 |
Description  |
|
|
BACKGROUND OF THE INVENTION
The present invention relates to digital communications, and more
particularly to a method and apparatus for recovering carrier phase in an
adaptive equalizer without the use of phase rotation or de-rotation.
Digital data, for example digitized video for use in broadcasting high
definition television (HDTV) signals, can be transmitted over terrestrial
very high frequency (VHF) or ultra high frequency (UHF) analog channels
for communication to end users. Analog channels deliver corrupted and
transformed versions of their input waveforms. Corruption of the waveform,
usually statistical, may be additive and/or multiplicative, because of
possible background thermal noise, impulse noise, and fades.
Transformations performed by the channel are frequency translation,
nonlinear or harmonic distortion, and time dispersion.
In order to communicate digital data via an analog channel, the data is
modulated using, for example, a form of pulse amplitude modulation (PAM).
Typically, quadrature amplitude modulation (QAM) is used to increase the
amount of data that can be transmitted within an available channel
bandwidth. QAM is a form of PAM in which a plurality of bits of
information are transmitted together in a pattern referred to as a
"constellation", which can contain, for example, sixteen or thirty-two
points.
In pulse amplitude modulation, each signal is a pulse whose amplitude level
is determined by a transmitted symbol. In 16-QAM, symbol amplitudes of -3,
-1, 1 and 3 in each quadrature channel are typically used. In bandwidth
efficient digital communication systems, the effect of each symbol
transmitted over a time-dispersive channel extends beyond the time
interval used to represent that symbol. The distortion caused by the
resulting overlap of received symbols is called intersymbol interference
(ISI). This distortion has been one of the major obstacles to reliable
high speed data transmission over low background noise channels of limited
bandwidth. A device known as an "equalizer" is used to deal with the ISI
problem.
In order to reduce the intersymbol interference introduced by a
communication channel, rather precise equalization is required.
Furthermore, the channel characteristics are typically not known
beforehand. Thus, it is common to design and use a compromise (or a
statistical) equalizer that compensates for the average of the range of
expected channel amplitude and delay characteristics. A least mean square
(LMS) error adaptive filtering scheme has been in common use as an
adaptive equalization algorithm for over 20 years. This algorithm is
described in B. Widrow and M. E. Hoff, Jr., "Adaptive Switching Circuits"
in IRE Wescon Conv. Rec., Part 4, pp. 96-104, Aug. 1960. The use of the
LMS algorithm in an adaptive equalizer to reduce intersymbol interference
is discussed in S. U. H. Qureshi, "Adaptive Equalization", Proc. IEEE,
Vol. 73, No. 9, pp. 1349-1387, September 1987.
In an LMS equalizer, the equalizer filter coefficients are chosen to
minimize the mean square error, i.e., the sum of squares of all the ISI
terms plus the noise power at the output of the equalizer. Therefore, the
LMS equalizer maximizes the signal-to-distortion ratio at its output
within the constraints of the equalizer time span and the delay through
the equalizer. Before regular data transmission begins, automatic
synthesis of the LMS equalizer for unknown channels may be carried out
during a training period. This generally involves the iterative solution
of a set of simultaneous equations. During the training period, a known
signal is transmitted and a synchronized version of the signal is
generated in the receiver to acquire information about the channel
characteristics. The training signal may consist of periodic isolated
pulses or a continuous sequence with a broad, uniform spectrum such as a
widely known maximum length shift register or pseudo-noise sequence.
An important aspect of equalizer performance is its convergence, which is
generally measured by the amount of time in symbol periods required for
the error variance in the equalizer to settle at a minimum level, which is
ideally zero. In order to obtain the most efficient operation for a data
receiver, the equalizer convergence time must be minimized.
