U.S. patent application number 10/758355 was filed with the patent office on 2004-08-05 for method and apparatus for canceling a transmit signal spectrum in a receiver bandwidth.
Invention is credited to Masenten, Wesley K..
Application Number | 20040151238 10/758355 |
Document ID | / |
Family ID | 23935562 |
Filed Date | 2004-08-05 |
United States Patent
Application |
20040151238 |
Kind Code |
A1 |
Masenten, Wesley K. |
August 5, 2004 |
Method and apparatus for canceling a transmit signal spectrum in a
receiver bandwidth
Abstract
A method and apparatus for attenuating a transmit signal
spectrum in a receiver bandwidth. The apparatus relates to an
adaptive, digital, coherent, spectral canceller. The method
comprises digitizing both a corrupted receiver signal and a
reference transmit signal and then digitally implementing an
adaptive coherent spectral canceller module to suppress a residue
transmit spectral signal power within the receiver bandwidth.
Inventors: |
Masenten, Wesley K.;
(Irvine, CA) |
Correspondence
Address: |
KNOBBE MARTENS OLSON & BEAR LLP
2040 MAIN STREET
FOURTEENTH FLOOR
IRVINE
CA
92614
US
|
Family ID: |
23935562 |
Appl. No.: |
10/758355 |
Filed: |
January 15, 2004 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10758355 |
Jan 15, 2004 |
|
|
|
09487396 |
Jan 18, 2000 |
|
|
|
6704349 |
|
|
|
|
Current U.S.
Class: |
375/219 |
Current CPC
Class: |
H04B 1/525 20130101;
H04B 1/123 20130101 |
Class at
Publication: |
375/219 |
International
Class: |
H04B 001/38 |
Claims
What is claimed:
1. A transceiver comprising: a receiver direct converter
translating a received signal to a baseband of the received signal
and digitizing the translated, received signal; an adaptive
canceller comprising a reference direct converter, the reference
direct converter outputting a digitized transmit signal reference
of a spectral energy of a transmitter within the bandwidth of a
receiver; and a matched filter, wherein the receiver direct
converter, the reference direct converter, and the matched filter
suppress the spectral energy of the transmitter within the
bandwidth of the receiver.
2. The transceiver of claim 1, wherein the transceiver is a full
duplex transceiver.
3. The transceiver of claim 1, further comprising a transmit and
receive antenna radiator.
4. The transceiver of claim 1, further comprising a transmit
antenna radiator and a receive antenna radiator.
5. The transceiver of claim 1, where the receiver direct converter,
the reference direct converter, and the matched filter have
approximately 90 dB attenuation.
6. The transceiver of claim 1, wherein the receiver direct
converter has a sampling rate approximately equal to that of the
carrier frequency of interest.
7. The transceiver of claim 1, wherein the reference direct
converter has a sampling rate approximately equal to that of the
carrier frequency of interest.
8. The transceiver of claim 1, wherein the canceller further
comprises an adaptive digital transversal filter adapted to align
an amplitude and a phase of the digitized transmit signal reference
in a reference path with a transmit signal in a leakage receiver
path, the adaptive digital transversal filter outputting an
compensated digitized transmit signal reference.
9. The transceiver of claim 1, wherein the transceiver is adapted
to cancel interference from other co-sited transmit antennas.
10. A method of attenuating a transmitter signal spectrum within a
bandwidth of a receiver, the method comprising: digitizing a
received signal which is corrupted by components of a transmit
signal; creating a digitized reference transmit signal of the
transmit signal within the bandwidth of the receiver; aligning the
digitized reference transmit signal in amplitude, phase and time
delay with the digitized received signal; subtracting the digitized
reference transmit signal from the digitized received signal to
form a residue; and suppressing a transmitter spectral signal power
of the residue within the bandwidth of the receiver.
11. The method of claim 10, further comprising adjusting the
transmit signal based on the residue determined by subtracting the
digitized reference transmit signal from the digitized received
signal.
12. A transceiver comprising: a duplexer coupled to an antenna; a
receiver receiving a first signal from the duplexer; a transmitter
sending a second signal to the duplexer; and an adaptive, digital,
coherent spectral canceller coupled to the receiver and the
transmitter, the canceller attenuating a signal spectrum leakage of
the second signal within a bandwidth of the first signal.
