfer register 85. This CLEAR signal causes the transfer register 85 to clear itself so its output lines assume an "all zero" state corresponding to the prohibited "all zero" k-bit digital code word which was modified in the encoding system. If a different prohibited k-bit digital code word is modified in the encoding system, the output of pulse generator 135 would be used to load transfer register 85 with the appropriate prohibited data word. Transfer register 85 is arranged so that only positive outputs of correlation detector will cause it to parallel load data from the reference code generator 83 so that whenever the "inverted" pseudorandom code word is detected transfer register 85 is not loaded with the output of reference code generator 83 but is merely cleared by the output of pulse generator 135 (or loaded with appropriate data by the output of pulse generator 135 if a prohibited k-bit data word is other than an all zero sequence).

0045If correlation detector 53 were directly connected to the output of the high frequency receiver 43, as would be the case in a normal correlation decoding system, the envelope of its output signal would be as shown in Fig. 4a, varying in amplitude and polarity in accordance with a sinusoidal signal component which is impressed upon the correlation signal output, having a frequency corresponding to that of the frequency translation error introduced during transmission and reception of the incoming digital code words. As shown in Fig. 4a, the nulls of the sinusoidal signal are such as to possibly result in a correlation output signal indicating a match being "missed" because the signal level is excessively low.

0046To eliminate this problem, the decoding system of the invention applies a frequency rotative component to the input of the correlation detector which, in effect, cancels the rotative component caused by the frequency translation error, thereby causing the correlation detector output phase to be stably maintained to produce an output signal of a relatively high level. This is accomplished by a feedback loop which includes peak finder 63, low pass filter 65, saw-tooth generator 67, voltage control oscillator 73, balance modulator 49, and low pass filter 51.

0047The feedback loop utilizes sawtooth generator 67 to provide a sawtooth waveform at the control input to voltage control oscillator 73 which varies the output frequency of oscillator 73 approximately symmetrically about the nominal chipping frequency f by an amount sufficient to encompass any anticipated translation error. For example, a 1 Hz sawtooth can be used which sweeps voltage control oscillator 73 over f_c+ 100 Hz.

0048The effect of the feedback system is to find and maintain a first local carrier frequency (output of VCO 73) whose frequency and phase maintain the output of correlation detector 53 at a level corresponding to the peak of the pseudorandom binary code word's correlation function, thereby eliminating effects of the frequency translation of the input signal.

0049The peak finder 63 is a circuit which compares successive outputs from correlator detector 53 and generates an output which instructs the saw-tooth generator 67 to continue generating an increasing output waveform as long as the pulses from peak finder 63 are positive and the nth pulse is equal to or greater in amplitude than the n-lth pulse. This indicates that the output of the correlator 53 is at or is climbing towards a positive peak on the transfer frequency sinusoid (Fig. 4a). When the nth pulse amplitude is less than the n-lth pulse, the saw-tooth generator is stopped and, if the decline continues, the output from the saw-tooth generator 67 will be reduced until a stable operation at or near the envelope peak illustrated in Fig. 4a is achieved. The correlation detector 53 output, when the feedback loop is propertly locked is illustrated in Fig. 4b.

0050Low pass filter 65, provided between peak finder 63 and saw-tooth generator 67, smooths the output from the peak finder 63 and makes the frequency generated by the voltage control oscillator 73 a function of the short term average of several pulses to prevent "hunting" and instability of the control loop.

0051An exemplary peak finding circuit 63 which may be, used is illustrated in Fig. 5. As shown therein, two sample and hold circuits 91 and 93 are provided for respectively receiving the nth and n-lth outputs from correlation detector 53. The outputs of the two sample and hold circuits are then fed to a comparator 95 which provides a positive output whenever the nth sample is greater than the n-lth sample and a negative output whenever the reverse is true. No output is provided if the two samples are equal. The output from comparator 95 is fed to low pass filter 65.

0052A mathematical analysis of the operation of the detranslation loop is as follows.

0053The incoming pseudorandom binary code words can be considered as an amplitude signal component, w (actually there are many components consisting of harmonics of the word repetition rate and the analysis applies to each component), and a w_t component which is the translation error component. Thus, the data exiting from high frequency receiver 43 can be expressed as:<maths id="math0001" num=""><img file="EP0102918A2_D0002.tif" /></maths>where w_t is a translation error component, w is a signal frequency component, and A is the amplitude of the signal component.

0054As result of modulating this signal on the output of voltage control oscillator 73, the following signal is produced:<maths id="math0002" num=""><img file="EP0102918A2_D0003.tif" /></maths>where w_c represents the nominal center frequency of voltage control oscillator 73 and w_x is the deviation from w_ccaused by the feedback loop.

