WO1989002200A1

WO1989002200A1 - Adaptive jitter-tracking method and system

Info

Publication number: WO1989002200A1
Application number: PCT/US1988/002957
Authority: WO
Inventors: Joel A. Wolensky
Original assignee: Case/Datatel
Priority date: 1987-08-28
Filing date: 1988-08-29
Publication date: 1989-03-09

Abstract

A noise-immune jitter-tracking mechanism is implemented through the use of regenerative signal processing techniques to generate a first signal that tracks frequency offset (20) and very slow distortions in the phase of the incoming carrier signal (A), and a second signal that tracks phase distortions (30) of the incoming signal caused by phase jitter; these two signals are combined to generate a signal that can demodulate (10) with respect to the actual instantaneous frequency and phase shift of the incoming carrier (A), even taking into account distortions caused by phase jitter.

Description

ADAPTIVE JITTER-TRACKING METHOD AND SYSTEM

BACKGROUND OF THE INVENTION

Field of the Invention The invention relates to systems that perform signal demodulation, and more specifically to systems that are jitter-tolerant.

Discussion of Related Art Various known teachings that are believed to be related to various ones of the innovations disclosed in the present application will now be discussed. However, Applicant specifically notes that not every idea discussed in this section is necessarily prior art. For example, the characterizations of the particular patents and publications discussed may relate them to inventive concepts in a way that is itself based on knowledge of some of the inventive concepts. Moreover, the following discussion attempts to fairly present various technical alternatives (to the best of Applicant's knowledge), even though the teachings of some of those technical alternatives may not be "prior art" under the patent laws of the U.S. or of other countries. Similarly, the Summary of the Invention section of the present application may contain some discussion of prior art teachings, interspersed with discussions of generally applicable innovative teachings and/or specific discussion of the best mode as presently contemplated, and Applicant specifically notes that statements in the Summary section do not necessarily delimit the various inventions claimed in the present application or in related applications.

A typical data communications system may transmit data over long distances via transmission lines such as, e.g., telephone cables or the like. The typical frequency bandwidth of the telephone-type transmission line is in the range of 3,000 Hz, which is adequate for transmitting voice data. Some transmission lines may happen to provide a wider bandwidth; however, the widely-used existing standards do not require this wider bandwidth. Therefore, various encoding and modulating schemes are employed to transmit a high rate of information within the relatively small bandwidth allowed by the conventional transmission line, to better utilize high-speed data communications equipment. This band- width is defined by the immense existing investment in multiplexers, data interfaces, etc., and will not change soon. Even if wider bandwidth services become available, the 3000-Hz voice lines will remain widely and cheaply available for many years to come.

One class of such scheme, to which the present invention is primarily directed (but not necessarily limited) is quadrature amplitude modulation ("QAM"). In QAM, two signals may be modulated onto the same carrier frequency, but in phase quadrature to form a single QAM signal. A typical carrier frequency for a conventional telephone-type transmission line may be 1800 Hz.

In addition to enabling transmission of two signals over the same bandwidth as required for one signal, the information transmission rate is often further increased through the use of M-ary QAM. In general, M-ary communication encodes each pulse as one of a pre-defined set of M symbols, and a single M-ary pulse can transmit the information of log₂M binary digits. For example, for M=128, each QAM pulse can transmit the information contained in log₂(128)=8 binary data bits, thus enabling transmission of 8 times the information that could be transmitted in binary form. The M- ary QAM pulse can be expressed in terms of its phase arid amplitude. Thus, for M=128, there are 128 corresponding pairs of values for the phase and amplitude of the data pulse.

Typically, these corresponding pairs of values are predetermined, and form what is commonly referred to as the constellation of the signal. Figure 3 represents an oscilloscope trace of a 16-point QAM constellation. Data points are evaluated by comparing a received pulse to the points in the constellation, and choosing the constellation point nearest the received pulse as the received data value. A discussion of constellation-type data detection methods can be found in Digital Communications, by J. Proakis, at 139-60 (McGraw-Hill, 1983), which is incorporated herein by reference.

A higher value of M creates more points in the constellation, i.e., a higher "constellation density."

The advantage of choosing a high value for M is that more information can be transmitted over existing conventional transmission lines, at very high speeds. There are, however, accompanying disadvantages. For instance, as M increases, the corresponding increase in constellation density causes the allowable margin for variance in the received signal (the maximum amount the received signal may vary in phase or amplitude without causing the receiver to choose the wrong point in the constellation, thus causing data errors) to decrease.

To minimize the data errors, therefore, a receiver must be able to detect and compensate for distortion in the signal. The data signal is generally demodulated at the receiver by multiplying the incoming data signal with a sinusoidal signal that matches the phase and frequency of the carrier. To detect the frequency and phase of the carrier signal, a phase locked loop ("PLL") is commonly used. A discussion of the use of PLLs for demodulation is found in Phaselock Techniques, 2nd Edition 167-77, by F.M. Gardner (John Wiley & Sons 1979), which is incorporated herein by reference.

The PLL usually employs a phase-comparator, a loop filter and a voltage-controlled oscillator ("VCO"). Generally, the phase of the VCO output is compared with that of the incoming signal to generate an error signal. This error signal is then filtered (in a particular filter example, the error is averaged over some length of time to suppress noise), and used to control the VCO. As the error signal is minimized, the VCO becomes locked to the phase and frequency of the incoming signal. The VCO can thus function as a reference by which to demodulate the incoming signal. As mentioned, the phase error signal in conventional PLLs may be filtered so that the VCO can operate stably. This filtering also has the advantage of providing good noise immunity. Since noise is suppressed by the loop filter, it does not appear on the VCO control signal; therefore, the demodulating signal generated with reference to the VCO output is largely unaffected by noise on the incoming signal. One way to achieve satisfactory oscillator stability and loop noise rejection is to set the loop filter gain very low, and include an integrating operation in the filter transfer function. In this way the PLL can effectively track slow deviations in the phase of the incoming carrier, and can also track phase deviations caused by frequency offset. However, this loop filter characteristic has the disadvantage of reducing the PLL's ability to track rapidly changing phase distortion in the incoming signal.

Distortion in the received signal may have several causes. For instance, the transmission line may have inherent characteristics that attenuate certain frequency components of the signal, and that cause a relatively constant phase shift. Also, the signal may be distorted by noise, especially where the transmission line path travels through an area close to large electric induction motors, power transformers, power transmission lines, and the like. In addition to noise, phase jitt.er may appear on the transmission line and distort the incoming signal. This phase jitter may typically be caused by power line "hum" of 60-120 Hz (50-100 Hz in Europe), or ringing tones, e.g., 20 Hz. This jitter rotates the phase of the incoming signal, and if sufficient jitter is present, it can cause data errors even in the absence of any random noise.