Description
CLAIM OF PRIORITY
[0001] This application is a continuation of, and claims priority
from U.S. patent application Ser. No. 09/487,396, filed Jan. 18,
2000, which is incorporated in its entirety by reference
herein.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention relates to wireless communications.
More specifically, the invention relates to canceling transmitter
signal energy in a receiver bandwidth.
[0004] 2. Description of the Related Art
[0005] There is a growing need for low cost, high performance radio
transceivers that can operate in a full duplex mode in support of
such applications as code division multiple access (CDMA), global
system for mobile communications (GSM) and time division multiple
access (TDMA) remote units and base stations.
[0006] Currently, a standard transceiver comprises a transmitter, a
receiver and a duplexer. One problem with standard transceivers is
some of the transmit signal leaks into the receiver, which corrupts
the received signal processed by the receiver.
SUMMARY OF THE INVENTION
[0007] In some transceivers, the duplexer functions to reduce the
transmit signal received by the receiver through the use of one or
more circulators, receive filters and/or transmit filters. For
example, a circulator within the duplexer directs the transmit
signal toward an antenna radiator, while a receive filter rejects
the residual transmit energy directed toward the receiver that
falls outside of the receive filter bandwidth. Alternatively, some
duplexers do not use a circulator.
[0008] Imperfect duplexers create two problems. First, the
high-level transmit signal from the transmitter drives the receiver
into a non-linear operating region. This problem can be solved by
increasing the rejection characteristics of the receive filter to
more effectively attenuate the main spectral lobe of the
transmitter signal. Alternatively, this problem may be mitigated
through the use of a high dynamic range, direct conversion, digital
receiver that can implement the requisite filter characteristics
with lossless digital filtering.
[0009] A second problem is noise from the transmitter leaks into
the receiver and raises its noise figure. The second problem can be
mitigated by further filtering the transmitter output and by
reducing the spectral sidelobe energy through signal design and
linearization of a transmit amplifier. These changes, however, are
undesirable from the standpoint of transmitter efficiency.
[0010] Canceling an undesired signal with an adaptive canceller was
introduced by B. Widrow et al. in Adaptive Noise Canceling
Principles and Applications, Proc. IEEE, Vol. 63, pp. 1269-1716,
December 1975, which is hereby incorporated herein by reference.
Specific applications for adaptive noise canceling and echo
cancellation in data transmission over telephone channels are given
by J. G. Proakis et al. in Advanced Digital Signal Processing,
Macmillan Publishing Co., New York, 1992, pp. 322-327, 331-332.
[0011] The present invention relates to a method and apparatus for
adaptive digital cancellation of a transmit signal spectrum in a
receiver bandwidth. The apparatus relates to a digital coherent
spectral canceller that attenuates the spectral components from the
transmitter that fall within the bandwidth of the receiver. The
digital adaptive coherent spectral canceller digitizes both a
corrupted receiver signal and a reference transmit signal and then
digitally implements an adaptive coherent spectral canceller
adaptation module.
[0012] One embodiment of the invention is implemented with a direct
conversion digital receiver, which achieves an image-free, high
dynamic range without the use of automatic gain control. Automatic
gain control used to extend the dynamic range of a receiver may be
undesirable for spectrally crowded applications such as cellular
communications because the automatic gain control may make the
receiver sensitivity dependent upon signals and interference that
are outside the signal channel. For example, it is possible for a
strong signal in an adjacent channel to capture the receiver front
end desensitize the receiver such that a weak signal in the channel
of interest is undetectable. This is particularly harmful in a base
station receiver where the receiver receives incoming signals from
multiple remote units. Furthermore, the use of automatic gain
control will likely require the digital coherent canceller to track
the gain changes which will introduce errors and noise.
[0013] One embodiment of the invention uses an adaptive transversal
filter in an all digital implementation to compensate for the
amplitude and phase differences between the transmitter-to-receiver
leakage path and the transmit signal path utilized as a reference
over the receiver bandwidth or bandwidth of interest.
[0014] An advantage of the invention is reduced complexity and
cost, which is achieved through: (1) reduced performance
requirements of the duplexer, (2) reduction in the requirement for
transmit signal filtering, and (3) reduction in the requirements
for high linearity or linearization of the transmit amplifier.
