Adaptive multiharmonic phase jitter compensation.

BACKGROUND

0001This invention relates to the correction of phase jitter which is experienced by digital data signals during transmission over media of limited frequency bandwidth. More particularly, this invention relates to phase jitter correction via the use of a multiharmonic adaptive phase compensator.

0002It has been well known in the art for some time to utilize limited frequency band channels for the transmission of data over telephone voice lines. Recently, the technology has advanced to the point where transmission of data at 14,40̸0̸ and even 19,20̸0̸ bits of dat per second is accomplishable. In providing such high data rates it has been necessary to increase the number of points in the transmitted constellation while maintaining the same average power. With more constellation points, it has become advisable to correct for and/or eliminate as many impairments as possible because given the identical bit rate and amount of noise and impairments, fewer errors would be expected to occur in a system with fewer constellation points. In other words, the closer the constellation points are one to the other, the smaller is the transmission impairment which will cause an error.

0003In the modems of the art, it is well known that an incoming signal is demodulated and phase corrected. The so-corrected signal is then passed to circuitry which decides to which constellation points it corresponds. By comparing in a phase detector the complex signal being fed into the decision circuitry and the complex signal put out by the decision circuitry, a waveform representative of the phase may be obtained. The resulting phase waveform may then be fed back via a phase jitter canceller to the phase corrector which corrects the phase of the incoming signal.

0004It will be appreciated that the phase waveform resulting from the comparison of the complex signals may be said to take the form of: <maths id="math0001" num=""><math display="inline"><mrow><msub><mrow><mtext>ψ = Δα sin(w</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>t + φ) + n(t)   (1)</mtext></mrow></math><img file="EP0340014B1_D0001.tif" /></maths> where Δα is the amplitude of the phase deviation in degrees or radians, w_j is the jitter frequency, φ is the initial phase, and n(t) is the noise. In correcting for phase jitter, it is evident that it is desirable to drive the phase ψ to zero. Indeed, various approaches which attempt to accomplish this correction are known in the art. For example, in U.S. Patent #4,320̸,526 to Gitlin, an adaptive phase jitter compensator seen in Figure 1 is provided. The compensator 20̸ obtains signals which are sent to decision circuitry 22, conducts in phase detector 24 a phase comparison on the incoming signals with the complex signals output by the decision circuitry 22, and sends the determined phase error to the input 26 and the coefficient update 28 of an adaptive infinite impulse response filter 30̸ which is essentially comprised of a finite impulse response filter 32 with feedback 34. The IIR 30̸, in accord with techniques well known in the art automatically tunes itself to the predominant phase jitter frequency and provides a signal indicative of the predominant phase jitter frequency and amplitude of the phase jitter. The sine and cosine of the output signal of the IIR (itself a sine wave) are then taken at means 36 such that when the values are multiplied by the equalized signal, the complex equalized signal is corrected and the phase error minimized. Thus, the IIR provides a signal with automatically adjusted amplitude, phase, and frequency.

0005A second arrangement for phase jitter compensation is seen in Figure 2 herein which substantially amounts to an equivalent arrangement of that described in U.S. Patent #4,253,184 to Gitlin et al. In Figure 2, it is seen that a digital oscillator 50̸ is provided and outputs the sine and cosine function for a given frequency and for second and third harmonics of the same. The sine and cosine functions for each harmonic are input into adaptive amplitude-phase correctors 52a, 52b, and 52c which also receive phase comparison information from a phase detector 54 which compares signals which are sent to decision circuitry 56 with the complex signals output by the decision circuitry 56. Given set frequencies w_j, 2w_j and 3w_j, the adaptive correctors 52 adjust the amplitude and initial phase so as to minimize the output of phase detector 54.

0006While the adaptive phase jitter compensation arrangement of U.S. Patent #4,320̸,526 does provide some phase correction, it will be appreciated that the provided arrangement assumes that all of the phase jitter is first order; i.e. that the phase jitter does not include any harmonics. When even small amounts of harmonics of the phase jitter frequency are present, however, the performance of the provided arrangement degrades significantly. On the other hand, while the phase jitter compensation arrangement of U.S. Patent #4,253,184 does account for harmonics and permits the phase and amplitude of the jitter to be adaptively found, the arrangement assumes a particular frequency for the jitter. However, if the jitter frequency is not known in advance, the compensation technique of Patent #4,253,184 cannot be implemented because it requires pretuning; i.e. it is not adaptive to jitter frequency.

0007Further, it should be appreciated that while a first arrangement provides a means for adapting to the frequency, amplitude and initial phase of the phase jitter, and a second arrangement provides a means for accounting for the phase jitter harmonics, the two arrangements cannot be easily combined. For example, the combination which would suggest itself would be to use additional adaptive IIR filters of U.S. Patent #4,320̸,526 in the parallel arrangement of U.S. Patent #4,253,184, with the phase error as decided by the phase detector as the input into each IIR filter in order to cancel the second and third harmonics. However, in providing such an arrangement, all three filters interact with each other and the entire system becomes unstable.