After any initial training period, the coefficients of an adaptive
equalizer may be continually adjusted in a decision directed manner. In
this mode, the error signal is derived from the final receiver estimate
(not necessarily correct) of the transmitted sequence. In normal
operation, the receiver decisions are correct with high probability, so
that the error estimates are correct often enough to allow the adaptive
equalizer to maintain precise equalization. Moreover, a decision directed
adaptive equalizer can track slow variations in the channel
characteristics or linear perturbations in the receiver front end, such as
slow jitter in the sampler phase.
Many transmission systems employ modulation schemes that are constructed
with complex signal sets. In other words, the signals are viewed as
vectors in the complex plane, with the real axis called the inphase (I)
channel and the imaginary axis called the quadrature (Q) channel.
Consequently, when these signals are subjected to channel distortion and
receiver impairments, cross talk between the I and Q channels occurs,
requiring a complex adaptive equalizer. In this case, the equalizer's
coefficients will be complex valued. If, as noted above, the channel
distortion is unknown by the receiver, then the coefficients must be
adjusted after the system has been in operation to cancel the channel
distortion. The term "adaptive" in a complex adaptive equalizer signifies
the ongoing adjustment of the coefficients.
In many practical transmission systems, some method must be provided to
derive a reference signal at the receiver's demodulator that is phase
coherent with the received signal. Such coherent demodulators are used to
demodulate signals containing information in their phase. For example, in
binary phase shift keying (BPSK), modulation of a digital "one" is
represented by a phase of zero degrees and modulation of a "zero" is
represented by a phase of 180 degrees in the modulated signal. Data
modulated using QAM techniques is demodulated on the basis of similar,
although more complicated, phase relationships. Thus, demodulators for
such data rely on a reference signal that must be synchronized in phase
with the data carrier. This process is known as carrier phase recovery
(CPR).
A phase locked loop (PLL) is a common and well known method used to recover
the carrier in signal demodulators. When used in such applications, the
PLL is sometimes referred to as a carrier recovery loop (CRL). When an
adaptive equalizer is employed, it has been common practice to locate the
CRL after the equalizer in the receiver. A free running oscillator is used
to translate the input signal frequency to baseband, and a phase rotator
is required to recover the carrier phase. In addition, a phase de-rotator
is required in the adaptive equalizer to provide a correctly phased error
signal for use in updating the filter coefficients. The requirement for a
phase rotator and de-rotator complicates the receiver design, and adds
expense to the receiver circuitry.
It would be advantageous to provide a method for recovering carrier phase
in systems employing adaptive equalization without the need for phase
rotation and de-rotation hardware. It would be further advantageous to
provide an adaptive equalizer for a communications receiver that can
initially adjust the equalizer coefficients in the absence of carrier
phase recovery, thereby reducing the acquisition time of the system.
Reduction of the system complexity by using self-recovering equalization
algorithms that do not require a training sequence would be further
advantageous. Such a system would be able to commence equalization without
waiting for carrier recovery to occur.
The present invention provides a method and apparatus enjoying the
aforementioned and other advantages.
SUMMARY OF THE INVENTION
In accordance with the present invention, a method is provided for
adaptively equalizing data signals in a communications receiver An
unequalized data signal is demodulated. The demodulated data signal is
filtered in an adaptive equalizer that initially updates adaptive filter
coefficients using error signals derived from a first algorithm. A carrier
lock signal is generated when a phase error of a filtered signal output
from the adaptive equalizer reaches a threshold value. The adaptive filter
coefficients are updated in the adaptive equalizer using error signals
derived from a second algorithm instead of the first algorithm in response
to the carrier lock signal.
In a preferred embodiment, the phase error is monitored during the
operation of the adaptive equalizer, and the first algorithm takes over if
it is determined during the monitoring step that the phase error no longer
meets the threshold Advantageously, the first algorithm will be a
self-recovering equalization algorithm, such as a constant modulus
algorithm. The second algorithm is advantageously a decision directed
algorithm
The phase error threshold is reached when at least a minimum percentage of
samples of the filtered signal taken over time fall within a predetermined
range. In an embodiment where the demodulated data signal comprises
coordinates representing an N-bit constellation pattern for a demodulated
N-bit quadrature amplitude modulated signal, the range can comprise a
plurality of separate fixed areas, each area enclosing one of the
constellation points. The separate fixed areas can comprise, for example,
an ellipse surrounding a constellation point. In an illustrated
embodiment, the ellipse is aligned with a corresponding radius extending
from an origin of the constellation pattern to the constellation point the
ellipse surrounds.
