U.S. patent application number 12/945104 was filed with the patent office on 2011-03-10 for dynamic crest factor reduction system.
This patent application is currently assigned to Microelectronic Technologies, Inc.. Invention is credited to Khiem V. Cai, Samuel Davis Kent, III.
Application Number | 20110059710 12/945104 |
Document ID | / |
Family ID | 39684402 |
Filed Date | 2011-03-10 |
United States Patent
Application |
20110059710 |
Kind Code |
A1 |
Cai; Khiem V. ; et
al. |
March 10, 2011 |
DYNAMIC CREST FACTOR REDUCTION SYSTEM
Abstract
A system and method for performing digital crest factor
reduction. In one embodiment, the method is devised to suppress the
signal amplitude to maintain a low signal peak to average ratio
(PAR), while maintaining a desirable Error Vector Magnitude (EVM).
This technique may be designed to operate in highly dynamic signal
conditions.
Inventors: |
Cai; Khiem V.; (Brea,
CA) ; Kent, III; Samuel Davis; (Long Beach,
CA) |
Assignee: |
Microelectronic Technologies,
Inc.
Hsinchu
CN
|
Family ID: |
39684402 |
Appl. No.: |
12/945104 |
Filed: |
November 12, 2010 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11784433 |
Apr 5, 2007 |
7839951 |
|
|
12945104 |
|
|
|
|
Current U.S.
Class: |
455/232.1 |
Current CPC
Class: |
H04L 27/2623
20130101 |
Class at
Publication: |
455/232.1 |
International
Class: |
H04B 7/00 20060101
H04B007/00 |
Claims
1. A crest reduction system for gain leveling, the system
comprising: an estimator for estimating a root-mean-square (RMS)
level of a first signal carrier and an RMS level of a second signal
carrier, each of the first and second signal carriers having an
amplitude; a leveling setter for receiving the estimated RMS levels
of the first and second signal carriers, for producing a first gain
leveling factor and a second gain leveling factor for the first
signal carrier by using the estimated RMS level of the first signal
carrier, and for producing a third gain leveling factor and a
fourth gain leveling factor for the second signal carrier by using
the estimated RMS level of the second signal carrier; a first
leveler for receiving the first signal carrier and the first gain
leveling factor and for changing the amplitude of the first signal
carrier by adjusting the first signal carrier according to the
first gain leveling factor; a second leveler for receiving the
second signal carrier and the third gain leveling factor and for
changing the amplitude of the second signal carrier by adjusting
the second signal carrier according to the third gain leveling
factor; and a filter configurator for receiving the second and
fourth gain leveling factors, for producing one or more first
filter coefficients to further change the amplitude of the first
signal carrier according to the second gain leveling factor, and
for producing one or more second filter coefficients to further
change the amplitude of the second signal carrier according to the
fourth gain leveling factor, wherein the second gain leveling
factor is substantially equal to the reciprocal of the first gain
leveling factor, and wherein the fourth gain leveling factor is
substantially equal to the reciprocal of the third gain leveling
factor.
Description
CROSS-REFERENCE TO RELATED APPLICATION
[0001] The present invention is a continuation application of U.S.
patent application Ser. No. 11/784,433 of CAI et al., entitled
"DYNAMIC CREST FACTOR REDUCTION SYSTEM," filed on Apr. 5, 2007, now
allowed, and which contains subject matter that is related to U.S.
patent application Ser. No. 11/246,027 of CAI et al., entitled
"SYSTEM AND METHOD FOR CREST FACTOR REDUCTION," filed Oct. 7, 2005,
now U.S. Pat. No. 7,738,573, the entire disclosures of all of which
are hereby incorporated by reference herein.
BACKGROUND
[0002] 1. Field of the Invention
[0003] The present invention relates to electrical and electronic
circuits and systems. More specifically, the present invention
relates to systems and methods for reducing crest factor in
electrical and electronic circuits and systems.
[0004] 2. Description of the Related Art
[0005] In Multi-Carrier Power Amplifier (MCPA) communication
transmission applications, multiple carriers are typically combined
in the baseband, intermediate frequency (IF) or radio frequency
(RF) frequency range and the resulting signal is transmitted using
a single power amplifier. An aspect for MCPA transmission is to
transmit a signal at a very high efficiency while maintaining a low
Adjacent Channel Power Ratio (ACPR) to meet spectral mask
requirements. ACPR is defined as the ratio of power in a bandwidth
away from the main signal (the distortion product) to the power in
a bandwidth within the main signal. The bandwidths and locations
are functions of the standards being employed.
[0006] To achieve high efficiency power amplifier (PA)
transmission, it is desirable to use semi-non-linear PAs, such as
Class A/B PAs. A challenge for MCPA signal transmission is due to
the fact that the combined signal has a high crest factor (ratio of
peak power to average power), where the peak power is significantly
higher than the average power. A small portion of the combined
signal can have very high peaks and when transmitted at high PA
efficiency, these high-level signals reach into the saturated
region of the PA's transfer function and the output of the PA has
high intermodulation distortion (IMD). The high IMD level raises
the ACPR levels.
[0007] To maintain low ACPR without any linearization techniques,
the transmit signal level must be decreased sufficiently so that
the peak amplitudes are not in the saturated zone of the PA, but
this reduces the amplifier efficiency. For example, a four carrier
W-CDMA (wideband code division multiple access) signal can have a
crest factor exceeding 13 dB. If the crest factor is reduced by
about 6 dB, the average power can be increased by 6 dB thus
increasing the power efficiency by a factor of 4.
[0008] One approach to this problem is to limit the amplitude of
either the baseband signal or the RF signal output of each channel
using a look-ahead approach. However, it is difficult to generate
signals with low crest factor and low ACPR inasmuch as limiting the
amplitude increases out of band emissions (e.g. sidelobes) and
thereby raises the ACPR level. Similarly, efforts to reduce the
ACPR levels generally increase crest factor.
[0009] Another approach involves the use of unused CDMA codes to
reduce the crest factor in the output signals. However, this
approach requires knowledge of what is being transmitted so that
the unused codes can be identified. This adds to the complexity,
storage requirements and cost of the system.
[0010] Hence, a need remains in the art for an improved system or
method for reducing the crest factor in communications systems
while maintaining a low ACPR therefor.
SUMMARY OF THE INVENTION
[0011] Aspects of embodiments of the present invention are directed
to systems and methods for reducing crest factor in electrical and
electronic circuits and systems.
[0012] In one embodiment of the present invention, a crest
reduction system for gain leveling includes: an estimator for
estimating a root-mean-square (RMS) level of a first signal carrier
and an RMS level of a second signal carrier, each of the first and
second signal carriers having an amplitude; a leveling setter for
receiving the estimated RMS levels of the first and second signal
carriers, for producing a first gain leveling factor and a second
gain leveling factor for the first signal carrier by using the
estimated RMS level of the first signal carrier, and for producing
a third gain leveling factor and a fourth gain leveling factor for
the second signal carrier by using the estimated RMS level of the
second signal carrier; a first leveler for receiving the first
signal carrier and the first gain leveling factor and for changing
the amplitude of the first signal carrier by adjusting the first
signal carrier according to the first gain leveling factor; a
second leveler for receiving the second signal carrier and the
third gain leveling factor and for changing the amplitude of the
second signal carrier by adjusting the second signal carrier
according to the third gain leveling factor; and a filter
configurator for receiving the second and fourth gain leveling
factors, for producing one or more first filter coefficients to
further change the amplitude of the first signal carrier according
to the second gain leveling factor, and for producing one or more
second filter coefficients to further change the amplitude of the
second signal carrier according to the fourth gain leveling factor.
The second gain leveling factor is substantially equal to the
reciprocal of the first gain leveling factor, and the fourth gain
leveling factor is substantially equal to the reciprocal of the
third gain leveling factor.
[0013] The estimator may include a low pass filter configured by
one or more scale factors, the low pass filter being for estimating
at least one of the RMS level of the first signal carrier or the
RMS level of the second signal carrier.
[0014] The low pass filter may include an infinite impulse response
filter.