An ideal incoming signal could be thought of as containing only two components: (1) data, and (2) a carrier, onto which the data signal is modulated. For demodulation, the incoming signal would be multiplied by exp(jH_c(t)), where H_c(t) is a time-variant function that represents the ideal instantaneous value of the carrier signal frequency and phase shift at a given time, t; thus:

H_c(t) = W_ct + θ,

where w_c is the carrier frequency, and θ represents the phase shift inherent in the transmission line.

In the case of a non-ideal signal, however, the carrier may be distorted by noise, frequency offset, a constant phase shift corresponding to the channel characteristic, relatively slow changes in the channel characteristic, and more rapidly-changing phase deviations caused by phase jitter. The PLL techniques described above can be (and commonly are) used to construct a demodulating signal that is relatively unaffected by random noise, and that tracks the phase and frequency of the carrier, including the phase shift and slowly-varying phase distortions. However, as previously mentioned, conventional PLL techniques do not facilitate generation of a demodulating signal that effectively tracks distortion caused by phase jitter.

The presence of phase jitter causes the data signal to deviate from its correct phase at a frequency corresponding to that of the phase jitter. Thus, for example, in the case of phase jitter caused by power line hum of 60 Hz, the phase of the data signal deviates from the correct phase, e.g., by ±10° at a frequency of 60 Hz. In Figure 3, for example, if 76 is a received data point distorted only by phase jitter of approximately ±7º, it will vary sinusoidally between the points

74 and 78. This phase deviation occurs in addition to the relatively slowly-varying random phase distortion caused by other sources. If the constellation is sufficiently dense, a very small uncorrected phase jitter (e.g., as little as ±5º) may be sufficient to cause data errors.

Further, users have varying needs, diverse modulation schemes, and various phase jitter rejection specifications that must be met. Therefore, certain users are highly sensitive to phase jitter, and require affordable, reliable, high-speed systems that are immune to phase jitter. In order to use a conventional PLL to track the periodic phase deviation of the incoming data signal caused by phase jitter, the PLL's filter gain must be increased. However, this approach is not feasible because of at least two distinct disadvantages: (1) an increase in filter gain sufficient to pass the phase jitter is also sufficient to pass noise, with the result of significantly degrading the PLL's noise immunity; and (2) the increase in random noise on the VCO control signal significantly reduces the PLL's stability, creating a greater risk that the PLL will be unable to lock onto the carrier of the incoming signal. Thus, the conventional PLL, although highly effective in tracking an incoming signal that is not distorted by phase jitter, is ineffective in tracking the incoming signal when it is distorted by phase jitter.

An alternate approach to the PLL is to implement a separate phase-predicting operation, and to apply phase corrections to the signal in a separate step from demodulation. Typically, the "correct" phase of the incoming signal (i.e., the phase of the incoming signal were it not for phase jitter) is predicted based on sequential phase samples. Based on this prediction, the incoming signal is rotated to the "correct" phase. An example of a device that embodies some of the above techniques is disclosed in U.S. Patent 4,639,939 to Hirosaki et al. Hirosaki et al. discloses a device that predicts the correct phase by statistically analyzing sequential instantaneous phase samples of the received complex baseband signal, and correlating the samples with error detector feedback. The received complex baseband signal is then phase rotated to minimize deviation from the predicted phase. Hirosaki et al. specifically discloses detection of the "correct" phase by statistical analysis of sequential instantaneous phase samples of the received complex baseband signal. Also, the phase correction operation is performed on the received baseband signal as a separate step from demodulation. Further, Hirosaki et al. specifically disclose generating the signal corresponding to the "correct" phase of the incoming signal from the instantaneous phase samples by statistically weighing the phase samples and adding them together. This approach has several distinct disadvantages.

A first disadvantage is that the operations of sampling, statistically weighting, and summing the detected phase samples results in a transfer function similar to that of a conventional finite impulse response filter, rather than the transfer function of a PLL. As a result, the detected "correct" phase signal is distorted by any noise present on the incoming demodulated baseband signal. As an example, one such commercially available system (the NEC Modem Model No. DSP 14400MII) has exhibited a four dB signal to noise ratio degradation when a 10° peak-to-peak, 60 Hz phase jitter is added to distort the received signal. These results were achieved in Applicants' testing laboratory. The "filter" transfer function of this type of system leads to another problem. In the specific case of phase jitter at 60Hz, for example, the phase jitter component is a 60Hz sinusoid. Therefore, the ideal jitter correction signal must likewise be a 60Hz sinusoid. The above-mentioned transfer function, however, is not capable of generating an exact sinusoid unless it is provided with an exact sinusoid.

The phase-sampling does not provide an exact sinusoid corresponding to phase jitter, so the system cannot generate an exact phase jitter estimate. Thus, thes jitter correction signal, ideally a sinusoid, suffers waveform distortion. As a result, the system's jitter-tracking performance is impaired by its inability tα create an exact phase jitter estimate.

Moreover, when compared with, e.g., conventional PEL methods, systems that feature phase rotation as a separate step from demodulation require slightly more sigjial processing.

In particular, systems that utilize a technique similar to that disclosed by Hirosaki et al. are relatively computation-intensive; consequently, they require more processing and more memory than PLL-type systems.

SUMMARY OF THE INVENTION

In this section, various ones of the innovative teachings presented in the present application will be discussed, and some of their respective advantages described. Of course, not all of the discussions in this section define necessary features of the invention (or inventions), for at least the following reasons: (1) various parts of the following discussion will relate to some (but not all) classes of novel embodiments disclosed; (2) various parts of the following discussion may relate to innovative teachings disclosed but not claimed in this specific application as filed; (3) various parts of the following discussion will relate specifically to the "best mode contemplated by the inventor of carrying out his invention" (as expressly required by the patent laws of the United States), and will therefore discuss features which are particularly related to this specific subclass of embodiments, and are not necessary parts of the claimed invention; and (4) the following discussion is generally quite heuristic, and therefore focuses on particular points without explicitly distinguishing between the features and advantages of particular subclasses of embodiments and those inherent in the invention generally.

The present invention provides several advantages and improvements over the devices and concepts known in the art.

As noted, phase jitter distortion in data signals is a basic problem in high-speed data communication. The presence of phase jitter of sufficient amplitude can cause errors even in the hypothetical "noiseless" transmission system. In the particular context of M-ary QAM, uncorrected phase jitter limits the constellation density; this limitation reduces the attainable information communication rate.