An adaptive equalizer for a communications receiver in accordance with the
present invention comprises means for demodulating an unequalized data
signal. An equalizer loop contains a filter coupled to receive demodulated
data from the demodulating means, an error signal generator coupled to
receive filtered data from the filter, and means responsive to error
signals from the error signal generator for updating coefficients for
input to the filter. A carrier recovery loop comprises a phase detector
coupled to receive the filtered data and provide a first phase error
signal for controlling the demodulator. Means, coupled to receive a second
phase error signal from the phase detector, generate a carrier lock signal
when the second phase error signal meets a threshold. The error signal
generator is responsive to the carrier lock signal for generating error
signals from a first algorithm when the second phase error signal fails to
meet the threshold and for generating error signals from a second
algorithm when the second phase error signal meets the threshold
The error signal generator can comprise, for example, a memory storing a
first set of error signals computed using the first algorithm and a second
set of error signals computed using the second algorithm. In such an
embodiment, the filtered data and the carrier lock signal are used to
address the memory to output error signals
Preferably, the first algorithm will be a self- o recovering equalization
algorithm, such as a constant modulus algorithm. The second algorithm can
be a decision directed algorithm. The phase error threshold will be met
when at least a minimum percentage of samples of the filtered signal taken
over time fall within a predetermined range. In an illustrated embodiment,
the demodulated data signal comprises coordinates representing an N-bit
constellation pattern for a demodulated N-bit quadrature amplitude
modulated signal. The predetermined range for determining whether the
threshold is met comprises a plurality of separate fixed areas, each
enclosing one of the constellation points. Each of the separate fixed
areas can comprise an ellipse surrounding a constellation point, each
ellipse being aligned with a corresponding radius extending from an origin
of the constellation pattern to the constellation point the ellipse
surrounds.
The present invention also provides an adaptive equalizer for a
communications receiver wherein an adaptive filter is provided for
filtering unequalized data representative of coordinates in a
constellation pattern. An error signal generator converts the filtered
data from the filter to error signals based on a first or second
algorithm. Means are coupled to receive error signals output from the
error signal generator for updating coefficients for the adaptive filter A
phase detector converts filtered data from the filter to phase error
signals. Means responsive to the phase error signals control the error
signal generator to provide error signals according to the first algorithm
when a phase error represented by the phase error signals is above a
predetermined threshold Error signals are provided according to the second
algorithm when the phase error is below the predetermined threshold.
The error signal generator of the adaptive equalizer can comprise a look-up
table containing error signal data computed under the first and second
algorithms. The look-up table is addressed by the filter data and the
control means to output the error signals. The phase detector can also
comprise a look-up table. This table would contain phase error data and be
addressed by the filter data to output the phase error signals.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of a prior art communications receiver
illustrating an adaptive equalizer followed by a carrier recovery loop
including a phase rotator;
FIG. 2 is a block diagram illustrating a communication system incorporating
an adaptive equalizer in accordance with the present invention;
FIG. 3 is a block diagram illustrating the adaptive equalizer of the
present invention in greater detail;
FIGS. 4a to 4c provide three scatter plots illustrating the output of the
equalizer of the present invention at different points in time;
FIG. 5 is a graph illustrating the mean square error over time of an
adaptive equalizer in accordance with the present invention;
FIG. 6 is a graph illustrating a carrier lock signal provided to the
adaptive equalizer of the present invention; and
FIG. 7 is a graphical representation of a constellation pattern for
sixteen-bit QAM data, illustrating fixed elliptical areas used to
determine when a phase error threshold has been met.
