Receiver for a Spread Spectrum Communication System Having a Time-Multiplexed Convolver

1. Field of the Invention

The invention is related to surface acoustic wave convolvers and in particular to a time multiplexed convolver for M-ARY orthogonal keyed decoding.

2. Prior Art

M-ary Orthogonal Keying data modulation is a technique used in spread spectrum communications where one basic spreading function (SF) such as a pseudo random noise (PN) sequence is phase modulated on a carrier. The basic spreading function is modulated by certain other functions, such as Walsh functions, to cause a modulation orthogonal to the basic spreading function and every other function used. The resultant response of a signal in a channel, match filtered to an orthogonally modulated signal, is zero while the matched response is a maximum.

This modulation technique is used extensively in secure communications, both for continuous communications and short message formats. Spreading the small amount of information pseudo randomly over a wide bandwidth allows for a low probability of intercept since the signal power density can be below the noise power density. Only by knowing the spreading function can the signal energy be integrated above the noise.

Conventionally, orthogonal-keyed data is recovered by implementing parallel decoding devices such as correlators or convolvers and then comparing the filtered amplitudes as in a Mark and Space receiver. This is sometimes implemented by parallel processing the orthogonal modulations using partial sums from a single decoder of the basic pseudo-random noise (PN) spreading function. Convolvers are generally used for maximum bandwidth modulation where the spreading function is changing after short intervals. However, convolvers are capable of much wider bandwidths than can be obtained in base band processing. Therefore, convolvers have a lot of excess bandwidth over that required for modulation.

This invention takes advantage of the excess bandwidth of the convolver by time multiplexing more than one reference signal and applying the time multiplexed reference signal to the signal via the reference input of the convolver. To prevent the summing bus of the convolver from simultaneously containing the product of the input signal with all of the time multiplexed reference signals, the resultant signal from the input signal and one of the time multiplexed reference signals is kept separated from the other resultant signals by segmenting the summing bus of the convolver.

The invention is a time multiplexed convolver for convolving an input signal with a time reversed-time multiplexed reference signal. The convolver has a first interdigital transducer disposed on a piezoelectric substrate for launching a first surface acoustic wave along a first path in response to the input signal, a second interdigital transducer disposed on the substrate, displaced from said first interdigital transducer along said first path for launching a second surface acoustic wave towards said first interdigital transducer in response to time reversed-time multiplexed reference signal, said first path defining an interaction region between said first and second surface acoustic waves, and a segmented summing bus disposed on said substrate between said first and second interdigital transducers for generating time multiplexed samples from the convolution of said first and second surface acoustic waves, said segmented summing bus having a plurality of equally spaced fingers extending into said interaction region and a connector bus connected to said plurality of fingers outside said interaction region. In the preferred embodiment the convolver has a pair of acoustic horns, one disposed intermediate said segmented summing bus and each of the interdigital transducers and a ground bus having a plurality of segments extending into the interaction region intermediate said fingers.

The advantage of the time multiplexed convolver is that it replaces a plurality of convolvers. Another advantage is that it eliminates parallel decoding systems minimizing weight and power consumption. These and other advantages will become more apparent from reading the detailed description of the invention in conjunction with the drawings.

FIG. 1 is a drawing illustrating the details of a conventional surface acoustic wave convolver.

FIG. 2 is a drawing illustrating the details of the time-multiplexed convolver.

FIG. 3 is a drawing illustrating the details of a first alternate embodiment of the time multiplexed convolver.

FIG. 4 is a drawing illustrating the details of a second alternate embodiment of the time multiplexed convolver.

FIG. 5 is a block diagram showing the details of the spread function communication system.

FIG. 6 is a block diagram of the transmitter.

FIG. 7 illustrates the waveforms of the spreading function and Walsh functions used to describe the operation of the communication system.

FIG. 8 is a circuit diagram of the receiver input.

FIG. 9 is a circuit diagram of the Reference Signal Generator Means.

FIG. 10 is a circuit diagram of the Demultiplexer Means.

FIG. 11 illustrates the waveforms generated by the Reference Signal Generator Means.