Further, technology is continually introducing devices that can process binary data at ever-increasing speeds. Cost-effective utilization of such high-speed devices depends upon the ability to use existing transmission lines for high-speed data communication, rather than to construct dedicated transmission lines, which would be very costly.

The above factors translate the presence of uncorrected phase jitter in data communication signals to higher information transmission costs. In particular, a QAM data communication system capable of achieving a high data rate is relatively intolerant of uncorrected phase deviations in the transmitted signal (because of the high constellation density).

It is therefore imperative that such a system have the ability to effectively track such phase jitter. Given an incoming signal, the present invention precisely tracks the following: (1) the ideal carrier frequency; (2) the constant phase shift caused by the channel characteristic; (3) slow changes in the channel characteristic; and (4) phase jitter. A preferred embodiment of the present invention generates a demodulating signal that precisely tracks the incoming signal, including any phase deviations caused by phase jitter.

To illustrate the operation of the present invention, the incoming signal, A, can be represented as follows:

A = X_n + X_d(t) [exp(j (W_ct + θ + H_r + H_j (t)))]

Equation 1 where: X_n is random noise,

X_d(t) is time-variant complex data, W_c (omega_c) is carrier frequency, θ is the channel characteristic, H_r is slow variation of the channel characteristic, and H_j (t) = A_j sin (W_jt), where

A_j is jitter amplitude, W_j is jitter frequency. As noted, conventional demodulating PLL techniques are capable of effectively rejecting the noise term, X_n, while tracking the carrier phase and frequency, W_ct + θ, and the random phase distortion, H_r. A preferred embodiment of the present invention adapts the concept of a conventional demodulating PLL to track these terms and to generate a corresponding "first phase estimate".

In addition, the present invention supplements the first carrier-tracking operation by separately tracking the phase jitter term, H_j(t). The result of the carrier-tracking operation is then combined with that of the jitter-tracking operation to control the principal demodulation. Combining the terms thus enables demodulation with respect to:

W_ct + θ + H_r + H_j(t),

i.e., the ideal carrier frequency and phase shift (W_ct + θ), as distorted by both phase jitter (Hj(t)) and channel deviations H_r. By utilizing PLL techniques, and using the jitter- tracking information in controlling the principal demodulation, a preferred embodiment of the present invention effectively tracks and compensates for phase jitter while avoiding problems associated with various known systems, and particularly while avoiding the previously identified disadvantages.

This enhanced jitter-tracking ability enables the use of a denser constellation in QAM and M-ary QAM data communication. A denser constellation allows the system to communicate a greater amount of information for a given time period, without compromising the data integrity. Moreover, this increased transmission data rate can be accomplished using conventional existing transmission lines, thus avoiding the enormous (and possibly prohibitive) expense of constructing new dedicated transmission lines with wider frequency bandwidths. Thus, the present invention may be used (and in the presently preferred embodiment, is used) as a key component in high-speed QAM modems.

Another key advantage of the present invention is its relative simplicity of implementation; in a preferred embodiment, the present invention can be implemented in one DSP chip with appropriate software (together with appropriate interface elements). The relative simplicity of the signal processing techniques used in this preferred embodiment would allow for the complete demodulation operation to be executed by one digital signal processing integrated circuit chip (in addition to the other digital signal processing (DSP) chips which would normally be used for other system functions in a complete modem). Some commercially available systems that are known to use the "predict phase, then phase- rotate signal, then demodulate signal" technique require slightly more extensive and complex signal processing. The ease and low cost of implementing the present invention is made possible by the relative simplicity of the method that may be used to generate the principal demodulating signal. To illustrate, in a preferred embodiment of the present invention, the principal demodulating operation is followed by a constellation-type data detection routine, which may be similar to, e.g., that described in Digital Communications (previously referenced). In the particular case of M-ary data encoding, the result of the data detection operation represents received M- ary encoded data, and can be used as the reference by which phase deviations of the incoming signal are measured. A significant teaching of the present application is generating a first phase-error approximation (which approximately corresponds to the unfiltered error signal in a conventional demodulating PLL), and performing further operations to derive an accurate jitter-tracking signal from this approximation.

The first phase-error approximation, E_Φap, contains basically four types of information: (1) random noise;

(2) the channel characteristic (constant phase shift);

(3) phase distortion of the incoming signal caused by phase jitter; and (4) phase distortion caused by other sources. The jitter-tracking signal is created by extracting the phase jitter information.

There are various ways to generate the first phase- error approximation; as mentioned, the preferred embodiment adapts the concept of the conventional demodulating PLL.

The jitter-tracking signal generation is implemented by using a signal regenerative function (prefer ably a phase-locked loop operation). The signal regenerative function ensures that the jitter tracking signal is not affected by random noise on the first phase-error approximation. In the presently preferred embodiment, a first phase estimate, Φ₁(t), (which approximately corresponds to the filtered error signal in a conventional demodulating PLL) is produced by performing a filtering operation (using a second-order integrating filter characteristic) on the first phase-error approximation, E_Φap. The filtering operation extracts the components of the first phase-error approximation which relate to the carrier frequency and channel characteristic, and slowly varying phase distortion caused by sources other than phase jitter.

As discussed above, this filtering removes the phase jitter information from the first phase-error approximation. The resulting first phase estimate, Φ₁(t), contains information regarding the ideal instantaneous value of carrier frequency and phase shift (W_ct + θ), and random phase distortion (H_r), i.e., phase distortion of the incoming signal caused by sources other than phase jitter (such as slowly-varying phase distortion caused by changing channel characteristics). Alternatively, the first phase estimate, Φ₁ (t), could be derived in another fashion, and advantages would still be derived from the teachings related to tracking the jitter component using a signal-regenerative operation. Nevertheless, the additional functionality just described provides additional advantages, and is therefore preferred.

This information is utilized in generating a first demodulating signal, which is used for a "dummy" demodulating step. In the preferred embodiment, a periodic function generating operation (e.g., sine/cosine lookup) produces the "dummy" demodulating signal. Since the carrier frequency of the incoming signal is generally known (a typical carrier frequency for QAM signals is 1800 Hz), the periodic function generating operation is preferably initialized to the known carrier frequency.

Since the first phase estimate, Φ₁(t), contains an insignificant magnitude of noise, it can advantageously be used as a noise-immune component in generating the principal demodulating signal. However, the "dummy" demodulating operation does not track phase jitter. Therefore, to generate an accurate, noise-immune principal demodulating signal, the first phase estimate must be supplemented with the noise-immune jitter- tracking signal.

In terms of equation 1, the first phase estimate, Φ₁(t), can be expressed as:

Φ₁(t) =W_ct + θ + H_r.