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 illustrates a prior art data transmission/reception system in which
the communication receiver includes an adaptive equalizer followed by a
carrier recovery loop using a phase rotator. Modulated digital data is
input to a transmitter 12 via an input terminal 10 in a conventional
manner. The transmitter broadcasts the data via a channel 14 that
introduces amplitude and/or delay (phase) distortion. When the modulated
data comprises multilevel pulse amplitude modulated data, such as QAM
data, intersymbol interference takes place within the channel. An adaptive
equalizer 20 is provided in the receiver to compensate for the intersymbol
interference. The equalizer is essentially a filter with coefficients
chosen to cancel the effects of the channel distortion.
The data received from channel 14 is demodulated at the receiver in a
demodulator 16 that is controlled by a free running oscillator 18. In the
illustrated embodiment, a quadrature demodulator is used to receive
complex QAM data. The received data is demodulated to recover the real and
imaginary complex components. These components are input to an adaptive
filter 24 of adaptive equalizer 20. The filtered output from filter 24 is
input to an independent carrier recovery loop 22. A phase rotator 26 in
the carrier recovery loop shifts the phase of the filtered signals by an
estimate of the phase error between a transmitted signal and a received
signal. A phase detector 30 coupled to the output of phase rotator 26
generates an error signal indicative of the difference between the
estimated phase shift and the actual shift introduced by channel 14. The
error signal is filtered by a loop filter 32 and used as an input to a
numerically controlled oscillator 34 to adjust phase rotator 26 in a
manner that attempts to reduce the error signal to zero.
The output of phase rotator 26 is also coupled to an error signal generator
36 in the adaptive equalizer. An error signal is generated indicative of
the amount of intersymbol interference contained in the filtered,
demodulated input signals. The phase of the error signal is de-rotated in
a phase de-rotator 38, and input to a coefficients update calculation
circuit 40 for use in updating the adaptive filter coefficients. In this
manner, the intersymbol interference is reduced over time so that the
transmitted data can be accurately decoded by a conventional decoder 28.
A problem with the prior art structure illustrated in FIG. 1 is that it is
complicated and expensive, in particular because of the need to provide a
phase rotator in the carrier recovery, loop and a phase de-rotator in the
adaptive equalizer. A typical phase rotator requires four multiplies and
two additions to provide the desired phase correction. Similar operations
are required in the de-rotator. Therefore, by eliminating the phase
rotator and phase de-rotator, it is possible to save the hardware that
performs eight multiplies and four adds.
The present invention provides an adaptive equalizer that eliminates the
phase rotation and phase de-rotation components by locating the equalizer
inside of the carrier recovery loop. This is illustrated in general terms
in FIG. 2. As with the prior art, modulated data is input to a transmitter
52 via an input terminal 50. The data is broadcast over a channel 54, that
introduces the distortions which cause intersymbol interference in the
multilevel modulated data. A communications receiver in accordance with
the invention uses a carrier recovery loop 56 that incorporates a
demodulator 58, adaptive equalizer 60, and carrier recovery circuit 62. In
the illustrated embodiment, 16-QAM data is received, and demodulator 58 is
a quadrature demodulator that recovers the real and imaginary complex
components from the 16-QAM data. Since complex data is provided, adaptive
equalizer 60 is a complex adaptive equalizer. Carrier recovery circuit 62
provides a phase error signal to demodulator 58 and also provides a
"carrier lock" signal to adaptive equalizer 60. The carrier lock signal,
as discussed in greater detail below, is used to select between an
intersymbol interference error signal derived from a first,
self-recovering equalization algorithm such as the Constant Modulus
Algorithm and a second decision directed algorithm for use in updating
filter coefficients for the equalizer. A conventional decoder 64 is
provided to recover individual data bits from the equalized channel data
output from the adaptive equalizer.
FIG. 3 illustrates the carrier recovery loop 56 in greater detail A phase
locked loop, consisting of a phase detector 76, loop filter 80, and
voltage controlled oscillator (VCO) 82 surrounds adaptive equalizer 60.