[0015] The first signal carrier may have a first signal to
distortion ratio (SDR), and the second signal carrier may have a
second SDR different from the first SDR.
[0016] The first leveler may be adapted to adjust the first signal
carrier according to the first gain leveling factor by multiplying
the amplitude of the first signal carrier by the first gain
leveling factor.
[0017] The second leveler may be adapted to adjust the second
signal carrier according to the third gain leveling factor by
multiplying the amplitude of the second signal carrier by the third
gain leveling factor.
[0018] The second gain leveling factor may be a multiplicative
factor of at least one of the one or more first filter coefficients
produced by the filter configurator.
[0019] The fourth gain leveling factor may be a multiplicative
factor of at least one of the one or more second filter
coefficients produced by the filter configurator.
[0020] The crest reduction system may further include a leveled
carrier combiner for receiving the adjusted first signal carrier
from the first leveler, for receiving the adjusted second signal
carrier from the second leveler, and for producing a multi-carrier
signal from the adjusted first signal carrier and the adjusted
second signal carrier.
[0021] The leveled carrier combiner may be adapted to produce the
multi-carrier signal by coherently combining the adjusted first
signal carrier and the adjusted second signal carrier.
[0022] In another embodiment of the present invention, a method for
gain leveling in a crest reduction system may include: estimating a
root-mean-square (RMS) level of a first signal carrier and an RMS
level of a second signal carrier, each of the first and second
signal carriers having an amplitude; determining a first gain
leveling factor and a second gain leveling factor for the first
signal carrier by using the estimated RMS level of the first signal
carrier; determining a third gain leveling factor and a fourth gain
leveling factor for the second signal carrier by using the
estimated RMS level of the second signal carrier; changing the
amplitude of the first signal carrier by adjusting the first signal
carrier according to the first gain leveling factor; changing the
amplitude of the second signal carrier by adjusting the second
signal carrier according to the third gain leveling factor; and
determining one or more first filter coefficients to further change
the amplitude of the first signal carrier according to the second
gain leveling factor; and determining one or more second filter
coefficients to further change the amplitude of the second signal
carrier according to the fourth gain leveling factor. The second
gain leveling factor is substantially equal to the reciprocal of
the first gain leveling factor, and the fourth gain leveling factor
is substantially equal to the reciprocal of the third gain leveling
factor.
[0023] In another embodiment of the present invention, a crest
reduction system for amplitude limiting includes: a controller for
receiving a first signal including one or more signal carriers, for
determining a signal to distortion ratio (SDR) of the first signal,
and for producing a correction value by using the determined SDR
and a threshold SDR; and a dynamic amplitude clipper for receiving
the correction value and a second signal corresponding to the first
signal and for producing a clipped signal by limiting an amplitude
of the second signal according to a value corresponding to the
correction value such that a peak to average ratio (PAR) of the
clipped signal is not greater than a PAR of the second signal. The
clipped signal has phase characteristics substantially equal to
phase characteristics of the second signal.
[0024] The crest reduction system may further include a second
dynamic amplitude clipper coupled with the dynamic amplitude
clipper, the second dynamic amplitude clipper being for receiving
the correction value and a third signal corresponding to the
clipped signal and for producing a second clipped signal by
limiting an amplitude of the third signal according to a second
value corresponding to the correction value such that a PAR of the
second clipped signal is not greater than a PAR of the third
signal. The second clipped signal has phase characteristics
substantially equal to phase characteristics of the third
signal.
[0025] The phase characteristics of the second clipped signal may
be substantially equal to phase characteristics of the clipped
signal.
[0026] The dynamic amplitude clipper may include an estimator for
estimating a root-mean-square (RMS) level of the second signal. The
value corresponding to the correction value may further correspond
to the estimated RMS level of the second signal such that the PAR
of the clipped signal is substantially constant over time. The
second dynamic amplitude clipper may include an estimator for
estimating an RMS level of the third signal. The second value
corresponding to the correction value may further correspond to the
estimated RMS level of the third signal such that the PAR of the
second clipped signal is substantially constant over time.
[0027] The estimator for estimating the power of the second signal
may include an infinite impulse response filter.
[0028] The crest reduction system may further include a gain
corrector for receiving a third signal corresponding to the clipped
signal and for reducing an energy loss of the third signal
resulting from the limiting of the amplitude of the second signal
by the dynamic amplitude clipper.
[0029] The dynamic amplitude clipper may be adapted to be
controlled to produce an unclipped signal from the second signal,
and a PAR of the unclipped signal may be substantially equal to the
PAR of the second signal.
[0030] The crest reduction system may further include a timing and
control unit for controlling the dynamic amplitude clipper to
produce the unclipped signal.
[0031] The dynamic amplitude clipper may include a first multiplier
for multiplying the second signal and an inverse of the value
corresponding to the correction value and a second multiplier for
producing the clipped signal by multiplying a third signal
corresponding to the second signal and the value corresponding to
the correction value.
[0032] In another embodiment of the present invention, a method for
amplitude limiting in a crest reduction system includes: receiving
a first signal including one or more signal carriers; determining a
signal to distortion ratio (SDR) of the first signal; producing a
correction value by using the determined SDR and a threshold SDR;
and producing a second signal corresponding to the first signal;
producing a clipped signal by limiting an amplitude of the second
signal according to a value corresponding to the correction value
such that a peak to average ratio (PAR) of the clipped signal is
not greater than a PAR of the second signal. The clipped signal has
phase characteristics substantially equal to phase characteristics
of the second signal.
[0033] In another embodiment of the present invention, a crest
reduction system for composite filtering includes: a timing and
control unit for receiving a first signal including a plurality of
signal carriers and for processing a detected absence of one or
more of the signal carriers; a filter configurator for producing a
plurality of composite coefficients for filtering of the signal,
each of the composite coefficients being formed from a plurality of
coefficients, and for zeroing one or more of the coefficients
corresponding to the one or more signal carriers in response to the
detected absence of the one or more signal carriers; and a filter
generator for receiving the composite coefficients and for
implementing a filter for a second signal corresponding to the
first signal, the filter being configured to output a third signal
by filtering the second signal according to the composite
coefficients.
[0034] The filter may be further configured to output the third
signal by substantially filtering out one or more frequency
components of the second signal corresponding to the zeroed one or
more of the coefficients.
[0035] The filter configurator may include a plurality of
multiplexer units, each of the multiplexer units being adapted to
receive one or more of the coefficients corresponding to one of the
signal carriers and to output either the one or more of the
coefficients or one or more zero value coefficients in response to
the detected absence of the one of the signal carriers.
[0036] The filter configurator may further include a combiner for
receiving the respective outputs of the multiplexer units and for
producing the composite coefficients by linearly combining the
respective outputs.
[0037] The crest reduction system may further include a second
filter generator for receiving the composite coefficients and for
implementing a second filter for a fourth signal corresponding to
the third signal, the second filter being configured to output a
fifth signal by filtering the fourth signal according to the
composite coefficients.
[0038] The second filter may be further configured to output the
fifth signal by substantially filtering out one or more frequency
components of the fourth signal corresponding to the zeroed one or
more of the coefficients.
[0039] A first one of the signal carriers may have a first
passband, and a second one of the signal carriers may have a second
passband, the first passband and the second passband forming a
frequency well therebetween. The filter configurator may further be
for producing a plurality of nibble coefficients for the filtering
of the signal, the nibble coefficients being configured to
substantially fill the frequency well.
[0040] The filter configurator may include a multiplexer unit
adapted to receive one or more of the nibble coefficients
corresponding to the first one of the signal carriers and the
second one of the signal carriers and to output either the one or
more of the nibble coefficients or one or more zero value
coefficients in response to the detected absence of at least one of
the first one of the signal carriers or the second one of the
signal carriers.
[0041] In another embodiment of the present invention, a method for
composite filtering in a crest reduction system includes: receiving
a first signal including a plurality of signal carriers; processing
a detected absence of one or more of the signal carriers; producing
a plurality of composite coefficients for filtering of the signal,
each of the composite coefficients being formed from a plurality of
coefficients; zeroing one or more of the coefficients corresponding
to the one or more signal carriers in response to the detected
absence of the one or more signal carriers; and implementing a
filter for a second signal corresponding to the first signal, the
filter being configured to output a third signal by filtering the
second signal according to the composite coefficients.