Also from equation 1, the instantaneous frequency and phase shift of the incoming signal, H_in(t), can be (not counting data and noise) expressed as

H_in(t) = W_ct + θ +H_r+ H_j(t).

Therefore, the precise phase-tracking principal demodulating signal,

PD = exp[(j)(W_ct+ θ + H_r + H_j(t))], can be created from Φ₁(t) and JT(t), where JT(t) is the jitter-tracking signal, and

JT(t) = H_j(t).

The next step in implementing the present invention, therefore, is to extract the phase jitter information from the first phase-error approximation. Once the phase jitter information is extracted, it may then be used to generate a noise-immune jitter-tracking component.

As mentioned above, such a noise-immune jitter- tracking component assures that the principal demodulating operation is accurately referenced to the phase distortion in the incoming signal caused by phase jitter.

In accordance with some embodiments of the present invention, a signal regenerative operation is preferably used to extract the phase jitter information from the first phase-error approximation.

Phase jitter is periodic (e.g. at 60 Hz), whereas the other common type of phase distortion varies randomly, and relatively slowly. One particular function that is well-adapted to track a discrete frequency component while providing good noise rejection is a phase-locked loop. Accordingly, a preferred embodiment of the present invention uses PLL methods to extract the phase jitter component from the first phase- error approximation. The carrier-tracking operation detects non-data shifts in phase and frequency (i.e., produces the first phase error approximation) by comparing its (demodulated) input to a data detect output. However, the jitter-tracking operation input (the first phase-error approximation) does not have a corresponding requirement for a data detection routine; therefore, the jitter- tracking operation preferably utilizes a different phase-detection function than that used by the carrier- tracking operation.

In the preferred embodiment of the present invention, the phase-detect function in the jitter-tracking operation is implemented by multiplying the first phase- error approximation by a sine wave. Because the phase- jitter may vary at one of several frequencies (e.g., 20 Hz for ringing tones, 60 or 120 Hz for power line "hum" in the U.S., 50 or 100 Hz in Europe), the jitter- tracking operation must be capable of tracking a wide range of jitter frequencies. Therefore, the sine wave of the phase-detect function is monitored, and is maintained within a certain frequency range (e.g., 0-300 Hz). Using this monitoring technique, the system can track any single frequency component of the phase-jitter within that range. The sine wave is generated in a periodic function generating operation (e.g., a sine look-up).

The result of the multiplication is filtered, and this filtered result subsequently drives the periodic function generating operation. The filtered multiplication result (the "jitter phase estimate," Φ_j) represents the instantaneous frequency and phase of a sine wave at the frequency of, and in phase quadrature, with the phase jitter. That is,

Φ_j(t) = W_jt - π/2 Note that much of the present application describes the presently preferred embodiment, which uses a single jitter-tracking operation to track a single jitter frequency term; however, the present invention can simultaneously track several jitter frequency terms simply by using several corresponding jitter-tracking operations. The jitter estimates of the respective jitter-tracking operations can then be combined (e.g., by summing them) to provide an improved demodulating reference signal.

Applicant specifically notes that the above- described technique is only one of many ways in which the phase jitter tracking operation of the present invention may be implemented. These various alternative techniques are within the scope of the teachings of the present invention.

The jitter phase estimate Φ_j is preferably used to generate a signal that is both phase- and frequency- matched to the phase jitter. This is most easily accomplished by providing the jitter phase estimate, Φ_j, to a cosine function generating operation. The cosine function compensates for the quadrature-shift effect of the above-described phase detection. Because of the PLL methods that may be used, the resulting jitter-locked signal:

JL(t) = cos(W_jt-π/2) = sin(W_jt)

is relatively unaffected by noise.

The above-described signal-regenerative techniques produce two relatively noise-immune signals: JL(t), the jitter-locked signal; and Φ₁(t), the first phase estimate. These two signals can thus control the principal demodulating operation.

The high phase accuracy of the resulting demodulating signal obviates the need to make any phase adjustments in the incoming data signal to compensate for phase jitter. Significantly, jitter-tracking systems similar to those discussed in the Background section above do not exhibit the ability to generate such a high accuracy demodulating signal. Note that previous attempts to track and compensate for phase jitter have included a phase-correction operation that is performed on the incoming signal separately from the demodulating step.

Although the jitter-locked signal (produced by the cosine function generating operation) produces a signal that is locked to the phase and frequency of the phase jitter, the above-described jitter-tracking operation does not detect the amplitude of the phase jitter. In an optional embodiment of the present invention, therefore, a gain-control section is included to detect the amplitude of the phase jitter, and to adjust the amplitude of the jitter-locked signal accordingly.

The jitter-tracking signal, JT(t), is created as follows:

JT(t) = (G_n)(JL(t)) = A_jsin(W_jt)

where G_n is the gain control. This is done by first detecting the instantaneous phase deviations of the demodulated principal baseband signal B with reference to the received (detected) data D. The phase-error detect function 50 is preferably identical to that (24) from which the first phase-error approximation results. In contrast, however, the detected phase-error E_Φ essentially corresponds only to that portion of the phase distortion in the principal baseband signal, B, that is caused by phase jitter.

The jitter-tracking signal JT(t) is essentially a regenerated version of a sine wave matching the frequency and phase of the actual phase jitter. Therefore, a high correlation between the jitter-tracking signal, JT(t), and the jitter-induced phase-error, E_Φ, indicates that the principal demodulating operation is not effectively tracking phase jitter.

In an optional embodiment, the present invention uses this correlation to develop a gain control G_n, the amplitude adjustment of the jitter-tracking signal.

The correlation function includes a step by which the rate-of-change of G_n is made to depend on the product of the phase-error and the jitter-tracking signals. If there is no phase jitter in the incoming signal, or if the principal demodulating signal is effectively compensating for any phase jitter present, the phase-error will be very close to zero. Therefore, the amplitude of the jitter-tracking signal is very stable. To further stabilize the jitter-tracking operation, another optional embodiment of the present invention features an operation to detect the condition where the jitter-tracking signal is phase- and frequency-locked to the phase jitter. In this condition, the product of the first phase-error approximation and the jitter-tracking signal contains a strong DC component. The presence of the strong DC component in this product signal produces a phase-lock signal. This phase-lock signal is used to control the rate- of-change of the amplitude adjustment further, and also to control the gain in the filtering step of the jitter- tracking operation. Both this filter gain and the gain control rate-of-change are set higher when there is no lock condition. The higher gain permits faster acquisition of the phase, frequency and amplitude of any phase jitter present.

A significant additional advantage provided by the gain control is this: in the absence of any phase jitter on the incoming signal, the jitter-tracking signal is a constant with very low amplitude. Its amplitude is further reduced by the near-zero phase error between the principal baseband signal and the received detected data.