The adaptive equalizer uses two least mean square (LMS) algorithms to
adjust (i.e., update) the coefficients used by an adaptive filter 70. In
the illustrated embodiment, the first LMS algorithm used is the Constant
Modulus Algorithm (CMA) which is well known in the art and described, for
example, in D. N. Godard, "Self-Recovering Equalization and Carrier
Tracking in Two-Dimensional Data Communication Systems", IEEE Trans. on
Commun., Vol. COM-28, pp. 1867-1875, Nov. 1980. The second LMS algorithm
used by the adaptive equalizer is a decision directed algorithm (DDA). The
two coefficient update algorithms differ only in the way the error signal
used to update the coefficients is generated. The LMS algorithm is given
by:
C(k+1)=c(k)+.DELTA.E(k)X.sup.* (k)
where C(k) is the complex vector of coefficients, X(k) is the complex
vector of delayed data, hu * means complex conjugate, E(k) is the complex
error signal, and .DELTA. is a scale factor. For the CMA the error signal
is given by:
E(k).sub.cma ={.vertline.y(k).vertline..sup.2 -R.sub.2 }y(k)
where y(k) is the complex output of the adaptive equalizer and R.sub.2 is a
constant For the DDA the error signal is given by:
E(k).sub.dda =y'(k)-y(k)
where y'(k) is the "signal decision". The signal decision is based on a
determination as to which constellation point a received coordinate set is
closest to. Upon finding the closest constellation point to the received
data point, a decision is made that the received data point corresponds to
the nearest constellation point.
Adaptive equalizer 60 comprises an inner loop including error signal
generator 72, coefficients update calculation circuitry 74, and adaptive
filter 70. The error signal generator receives the filtered channel data
from adaptive filter 70, determines the error in the filtered data (i.e.,
the difference between the filtered data and an ideal constellation
pattern), and outputs an error signal indicative thereof for use by the
coefficients update calculation circuit In response to the error signal,
updated coefficients are provided to the adaptive filter 70, so that after
a period of time the equalized channel data output from filter 70 will be
restored to a condition from which the transmitted data can be recovered
by a conventional decoder
In accordance with the present invention, the CMA algorithm, which is a
self-recovering equalization algorithm (also known as a blind equalization
algorithm) that does not require initialization with a training sequence,
is first used to adjust the coefficients until the channel data is
sufficiently equalized so that carrier phase recovery can be achieved An
important aspect of the CMA algorithm for purposes of the present
invention is that it is independent of carrier phase recovery. The CMA
algorithm provides the initial equalization necessary for the outer
carrier recovery loop (phase detector 76, loop filter 80, and VCO 82) to
be operational.
Like error signal generator 72, phase detector 76 also monitors the
filtered data from adaptive filter 70. It determines the phase error
between the filtered data and the ideal constellation pattern for the
modulation scheme used. The phase error is quantized in a well known
manner to provide a first phase error signal on line 77 that is processed
for use by demodulator 58 in recovering the carrier phase An example of
such a quantizing scheme is provided in A. Leclert and P. Vandamme,
"Universal Carrier Recovery Loop for QASK and PSK Signal Sets," IEEE
Trans. on Communications, Vol. COM-31, No. 1, Jan. 1983, pp. 130-136. The
operation of the phase detector is explained in greater detail below For
the present, it is noted that one output of the phase detector is coupled
to loop filter 80 and VCO 82 in a conventional manner to control
quadrature demodulator 58 based on the detected phase error. The operation
of the carrier recovery loop will drive demodulator 58 to a point where
the phase error is minimized in view of the feedback provided by the loop.
Phase detector 76 also quantizes the phase error using a second quantizing
scheme to produce a second phase error signal on line 75 that is input to
a carrier lock generator 78 in accordance with the present invention.