Shown in FIG. 1 is a conventional broadband width surface acoustic wave convolver 1 such as disclosed by Michaels in U.S. Pat. No. 4,428,062. A pseudo-random noise, phase modulated input surface acoustic wave is launched by an interdigital transducer 2 onto a piezoelectric substrate 3, such as a lithium niobate substrate. The launched surface acoustic wave is compressed and focused by an acoustic horn 4 along a path covered by a summing metal strip 5 having an output terminal 6. This compression increases the energy density of the acoustic wave thereby enhancing the non-linear interactions in the region of summing strip 5.

A reference surface acoustic wave, identical in form with the input acoustic wave, but time reversed is launched by interdigital transducer 7 onto the substrate 3 and focused by acoustic horn 8 along the path covered by the summing strip 5. The reference surface acoustic wave propagates in a direction opposite from the input acoustic wave. The distortion of the substrate along counter propagating paths of the input and reference acoustic waves produces a fluctuating voltage, due to the piezoelectric properties of the substrate, which is the sum of the energy of the two acoustic waves at each point along the path. These voltages are picked up and summed by the summing strip 5 to produce a waveform at output terminal 6 which is proportional to the convolution of the two surface acoustic waves, except that the produced waveform is compressed 2 to 1 in time. The launching of the reference acoustic wave at a 2 to 1 duty cycle causes a convolution of the input signal at the output terminal 6 continuous with time. The waveform at the output terminal 6 is compressed 2 to 1 and the desired portion of the waveform has a 50% duty cycle, the other 50% is discarded.

If, for example, the input signal has one of four different orthogonal phase modulated sequences, such as a 4-ary-OK (orthogonally keyed) modulation, four convolvers such as shown in FIG. 1 using different reference signals would have to be used in parallel. The four outputs are then compared to determine which one produces the largest signal at its output terminal at the compressed peak time.

In order to reduce the size, weight, and power consumption, a single time-multiplexed convolver 10 of the type shown on FIG. 2 may be used in place of multiple convolvers of the type shown in FIG. 1. The advantage of the convolver of FIG. 2 is that it is capable of time multiplexing the multiple reference signals by sampling each reference signal at the chip rate for a predetermined fraction of a chip width.

The samples are staggered in time for each time-multiplexed reference signal. To prevent sampling loss, the samples may be taken at the optimum phase of the reference waveform and the time difference between the optimum phase of the modulated input signal and the time-multiplexed position of the reference signal may be compensated before amplitude comparison.

To prevent the summing bus output from being a function of the multiplexed reference signals, a segmented summing bus 9 as shown on FIG. 2 is used. The convolver of FIG. 2 is similar to that shown on FIG. 1 having first and second interdigital transducers 2 and 7, acoustic horns 4 and 8 disposed on a piezoelectric substrate 3, except that in the interaction area of the two signals, the summing bus 9 is segmented having a plurality of summing fingers 12 selecting the product of the input signal and the samples of the desired reference signal. The fingers 12 have a width corresponding to 1/2M of the chip width of the surface acoustic wave in the substrate 3 where M is equal to the number of orthogonal phase modulations of the reference signal. For example, for a 4-ary OK orthogonal phase modulated reference signal, the width of the fingers would be one eighth of the chip width. The fingers 12 are spaced at chip-width intervals for sampling once per each chip. The fingers 12 are connected to a connector bus 14 disposed outside of the surface acoustic beam interaction region so that the signal induced in the fingers 12 is a sample of the input signal at one point in the convolution with one of the multiplexed reference signals. The next sample of the input signal is the signal induced in the fingers at another point in the convolution with another of the multiplexed reference signals. For example, with a 4-ary OK modulated input signal, there would be four time-multiplexed reference signals and four samples would be extracted, one corresponding to each of the four reference signals.

In order to prevent the fingers 12 from distorting the surface acoustic wave beam in the interaction area, filler segments 16 are disposed between the fingers 12, as shown, to give continuity to the wave guide like medium. Each of these filler segments 16 is connected to a grounded bus 18 outside of the interactive region of the beams.