[0042] In another embodiment of the present invention, a crest
reduction system includes: a controller for receiving a first
frequency and a first bandwidth of a first baseband symbol stream
and for receiving a second frequency and a second bandwidth of a
second baseband symbol stream; a first carrier processor for
receiving the first baseband symbol stream and for converting the
first baseband symbol stream to a first signal having the first
bandwidth centered about the first frequency; a second carrier
processor for receiving the second baseband symbol stream and for
converting the second baseband symbol stream to a second signal
having the second bandwidth centered about the second frequency;
and one or more signals processors for receiving the first and
second signals and for adjusting a respective signal to distortion
ratio (SDR) of each of the first and second signals while a peak to
average ratio (PAR) of a composite signal produced from the first
and second signals is reduced.
[0043] The first frequency may have a value different from a value
of the second frequency.
[0044] The first bandwidth may have a frequency range different
from a frequency range of the second bandwidth.
[0045] The first bandwidth may have a frequency range substantially
equal to a frequency range of the second bandwidth.
[0046] The first baseband symbol stream may correspond to a WCDMA
signal, and the second baseband symbol stream may correspond to a
cdma2000 signal.
[0047] The crest reduction system may further include a combiner
for receiving the first and second signals and for producing a
multi-carrier signal from the first and second signals.
[0048] The combiner may be adapted to produce the multi-carrier
signal by coherently combining the first and second signals.
[0049] The first carrier processor may be adapted to increase a
sample rate of the first baseband symbol stream according to a
first rate to produce a first intermediate signal, and the second
carrier processor may be adapted to increase a sample rate of the
second baseband symbol stream according to a second rate to produce
a second intermediate signal. The first rate may have a value
different from a value of the second rate. The first carrier
processor may include a first resampler for resampling the first
intermediate signal according to a third rate to produce the first
signal, and the second carrier processor may include a second
resampler for resampling the second intermediate signal according
to a fourth rate substantially equal to the third rate to produce
the second signal.
[0050] In another embodiment of the present invention, a method of
signal converting in a crest reduction system includes: receiving a
first frequency and a first bandwidth corresponding to a first
baseband symbol stream; receiving a second frequency and a second
bandwidth corresponding to a second baseband symbol stream;
converting the first baseband symbol stream to a first signal
having the first bandwidth centered about the first frequency;
converting the second baseband symbol stream to a second signal
having the second bandwidth centered about the second frequency;
and adjusting a respective signal to distortion ratio (SDR) of each
of the first and second signals while a peak to average ratio (PAR)
of a composite signal produced from the first and second signals is
reduced.
BRIEF DESCRIPTION OF THE DRAWINGS
[0051] The accompanying drawings, together with the specification,
illustrate exemplary embodiments of the present invention, and,
together with the description, serve to explain the principles of
the present invention.
[0052] FIG. 1 is a block diagram of a Multi Carrier Power Amplifier
(MCPA) transmission system with a Crest Factor Reduction (CFR)
Processor according to an embodiment of the present invention.
[0053] FIG. 2 is a block diagram of the CFR Processor of FIG. 1
according to one embodiment of the present invention.
[0054] FIG. 3 shows an arrangement of Dynamic Amplitude Clippers
and CFR Filters according to one embodiment of the present
invention.
[0055] FIGS. 4(a), 4(b), 4(c), 4(d) and 4(e) show an example of
signal processing using Carrier Leveling Mode.
[0056] FIGS. 5(a), 5(b), 5(c), 5(d) and 5(e) show an example of
signal processing when Carrier Leveling Mode is bypassed.
[0057] FIG. 6 is a block diagram of a Carrier Processor according
to one embodiment of the present invention.
[0058] FIG. 7 is a block diagram of a Carrier RMS Estimator
according to one embodiment of the present invention.
[0059] FIG. 8 is a block diagram of an infinite impulse response
(IIR) filter according to one embodiment of the present
invention.
[0060] FIG. 9 is a block diagram of a Dynamic Amplitude Clipper
according to one embodiment of the present invention.
[0061] FIG. 10 is a graph showing signal to distortion ratio (SDR)
versus gain (or gain value) g.sub.s.
[0062] FIG. 11 is a graph showing values of a Gain look-up table
(LUT) according to an embodiment of the present invention.
[0063] FIG. 12 shows a CFR filter according to one embodiment of
the present invention.
[0064] FIG. 13 is a block diagram of a CFR Filter Configurator
according to one embodiment of the present invention.
[0065] FIG. 14 is a block diagram of a CFR Filter Configuration
Processor according to one embodiment of the present invention.
[0066] FIG. 15 shows a processing sequence of Nibble Filter
Generation according to one embodiment of the present
invention.
[0067] FIG. 16 is a block diagram of a Gain Correction Processor
according to one embodiment of the present invention.
[0068] FIG. 17 is a block diagram of a Signal to Distortion (SDR)
Controller according to one embodiment of the present
invention.
DETAILED DESCRIPTION
[0069] In the following detailed description, only certain
exemplary embodiments of the present invention are shown and
described, by way of illustration. As those skilled in the art
would recognize, the described exemplary embodiments may be
modified in various ways, all without departing from the spirit or
scope of the present invention. Accordingly, the drawings and
description are to be regarded as illustrative in nature, and not
restrictive.
[0070] FIG. 1 shows a block diagram of a Multi Carrier Radio
Transmitter 50 using a Non Linear Power Amplifier 70. This is
referred to as a Multi Carrier Power Amplifier system (MCPA). The
challenge for the MCPA signal transmission is that the combined
signal can have a high crest factor (the ratio of peak power to
average power), where the peak power is significantly higher than
the average power. A small portion of the combined signal can have
very high peaks, and when transmitted at high power amplifier (PA)
efficiency these high-level signals are saturated, and the output
of the PA has high intermodulation distortion (IMD) that raises the
adjacent channel power ratio (ACPR) levels. To improve ACPR to some
acceptable level for transmission, the IMD should be effectively
filtered at the output of the power amplifier or the high-level
signals should be reduced prior to amplification. This reduction is
one task of the Crest Factor Reduction Processor 10. In one
embodiment of the present invention, the crest factor reduction
processor 10 should be designed to support dynamic changes of the
signal and to maintain low PAR and low ACPR.
[0071] One aspect of the present invention is directed towards
reducing the crest factor of a signal in a manner that delivers low
ACPR and low EVM for dynamic signals.
[0072] FIG. 2 shows a block diagram of the Crest Factor Reduction
Processor 10 according to one embodiment of the present invention.
I and Q symbol streams for N carriers are received by the Transmit
Modem Interface 100 to format the symbol stream x.sub.1[m],
x.sub.2[m], . . . , x.sub.N[m] to the Carrier Processors 200. The
signal x.sub.k[m] is the symbol stream to be transmitted on the
frequency f.sub.k. The Transmit Modem Interface 100 also detects
the frame sync timing information of the transmission, and
generates the frame sync signal, which is sent to the Timing and
Control 1300 to coordinate the timing of the whole system.
[0073] The Carrier Processors 200 support N channels. Each Carrier
Processor 200 has a different center frequency, but can have the
same or different filter types to shape similar or different signal
modulations. For example, Carrier Processor 200 for Channel 1 can
be for WCDMA having 3.84 MHz bandwidth; Carrier Processor 200 for
Channel 2 can be for cdma2000 having 1.2288 MHz bandwidth; Carrier
Processor 200 for Channel 3 can be bypassed to accommodate signals
that are already shaped; and Carrier Processor 200 for Channel 4
can be for OFDM having 5 MHz bandwidth. Likewise, each channel
processor can have identical bandshapes.
[0074] Each channel carrier processor converts the baseband symbol
stream x.sub.k[m] to an intermediate frequency (IF) signal
y.sub.k[m] center at the frequency f.sub.k with reference to FIG.