[0015] One aspect of the invention relates to an adaptive,
coherent, digital canceller system. The canceller system is
configured to attenuate a signal spectrum from a transmitter which
falls within a bandwidth of a receiver. The canceller system
comprises a reference bandpass filter, a reference direct
converter, a cross correlation measurer, an adaptation coherent
spectral canceller algorithm module executed, for example, on a
microcontroller, an adaptive digital transversal filter, and a
combiner.
[0016] The reference direct converter is adapted to output a
digitized transmit signal reference of a spectral energy of the
transmitter within the bandwidth of the receiver. The adaptive
digital transversal filter is adapted to align an amplitude and
phase of a digitized transmit signal reference in a reference path
with a transmit signal in a leakage receiver path. The adaptive
digital transversal filter outputs a compensated or equalized
digitized transmit signal reference. The combiner is adapted to
coherently subtract the compensated, digitized transmit signal
reference from a corrupted, digitized receiver signal to form a
residue, having transmitter spectral signal power within the
bandwidth of a receiver is suppressed.
[0017] Another aspect of the invention relates to a method of
attenuating a transmit signal spectrum in a bandwidth of a
receiver. The method comprises digitizing a received signal which
is corrupted by components of a transmit signal, creating a
digitized reference transmit signal of the transmit signal within
the bandwidth of the receiver, aligning the digitized reference
transmit signal in amplitude, phase and time delay with the
digitized received signal, subtracting the digitized reference
transmit signal from the digitized received signal to form a
residue, and suppressing a transmitter spectral signal power of the
residue within the bandwidth of the receiver.
BRIEF DESCRIPTION OF THE DRAWINGS
[0018] FIG. 1 illustrates one embodiment of a transceiver and a
processor in accordance with the present invention.
[0019] FIG. 2A illustrates transmitter spectral spillage into a
receiver bandwidth with a single channel.
[0020] FIG. 2B illustrates transmitter spectral spillage into a
receiver bandwidth with multiple channels.
[0021] FIG. 3 illustrates one embodiment of a communication system
with a transceiver in accordance with the present invention.
[0022] FIG. 4 illustrates an exemplary spectral density at an input
and output of an adaptive coherent spectral canceller within the
transceiver of FIG. 1.
[0023] FIG. 5A illustrates one embodiment of a transceiver with a
shared transmit and receive antenna radiator.
[0024] FIG. 5B illustrates one embodiment of a transceiver with
separate receive and transmit antenna radiators.
[0025] FIG. 6 illustrates one embodiment of a digital coherent
spectral canceller within the transceiver of FIGS. 5A and 5B.
[0026] FIG. 7 illustrates one embodiment of a direct converter
within the transceiver of FIGS. 5A and 5B.
[0027] FIG. 8 illustrates another embodiment of a direct converter
within the transceiver of FIGS. 5A and 5B.
[0028] FIG. 9A illustrates exemplary baseband responses of one
embodiment of a receive filter and a transmit reference filter
within the transceiver of FIG. 5A.
[0029] FIG. 9B illustrates cancellation ratios of spectral leakage
from the transmitter as a result of a digital coherent spectral
canceller of FIG. 5A having varying number of taps with the filter
responses of FIG. 9A.
[0030] FIG. 10A illustrates exemplary baseband responses of another
embodiment of the receive filter and transmit reference filter
within the transceiver of FIG. 5A.
[0031] FIG. 10B illustrates exemplary outputs of one embodiment of
a digital spectral cancellers of FIG. 5A having varying numbers of
taps with the filter responses of FIG. 10A.
[0032] FIG. 11 illustrates another embodiment of a direct converter
within the transceiver of FIGS. 5A and 5B.
[0033] FIG. 12 illustrates another embodiment of a direct converter
within the transceiver of FIGS. 5A and 5B.
[0034] FIG. 13 illustrates one embodiment of a translating
delta-sigma modulator within the direct converter of FIGS. 7 and
11.
[0035] FIGS. 14A-14E are spectral plots used to illustrate the
operation of various embodiments of the transceiver of FIG. 1.
[0036] FIG. 15 is a block diagram showing one embodiment of a
component within FIGS. 11 and 12.