Thus, the phase jitter tracking signal generation system "vanishes" in the absence of any phase jitter on the incoming signal. This assures that the system will not contribute to performance degradation of the receiver when the jitter-tracking system is not needed.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 shows a flow chart representing one method of implementing a preferred embodiment of the present invention.

Figure 2 shows a detailed schematic block diagram of a receiver containing the present invention.

Figure 3 is a complex phasor illustration representing the QAM constellation, and the effects of phase jitter on received data.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

As previously noted, the preferred embodiment of the present invention is implemented in a DSP. A preferred embodiment of one system for implementing the present invention is therefore most accurately described with reference to the flow chart of Figure 1. Nevertheless, for illustrative purposes, and to suggest other systems that may be used for implementing certain elements of the various embodiments of the present invention, reference is also made to the detailed schematic block diagram depicted in Figure 2.

Referring now to Figure 1, the first step (step 0.0) in implementing the present invention is to initialize certain system values. The initial gain, G_(-1) (which adjusts the jitter-locked signal to create the jitter-tracking signal), is set to zero; this initializes the jitter-tracking signal JT(t), to zero.

By setting the first phase estimate to W_ct, the "dummy" demodulating operation and the principal demodulating operation are initialized to the carrier frequency, W_c, which, as mentioned, is a known value.

The frequency of the sine function generator in the jitter-tracking operation is monitored and maintained within a certain range (e.g., 0-300 Hz), as previously described. It may be desirable to track several frequencies using jitter-tracking operations, in which case the frequency of each sine function generator would be maintained according to the corresponding jitter frequency component.

Once initialized, the system is ready to receive the incoming signal, as shown at step 1.0. Various methods for receiving a modulated signal may be used. In a preferred embodiment, the present invention receives an M-ary QAM signal, and provides it to the DSP in the following manner: An analog-to-digital converter ("A/D") first receives the signal. The A/D outputs the digital version of the incoming signal through a filter, to a Hubert transform. The Hubert transform shifts the signal by -π/2, then passes it to an adaptive equalizer. The equalizer amplifies frequency components of the incoming signal that have been attenuated by the transmission line. The resulting digital unmodulated and equalized signal is then provided to the demodulating section of the receiver, which includes the present invention.

As shown in Block 2.0, the next step in implementing the present invention is to demodulate the incoming signal. Specifically, Block 2.0 indicates generation of the precise instantaneous phase value of the incoming signal, H_in(t), by combining the first phase estimate with the jitter-tracking signal:

H_in(t) = Φ₁(t) + JT(t)

= W_ct + θ + H_r + H_j(t) where JT is the jitter-tracking signal, and JT(t) = A_jsin(W_jt) = H_j(t)

The jitter-tracking operation is not yet active at the first instant of receiving the incoming signal; i.e., the jitter-tracking signal, JT(0) = 0. During the "start-up" interval therefore, the system demodulates the incoming signal with respect only to the initialization value of the principal demodulator (the carrier frequency, W_c). Hence, the initial principal demodulation consists of multiplying the incoming signal by exp(_jw_ct). Recalling the representation of the incoming signal;

A = X_n + X_d(t) [exp(j(W_ct+ θ + H_j(t) + H_r))],

it can be seen that the initial principal demodulation does not reflect any phase deviation (θ, H_j, H_r, or X_d) of the incoming signal. However, the system tracks the remaining three terms through successive iterations.

The step following the principal demodulation is the detection of data from the principal demodulator output, as indicated at Block 5.0. The principal demodulator outputs a principal baseband signal, B. In the particular context of M-ary QAM, this output is commonly referred to as the "constellation point" of the modem, and is sometimes displayed on an oscilloscope. One such display, a 16-point constellation pattern with received baseband data point B, is shown in Figure 3. The points B, C and D represent instantaneous samples of principal baseband signal B, baseband approximation C and data output D, respectively.

The principal baseband signal, B, can be expressed in terms of two components:

B = B' + N,

where B' represents the actual transmitted data; and N represents noise, phase distortion, etc. The signal B' is a complex signal, and can therefore be expressed as:

B' = B'_re + jB'_im where B'_re represents the real component, and B'_im represents the imaginary component (in the DSP implementation of the invention, complex numbers are represented as ordered pairs). The values of B'_re and B'_im are generated by the transmit modem and usually chosen from a finite set of integer values (i.e., the constellation). The data detector receives the signal

B = B' + N,

and must estimate B' from B by choosing the constellation point nearest B. As mentioned, the above- referenced Digital Communications discloses a typical method that may be used for the data detection. If the magnitude of N is not too large, the data detector functions without making any errors. On the other hand, uncorrected phase jitter increases the magnitude of N, thereby causing more data detection errors for a given level of background noise. The next step (Block 6.0) is detecting instantaneous phase deviation, E_Φ, of the principal baseband signal. The complex signal, D (data output, also ideally equal to B'), output by the data detector, and the principal baseband data point, B, can be represented as complex phasors, as shown in Figure 3. The angle E_Φ represents phase error, assuming that the data detector chose the correct point.

If B = B_re + jB_im, and D = D_re + jD_im, then the phase error, E_Φ, can be approximated by the following formula:

E_Φ represents the instantaneous amplitude of the phase jitter, and is used in a subsequent step (Block 12.0) to set the jitter-tracking signal amplitude.

The first phase-error approximation, E_Φap' is similarly obtained (see Block 8.0), where D is compared with the output of the "dummy" demodulator, C. In that case the first phase-error appoximation, E_Φap, is determined as follows (see Figure 7):

In the preferred embodiment, the incoming signal is next provided to the "dummy" demodulator (Block 7.0). The incoming signal (the output of the passband equalizer) is a complex signal with a real component and an imaginary component. To illustrate, the incoming signal. A, can be expressed using complex notation as

A = A_re + jA_im,

where A_re represents the real component, and A_im represents the imaginary component. The dummy demodulation is accomplished by multiplying the incoming signal by the complex exponential

exp(jΦ₁(t)),

where Φ₁(t) is the first phase estimate. Using the substitution

exp(jx) = cos(x) + jsin(x), the preferred embodiment of the present invention utilizes a "look-up" technique to generate the first demodulating signal. This technique utilizes an input corresponding to a value between 0 and 2π (i.e., corresponding to the phase estimate). The "look-up" then outputs the values of the sine function and the cosine function corresponding to the input value by searching through a memory table in which various values for these functions are stored.