2. The carriers are then combined by the Carrier Combiner 1100 to
produce a composite signal s[m] that contains multi-carriers. This
signal may then be used to control the composite Signal to
Distortion Ratio (SDR) and thus the EVM of each carrier.
[0075] Carrier Leveling Mode is controlled by the Carrier Leveling
Setter 400, Carrier Levelers 500 and the Leveled Carrier Combiner
600. Carrier Leveling Mode provides a mechanism for setting and
maintaining possibly different signal to distortion ratios for each
individual carrier. In the main path, the carriers y.sub.k[m] are
adjusted by gains 1/g.sub.k in the Carrier Levelers 500 to force
the carrier amplitudes to the desired levels. The outputs of the N
Carrier Levelers, gained carriers z.sub.1, z.sub.2, . . . , z.sub.N
are combined using the Leveled Carrier Combiner 600 to produce the
signal u[m]. This signal is fed to the Dynamic Amplitude Clippers
and CFR Filters 1500 for peak amplitude suppression and to the
Signal to Distortion Controller 1200 to control the peak amplitude
suppression to maintain the desired SDR and thus the carriers'
EVM.
[0076] An aspect of the Carrier Leveling Setter 400 is to control
the amplitude of each carrier channel signal to establish the
desired channel distortion or EVM. The power of each carrier is
estimated using the Carrier RMS Estimator 300. Based on the carrier
powers, the Carrier Leveling Setter 400 determines gain g.sub.k and
1/g.sub.k that will be applied at the CFR Filter Configurator 900
and the Carrier Levelers 500. The Carrier Levelers can be bypassed
when adjustments to control SDR for individual carriers are not
required.
[0077] Occasionally, a first Dynamic Amplitude Clipper 700 or a
second Dynamic
[0078] Amplitude Clipper 700' or both contained in Dynamic
Amplitude Clippers and CFR Filters 1500 (see, for example, FIG. 3)
can be disabled momentarily to produce a high PAR signal if the
digital predistortion processor (DPD) needs a high peak signal to
characterize the PA. Upon instruction the amplitude clipper or
clippers occasionally halt the clipping process for a very short
period of time to produce a non-clipped signal. This high-PAR
signal allows the DPD system to characterize the PA at high power
for effective linearization. Because the duration of this high PAR
signal is short, it does not degrade the power amplifier
performance significantly.
[0079] In one embodiment, with reference to FIG. 3, the Dynamic
Amplitude Clippers 700, 700' are processes that suppress the peak
amplitude of the composite transmit signal, without phase
distortion. The Dynamic Amplitude Clippers can operate in either
Static Mode or Dynamic Mode. In Static Mode, clipping is performed
at a fixed level, regardless of the RMS level of the signal being
clipped. Thus, as the signal's RMS level increases, more of the
signal is clipped, and as it decreases, less of the signal is
clipped. In Static Mode, the peak to average ratio of the signal
will fluctuate with the RMS level of the signal. In Dynamic Mode,
the ratio of the clipping level to the signal's RMS level is held
constant, which leads to a constant peak to average ratio of the
output signal regardless of the RMS level of the clipped
signal.
[0080] Whether in Static or Dynamic Mode, this amplitude clipping
function may produce both in-band and out-of-band distortion. The
in-band distortion should be maintained to a desired SDR level, and
the out-of-band distortion should be suppressed to below the
spectrum emission mask (SEM) requirements. The Signal to Distortion
Controller 1200 maintains the SDR level by setting the amplitude
clipping level. The CFR Filter Configurator 900 together with the
CFR filters 800, 800' suppress the out-of-band distortion.
[0081] The CFR filters can be programmable filters which perform
filtering to remove the out-of-band noise. To support changing
signal characteristics, the filter coefficients are generated using
the CFR Filter Configurator 900.
[0082] In one embodiment, the CFR Filter Configurator 900
implements filter coefficient combiner processes that determine the
filters based on the amplitude of the carriers, carrier
frequencies, carrier bandwidths, filter shapes and which carriers
are on or off.
[0083] The Signal to Distortion Controller 1200 monitors the signal
to distortion ratio by comparing the input signal with the output
signal. The SDR is computed, and a correction factor is determined,
and is used to adjust the clipping thresholds in the Dynamic
Amplitude Clippers 700, 700'. If the SDR is too low then the
amplitude clippers' thresholds are increased to improve the SDR to
the required level. For example, if the measured SDR is lower than
a desired level then a correction factor larger than 1 is applied.
This would increase the clipping threshold to improve the SDR to
the correct level; otherwise, a factor smaller than 1 is applied,
and this would reduce the clipping threshold. In this fashion, the
SDR can be maintained at the desired level. The Signal to
Distortion Controller 1200 can also be programmed to adjust the
factor if the SDR is below the desired level, and not adjust the
clipping threshold if SDR is higher than the level. Thus, the SDR
Controller can prevent a small SDR while allowing fluctuations of
SDR provided they remain above the desired level.
[0084] In one embodiment, the Dynamic Amplitude Clippers and CFR
Filters contains back-to-back clipper/filter pairs. For example,
with reference to FIG. 3, Clipper/Filter Pair 1 (i.e., clipper 700
and filter 800) performs the majority of the crest factor reduction
and noise filtering. As the highly clipped signal from the first
Dynamic Amplitude Clipper 700 passes through the first CFR Filter
800, the signal experiences some PAR regrowth. Clipper/Filter Pair
2 (i.e., clipper 700' and filter 800') removes some of this
regrowth. Accordingly, the second Dynamic Amplitude Clipper 700'
may be programmed with a slightly higher clipping level than that
used in the first Dynamic Amplitude Clipper 700. The out-of-band
noise introduced by the second clipper is removed with the second
CFR Filter 800' coefficients, which are designed to minimize PAR
regrowth due to the second filter.
[0085] Optionally included at the interfaces of each block are rate
changing interpolators or decimators with the purpose of maximizing
implementation efficiency.
[0086] As the result, the Carrier Processor and Crest Factor
Reducer produce the multi carrier signal and provide the following
features: [0087] (1) Produces a low PAR signal; [0088] (2)
Maintains low ACPR through filtering; [0089] (3) Maintains a
desirable SDR or EVM; and [0090] (4) Maintains performance despite
signal fluctuations, power transitions, varying statistics, and
carrier blanking.
Comparative Examples
[0091] FIG. 4(a) shows four carriers that are transmitted with
different power levels. The
[0092] Carrier Levelers 500 are used to adjust the carrier gain to
bring all carriers' powers to the same level (see FIG. 4(b)). The
Amplitude Clippers 700, 700' clip the signal to reduce the PAR;
however this process introduces clipping noise that has
approximately constant spectral density (see FIG. 4(c)). Hence with
carrier gain applied, the signal to clipper distortion can be
maintained to the desired level. In this example, the carriers have
the same SDR; however different SDRs may be achieved if desired. To
reverse the carrier gain, at least one of the CFR Filters 800, 800'
should have the filter gain shape having the reverse gain (see FIG.
4(d)) so that when this filter is applied to the Amplitude Clipper
or Clippers output, the resulting signal will have the same
amplitude and spectral distribution as in the input (see FIG.
4(e)). It is noted that this signal would have the desired SDR that
is due to clipping noise.
[0093] FIGS. 5(a), 5(b), 5(c), 5(d) and 5(e) show the signal
processing according to another example when Carrier Leveling Mode
is bypassed, where g.sub.k=1. FIG. 5(a) and FIG. 5(b) show four
carriers transmitted with different power levels. After the
amplitude clipping process, the clipping noise has approximate
constant spectral density (see FIG. 5(c)). FIG. 5(d) shows the
filter gain shape having the same amplitude gain level. When this
filter is applied, the output of the CFR filter will have the
spectral shape as shown in FIG. 5(e). It is noted that this signal
will not have equal SDR across all carriers due to clipping noise
when the carriers have different amplitudes.
Transmit Modem Interface
[0094] The Transmit Modem Interface 100 provides the interface
between the Communication Modem System with the Crest Factor
Reduction Processor 10. This interface may be specifically designed
to support the desired interface.