[0037] FIG. 16 is a block diagram showing one embodiment of a clock
generator of FIG. 7.
[0038] FIG. 17 illustrates another embodiment of a modulator within
the direct converters of FIGS. 7 and 11.
DETAILED DESCRIPTION OF THE EMBODIMENTS
[0039] FIG. 1 illustrates an exemplifying communication system in
which the present invention may be implemented. Alternatively, the
present invention may be used in other systems and applications.
FIG. 1 illustrates one embodiment of a transceiver 200 and a
processor 900 in accordance with the present invention. In one
embodiment, the transceiver 200 is a full duplex transceiver,
although the invention can be used in communication systems in
which the base station concurrently transmits and receives
regardless of whether the communication system provides full duplex
operation.
[0040] In one embodiment, the transceiver 200 is for code division
multiple access (CDMA) mobile and base station transceivers. In
another embodiment, the transceiver 200 is for a time division
multiple access (TDMA) transceiver, such as a transceiver used for
GSM and IS-54 base stations. In FIG. 1, the transceiver 200
comprises transmit and receive antenna radiator 102, a duplexer
210, a receiver 106, a transmitter 280 and a digital coherent
spectral canceller 230. In one embodiment, the receiver 106 is a
direct conversion digital receiver described herein. In another
embodiment, the receiver 106 is a modified direct conversion
digital receiver described in U.S. Pat. No. 5,557,642, entitled
"Direct Conversion Receiver For Multiple Protocols" by Williams
modified to exclude automatic gain control.
[0041] In FIG. 1, the transmitter 280 and the digital coherent
spectral canceller 230 are coupled to the processor 900. The
transmit and receive antenna radiator 102 is coupled to the
duplexer 210, which is coupled to the receiver 106, the digital
coherent spectral canceller 230 and the transmitter 280. The
processor 900 sends data and control signals to the transmitter
280, which sends data to the duplexer 210 for transmission via the
antenna radiator 102. The duplexer 210 sends a reference transmit
signal to the canceller 230. The duplexer 210 also receives data
from the radiator 102 and sends data to the receiver 106, which
sends the received data to the canceller 230. The canceller 230 and
the processor 900 use the reference transmit signal from the
duplexer 210 to cancel noise from the transmitter 280 that leaked
into the receiver bandwidth.
[0042] In some conventional systems, noise from the transmitter 280
that leaks into the receiver 106 can be mitigated by further
filtering the transmitter output and by reducing the spectral
sidelobe energy through signal design and linearization of a
transmit amplifier. These changes, however, are undesirable from
the standpoint of transmitter efficiency.
[0043] The digital adaptive coherent spectral canceller 230 of the
present invention attenuates the spectral components from the
transmitter 280 that fall within the bandwidth of the receiver 106.
The digital canceller 230 digitizes both a corrupted receiver
signal and a reference transmit signal and then digitally
implements an adaptive coherent spectral canceller adaptation
module with the processor 900.
[0044] The receiver 106 is preferably a direct conversion digital
receiver that achieves an image-free, high dynamic range without
the use of automatic gain control. Automatic gain control used to
extend the dynamic range of a receiver may be undesirable for
spectrally crowded applications such as cellular communications
because the automatic gain control may make the receiver
sensitivity dependent upon signals and interference that are
outside the signal channel.
[0045] The digital coherent spectral canceller 230 of the present
invention and the methods of using the canceller 230 provide
several advantages over conventional systems. These advantages
include (1) reducing performance requirements of the duplexer 210,
(2) reducing the requirement for transmit signal filtering, and (3)
reducing in the requirements for high linearity or linearization of
the transmit amplifier.
[0046] FIG. 2A illustrates a single-channel system with a
transmitted signal power spectrum (spectral density) 400, denoted
as {overscore (S)}.sub.T(f) and centered about a transmit frequency
f.sub.T, and a received communication signal power 404, denoted as
{overscore (S)}.sub.R(f) and centered about a receive frequency
f.sub.R. In FIGS. 2A, 2B and 4, the horizontal axis represents
frequency (f), such as in units of Gigahertz, and the vertical axis
represents signal power amplitude, such as in units of decibels.
The horizontal axis has been segmented so that more of the signal
energy can be shown.