Although the quantizing schemes used to generate the first and second
phase error signals can be the same, it is preferable to use different
schemes, wherein the scheme used to generate the first phase error signal
is selected for its ease of implementation and to reduce the possibility
of false lock points for QAM. The second quantization scheme is selected
to provide an early indication as to when the CMA algorithm has converged
Such a quantization scheme is described below in connection with FIG. 7.
Generator 78 uses a sliding average technique to determine when the phase
error drops below a predetermined threshold When the threshold is met, the
phase of the filtered data signal will be sufficiently close to that of
the transmitted signal that accurate data recovery can commence At this
point, the CMA algorithm will have served its function of self-recovery,
and the DDA algorithm can be substituted to provide more efficient
equalizer operation. Thus, lock generator 78 outputs a carrier lock signal
to error signal generator 72 when the threshold is met. In response to the
carrier lock signal, error signal generator 72 switches from the CMA
method of calculating the error signals to the DDA method. In the event
that the threshold is no longer met at some time during the operation of
the equalizer, the carrier lock signal will turn off, and the error signal
generator will switch back to the CMA algorithm. Thus, the system will
automatically operate in the CMA mode when necessary, and switch over to
the DDA mode as soon as the phase error has been reduced below the
predetermined threshold value.
In a preferred embodiment, error signal generator 72 and phase detector 76
both comprise programmable read-only memory (PROM) devices to enable high
speed operation of the equalizer e.g at symbol rates on the order of 5
MHz. The PROM used for the error signal generator will contain two sets of
values. One set will comprise error signal values computed using the CMA
algorithm. The other set will comprise error signal values computed using
the DDA method. The filtered data input to the error signal generator PROM
from adaptive filter 70 is used to address the memory and output the error
signals that have been precomputed for the specific filtered data values.
The carrier lock signal input to the error signal generator PROM provides
an additional address signal to select between the first set of values
(CMA) or second set of values (DDA) depending on whether the phase error
threshold has been met.
The phase detector PROM stores two sets of precomputed phase error values
corresponding to the possible filtered data values output from adaptive
filter 70. One set of phase error values represents quantized values
according to the first quantizing scheme discussed above, and the other
set of phase error values corresponds to the quantized values provided by
the second quantizing scheme discussed above and explained in further
detail in connection with FIG. 7. The filtered data values are used to
address the phase detector PROM, and output the first and second phase
error signals associated with the particular filtered data values. Lock
generator 78 computes a sliding average of the second phase error signals
based on a relatively large number of samples. For example, lock generator
78 can comprise an accumulator that accumulates the error signals output
from phase detector 76 for one thousand samples of the filtered data
output from filter 70. In the event a particular data coordinate set
output from adaptive filter 70 represents a point that falls within a
predetermined area of the constellation pattern, the phase detector 76 can
output a second phase error signal that is, e.g., a "+1". On the other
hand, if the data coordinates represent a point falling outside of the
predetermined area in the constellation pattern, a "-1" can be output as
the second phase error signal. If the last one thousand error signal
samples input to lock generator 78 have an average value of, say, zero or
above, lock generator 78 will output the carrier lock signal to actuate
error signal generator to switch from the CMA mode to the DDA mode.
FIG. 7 illustrates a preferred embodiment of the phase error detection
scheme (i.e., the second quantization scheme) used to generate the second
phase error signals that are output from phase detector 76. In the
illustrated embodiment, 16-QAM is used to transmit the data. Accordingly,
constellation pattern 120 includes sixteen points. Each point is
surrounded by a predetermined elliptical area, such as areas 126 and 132
illustrated. Ellipse 126 surrounds constellation point 122, and is aligned
with a corresponding radius 124 extending from the origin of the
constellation pattern to the constellation point 122. Similarly, ellipse
132 surrounds constellation point 128, and is aligned with a radius 130
extending from the constellation pattern origin to point 128. Similarly
aligned elliptical areas (not shown) are defined around each of the other
points of the constellation pattern.
In the ev | | |