Carrier Processor
[0095] In one embodiment, the Carrier Processor 200 provides the
functions of digital upsampling, filtering and frequency
translation. The upsampling is used to increase the sample rate of
the incoming signal. The filtering is used to remove the aliases
caused by upsampling and to provide spectral shaping of the
carrier.
[0096] Upsampling and filtering is performed first. With reference
to FIG. 6, the signal x.sub.1,k[m]=I.sub.1,k[m]+jQ.sub.1,k[m] is
processed with zero-padding at processor 210 to produce the signal
x.sub.2,k[m]=I.sub.2,k[m]+jQ.sub.2,k[m] that has a sampling rate of
R.sub.2=KR.sub.1, where I.sub.1,k[m] is the in-phase component,
Q.sub.1,k[m] is the quadrature component and R.sub.1 is the
sampling rate of x.sub.1,k[m]. The resulting in-phase and
quadrature-phase components of the output are then separately
filtered with a shaped low pass filter, Shaped FIR 220, to produce
the signal x.sub.3,k[m]
x 3 , k [ m ] = I 3 , k [ m ] + j Q 3 , k [ m ] = i = 0 L 1 - 1 c i
I 2 , k [ m - i ] + j i = 0 L 1 - 1 c i Q 2 , k [ m - i ] ( 1 )
##EQU00001##
where c.sub.i are the coefficients of the shaped filter. The filter
has a low pass response and the coefficients c.sub.i are symmetric.
Thus, the signal I.sub.2,k[m] and Q.sub.2,k[m] can be pre-summed
before filtering to reduce the number of multiplications.
Additional upsampling and filtering can be performed by cascading
multiple zero-pad and filtering operations.
[0097] After the signals I.sub.3,k[m] and Q.sub.3,k[m] are
sufficiently upsampled and filtered, they are then frequency
shifted with the Frequency Shifter 240 to the desired carrier IF
frequency as follows:
y.sub.k[m]={I.sub.3,k[m]+jQ.sub.3,k[m]}{cos(2.pi.f.sub.km/T+.phi..sub.k)-
+j sin(2.pi.f.sub.km/T+.phi..sub.k)} (2)
where f.sub.k is the carrier IF frequency, T is the sample period,
cos(2.pi.f.sub.km/T+.phi..sub.k)+j sin(2.pi.f.sub.km/T+.phi..sub.k)
are the outputs of the numerically controlled oscillator (NCO) 230
and .phi..sub.k are phase offsets of each channel.
[0098] If all carriers have the same modulation, then all the
Carrier Processors 200 would be the same in structure, but if the
carriers are different then the Carrier Processors 200 would have
different up-sampling and filter coefficients, and a Resampler 250
may be required to provide the N carriers with the same sampling
rate prior to combining the individual carriers into a single
signal. Each carrier processor's Resampler resamples at the rate
appropriate for the input signal y.sub.k[m] to produce r.sub.k[m]
such that all r.sub.k[m] across all Carrier Processors 200 are at
the same sample rate.
Carrier RMS Estimator
[0099] With reference to FIG. 7, in one embodiment, the Carrier RMS
Estimator 300 estimates the carrier powers of the individual
carriers {C.sub.1, C.sub.2, C.sub.3, . . . , C.sub.N} to produce
the RMS amplitude of the carriers {A.sub.1, A.sub.2, A.sub.3, . . .
, A.sub.N}. The equation,
A k = LPF ( I 1 , k 2 [ m ] ) + LPF ( Q 1 , k 2 [ m ] ) .about. 4
.pi. LPF { .alpha. max ( I 1 , k [ m ] , Q 1 , k [ m ] ) + .beta.
min ( I 1 , k [ m ] , Q 1 , k [ m ] ) } ( 3 ) ##EQU00002##
where .alpha. and .beta. are constants chosen to provide a good
estimate of the magnitude A.sub.k provides the definition of the
RMS amplitude estimator and the complexity-saving approximations
employed with reference to FIG. 7. The constants .alpha. and .beta.
may be chosen to allow the multipliers 330, 335 to be implemented
with simple shifts and adds. The multiplier 350 may be omitted when
the absolute value of A.sub.k[m] is not required.
[0100] With reference to FIG. 8, the LPF can be implemented using
an integrate-and-dump filtering over L samples, or an infinite
impulse response filter (IIR) having a transfer function
H ( z ) = .beta. 1 - ( 1 - .beta. ) z - 1 ( 4 ) ##EQU00003##
where .beta. is a filter bandwidth control parameter that is much
less than 1.
[0101] The time constant for the IIR filter is 1/.beta.R.sub.s. If
.beta. is set small then the IIR converges slowly but has good
accuracy, and if .beta. is large then the IIR converges rapidly but
has low accuracy due to the variation of the signal. If .beta. is
conveniently set to 2.sup.-k, where k is some integer, then the IIR
multipliers can be replaced by bit shifters.
Carrier Leveling Setter
[0102] An aspect of the Carrier Leveling Setter 400 is to determine
the gains 1/g.sub.1, 1/g.sub.2, . . . , 1/g.sub.N that are applied
at the Carrier Levelers 500 for N carriers, and the gains g.sub.1,
g.sub.2, . . . , g.sub.N that are applied at the CFR Filter
Configurator 900.
[0103] When the Dynamic Amplitude Clippers 700, 700' are applied,
the distortion is distributed over all carriers, and the
distortion's spectral density is approximately a constant over the
carriers. Therefore if a carrier is strong, the signal to
distortion (SDR) for this carrier is high, and if the carrier is
weak, the SDR for this carrier is low. In practice each carrier
type has a specific SDR, and the SDR for different signals can be
different.
[0104] Table 1 shows the case where Carrier Leveling is disabled,
and no gain is applied to the individual carriers.
TABLE-US-00001 TABLE 1 Example SDR margin for when the Carrier
Level is not used Signal Clipper Carrier Transmitted Bandwidth
Density Distortion SDR Required SDR Margin # Carrier Type Power
(dBm) (MHz) (dBm/Hz) Density (dBm/Hz) (dB) SDR (dB) (dB) 1 cdma2000
37 1.25 -23.97 -46.79 22.82 18 4.82 2 WCDMA 40 4 -26.02 -46.79
20.77 20 0.77 3 OFDM 40 5 -26.99 -46.79 19.80 30 -10.20 Total Power
43.98
[0105] According to this example, the cdma2000 has 4.8 dB SDR
margin, WCDMA meets the required margin, and the OFDM has -10.2 dB
margin. The different signal powers and the constant distortion
power produces a SDR imbalanced design.
[0106] Table 2 shows the case where Carrier Leveling mode is
enabled, and desirable gains are applied to the individual
carriers.
TABLE-US-00002 TABLE 2 Example SDR margin for when the Carrier
Level is used Signal Clipper Carrier Transmitted Bandwidth Carrier
Gain Leveled Density Distortion SDR Required SDR Margin # Carrier
Type Power (dBm) (MHz) i.e., 1/gk(dB) Power (dB) (dBm/Hz) Density
(dBm/Hz) (dB) SDR (dB) (dB) 1 cdma2000 37 1.25 -5.00 32.00 -28.97
-47.03 18.07 18 0.07 2 WCDMA 40 4 -1.00 39.00 -27.02 -47.03 20.01
20 0.01 3 OFDM 40 5 10.00 50.00 -16.99 -47.03 30.05 30 0.05 Total
Power 43.98 Leveled 50.40 Power
[0107] According to this example, the cdma2000 has 0.07 dB SDR
margin, WCDMA has 0.01 dB SDR margin and the OFDM has 0.05 dB
margin. This is clearly an SDR-balanced design, where all carriers
meet the SDR requirements.
[0108] As shown in the above example, the individual carriers may
have to be adjusted in amplitude so that the Crest Factor Reduction
has lowest PAR while meeting all carriers' SDR requirements.