[0047] FIG. 2A also illustrates a receiver bandwidth 402 from a
receiver bandpass filter, such as the receiver bandpass filter 214
described below with reference to FIG. 5A. The receiver bandwidth
402 is designed to pass the received signal 404 but not the main
transmit signal 400. The transmitted signal power spectrum 400 and
the receiver bandwidth 402 are shown in dashed lines. FIG. 2A also
shows transmit spectral spillage 406 in the receiver bandwidth 402
from a single channel. In other words, noise 406 from the
transmitter 280 (FIG. 1) leaks (spills) into the receiver 106 and
raises its noise figure. The canceller 230 of the FIG. 1 is
designed to cancel this transmit noise 406 within the receiver
bandwidth 402.
[0048] FIG. 2B illustrates a multiple-channel system with a
plurality of transmitted signal powers 404', 404", 404'", a
receiver bandwidth 402 and a plurality of received communication
signal powers 404', 404", 404'". FIG. 2B also shows transmitter
spectral spillage 406 in the receiver bandwidth 402 from multiple
channels (multiple transmit signals and multiple receive signals).
In FIGS. 2A and 2B, the plots represent signals which have passed
through an amplifier in the receiver 106 of FIG. 1, such as the low
noise amplifier 216 shown in FIG. 5A.
[0049] FIG. 4 illustrates an exemplary spectral density at an input
and output of the adaptive coherent spectral canceller 230, as
shown in FIG. 1 and described in greater detail below with
reference to FIGS. 5A, 5B and 6. FIG. 4 is based on the plots
within the receiver bandwidth 402 of FIG. 2B. In FIG. 4, the
bandpass transfer function 402 of the receiver bandpass filter 214
(FIG. 5A) passes the three received signal power densities 404',
404", 404'" and the transmit noise spectral density 406.
[0050] In FIG. 4, a direct converter filter, such as the direct
converter 220 described below with reference to FIG. 5A, has an
even more selective transfer function 408 than the receiver
bandpass filter transfer function 402. The direct converter filter
transfer function 408 passes only two 404', 404" of the three
signal spectral densities 404', 404", 404'" and a portion of the
transmit noise spectral density 406. In other words, only two
received spectral densities 404', 404" are efficiently passed
through the direct converter 220 and on to the canceller 230 (FIGS.
1 and 5A). As shown in FIG. 4, the canceller 230 reduces the
transmit noise spectral density 406 to a residual spectral density
410 or lower.
[0051] FIG. 3 illustrates some of the inputs, outputs, signal flow
and functions of the processor 900 and the transceiver 200 of FIG.
1. FIG. 3 illustrates an antenna radiator 102, a transceiver 200, a
processor 900 and a transmitter frequency synthesizer and VCO 600.
The transceiver 200 comprises an adaptive digital canceller 230, as
shown in FIGS. 1, 5A, 5B and 6, in accordance with the present
invention.
[0052] In FIG. 3, the processor 900 comprises a canceller
adaptation algorithm module 910, a digital demodulator module, a
frequency control module and a signal waveform clock module. The
modules may comprise software, firmware, hardware or any
combination thereof or may be implemented with discrete logic or in
an application specific integrated circuit (ASIC). In an
alternative embodiment, the canceller adaptation module 910 may be
executed by a dedicated element, which is separate from the
processor 900. In one embodiment, the processor 900 comprises a
collection of processing elements such as an ARM or a Power Point
microprocessor and a dedicated digital modem such as the Qualcomm
MSM 3000. Alternatively, in other embodiments, the configurations
of the embodiments, the configuration of the processor 900 may be
different.
[0053] In FIG. 3, the antenna radiator 102 is coupled to the
transceiver 200, which is coupled to the processor 900. The
processor 900 is coupled to the transmitter frequency synthesizer
and VCO 600, which is coupled to the transceiver 200. The processor
900 may be coupled to other components that transfer data and/or
control signals to the processor 900 or receive signals from the
processor 900.
[0054] As shown in FIG. 3, the transceiver 200 transmits and
receives signals via the antenna radiator 102. The transceiver 200
passes various data to the processor 900, such as an I & Q
filtered signal and correlation measurement data, C(j), as
described below with reference to FIG. 5. In this application, the
underlining of a variable, such as C(j), denotes that the variable
is a complex variable, i.e., the variable consists of an in-phase
(I) component and a quadrature component (Q).