In order to reduce spurious noise signals in the time multiplexed convolver, a second parallel convolver may be formed in parallel with the time multiplexed convolver shown on FIG. 2. The second convolver has a third interdigital transducer launching a third surface acoustic wave 180.degree. out of phase with the first surface acoustic wave launched by the first interdigital transducer of the time-multiplexed convolver. The second convolver may be of the type disclosed by Michaels as illustrated in FIG. 3 or may be a second time-multiplexed convolver as shown in FIG. 4. Referring first to FIG. 3, the first (lower) convolver is identical to that shown on FIG. 2 while the second (top) convolver is a conventional convolver of the type shown in FIG. 1. In this embodiment, the two first digital transducers 2 and 2a have a common terminal 200 receiving the input signal and the other two terminals 202 and 202a are grounded as shown. For simplicity, the identification numerals are identical with the numerals used with reference to FIGS. 1 and 2 except the elements of the second convolver are followed by the letter "a". As known in the art, the input signal may alternatively be applied to terminals 200 and 200a and the common terminal 200 grounded. The connections of the interdigital fingers of transducers 2a of the conventional convolver are reversed with respect to the terminal 200 so that the surface acoustic wave launched in response to the input signal is 180.degree. out of phase with the surface acoustic wave launched by transducer 2 of the time-multiplexed convolver. The length of the summing strip 5 of the conventional convolver is adjusted so that the surface acoustic waves launched by the first and third transducers 2 and 2a, respectively, are received by the second and fourth interdigital transducers 7 and 7a, respectively, remain 180.degree. out of phase.

The fingers of the second and fourth interdigital transducers 7 and 7a are connected identically between the terminals 204, 206 and 206a so that the surface acoustic waves launched by the second and fourth interdigital transducers 7 and 7a are in phase. With this arrangement the surface acoustic waves generated by the first and third interdigital transducers 2 and 2a are received by the second and fourth transducers 7 and 7a, and vice versa, are out of phase and cancel each other substantially reducing the generation of spurious signals in the convolver. In a like manner, the surface acoustic waves launched by the second interdigital transducers 7 and 7a are received out of phase by the first and third interdigital transducers 2 and 2a also reducing the generation of spurious signals by the first transducers in response to the received surface acoustic waves. The multiplex output samples of the convolvar are generated at the output terminal 6 of the first convolver as previously discussed.

Alternatively as shown in FIG. 4, the convolver may comprise two time-multiplexed convolvers of the type shown on FIG. 2. As in the embodiment shown on Figure 3, the fingers of the first and third interdigital transducers 2 and 2b are connected to terminal 200 displaced by 180.degree. so that the launched surface acoustic waves are 180.degree. out of phase to reduce the generation of spurious signals as discussed relative to the embodiment shown on FIG. 3. The time-multiplexed output of the convolver of FIG. 4 may be taken from the terminals 6 or 6b as desired or they may be added together by using an appropriate phase delay circuit (not shown) which will delay the phase of one or the other of the two output samples, placing the output samples of both convolvers back in phase. As is known, the connection of the fingers on the first and third interdigital transducers 2 and 2a may be identical and the connection of the fingers of second and fourth interdigital transducers 7 and 7a reversed to produce the same effect.

FIG. 5 is a block diagram of a spread spectum communication system embodying the time-multiplexed reference convolver of the type shown on FIGS. 2, 3 or 4. A Transmitter 20, for example, generates and transmits a 128 chip long binary phase shift keyed (BPSK) modulated signal with a pseudo-random noise spreading function. The transmitted signal has a M-ary orthogonally keyed data modulation superimposed on the chosen spreading function (SF). For example, the M-ary orthogonally keyed data modulation is a 4-ary orthogonally keyed (4-OK) data modulation in which the modulating waveforms are four Walsh functions such as Walsh function WAL (0,T), WAL (1,T), WAL (3,T), and WAL (7,T) shown on FIG. 7.

The 128 chip long phase shifted keyed modulated transmitted signal is received by a Receiver 30 and is converted to a convenient intermediate (IF) frequency in a Receiver Input 22. This intermediate frequency signal is applied to the input of the time-multiplexed Convolver 10 of the type shown on FIGS. 2, 3 or 4. A Reference Signal Generator 24 generates the required time-multiplexed reference signals which are applied to the opposite end of Convolver 10. The reference signals are reversed in time and have the same frequency as the intermediate frequency of the signal input to the Convolver 10 from the Receiver Input 22.