SUMMARY OF THE INVENTION

0008It is therefore an object of this invention to provide a multiharmonic adaptive phase jitter compensator.

0009It is a further object of the invention to provide a stable multiharmonic adaptive phase jitter compensator where second and third harmonics of an adaptively found fundamental phase jitter frequency are used as driving inputs to second and third IIRs.

0010In accord with the objects of the invention, a multiharmonic adaptive phase jitter compensator for a high speed modem generally comprises: <ul id="ul0001" list-style="none"><li>a) a primary filter means having an input, a coefficient update means, and an output, wherein the primary filter means is arranged to receive at its input and at its coefficient update means an error signal from a phase detector which compares an equalized phase corrected signal entering a decision means with the signal exiting the decision means, and wherein the primary filter means is arranged to adaptively provide at its output a wave indicative at least of the fundamental phase jitter frequency of the signal entering the decision means;</li><li>b) harmonic derivation means connected to the output of the primary filter means for obtaining at least one non-fundamental harmonic of the phase jitter frequency provided by the first filter means;</li><li>c) at least one secondary filter means, each secondary filter means having an input, a coefficient update means, and an output, wherein a particular non-fundamental harmonic is provided by the harmonic derivation means to an input of a corresponding particular secondary filter means while said error signal is provided to the coefficient update of each secondary filter means; and</li><li>d) summation means connected to outputs of said primary and secondary filter means for summing the outputs and applying the sum to a multiplier which multiplies a function of the sum with the equalized signals to provide a equalized and phase corrected signal which is sent to the decision means.</li></ul>

0011Preferably, the filter means are infinite impulse response filters (IIRs), and at least second and third IIRs are provided. The second IIR is supplied with the second harmonic and the third IIR is supplied with the third harmonic of the fundamental phase jitter frequency by the harmonic derivation means. The second and third harmonics train the IIRs to look for and correct for the second and third harmonics in the error signal. The second harmonic is essentially obtained by multiplying the output of the primary IIR <maths id="math0002" num=""><math display="inline"><mrow><msub><mrow><mtext>(A sinw</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>t)</mtext></mrow></math><img file="EP0340014B1_D0002.tif" /></maths> by itself to provide a value <maths id="math0003" num=""><math display="inline"><mrow><msub><mrow><mtext>A²/2 - (A²/2)cos2w</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>t</mtext></mrow></math><img file="EP0340014B1_D0003.tif" /></maths> , and by filtering to remove the d.c. component. The third harmonic is preferably obtained by cubing the output of the primary IIR to obtain <maths id="math0004" num=""><math display="inline"><mrow><msub><mrow><mtext>.75A³sinw</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><msub><mrow><mtext>t - .25A³sin3w</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>t</mtext></mrow></math><img file="EP0340014B1_D0004.tif" /></maths> , and subtacting therefrom the product of the primary IIR output and a scaled d.c. component of the square of the primary IIR output.

0012A better understanding of the invention, and additional advantages and objects of the invention will become apparent to those skilled in the art upon reference to the detailed description and the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

0013<ul id="ul0002" list-style="none"><li>Figure 1 is a block diagram of an adaptive phase jitter compensator of the art;</li><li>Figure 2 is a block diagram of an equivalent to a phase jitter compensator of the art which compensates for harmonics;</li><li>Figure 3 is a block diagram of the multiharmonic adaptive phase jitter compensator of the invention; and</li><li>Figure 4 is a schematic diagram of the harmonic computator of the multiharmonic adaptive phase jitter compensator of Figure 3.</li></ul>

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

0014Turning to Figure 3, a block diagram of the preferred multiharmonic adaptive phase jitter compensator is seen. As with the adaptive compensator of the art shown in Figure 1, the compensator 120̸ of Figure 3 obtains signals which are sent from an equalizer 118 which with other circuitry known to those skilled in the art (not shown) corrects for constant phase shift and frequency offset to decision circuitry 122. Phase detector 124 is provided to conduct a phase comparison on the incoming signals with the complex signals output by the decision circuitry 122. The signal resulting from the phase comparison information is then fed as an input into an adaptive infinite impulse response filter 130̸a (the "primary filter") which is essentially comprised of a finite impulse response filter 132a with feedback 134a, a summing junction 126a for the feedback and the filter input signal, and a coefficient update means 128a for the tap coefficients of filter 132a. Not only is the phase comparison signal sent to the summing junction 126a, but it is also sent to the coefficient update 128a of the primary filter. The phase comparison signal is further sent to the coefficient updates 128b and 128c of inifinite impulse response filters 130̸b and 130̸c (the "secondary filters"). However, as will be described hereinafter, the phase comparison signal is not sent as an input (i.e. to the summing junctions) to the secondary filters.