[0055] The processor 900 may pass various data to the transceiver
200, such as a clock waveform, CLK, canceller parameters, W(j), and
transmitter data. The processor 900 may also provide control
information, such as a receiver frequency control signal, to the
transceiver 200. The processor 900 may also provide control
information, such as a transmitter or calibration frequency control
signal, to the transmitter frequency synthesizer and VCO 600, which
transmits a signal to the transceiver 200.
[0056] FIG. 5A illustrates one embodiment of the transceiver 200 of
FIG. 3 with a shared receiver/transmitter (Rx/Tx) antenna radiator
102 and an adaptive, digital, coherent spectral canceller 230. In
FIG. 5A, the transceiver 200 comprises a duplexer 210, a low noise
amplifier (LNA) 216, a receiver direct converter 220, a digital
adaptive coherent spectral canceller 230, a matched or channel
filter 250 and a transmitter 280. The duplexer 210 comprises a
circulator 212, a receiver bandpass filter 214, a transmitter
bandpass filter 218 and a directional coupler 235. The canceller
230 comprises a combiner 232, an adaptive digital transversal
filter (or equalizer) 234, a reference bandpass filter 236, a
reference direct converter 238 and a cross correlation measurement
module 240.
[0057] In another embodiment, the duplexer 210 does not comprise a
directional coupler 235, and instead some of the energy passed to
the antenna radiator 102 is coupled to an input of the reference
bandpass filter 236 within the canceller 230.
[0058] In FIG. 5A, the transmitter 280 receives data and control
signals from a digital processor interface. The transmitter 280 is
coupled to the transmitter bandpass filter 218, which is coupled to
the directional coupler 235. The directional coupler 235 is coupled
to the circulator 212 and the reference bandpass filter 236. The
circulator 212 is coupled to the antenna radiator 102 and the
receiver bandpass filter 214, which is coupled to the LNA 216. The
LNA 216 is coupled to the receiver direct converter 220, which is
coupled to the combiner 232, the reference direct converter 238 and
the digital processor interface. The combiner 232 is coupled to the
matched or channel filter 250 and the adaptive digital transversal
filter 234. In one embodiment, the combiner 232 is also coupled to
the cross correlation measurement module 240. The matched filter
250 is coupled to the digital processor interface and the cross
correlation measurement module 240. The reference bandpass filter
236 is coupled to the reference direct converter 238, which is
coupled to the receiver direct converter 220, the digital processor
interface, the adaptive digital transversal filter 234 and the
cross correlation measurement module 240. Both the adaptive digital
transversal filter 234 and the cross correlation measurement module
240 are coupled to the canceller adaptation module 910, which is
executed by the digital processor 900 (FIG. 3) or by a separate
microcontroller.
[0059] The general use and operation of the transceiver 200 is
described with reference to FIGS. 1-5A. In FIG. 5A, the antenna
radiator 102 receives an incoming signal and transfers the incoming
signal to the circulator 212. In one embodiment, the incoming
signal is a high frequency signal, such as a digitally modulated RF
signal centered about a carrier frequency of approximately 2
Gigahertz (GHz). Alternatively, in other embodiments, other signals
and carrier frequencies may be used in accordance with the
invention. The circulator 212 transfers the received signal to the
receiver bandpass filter 214.
[0060] Concurrently, the transmitter 280 may be transmitting a
signal to the transmitter bandpass filter 218, which outputs a
transmitter signal S.sub.TX(t), to the circulator 212. The signals
described herein may be represented as either functions of time,
(t), or alternatively, as their Fourier equivalents, functions of
frequency, (.omega.). The circulator 212 directs the transmit
signal to the antenna radiator 102. As shown in FIGS. 2A, 2B and 4,
some of the energy associated with the transmit signal,
S.sub.TX(t), leaks into the receiver bandpass filter 214 because
the circulator 212 does not provide perfect isolation. This leakage
signal energy 406 is limited by the receiver bandpass filter 214,
as shown in FIGS. 2A, 2B and 4.