[0109] The following steps can be followed to compute Carrier Gains
to provide for equal SDR margin. [0110] (1) Compute the power
level, P.sub.m of each carrier. [0111] (2) Compute the power
spectral density, S.sub.n=P.sub.n/B.sub.n, of each carrier where
B.sub.n is the signal bandwidth. [0112] (3) Determine the
relationships between g.sub.k such that
[0112] S k g k 2 SDR req , k = D ( K L , P tot ) = D ( 5 )
##EQU00004##
where SDR.sub.req,k is the SDR required for carrier k to maintain
good EVM performance and D(K.sub.L,P.sub.tot) is the total
distortion power spectral density, which is a function of the
clipping level, K.sub.L and the total signal power, P.sub.tot,
entering the first Dynamic Amplitude Clipper 700. The distortion
power spectral density is approximately a constant, D, for all k.
[0113] (4) If the value for D is not known, scale all g.sub.k found
in the previous step such that the desired signal to distortion
ratio for all carriers meets the desired level.
Carrier Levelers
[0114] In one embodiment, the Carrier Levelers 500 multiply the
gain values 1/g.sub.1[m], 1/g.sub.2[m], . . . , 1/g.sub.N[m] from
Block 400 with the signals r.sub.1[m], r.sub.2[m], . . . ,
r.sub.N[m] to produce the signals z.sub.1[m], z.sub.2[m], . . . ,
z.sub.N[m].
Z k [ m ] = 1 g k [ m ] r k [ m ] ( 6 ) ##EQU00005##
[0115] The values of g.sub.k[m] may change as the waveform
changes.
Leveled Carrier Combiner
[0116] The Leveled Carrier Combiner 600 coherently combines the
carriers (separately in-phase and quadrature-phase) as follows
u [ m ] = I u [ m ] + jQ u [ m ] = k = 1 N z k [ m ] . ( 7 )
##EQU00006##
The resulting signal is a multi carrier signal with the individual
carriers amplitude adjusted.
Carrier Combiner
[0117] The Carrier Combiner 1100 coherently combines the carriers
(separately in-phase and quadrature-phase) as follows
s [ m ] = I s [ m ] + j Q s [ m ] = k = 1 N y k [ m ] ( 8 )
##EQU00007##
The resulting signal is a multi carrier signal with the individual
carriers not amplitude adjusted.
Dynamic Amplitude Clippers
[0118] An aspect of the Dynamic Amplitude Clippers 700, 700' is to
limit the amplitude without distorting the phase of the signal,
u[m], in a fashion that maintains the desired signal to distortion
ratio even when the signal's statistics are changing.
[0119] With reference to FIG. 9, in one embodiment, the processing
steps for the Dynamic Amplitude Clippers are as follows: [0120] (1)
Compute the amplitude of signal u[m] (Block 721). [0121] (2)
Perform low pass filtering (LPF) (Block 722) of the amplitude of
signal u[m] to estimate the signal envelope a[m]. This filter can
be implemented with an IIR (e.g., recursive filter
[0121] H ( z ) = .beta. 1 - ( 1 - .beta. ) z - 1 ) .
##EQU00008##
The filter coefficient .beta. determines the time response of the
signal envelope in dynamic signal conditions. [0122] (3) Instruct
the Dynamic Amplitude Clippers to operate in static or dynamic
clipping mode. For static clipping mode, Timing and Control Block
1300 selects the Mux 725 to pass value a to the multiplier 723. For
dynamic clipping mode, Timing and Control Block 1300 selects the
Mux 725 to pass the signal envelope a[m] to the multiplier 723.
[0123] (4) Adjust gain (or gain value) g.sub.s to maintain the
desired SDR or to otherwise control the SDR. This is achieved
without requiring the Gain look-up table (LUT) regeneration by
effectively adjusting the clipping threshold described in step 8
below.
[0124] Based on the gain correction value g[m] as produced by the
Signal to Distortion (SDR) Controller 1200, the processor
determines the gain value g.sub.s. This g.sub.s effectively is the
amount of adjustment on the threshold of the clipper. If
g.sub.s>1 then the threshold is effectively increased to relax
the clipping. If g.sub.s<1, then the threshold is effectively
reduced to force more clipping. Here, the Gain LUT physically
remains constant; the signal is raised or lowered instead.
[0125] FIG. 10 illustrates the effect of the gain value g.sub.s
versus the SDR. In this example, the measured SDR is 23 dB but the
PAR is 7 dB, and it is desirable for the system to reduce the PAR
while maintaining SDR>20 dB. To support this, the gain, g.sub.s,
can be decreased by a small value so that the signal is effectively
clipped harder to reduce the SDR to 20 dB and, as a result, to
reduce the PAR.
[0126] To prevent the situation of excessive clipping, the gain
value g.sub.s produced at the output of determining unit 724 may be
the larger of g.sub.s, with reference to FIG. 10, and a preset
threshold g.sub.T. A method to compute g.sub.s[m] is as follows
g s [ m ] = max { g T , g s , min + g s , max - g s , min SDR max -
SDR min ( SDR [ m ] - SDR min ) } . ( 9 ) ##EQU00009##
Here, as in FIG. 10, the relationship between SDR and g.sub.s is
approximated with a linear equation. [0127] (5) Multiply at
multiplier 723 signal envelope a[m] with g.sub.s to produce signal
b[m]. [0128] (6) Compute the signal 1/b[m] at computing unit 730
and multiply it with signal u[m] to produce signal u.sub.2[m] at
multiplier 711. [0129] (7) Compute amplitude of signal u.sub.2[m]
at unit 740. [0130] (8) Bypass the clipper if necessary. When the
Timing and Control unit 1300 sends a command to employ the Clipper
Bypass mode, the Clipper Bypass Processor 750 replaces the
amplitude with zero over a specified duration.
[0130] p [ m ] = u 2 [ m ] ; normal clipping operation = 0 ; if T /
C controller commands clipper bypass ( 10 ) ( 11 ) ##EQU00010##
[0131] (9) Send the amplitude p[m] to a Gain LUT 760. The output of
the LUT is a function of the input amplitude
[0131] G [ m ] = T 1 / p [ m ] ; p [ m ] > T 1 = 1 ; otherwise ,
( 12 ) ##EQU00011##
where T.sub.1 is the clipping level of the signal. Here, with
reference to FIG. 11, when the T/C Controller commands Clipper
Bypass mode, the amplitude is zero, the Gain LUT 760 will output
unity and no clipping will be applied. This allows the signal to
skip the clipping to produce a high PAR signal. [0132] (10) Delay
the signal u.sub.2[m] at delay unit 712 to produce the signal
u.sub.3[m]=u.sub.2[m-.tau..sub.1] that is aligned in time with gain
G. The signal u.sub.3[m] is multiplied at multiplier 713 with the
gain G to produce signal u.sub.4[m]. This effectively processes the
clipping of the signal amplitude. [0133] (11) Delay at delay unit
725 signal b[m] to produce signal b[m-.tau..sub.2] to compensate
for the signal delay in the main path. [0134] (12) Multiply at
multiplier 714 the signal u.sub.4[m] with signal b[m-.tau..sub.2]
to produce signal v[m]. Signal v[m] is the amplitude clipped
version of signal u[m].