The convolution of the four time-multiplexed reference signals with the signal received from Receiver Input 22 in Convolver 10 produces four time-multiplexed output sampled signals. The four output sampled signals are separated into four separate channels in Demultiplexer 26 where the output samples s.sub.1 through s.sub.4 are gated and filtered. The data is finally recovered by comparing the relative amplitudes of the four channels in Comparator 28.

A detailed block diagram of the Transmitter 20 is shown on FIG. 6. Referring to FIG. 6, a Spread Function (SF) Generator 32 serially generates a pseudo-random noise spreading function code, 128 chips long, in response to a transmit (TMIT) trigger signal as shown by waveform SF on FIG. 7. The pseudo-random noise spreading function is also known by the Receiver 30. The spreading function is modulated in the Exclusive OR gate 34 by the Walsh Function sequence generated in the Walsh Function Generator 36 to generate a 4-ary orthogonally keyed data modulated signal superimposed on the chosen 128 chip spreading function signal. The Walsh Function sequence is generated by Walsh Function Generator 36 in response to the transmit trigger signal and mofulated by the data to be transmitted. The modulated unipolar output of Exclusive OR 34 is converted to a bi-polar signal by BI-polar Generator 38 and transmitted to a Balanced Modulator 40 through Gate 42. The gate 42 is enabled by Walsh Function WAL (0,T), shown in FIG. 7, generated by Walsh Function Generator 36.

An Oscillator 44 generates a carrier frequency (f.sub.T) signal which is modulated by the bi-polar Ex-Or'ed Walsh Function modulated spreading function signal received from the Bi-Polar Generator 38 to produce a carrier signal modulated with a 128 chip long binary phase shift keyed (BPSK) signal. The modulated carrier signal generated by the Balanced Modulator 40 is amplified by Amplifier 46 and transmitted to the Receiver 30 by antenna 48.

The details of the Receiver Input 22 are shown in the block diagram illustrated in FIG. 8. Referring to FIG. 8, an antenna 50 receives the transmitted signal. This signal is amplified in a Preamplifier 52 and combined in Mixer Amplifier 54 with a signal from a Local Oscillator 56 to generate an intermediate frequency signal. A Coupler 58 couples either the intermediate frequency signal generated by Mixer Amplifier 54 or a test signal generated by the Reference Generator 24 to a time-multiplexed Convolver 10, of the type shown on FIGS. 2, 3 or 4.

The details of the Reference Generator 24 are shown in the block diagram of FIG. 9. Referring to FIG. 9, the same pseudo random noise spreading function (SF) code used in the Transmitter 20 is generated in a Spreading Function Generator 60 and is initiated by an internally generated spreading function validity internal (SFVI) trigger signal. The 128 chip spreading function code is clocked into a Serial-In-Parallel-Out (SIPO) Shift Register 62. A trigger signal generated by a Single Shot (SS) Multivibrator 64 delayed by the proper time from the SFVI trigger signal by a Monostable Multivibrator 66 is used to transfer the spreading function code in parallel from the SIPO Register 62 to a Parallel-In-Serial-Out (PISO) Shift Register 68 which permits the spreading function code to be extracted in a time reversed order.

Clock pulses having a frequency eight times the chip rate of the transmitted signal are received by a 2.sup.7 Counter 70 from a Clock 72. The overflow of the 2.sup.7 Counter 70 is received by a 2.sup.4 Counter 74. The clock pulses generated at eight times the chip rate causes the most significant bit of Counter 74 to have a half period equal to 128 chips which is the desired symbol length as shown by waveform WAL (0,T) of FIG. 7. The next 3 most significant bits of Counter 74 produce the waveforms WAL (1,T), WAL (3,T), WAL (7,T), respectively. The opposite polarity waveform available in Counter 74 has the same effect as time reversing these waveforms. These are the four time reversed Walsh Functions required for the four reference signals.

In actual practice, time reversing of the Walsh functions is not necessary since the net effect is simply to change the polarity of the modulated signal, and this processing is independent of the relative phase between the input signal and the reference signal. The time reversing of the Walsh function, described herein, is used to simplify the understanding of the process.