This technique requires more memory than other available methods; however, it is preferred for implementation of this invention because it is faster and simpler to accomplish in real-time than computational-type methods. The "dummy" demodulation occurs prior to the first phase-error approximation. Therefore, the first-loop "dummy" demodulation is executed with respect only to the ideal carrier phase, W_ct. In subsequent loops, however, the operation is executed with respect to

W_ct + θ + H_r,

where θ is the channel characteristic, and H_r is slowly- varying channel distortion. The first phase-error approximation follows the dummy demodulation (Block 8.0), and has already been described.

As noted, E_Φap contains information regarding phase jitter. Specifically, the phase jitter appears on the first phase-error approximation as a sinusoid (assuming that the phase jitter is sinusoidal) at the frequency of, and proportional to the amplitude of the phase jitter. The next step therefore (Block 9.0), is to extract the phase jitter information from E_Φap to create the jitter-locked signal, JL(t).

The preferred embodiment of the present invention extracts the jitter component with a signal regenerative operation that adapts the concept of a wide pull-in range PLL.

The phase detector function consists of multiplying the first phase-error approximation by sin W_ot, where W_o becomes W_j (the jitter frequency) as the signal regenerative operation acquires the phase jitter.

Multiplication by sin(W_jt) produces a signal that can be processed according to the second order integrating filter technique. The second order integrating filter technique is preferable because of its favorable noise rejection characteristic. The second order integration results in a signal that varies linearly at the frequency of, but in phase quadrature with the phase jitter. This signal is the jitter phase estimate, Φ_j(t):

Φ_j(t) = W_jt - π/2.

Therefore, this signal can drive the sine wave generating operation for the phase detector function, by providing it to a sine wave "lookup" operation.

Similarly, by supplying this jitter phase estimate, Φ_j(t), to a cosine "lookup" operation, the quadrature phase shift is corrected, and a reconstructed cosine wave at the frequency of, and in phase with the jitter signal is created. Because of the PLL methods utilized by the signal regenerating operation in the preferred embodiment of the present invention, the reconstructed cosine wave, the "jitter-locked signal" (JL(t)), has significantly less waveform distortion and noise than the first phase-error approximation.

Since the particular embodiment of the invention disclosed in this section is a digital implementation, the amplitude of the jitter-locked signal, JL(t), is always unity. An optional embodiment of the present invention therefore features a gain control function (Block 10.0) through which the amplitude of the jitter- locked signal is adjusted to correspond with that of the interfering phase jitter.

Accordingly, the phase-error, E_Φ, is combined with the jitter-locked signal, JL(t), to produce a gain signal, G_n, that makes the required adjustment in the amplitude of the jitter-locked signal. A high correlation between the phase error and the jitter-locked signal indicates a high amount of phase jitter. Therefore, the gain control, G_n, is generated according to the correlation between the phase error E_Φ, and the jitter-locked signal, JL(t); specifically, the two signals are multiplied, and the result is subtracted from the previous gain control, G_n-1. This is known as the "Least-Mean-Square" ("LMS") technique, and is preferably implemented as set forth below:

G_n = G_n-1 - A_j1(JL(t)) (E_Φ)

where Aj1 is the "gain" of the LMS.

The desirable feature of LMS is that the gain will automatically set to zero when no phase jitter is present. That is, the system of the present invention "vanishes" in these conditions; therefore, this characteristic ensures that the jitter-tracking system will never cause performance degradation, even when no phase jitter is present.

Another gain-control feature presented in an optional embodiment of the present invention is the lock-detecting operation (indicated at Block 11.0). A lock-indicating signal, A_j1 (the LMS "gain"), is generated in accordance with acquisition of the phase jitter. When the jitter-tracking operation has acquired the phase jitter information from the first phase-error approximation, the product of the jitter-locked signal, JL(t), and the first phase-error approximation, E_Φap, contains a strong DC component.

Accordingly, to implement the lock-detecting operation, these two signals are multiplied together. To remove the effect of the varying amplitude of the first phase-error approximation, the product is normalized through a hard limiting operation. Next, higher frequency components of the normalized product are removed. The resulting signal corresponds to the DC component of the normalized product. If this signal exceeds a certain threshold level, then phase-lock is declared, and A_j1 is set to a low value. The lock- indicating signal controls the value of A_{j 1} in the gain- control operation, and also controls the gain in the second order integrating filter function of the jitter- tracking operation. In an unlocked condition, both gains are set higher to permit faster rate-of-change of the adjusting signals, thus permitting faster acquisition of the phase jitter.

As noted, the "look-up" technique used to generate the demodulating signals requires an input that varies at the frequency of, and in phase with, the phase of the signal with respect to which demodulation is to be performed. The first phase estimate, Φ₁(t) , is created (Block 13.0) by performing a second-order integration on the first phase-error approximation (as schematically illustrated in Figure 2 by the loop filter).

However, to stably acquire the phase and frequency of the incoming carrier, this second order integrating filter operation must exhibit a low-gain, narrow bandwidth filter transfer function. (The low gain provides low noise, and the narrow bandwidth permits efficient matching to a known standard carrier frequency.) In addition to the stability consideration, this filter characteristic is also desirable because it rejects random noise, the presence of which would distort the phase data input to the demodulating generator, and significantly impair the accuracy of demodulation.

The disadvantage of this filter characteristic is that, in suppressing the higher-frequency components of the first phase-error approximation, the filter operation also suppresses information regarding phase jitter. As a result, the dummy demodulation is incapable of tracking phase jitter.

However, the first phase estimate, Φ₁(t), does contain information regarding the ideal carrier frequency and phase shift, W_ct + θ, and channel deviations, H_r:

Φ₁(t) = W_ct + θ + H_r.

Accordingly, returning through Block 1.0 to Block 2.0, the instantaneous phase of the incoming signal, H_in(t), is created by combining the first phase estimate with the jitter tracking signal:

H_in(t) =Φ₁(t) + JT(t)

= W_ct + θ + H_r + H_j(t).

As indicated at Block 12.0, the jitter-tracking signal JT(t), is created next. This is accomplished as follows:

JT(t) = (G_n)(JL(t)) = A_jsin(W_jt).

Referring now to Blocks 3.0 and 4.0, the principal demodulating operation can be described. The combined signal H_in provides an input to the demodulating signal

"look-up". The resulting principal demodulating signal,

PD, is:

PD = exp(jH_in),

where H_in is the actual instantaneous frequency and phase of the carrier signal. The principal demodulating signal, PD, is thus precisely referenced to for any instantaneous phase distortion in the incoming signal, whether caused by phase jitter or by other factors.

The system may now be described with reference to the schematic block diagram of Figure 2. The system may include principal demodulator 10, "dummy" demodulator 20, jitter-tracking operation 30, gain control 40, phase error detector 50, and data detector 60. The incoming signal A is the output of the passband equalizer; the signal A is input to the principal demodulator 10 and the "dummy" demodulator 20.