0055Equation (2) can be rewritten as:<maths id="math0003" num=""><img file="EP0102918A2_D0004.tif" /></maths>

0056Low pass filter 51 removes the higher frequency component (the first half of equation 3) resulting in:<maths id="math0004" num=""><img file="EP0102918A2_D0005.tif" /></maths>

0057By varying the frequency component w_x with the detranslation loop so that it equals w_t equation 4 can be converted to ½ A cos (w_c - w_m)t so that w_tis eli- minated. By using this signal to modulate a second carrier fixed in frequency at w , the following is obtained,<maths id="math0005" num=""><img file="EP0102918A2_D0006.tif" /></maths>

0058After filtering to remove the term containing 2w , the original cos w_mt term is obtained which will correlate with the locally generated reference pseudorandom code words without distortions due to translation error.

0059As a modification to the detranslation system illustrated in Fig. 2, balanced modulator 52 and low pass filter 54 can be provided upstream of balanced modulator 49 and low pass filter 51, as functionally achieved by connecting the output of oscillator 101 to the carrier frequency input of balanced modulator 52 and the output of VCO 73 to the carrier frequency input of balanced modulator 49. Or, balanced modulator 52 and low pass filter 54 can be eliminated from the decoding system and instead provided in the encoding system just upstream of transmitter 41, as illustrated in Fig. 3.

0060As evident from the above, the described decoding system includes a detranslation loop which improves the quality of the correlation detector output by eliminating the effect of frequency translation errors thereat induced during the transmission and reception of a DC coded signal.

0061Although the decoding system illustrated in Fig. 2 may be used with the encoding system illustrated in Fig. 1, the principles of operation of the decoding system allow its use with other types of code transmission systems which need not employ pseudorandom coding, so long as the leading k chips of a transmitted data code word used (or any other easily recognized set of k chips) are the same as the k bits of data used to select the transmitted code word. The encoding system illustrated in Fig. 1 does have advantages, however, in providing pseudorandom binary code words for transmission which have desirable correlation characteristics and in providing pseudorandom binary code words which have an equal number of ones and zeros to thereby maintain a DC balance for the transmitted signal which simplifies decoding circuitry.

0062An alternative to the above-described feedback loop technique for eliminating frequency translation errors is illustrated in Figs. 8 and 9 which respectively show modified encoding and decoding systems for use in the invention. Fig. 8 illustrates the encoding system of Fig. 1 with the addition of a minimum shift keying code generator 301, for generating an MSK code from the pseudorandom binary code words (either 2^k or 2^k-1 pseudorandom binary code words) produced by code generator 15. All of the structures and operations of the encoding system of Fig. 1 are retained, but the generated output thereof is further modified by code generator 301. The MSK code generator 301 employs a J-K flip-flop 303 which receives at its J,K inputs the output of code generator 15 and at its clock input the chipping clock frequency fc. The Q output of flip-flop 303 is a differential code version of the applied binary code word which is connected to one input of an exclusive or gate 305, which receives at its other input the chipping clock frequency fc. The output of exclusive OR gate 305 is a phase shift keyed version of the differential code which alternates between two frequencies f₁ and f₂ where f₁= <maths id="math0006" num=""><img file="EP0102918A2_D0007.tif" /></maths>and f₂= <maths id="math0007" num=""><img file="EP0102918A2_D0008.tif" /></maths>where n is a selected integer (n = 1 for the Fig. 8 circuit) and 1/T is the chip rate (fc). Figs. 10a, 10b, 10c and lOd respectively illustrate the pseudorandom binary code word input to J-K flip-flop 303, the differential code output of J,K flip-flop 303, the chipping clock frequency fc, and the chipping clock phase shift keyed (PSK) modulated by the differential code to produce an MSK signal at the output of exclusive OR gate 305.

0063The output of exclusive OR gate 305 is applied to transmitter 41 either directly, or through the illustrated line driver 307. It should be noted that the minimum shift keying code generator 301 may also be used in other environments than that specifically disclosed.

0064In the decoding system, (Fig. 9) the MSK code present at the output of HF receiver 43 is applied to a 1 bit delay line 311 the output of which is applied as one input to a balanced modulator 52 functioning as a multiplier. The other input to balanced modulator 52 is the MSK code signal applied to delay line 311. By multiplying the MSK code by its one-bit delayed version and then filtering the resultant product by low pass filter 54, the original pseudorandom binary code applied to differential code generator 301 is recovered in inverted form. The recovered inverted pseudorandom binary code may be then inverted back to the original pseudorandom binary code by inverter 315 and then applied to correlation detector 53 which operates, as described earlier, to recover the original k-bit code word applied to data register 13 in the encoding system. Fig. 10e shows the output of balanced modulator 52, while Fig. 10f shows the output of inverter 315.

0065The remaining decoding structures of Fig. 9 operate in the same manner as described earlier with reference to Fig. 2.

0066By providing a differential MSK coding apparatus in the encoding system and a complementary differential decoding apparatus in the decoding system, frequency translation errors can be minimized. The explanation of how this occurs is as follows.