The Data Selector 76 is controlled by the second and third lease significant bits of Counter 70 which causes each of the four outputs of Counter 74 to be sampled once during each chip width. The output of Data Selector 76 are segments of the four Walsh functions in a sequential order during each chip width as shown in FIG. 11 as waveforms RWAL (0,T,) RWAL (1,T), RWAL (3,T), and RWAL (7,T) and the time-multiplexed waveform (TMWF).

Referring now to FIG. 11, the waveform R-SF is the time reversed spreading function code output from PISO Register 68, the waveforms RWAL (0,T) through RWAL (7,T) are the time reversed Walsh functions available at the inverting output of Counter 74 and the waveform TMWF is the time-multiplexed Walsh function output from Data Selector 76.

To synchronize the Walsh function waveforms with the output of the spreading function code stored in PISO Shift Register 68, the positive going leading edge of the third least significant bit (LSB) of Counter 70 is applied to a single Shot Multivibratoe 78 to generate a clock trigger signal at the chipping rate of the transmitted signal. The output of Multivibrator 78 is applied to the clock input of PISO Register 68 through AND Gate 80. AND Gate 80 is enabled by the output of Flip Flop 82 which is set by the output of AND Gate 84 and reset by the spread function validity internal (SFVI) trigger signal which also activates Monostable Multivibrator 66. AND Gate 84 is enabled by the output of Monostable Multivibrator 66 and passes the signal generated by a Single Shot Multivibrator 86 to the set input of Flip Flop 82. The Single Shot Multivibrator 86 is triggered by the positive going leading edge of the inverted most significant bit of counter 74.

In operation, the spreading function validity internal (SFVI) trigger signal activates the Spreading Function Generator 60 to load the spreading function Generator 60 to load the spreading function code into the SIPO Register 62, resets Flip Flop 82 and activates Monostable Multivibrator 66 to generate an inhibit signal. After the spreading function code is loaded into the SIPO Register 62, the Monostable Multivibrator 66 recovers enabling AND Gate 84 and activating Single Shot Multivibrator 64 to generate a load signal which activates PISO Registers 68 and 88 to be loaded with the content of SIPO Registers 62. Upon the occurrence of the positive going leading edge of the inverted most significant bit of Counter 74 Single Shot Multivibrator 86 is activated to generate a signal setting Flip Flop 82 and enabling AND Gate 80. With AND Gate 80 enabled, PISO Registers are activated to shift out one bit for every eighth clock signal of Clock 72.

The spreading function code stored in SIPO Register 68 is shifted out one bit at a time in a reversed order at the chip rate coincident with the effectively time reversed time multiplexed Walsh functions from the Data Selector 76. The shifted out spreading function code bit and the time multiplex Walsh function segments are applied to separate inputs of an Exclusive OR Gate 90 where they are effectively multiplied together to form the desired time-multiplexed reference signal. The output from Exclusive OR Gate 90 is input to a Unipolar to Bipolar Converter 92 which converts the 1s and 0s of the unipolar output of Exclusive OR Gate 90 to a bipolar signal. The output of Unipolar to Bipolar Converter 92 is transmitted to a Balanced Modulator 96 through Gate 94 enabled by the inverted most significant bit (MSB) of Counter 74. The bipolar output of Unipolar to Bipolar Converter 92 modulates, in the Balanced Modulator 96, the frequency signal generated by an Oscillator 98. The frequency signal generated by Oscillator 98 has the same frequency as the intermediate frequency signal input to the other end of Convolver 10 from the Receiver Input 22. The modulated frequency signal generated in Balanced Modulator 96 is applied to Interdigital Tranducer 8 of the Convolver 10.

The convolution of the time-multiplexed reference signals with the signal received from Receiver Input 22 produces time-multiplexed output samples at output terminal 6 of Convolver 10 which are transmitted to the Demultiplexer 26.

The details of the Demultiplexer 26 are shown on FIG. 10. Referring to FIG. 10, the time multiplex output samples from time-multiplexed Convolver 10 are received by Demultiplexer Circuit 100 and are separated into four separate channels. Since the output of the Convolver 10 occurs in one half the time, the Demultiplexer Circuit 100 is synchronized with the Data Selector 76, FIG. 9, in the Reference Generator 24 by using the second least significant bit indicated as "X" output, and the least significant bit, indicated as "Y" output, of Counter 70. To compensate for the propagation time of the surface acoustic waves in the time-multiplexed Convolver 10, the second least significant bit "X" and least significant bit "Y" are phase delayed in a Phase Delay Circuit 102. This phase delay optimizes the switching of the received time-multiplexed output sample between the various channels.