0015When fed with the phase comparison information, the IIR 130̸a, automatically tunes itself to the predominant (fundamental) phase jitter frequency (i.e. the phase jitter frequency having the highest amplitude; the "first harmonic") and provides in response thereto an output wave signal 136a of a single frequency of the form <maths id="math0005" num=""><math display="inline"><mrow><msub><mrow><mtext>A₁ sin(w</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><msub><mrow><mtext>t + φ</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>)</mtext></mrow></math><img file="EP0340014B1_D0005.tif" /></maths> , where A₁ is the amplitude of the phase jitter, w_j is the fundamental phase jitter frequency, and φ_j is the phase offset. The output signal 136a of the IIR is fed to a harmonic derivation means 170̸ which computes at least one non-fundamenal harmonic of the predominant phse jitter frequency according to means which are discussed hereinafter with reference to Figure 4. The higher harmonics are then sent to the summing junction inputs 126b and 126c of IIR filters 130̸b and 130̸c which are further comprised of finite impulse response filters 132b and 132c, feedback paths 134b and 134c, and coefficient updates 128b and 128c respectively. With the higher harmonics being supplied as the inputs to the IIRs 130̸b and 130̸c, and with the phase comparison information from the phase detector 124 being supplied to the coefficient updates 128b and 128c, IIRs 132b and 132c tune themselves to their particular harmonics and provide output waves 136b and 136c which are indicative of the phase and amplitude of the harmonic frequencies of the jitter. The outputs from IIRs 130̸a, 130̸b and 130̸c are then summed by summation means 180̸, and the sine and cosine of the sum is obtained at generator 182. The sine and cosine of the sum are then applied by a multiplier 184 as a corrective signal to the signals (which include phase jitter) being sent from equalizer 118 to decision means 122.

0016According to the preferred embodiment at least second and third IIRs 130̸b and 130̸c are provided to correct for phase jitter. It will be recognized, however, that if desired, additional IIRs trained to the fourth and succeeding harmonics could be provided if desired.

0017Turning to Figure 4, the preferred manner of supplying the second and third harmonic signals to the secondary IIRs 130̸b and 130̸c is seen. If the output signal from primary IIR 130̸a is taken as <maths id="math0006" num=""><math display="inline"><mrow><msub><mrow><mtext>Asinw</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>t</mtext></mrow></math><img file="EP0340014B1_D0006.tif" /></maths> , multiplier 186 is used to provide the second harmonic. Thus, by squaring the primary output signal, it will be seen that by simple trigonometric identity, the value <maths id="math0007" num=""><math display="inline"><mrow><msub><mrow><mtext>(A²/2)(1 - cos2w</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>t)</mtext></mrow></math><img file="EP0340014B1_D0007.tif" /></maths> is obtained. By filtering, the d.c. component of the signal may be removed, and the remaining signal is a function of twice the fundamental harmonic frequency. The remaining signal may then be provided as the input to the IIR 130̸b. It will be recognized that as the output from primary IIR 130̸a changes (adapts), the input signal into IIR 130̸b will also adapt.

0018The third harmonic signal is preferably obtained by cubing the output of the primary IIR (multiplying at multiplier 188 the output signal from primary IIR 130̸a by the squared value of the primary IIR 130̸a obtained at multiplier 186) to obtain <maths id="math0008" num=""><math display="inline"><mrow><msub><mrow><mtext>.75A³sinw</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><msub><mrow><mtext>t - .25A³sin3w</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>t</mtext></mrow></math><img file="EP0340014B1_D0008.tif" /></maths> , and subtracting therefrom at summing means 190̸ the product of the primary IIR 130̸a output and a scaled d.c. component of the square of the primary IIR output <maths id="math0009" num=""><math display="inline"><mrow><msub><mrow><mtext>(.75A³sinw</mtext></mrow><mrow><mtext>j</mtext></mrow></msub><mtext>t)</mtext></mrow></math><img file="EP0340014B1_D0009.tif" /></maths> . The product of the primary IIR 130̸a output and a scaled d.c. component of the square of the pimary IIR output is obtained by using a low pass filter 192 to filter the signal obtained at multiplier 186 (thus obtaining the value A²/2), scaling the obtained d.c. component by multiplying at multiplier 194 by a value of 3/2 (thus obtaining the value 3A²/4), and multiplying at multiplier 196 the scaled d.c. component by the output of the primary IIR 130̸a. In this manner, while the signal from phase detector 124 is being used as an input into coefficient updates 128b and 128c, the harmonics of the phase jitter are supplied to the summing junction inputs of respective IIR filters 130̸b and 130̸c to permit the filters to properly train on the respective harmonics.

0019There has been described and illustrated herein a multiharmonic adaptive phase jitter compensator for a modem. While a particular embodiment has been described it is not intended that the invention be limited thereby as it is intended that the invention be broad in scope and that the specification be read likewise. Thus, for example, while particular harmonic derivation means were provided for generating second and third phase jitter harmonics for training the second and third IIR filters, it will be appreciated that other harmonic derivation means such as a Hilbert transformer means with corresponding trigonemetrical relationships could be utilized to accomplish the same. Therefore, it will be apparent to those skilled in the art that additional modifications can be made to the described invention without deviating from the scope of the invention as so claimed.