0067For MSK encoding, one code state is transmitted by a frequency, f_l, and the other code state is transmitted by a frequecy, f₂, such that<maths id="math0008" num=""><img file="EP0102918A2_D0009.tif" /></maths><maths id="math0009" num=""><img file="EP0102918A2_D0010.tif" /></maths>where n is a selected integer and is the bit (or chip) rate. For n=l, f₁= <maths id="math0010" num=""><img file="EP0102918A2_D0011.tif" /></maths>and f₂= <maths id="math0011" num=""><img file="EP0102918A2_D0012.tif" /></maths>.For n=1, the resulting code is differentially coherent and can be converted to the original binary code by multiplying the code by a delayed version of the code where the delay is the bit or chip interval. Any translation error on the signal will be elimated or reduced as now described.

0068Let the ONE state be transmitted by cos 2πf₁t = cos <maths id="math0012" num=""><img file="EP0102918A2_D0013.tif" /></maths>and the zero state by cos <maths id="math0013" num=""><img file="EP0102918A2_D0014.tif" /></maths>. When received with a frequency error of f_T' the two signals will be ^cos 2π(<maths id="math0014" num=""><img file="EP0102918A2_D0015.tif" /></maths> + f_T)t and cos 2π(1 T + f_T) t. If the received signal e_r is delayed by precisely T, the duration of one chip, and if the delayed signal, erd, and the undelayed signal are multiplied together they form the product ep = e_re_dr. There are four possible combinations of e and e_dr in terms of the two code states, ZERO and ONE, 0 x 0, 1 x 1, 0 x 1, and 1 x 0. The products may be evaluated by using the trigonometric identity, cos A cos B = 1/2 [cos(A+B) + cos (A-B)]. Accordingly, for multiplication of the (0 x 0) state the following equations hold.<maths id="math0015" num=""><img file="EP0102918A2_D0016.tif" /></maths><maths id="math0016" num=""><img file="EP0102918A2_D0017.tif" /></maths>

0069If the <maths id="math0017" num=""><img file="EP0102918A2_D0018.tif" /></maths>term is rejected by a low-pass filter (i.e. filter 54 in Fig. 10)<maths id="math0018" num=""><img file="EP0102918A2_D0019.tif" /></maths>

0070For multiplication of the (1 x 1), state the following equations hold.<maths id="math0019" num=""><img file="EP0102918A2_D0020.tif" /></maths><maths id="math0020" num=""><img file="EP0102918A2_D0021.tif" /></maths>

0071Filtering the <maths id="math0021" num=""><img file="EP0102918A2_D0022.tif" /></maths>term by a low pass filter leaves e_dr(1x1) = 1/2 cosπ(0 + π - 2πf_TT) = 1/2 cos 2πf_TT .. (13)

0072For multiplication of the (Oxl) state, the following equations hold.<maths id="math0022" num=""><img file="EP0102918A2_D0023.tif" /></maths><maths id="math0023" num=""><img file="EP0102918A2_D0024.tif" /></maths>

0073After low pass filtering to remove the <maths id="math0024" num=""><img file="EP0102918A2_D0025.tif" /></maths>term,<maths id="math0025" num=""><img file="EP0102918A2_D0026.tif" /></maths><maths id="math0026" num=""><img file="EP0102918A2_D0027.tif" /></maths>

0074Finally, for multiplication of the (1x0) state, the following equations hold.<maths id="math0027" num=""><img file="EP0102918A2_D0028.tif" /></maths><maths id="math0028" num=""><img file="EP0102918A2_D0029.tif" /></maths>

0075After low pass filtering the <maths id="math0029" num=""><img file="EP0102918A2_D0030.tif" /></maths>term<maths id="math0030" num=""><img file="EP0102918A2_D0031.tif" /></maths><maths id="math0031" num=""><img file="EP0102918A2_D0032.tif" /></maths>

0076Equation 21 is the same as equation 16 except for the direction of the phase shift caused by f_T. The term <maths id="math0032" num=""><img file="EP0102918A2_D0033.tif" /></maths>in equations 16 and 21 defines the sinusoidal transition from one code state to the next as required to produce a 0,1 or 1,0 sequence.

0077Equations 10, 13, 16 and 21 represent signals which should have phase angles of either 0, for equations 10 and 13, or π radians, for equations 16 and 21, so that the cosines of the angles would be respectively +1, the correct value for a code ONE, and a -1, the correct value for a code ZERO. The actual angles in the equations differ by 2 f_TT which is constant for a given value of f_T. When f_Tx T is small, i.e. when the sum of Doppler shifts phase frequency errors is small, compared to the bit or chip rate fc, the cosines of the angles differ only negligibly from the correct values. As f_T increases the effect also increases, with output, e.g. from the correlation detector, reduced to half power for f_T x T = 0.125 (a 45° phase shift) and to 0 for f_Tx T = 0.25 (a 90° phase shift). Thus, for values of f_Tx TS 0.125, the effect of translation errors is only to reduce a received signal to noise (S/N) ratio by not more than 3 db and not to cause a rotating phase which would degrade correlation detector performance.

0078Although representative embodiments of the invention have been illustrated, it should be clear that various modifications can be made to these embodiments without departing from the spirit and scope of the invention. Accordingly, the invention is not limited by the foregoing description, but is only limited by the claims which are appended hereto.