The separated output samples in the four separate channels are individually applied to associated Filters 104 through 110 respectively. The band widths of the Filters 104 through 110 are approximately equal to twice the reciprocal chip width to make the sample signal a continuous function of time. The separated output samples are then individually amplitude detected in demodulators 112 through 118. The four demodulated samples S1 through S4 are then applied through Gates 120 through 126 to the Comparator 28 wherein the data is recovered by comparing the relative amplitudes of the four channels.

Since the output of Convolver 10 is a function of two signals propagating at the same velocity in opposite directions, it has the effect of speeding up time so that the convolved output sample waveform occurs in one half of the time and there are spurious signals in the second half. To eliminate these spurious signals, the Gates "Z" is applied to the input of a phase Delay Circuit 128, through 126 are controlled by the second order Walsh function WF-1,T obtained from the second most significant bit of Counter 74 and inverted by Inverter 130 as shown on FIG. 9. The output of Inverter 130, labeled output "Z", is applied to the input of a phase Delay Circuit 128, which delays the second order Walsh function to compensate for the transit time of the acoustic wave signal in convolver 10.

To accurately compensate for variations in the delay through Convolver 10 for the reference signals versus the environment, a test signal is used to automatically set the proper phase of the control signals for Demultiplier Circuit 100. This is accomplished by Parallel-In-Serial-Out (PISO) Shift Register 88 shown on FIG. 9. PISO Register 88 is loaded with the spreading function code stored in SIPO Register 62 in response to the signal generated by Single Shot Multivibrator 64. In contrast to PISO Register 68 in which the spreading function code is shifted out in reverse order, the spreading function code stored in SIPO Register 88 is shifted out by the signal from AND Gate 80 in the same order in which it was generated.

The spreading function 0 and 1 bits as they are shifted out of PISO Register 88 are converted to a bipolar format in Unipolar to Bipolar Converter 132. The bipolar spreading function bits are then applied, through Gate 134, to a second Balanced Modular where they modulate the intermediate frequency signal generated by Oscillator 98. The Gate 134 is enabled from the output of AND Gate 138 in response to the simultaneous occurrence of a built-in-test equipment (BITE) signal and the inverted state of the most significant bit of Counter 74 received from Inverter 140. The BITE signal may be generated periodically using specially designed build-in-test equipment or manually generated to periodically correct the phase delay of Phase Delay Circuit 102 of the Demultiplexer 26.

The modulated intermediate frequency signal generated by the Balanced Modulator 136 is applied to the input (interdigital transducer 2) of the Convolver 10 through Coupler 58 as a test signal. The convolution of this test signal with the reference signals are then applied to the Demultiplexer 26. The signal output from AND Gate 120 indicative of the first order Walsh function output by the Demultiplexer 26 is applied to the input of the built in test BITE Receiver 142. The output of the (BITE) Receiver 142 is applied to the input of a Delay Variation Compensator 144 along with the delayed least significant bit of Counter 70 output from Phase Delay Circuit 102. The Delay Variation Compensator 144 compares the phase of the signal received from BITE Receiver 142 with the phase of the delayed least significant bit of Counter 70 and generates a phase compensation signal applied to the Phase Delay Circuit 102. The Phase Delay Circuit 102 in response to the compensation signal varies the delay of the second least and least significant bits (x and y) applied to the Demultiplexer Circuit 100 automatically maximizing the amplitude of the samples S-1 through S-4 of the separated Walsh functions channels. The BITE Receiver 142 and Delay Variation Compensator 144 constitute a feedback loop during the generation of the built-in-test signal for optimizing the sensitivity of the Demultiplier 26 to the samples output from Convolver 10.

It is not intended that the invention be limited to the embodiment illustrated in the drawings and discussed in the specification. A person skilled in the art will recognize that the invention is not limited to 4-ary OK modulation or that other circuit arrangements may be used to provide the desired functions as described herein and set forth in the following claims.