Crest Factor Reduction (CFR) Filter
[0135] The transmitted signal can be a combination of multiple
carriers, and thus the spectrum can be asymmetric. In these
applications multiple bandpass filters can be designed, one for
each carrier, followed by a filter combination process to realize
the combined filter with a single filter as engaged in the CFR
Filters 800, 800'. For example, let v[m]=I.sub.i[m]+jQ.sub.i[m] be
the input signal of the filter. The output of the filter can be
expressed as
w [ m ] = I 0 [ m ] + jQ 0 [ m ] = { I i [ m ] + j Q i [ m ] } * c
[ m ] = { I i [ m ] + j Q i [ m ] } * { h 1 [ m ] j.omega. 1 m T +
h 2 [ m ] j.omega. 2 m T + + h N c [ m ] j.omega. N c m T } = { I i
[ m ] + j Q i [ m ] } * { h 1 [ m ] cos ( .omega. 1 m T ) + h 2 [ m
] cos ( .omega. 2 m T ) + + h N c [ m ] cos ( .omega. N c m T ) + j
[ h 1 [ m ] sin ( .omega. 1 m T ) + h 2 [ m ] sin ( .omega. 2 m T )
+ + h N c [ m ] sin ( .omega. N c m T ) ] } ( 13 ) ##EQU00012##
where * indicates the convolution process,
x [ m ] * y [ m ] = i = 1 N c x ( t ) y ( i - t ) ( 14 )
##EQU00013##
and h.sub.i[m] is the lowpass version of the desired filter for
carrier i, and .omega..sub.i is the angular frequency of the
carrier to be passed through filter h.sub.i[m]. The filters
h.sub.i[m] can have different spectral shapes. .sub.The input
(I.sub.i[m]+jQ.sub.i[m]) and output (I.sub.o[m]+jQ.sub.o[m])
relationship of the filter is expressed as
I o [ m ] + j Q o [ m ] = { I i [ m ] + j Q i [ m ] } * c [ m ] = {
I i [ m ] + j Q i [ m ] } * { c c [ m ] + j c s [ m ] } and ( 15 )
c [ m ] = c c [ m ] + j c s [ m ] ( 16 ) c c [ m ] = h 1 [ m ] cos
( .omega. 1 m T ) + h 2 [ m ] cos ( .omega. 2 m T ) + + h N c [ m ]
cos ( .omega. N c m T ) ( 17 ) c s [ m ] = h 1 [ m ] sin ( .omega.
1 m T ) + h 2 [ m ] sin ( .omega. 2 m T ) + + h N c [ m ] sin (
.omega. N c m T ) ( 18 ) ##EQU00014##
In one embodiment, the process is to compute the filter h.sub.i[m],
shift it to the desired frequency .omega..sub.i, and then combine
in the above fashion to form the single complex filter c[m].
[0136] One embodiment of a CFR filter structure is shown in FIG.
12. The real and imaginary parts of the input signal
I.sub.i[m]+jQ.sub.i[m] are passed through tapped delay lines 510
and 511, respectively, of length M, selected to support the filter
h.sub.i[m]'s spectral requirements. Since the taps of c.sub.c[m]
are even symmetric and the taps of c.sub.s[m] are odd symmetric,
pre-summing of the data in the tapped delay line is possible. Thus,
the samples at the taps are fed to two summers: the adding summer
520, corresponding to c.sub.c[m], and the subtracting summer 521,
corresponding to c.sub.s[m]. The outputs of these summers are
passed to two multipliers 530 and 531 that form the products of the
pre-summed signal and the filter coefficients. Since the signal is
processed in this fashion, the number of multiplications required
is decreased by a factor of two. To reduce the amount of hardware
required to realize this filter, the pre-summing adders and
multipliers may be shared across many taps. Thus, as the inputs to
the adders sweep across the taps, the M/2 coefficients are changed
and the multipliers' outputs are combined with summers 540 and 541.
The Q.sub.i[m] signal is processed with identical circuitry. The
output from the I.sub.i[m] filter side and the Q.sub.i[m] filter
side are combined with adders (Blocks 550 and 551) to produce
I.sub.o[m] and Q.sub.o[m].
[0137] If the desired signal is symmetric over the origin (e.g., 0
Hz), then the filter is real. In that case c.sub.s[m]=0 and only
the filtering corresponding to c.sub.c[m] is required. In this
situation, the complexity of the filter is cut in half.
[0138] Depending on the spectral shapes of the filters h.sub.i[m],
the composite filter, c[m], may have spectral regions of excessive
attenuation where the edges of the passbands of adjacent filters
meet. This attenuation may degrade signal quality and increase the
peak to average ratio of the transmitted signal. To improve
performance, the attenuation may be removed by inserting nibble
filters. Nibble filters may be implemented with the same (or
substantially the same) hardware used for the noise suppression
filters. Their generation is described below and illustrated in
FIG. 15. For a system with N.sub.c carriers, there may be as many
as N.sub.c-1 nibble filters.
[0139] In one embodiment, nibble filter coefficients are determined
by computing the difference of an ideal filter response, i.e., one
without the offending attenuation, and the response of the
composite filter, c[m], in the vicinity of the attenuation.
Separate nibble filters should be created for each area of
undesired attenuation. In this way, nibble filters may be added or
removed as neighboring filters, h.sub.i[m], are added or
removed.
[0140] Including the nibble filters, g.sub.i[m], the expressions
for c.sub.c[m] and c.sub.s[m] become
c c [ m ] = h 1 [ m ] cos ( .omega. 1 m T ) + h 2 [ m ] cos (
.omega. 2 m T ) + + h N c [ m ] cos ( .omega. N c m T ) + g 1 [ m ]
cos ( .gamma. 1 m T ) + g 2 [ m ] cos ( .gamma. 2 m T ) + + g N c [
m ] cos ( .gamma. N c - 1 m T ) ( 19 ) c s [ m ] = h 1 [ m ] sin (
.omega. 1 m T ) + h 2 [ m ] sin ( .omega. 2 m T ) + + h N c [ m ]
sin ( .omega. N c m T ) + g 1 [ m ] sin ( .gamma. 1 m T ) + g 2 [ m
] sin ( .gamma. 2 m T ) + + g N c [ m ] sin ( .gamma. N c - 1 m T )
( 20 ) ##EQU00015##
where .gamma..sub.i is the frequency necessary to place the nibble
filters at the correct frequency.
CFR Filter Configurator
[0141] An aspect of the CFR Filter Configurator 900 is to compute
the set [C] of N.sub.c complex passband coefficients and N.sub.n
complex nibble coefficients that are best used for the CFR Filter
800, 800'. The passband coefficients form the passbands for each
carrier. The N.sub.c complex passband coefficients are mutually
orthogonal, i.e., an amplitude change on one frequency does not
affect other frequencies. The nibble coefficients fill in the gaps
in the filter response created when frequency-contiguous passband
filters are added together.
[0142] Because the carriers will be changing quickly, these
coefficients should be computed rapidly to adapt.
[0143] The transmission consists of N.sub.c carriers, each carrier
requiring a filter centered at frequency f.sub.k, having an
amplitude gain g.sub.k, bandwidth B.sub.k, a transition bandwidth
of .DELTA.f.sub.k and rejection attenuation of R.sub.k in decibels.
The number of filter coefficients could be computed as
L = Max k ( ( f s .DELTA. f k ) ( R k 22 ) ) ( 21 )
##EQU00016##
[0144] The filter design task is simplified if the N.sub.c carriers
have M different types of signals, where each type has the same
signal bandwidth, transition bandwidth, and rejection attenuation.
In that case, M basic filters of length L can be configured.
[0145] With reference to FIGS. 13 and 14, the processing steps for
the CFR Filter Configurator are as follows: [0146] (1) Compute the
basic passband filters of the M types of signals. The filter
coefficients are stored in the Basic Filter RAM 910. This process
is described in the Basic Filter and Nibble Generation section
below. [0147] (2) Compute the basic nibble filters. The filter
coefficients are stored in the RAM 910. This process is described
in the Basic Filter and Nibble Generation section below. [0148] (3)
With reference to FIG. 14, the passband and nibble filter
coefficients are frequency-shifted at shifter and scaler 921 to the
correct frequencies {f.sub.1, f.sub.2, . . . }. The frequencies are
provided from the Carrier Processor Controller 1400. [0149] (4) The
coefficients of the frequency-shifted filters are stored in RAMs
922 for rapid processing. [0150] (5) The coefficients for the
filters from the RAM are multiplied at multipliers 923 with the
gains g.sub.1, g.sub.2, . . . , g.sub.Nc to compensate for the
carrier leveling process. The gains g.sub.k are provided by the
Carrier Leveling Setter 400. [0151] (6) Let m be the index of the
nibble filter corresponding to neighboring passband filters k and
k+1. If the ratio of the gains exceeds a threshold,
[0151] 1/T>g.sub.k/g.sub.k+1 or g.sub.k/g.sub.k+122 T
the gain value, g.sub.m, for the corresponding nibble filter is set
to zero. Otherwise, g.sub.m may be set to unity. In a dynamic
signal environment, these ratio tests should be recomputed when the
gains are changed. [0152] (7) A MUX 930 is used to set the filter
coefficients of carrier k to zero when Timing and Control unit 1300
sends an interrupt to indicate carrier k is not transmitted. When a
carrier is not present, Timing and Control will also turn off the
corresponding nibble filters. [0153] (8) The filter coefficients
from all carriers are linearly combined in the Coefficient Combiner
940 as
[0153] C = { c ( 1 ) c ( 2 ) c ( L ) } = { k = 1 N C + N N c k ( 1
) k = 1 N C + N N c k ( 2 ) k = 1 N C + N N c k ( L ) } ( 22 )
##EQU00017##
where c.sub.k(i) is the i.sup.th element passband or nibble filter
coefficient of the k.sup.th carrier or nibble, N.sub.c is the
number of passband filters, N.sub.N is the number of nibble
filters, L is the number of taps for the carrier filters. Also,
c(i) is the i.sup.th element filter coefficient of the CFR filter.