The principal demodulator 10 outputs the principal baseband signal B from which the data detector 60 produces data output D. The signals B and D are input to phase-error detector 50. Detector 50 outputs the phase error E_Φ, which corresponds to the amplitude of the phase jitter (and noise).

The "dummy" demodulator 20, which outputs the first phase error approximation E_Φap, and the first phase estimate, Φ₁(t), may include mixer 22, phase error detector 24, second-order integrating filter 26, and sin/cos generator 28. The sin/cos generator 28 produces the demodulating signal

exp(jΦ₁(t)) = exp(j(W_ct + θ + H_r)) .

This demodulating signal is multiplied with the incoming signal A at mixer 22 to produce a baseband approximation C. The phase error detector 24 compares signal C to data output D; the detected phase error is the first phase error approximation, E_Φap. This approximation is then filtered and integrated by filter 26 to produce the first phase estimate, Φ₁(t), which corresponds to the carrier frequency and phase shift, and tracks frequency offset and slow changes in the channel characteristic. The first phase estimate Φ₁(t) is input to the sin/cos generator 28, and to summer 6 to the principal demodulator 10.

The first phase-error approximation E_Φap is also input to jitter-tracking operation 30. Jitter-tracking operation 30 generates the jitter-locked signal JL(t), and may include mixer 32, second-order integrating filter 34, sine generator 36 and cosine generator 38. The loop consisting of 32, 34, and 36 creates the jitter-phase estimate Φ_j (t):

Φ_j (t) =W_jt - π/2

Cosine generator 38 compensates for the π/2 phase shift, and outputs the jitter-locked signal JL(t) in response to the input Φ_j(t). The amplitude of JL(t) is then adjusted at gain control 40 to correspond with the amplitude of the phase jitter.

Gain control 40 may include lock detector 42, mixers 44 and 48, summer 45 and delay 46. Gain control 40 produces the jitter-tracking signal JT(t) by multiplying gain signal G_n with the jitter-locked signal JL(t) at mixer 48:

JT(t) =G_n(JL(t)).

The gain signal G_n is produced by mixer 44, summer 45 and delay 46 according to the LMS:

G_n= G_n-1 -A_j1(JL(t))(E_Φ).

Lock detector 42 also controls the gain of filter 34 according to whether jitter-tracking operation 30 has acquired the phase jitter. The filter gain is kept high until phase-lock is achieved; this allows faster acquisition of the jitter and a more stable lock once the jitter is acquired.

The two signals Φ₁(t) and JT(t) are combined in summer 6 of principal demodulator 10. The combined signal:

H_in(t) = W_ct + θ + H_r + H_j(t)

is provided to sin/cos generator 4, which produces the principal demodulating signal PD:

PD = exp[j(W_ct + θ + H_r + H_j(t))]

The incoming signal A is demodulated by the principal demodulating signal PD at mixer 2. This multiplication produces the principal baseband signal B.

It should be understood that the foregoing embodiments are representations of the present invention, and the full extent of the present invention is defined only by the claims.

Claims

CLAIMSWhat is claimed is:

1. A receiver system, comprising: a first signal generating operation which generates a first phase-error approximation corresponding to phase deviations of the incoming signal; a jitter-tracking operation, which generates a jitter-tracking signal in accordance with said first phase-error approximation; and a principal demodulating operation, which outputs a principal baseband signal corresponding to the incoming signal demodulated with respect to a combined signal which includes said jitter-tracking signal.

2. The system of Claim 1, wherein said jitter- tracking operation: generates a jitter-locked signal corresponding to phase jitter on the incoming signal; adjusts the amplitude of said jitter-locked signal in accordance with phase deviations of said principal baseband signal; and outputs said jitter-tracking signal in accordance with the amplitude-adjusted jitter-locked signal.

3. The system of Claim 1, wherein said jitter- tracking operation comprises: a signal regenerative operation, which generates a jitter-locked signal corresponding to phase jitter on the incoming signal; and a gain control function, which adjusts the amplitude of said jitter-locked signal in accordance with phase deviations of said principal baseband signal.

4. The system of Claim 2, wherein: said jitter-locked signal corresponds to the phase and frequency of phase jitter on the incoming signal.

5. The system of Claim 3, wherein: said jitter-locked signal corresponds to the phase and frequency of phase jitter on the incoming signal.

6. The system of Claim 1, wherein: said jitter-tracking signal corresponds to the phase, frequency and amplitude of phase jitter on the incoming signal.

7. The system of Claim 1, wherein said first signal generator: filters said first phase-error approximation; and outputs a first phase estimate corresponding to the filtered first phase-error approximation.

8. The system of Claim 7, wherein: said first phase estimate corresponds to phase distortions of the incoming signal caused by sources other than phase jitter.

9. The system of Claim 1, wherein said principal demodulating operation: combines said jitter-tracking signal with said first phase estimate; provides the combined jitter-tracking signal and first phase estimate as an input to a principal generating means, said means being adapted to generate a complex periodic function in accordance with the input to said generating means; generates a principal demodulating signal in accordance with the output of said principal generating means; demodulates the incoming signal with respect to said principal demodulating signal; and outputs a principal baseband signal corresponding to the incoming signal demodulated with respect to said principal demodulating signal.

10. The system of Claim 1 , wherein said principal demodulating operation comprises: means for combining said jitter-tracking signal and said first phase estimate; a principal generating means, said generating means being adapted to generate a complex periodic principal demodulating signal in accordance with the input to said generating means, and said generating means being operative upon the output of said means for combining as an input to said generating means; and a mixer, said mixer being connected to demodulate said incoming signal with respect to said principal demodulating signal, and to provide an output corresponding to said principal baseband signal.

11. The system of Claim 10, wherein said means for combining comprises: a summing operation that sums said jitter- tracking signal and said first phase estimate.

12. The system of Claim 10, wherein said mixer comprises: a multiplication operation that multiplies said incoming signal by said principal demodulating signal.

13. The system of Claim 1, wherein said first signal generating operation comprises: a first phase-locked loop operation.

14. The system of Claim 1, wherein: said first phase-error approximation corresponds to phase distortion in the incoming signal caused by phase jitter, and to phase distortion of the incoming signal caused by sources other than phase jitter.

15. The system of Claim 1, wherein: said principal demodulating signal corresponds to the actual instantaneous frequency and phase shift of the incoming signal, including phase distortions caused by both phase jitter and by sources other than phase jitter.