[0154] (9) The filter coefficients {c(1), c(2), . . . , c(L)} are
sent to the CFR Filters 800, 800 to remove the out of band
distortion.
Basic and Nibble Filter Generation
[0155] An aspect for the Basic Filter Generator 910 is to produce a
filter with very small in-band ripple, high out-of-band rejection,
and while meeting transition bandwidth requirements. Any valid
technique can be used to generate these filter coefficients.
[0156] An aspect for the Nibble Filter Generator 910' is to produce
a compensating filter to fill in the gaps in the filter response
created when frequency-contiguous passband filters are added
together. FIG. 15 illustrates the generation of Nibble Filters in
one embodiment of the present invention. First,
frequency-contiguous filter pairs are combined to produce the
coefficients s.sub.k[m]. Next, the desired filter response for this
pair is computed to produce the coefficients d.sub.k[m]. Finally,
the difference of these two are computed to produce the nibble
filter coefficients n.sub.k[m]=d.sub.k[m]-s.sub.k[m].
Gain Correction Processor
[0157] The clipping of the signal amplitude will introduce loss of
signal power. An aspect of the Gain Correction Processor 1000 is to
correct for the loss of energy due to the clipping and filtering
process to maintain the power accuracy of the transmit signal.
[0158] With reference to FIG. 16, in one embodiment, the processing
steps for the Gain Correction are as follows: [0159] (1) Compute
the amplitude squared at units 1001 and 1002 of the input signal
s[m] and perform low pass filtering (LPF1) at filter 1003 to
produce the signal envelope S[m]. In the case that gain leveling is
disabled, the signal s[m] can be replaced with the signal u[m] from
Leveled Carrier Combiner 600. [0160] (2) Compute the amplitude
squared of the input signal w[m] and perform low pass filtering
(LPF1) to produce the signal envelope W[m]. [0161] (3) Compute at
unit 1004 the ratio signal R[m]=S[m]/W[m]. This represents the
power ratio of the input and the output signal. [0162] (4) Perform
low pass filtering (LPF2) at filter 1005 to produce signal .GAMMA..
[0163] (5) Compute at unit 1006 amplitude gain .gamma.[m]= {square
root over (.GAMMA.[m])}. [0164] (6) Optionally delay, at delay unit
1007, signal w[m] to compensate for the delay of the gain
correction processing. [0165] (7) Multiply at multiplier 1008 gain
.gamma.[m] with the CFR output w[m-.tau..sub.3] to produce signal
q[m]. This signal has the same power with the source combined
signal s[m]. This signal is sent to the DPD system for
linearization.
Signal to Distortion (SDR) Controller
[0166] FIG. 17 shows a block diagram of the Signal to Distortion
(SDR) Controller 1200 in one embodiment of the present invention.
An aspect of the SDR Controller is to determine the correction gain
g to adjust the threshold at the Dynamic Amplitude Clippers 700,
700' based on the computed SDR and the threshold SDR.sub.T that is
set by the Carrier Processor Controller 1400.
[0167] With reference to FIG. 17, in one embodiment, the processing
steps to compute the correction gain g are as follows: [0168] (1)
Delay at delay unit 1201 the Carrier Combiner 1400 signal s by a
delay value t to align the signal r and w, the output of the second
CFR Filter 800'. [0169] (2) Compute at unit 1202 the amplitude r of
input signal: r[m]=ABS(s[m-.tau.]). [0170] (3) Estimate power S at
unit 1208 by squaring the amplitude r[m] and passing through a low
pass filter 1209. [0171] (4) Compute at unit 1210 the amplitude k
of CFR filter output signal:
[0171] k[m]=ABS(w[m]). [0172] (5) Estimate at unit 1212 power K by
squaring the amplitude k[m] and passing through a low pass filter
1213. [0173] (6) Compute at unit 1211 the signal p[m];
[0173] p [ m ] = c [ m ] k [ m ] = S K k [ m ] ##EQU00018##
so that the amplitude signals p[m] and r[m] have the same power.
[0174] (7) Compute at unit 1203 the distortion d[m]=r[m]-p[m].
[0175] (8) Estimate at unit 1204 distortion power D by squaring the
amplitude d[m] and passing through a low pass filter 1205. [0176]
(9) Compute at unit 1206 signal to distortion ratio
[0176] SDR = S D . ##EQU00019## [0177] (10) Because the distortion
is proportional to the 3.sup.rd order intermodulation product, the
correction factor can be established at unit 1207 as
[0177] g = ( SDR SDR T ) 1 3 ##EQU00020##
where the square root term translates the power ratio to a voltage
ratio. [0178] (11) Send the correction gain g to the Dynamic
Amplitude Clippers 700, 700' to adjust the amplitude clipping
threshold.
Carrier Processor Controller
[0179] An aspect of the Carrier Processor Controller 1400 is to
coordinate operation of the carrier processors and the CFR engine.
In one embodiment of the present invention, this includes the
following tasks. [0180] (1) Determine signal type of each carrier.
This information is used to reconfigure the carrier processors for
different types of signals that may be present (e.g., WCDMA,
cdma2000, etc.). This information is also sent to the CFR Filter
Configuration Processors 920 which determines CFR filter
coefficients. [0181] (2) Determine frequency of each carrier. This
information is sent to the NCO 230 to frequency-shift each carrier
and to the CFR Filter Configuration Processors 920 to set the CFR
filter center frequencies. [0182] (3) Activate and deactivate
carrier processors. This information turns off carrier processors
and this information is sent to the CFR Filter Configuration
Processor to reconfigure the CFR filter coefficients. [0183] (4)
IPDL processing. During Idle Periods in the Downlink (IPDL)
carriers are momentarily turned off by the Communications Modem
System by either zeroing the signal for the affected carrier or by
sending off/on signals to the Transmit Modem Interface 100. By
either detecting the absence of a carrier by the Carrier RMS
estimators 1400 or by using the off/on signals, the Carrier
Processor Controller triggers filter recomputations in the CFR
Filter Configurator 900. During IPDL periods, the CFR Filter
Configurator 900 will zero out the appropriate filters and nibbles
using the MUXes 930.
Timing and Control
[0184] In one embodiment of the present invention, the Timing and
Control Processor 1300 coordinates timing-critical tasks within the
Dynamic Crest Factor Reduction system. The Timing and Control
Processor 1300 rapidly detects control information from the
Transmit Modem Interface 100 such as Frame sync and control words.
This processor may also coordinate with the time structure of the
signal and at the correct time interrupt the Dynamic Amplitude
Clippers 700, 700' to pause the clipping process Timing and Control
may also coordinate the transmit carrier disable. Based on this
information, Timing and Control sends interrupts to the CFR Filter
Configurator 900 to reconfigure the CFR filter coefficients.
[0185] While the present invention has been described in connection
with certain exemplary embodiments, it is to be understood that the
invention is not limited to the disclosed embodiments, but, on the
contrary, is intended to cover various modifications and equivalent
arrangements included within the spirit and scope of the appended
claims, and equivalents thereof.
* * * * *