16. The system of Claim 13, wherein said first phase-locked loop operation comprises: a first demodulating operation, said demodulating operation being operative to demodulate the incoming signal with respect to a second phase-error approximation; a first phase-error approximating operation to generate said first phase-error approximation; a first loop filter function, said loop filter function being operative upon said first phase-error approximation to generate said second phase-error approximation; and a first periodic-function generating means, said generating means being operative upon said second phase-error approximation to generate a first demodulation signal corresponding to said second phase-error approximation.

17. The system of Claim 3, wherein said signal regenerative operation comprises: a phase-locked loop operation.

18. The system of Claim 17, wherein said phase- locked loop operation comprises: a phase detection function, said function being operative upon said first phase-error approximation; a loop filter function, said loop filter function being operative upon the result of said phase detection function; a second periodic-function generating means, said generating means being operative upon the result of said loop filter function to generate a control signal, said control signal being provided to said phase detection function; and a third periodic-function generating means, said third generating means being operative upon the result of said loop filter function to generate said jitter-locked signal.

19. The system of Claim 18, wherein: said jitter-locked signal corresponds to a periodic function in phase quadrature with the phase jitter.

20. The system of Claim 1, wherein: the amplitude of said jitter-tracking signal varies in accordance with phase deviations of said principal baseband signal.

21. The system of Claim 1, further comprising: a data detector, which generates a data output in accordance with information in said principal baseband signal.

22. The system of Claim 3, wherein said gain control function comprises: a data detector, which generates a data output in accordance with information in said principal baseband signal; and a phase-error detecting function that detects instantaneous phase deviations of said principal baseband signal relative to said data output, and generates a phase-error signal.

23. The system of Claim 3, wherein said gain control function: detects instantaneous phase deviations of said principal baseband signal; generates a phase-error signal in accordance with the instantaneous phase deviations of said principal baseband signal; and adjusts the amplitude of said jitter-locked signal in accordance with correlations between said jitter-locked signal and said phase-error signal.

24. The system of Claim 23, further comprising a data detector, which generates a data output in accordance with information in said principal baseband signal, wherein: said phase-error detecting function detects instantaneous phase deviations of said principal baseband signal relative to said data output.

25. The system of Claim 3, further comprising a data detector, which generates a data output in accordance with information in said principal baseband signal, wherein said gain control function: generates a gain signal in accordance with said phase-error signal and said jitter-locked signal; and mixes said gain signal with said jitter- locked signal; and outputs a jitter-tracking signal in accordance with the mixed gain signal and jitter-locked signal.

26. The system of Claim 25, wherein: said gain control function generates said gain signal with respect to correlations between said phase- error signal and said jitter-locked signal.

27. The system of Claim 3, further comprising a data detector, which generates a data output in accordance with information in said principal baseband signal, wherein said gain control function comprises: a phase-error detector that detects instantaneous phase deviations of said principal baseband signal relative to said data output, and outputs a phase-error signal; a gain signal generating function that generates a gain signal in accordance with correlations between said phase-error signal and said jitter-locked signal; and a signal mixer that combines said gain signal with said jitter-locked signal to generate said jitter- tracking signal.

28. The system of Claim 27, wherein: said mixer performs a multiply operation.

29. The system of Claim 1, wherein: the amplitude of said jitter-tracking signal varies in accordance with the presence or absence of phase jitter.

30. The system of Claim 29, wherein: the amplitude of said jitter-tracking signal varies in accordance with correlations between said first phase-error approximation and said jitter-locked signal.

31. The system of Claim 3, further comprising: a lock-detecting function, said lock-detecting function being adapted to generate a signal corresponding to the jitter-tracking operation's acquisition of phase jitter.

32. The system of Claim 3, further comprising: a lock-detector, which generates a signal in accordance with correlations between said first phase- error approximation and said jitter-locked signal.

33. The system of Claim 3, further comprising: a lock-detector, which generates a lock- indicating signal corresponding to the presence or absence of phase jitter, said lock-indicating signal being operative to control the gain of said loop filter.

34. The system of Claim 3, further comprising: a lock-detector, which generates a lock- indicating signal in accordance with correlations between said first phase-error approximation and said jitter-locked signal, said lock-indicating signal being operative to adjust the amplitude of said jitter-locked signal.

35. A method of demodulating an incoming signal, comprising the steps of: generating a first phase-error approximation corresponding to phase deviations of the incoming signal; generating a first phase estimate from said first phase-error approximation; generating a principal demodulating signal in accordance with said first phase-error approximation and said first phase estimate; and generating a principal baseband signal corresponding to the incoming signal demodulated with respect to said principal demodulating signal.

36. The method of Claim 35, wherein said step of generating said principal demodulating signal comprises the steps of: providing said first phase-error approximation to a jitter-tracking operation, said jitter-tracking operation comprising a regenerative signal processing function; generating a phase-error signal that corresponds to phase deviations in said principal baseband signal; providing said phase-error signal to the output of said jitter-tracking operation to control the amplitude of said jitter-tracking signal; providing said jitter-tracking signal and said first phase estimate to a principal demodulator.

37. The method of Claim 36, comprising the further steps of: summing said jitter-tracking signal and said first phase estimate to create a principal control signal; and providing said principal control signal to a principal generating means, said means being adapted to generate a complex periodic function in accordance with the input to said generating means.

38. A method of tracking phase jitter in an incoming signal, comprising the steps of: generating a jitter-locked signal that corresponds to the phase and frequency of any one of a plurality of phase jitter frequency components; adjusting the amplitude of said jitter-locked signal in accordance with the amplitude of the phase jitter; and providing the amplitude-adjusted jitter-locked signal as a jitter-tracking component.

39. The method of Claim 38, wherein said generating step comprises: demodulating the incoming signal with respect to the ideal carrier frequency and phase shift; detecting data from the demodulated signal; generating a first phase-error approximation corresponding to phase deviations of the demodulated signal relative to detected data; extracting information regarding phase jitter from said first phase-error approximation; and regenerating said jitter-locked signal in accordance with said phase jitter information.

40. The method of Claim 38 wherein: said adjusting step drives the amplitude of said jitter-locked signal to zero in the absence of phase jitter.

41. The method of Claim 38, wherein said ampli- tude-adjusting step comprises the steps of: demodulating the incoming signal; detecting data from the demodulated signal; detecting phase error in the incoming signal, relative to detected data, caused by phase jitter; determining correlations between said jitter- locked signal and said phase error; and adjusting said jitter-locked signal in accordance with said correlations.

42. The method of Claim 38, comprising the further steps of: detecting a phase-locked condition of said jitter-locked signal; and further adjusting the amplitude of said jitter-locked signal in accordance with the presence or absence of said phase-locked condition.