[0061] The resultant output of the receiver bandpass filter 214 due
to S.sub.TX(t) and the received signal is input to the LNA 216. The
LNA 216 amplifies the signal and passes the signal to the receiver
direct converter 220. The receiver direct converter 220 converts
the filtered signal 217 to a baseband, digitized and filtered
signal to produce an output, X(n).
[0062] In FIG. 5A, the digital coherent spectral canceller 230
adaptively suppresses the transmitter spectral spillage 406 in the
receiver band 402 shown in FIG. 4. Specifically, the directional
coupler 235 provides a reference signal to the digital coherent
spectral canceller 230. The reference signal represents the
transmit signal, S.sub.TX(t). Within the canceller 230, the
reference signal is passed to the reference bandpass filter 236.
The transfer function H.sub.REF(.omega.), of the reference bandpass
filter 236 preferably matches the transfer function
H.sub.RX(.omega.), of the receiver bandpass filter 214 in the
previously described leakage path to the receiver direct converter
220. Alternatively, one of the filters 214, 236 may be designed to
cancel a smaller bandwidth (more rejection) compared to the other
filter 214, 236. The reference bandpass filter 236 outputs a signal
to the reference direct converter 238, where the reference signal
is converted to a digitized, baseband signal.
[0063] The reference direct converter 238 creates a digitized
reference of the spectral energy of the transmitter 280 within the
bandwidth 402 of the receiver (FIG. 4) to produce an output, Z(n).
The reference direct converter 238 transfer the signal Z(n) to the
adaptive digital transversal filter 234 and the cross correlation
measurement module 240.
[0064] The adaptive digital transversal filter 234 compensates for
the amplitude and phase differences between the
transmitter-to-receiver leakage path (the receiver filter 214 and
direct converter 220) and the reference transmit signal path (the
reference filter 236 and direct converter 238) used as a reference
over the receiver bandwidth or bandwidth of interest. In a
preferred embodiment, the adaptive digital transversal filter 234
aligns and shapes the digitized transmit signal reference, Z(n),
from the reference path in amplitude and phase to substantially
match the amplitude and phase of the transmit leakage signal
through the leakage receiver path. The adaptive digital transversal
filter 234 outputs an output signal, U(n), to the combiner 232.
[0065] In FIG. 5A, the combiner 232 uses two summers 424, 426 to
coherently subtract the output, U(n), of the adaptive digital
transversal filter 234 from the digitized receiver output, X(n), to
form a signal, Y(n), whose transmitter spectral spillage signal
power (the error) within the receiver bandwidth is suppressed, as
shown in FIG. 4. The signal Y(n) is passed to the matched or
channel filter 250, which outputs a signal R(n). The signals Y(n)
and/or R(n) may be referred to as the `residue.`
[0066] The matched or channel filter 250 passes the residue signal
R(n) to the cross-correlation measurement module 240. The
cross-correlation measurement module 240 identifies the common
signal characteristics of the receiver path and the reference path
by cross-correlating the signal inputs. In one embodiment, the
cross correlation module 240 identifies the common signal
characteristics between R(n) and Z(n). The cross correlation
measurement module 240 outputs parameters C(j) to the canceller
adaptation module 910.
[0067] The canceller adaptation module 910 executes an adaptive
cancellation algorithm which is described below and outputs
adaptive filter coefficients W(j) to the adaptive digital
transversal filter 234. The canceller adaptation module 910 adjusts
the adaptive filter coefficients, W(j) such that the
cross-correlation between the digitized transmit reference signal,
Z(n), and the residue, R(n), are minimized in a least mean-square
sense over a bandwidth of interest.
[0068] Alternatively, in another embodiment, Y(n) may be used
instead of R(n) as the residue, which is input into the cross
correlation module 240. The cancellation bandwidth can be expanded
to the bandwidth of the receiver bandpass filter 214 and reference
bandpass filter 236, F(.omega.)=1, by utilizing the output of the
combiner 232, Y(n), for the residual signal instead of R(n).
[0069] The digital adaptive coherent spectral canceller 230 is
preferably implemented with two high dynamic range direct
converters 220, 238. The direct converters 220, 238 provide high
dynamic range, substantially identical performance and no automatic
gain control. Digitization allows use of high performance, easily
repeatable digital filters, such as the adaptive digital
transversal filter 234. In general, digital filters are more
repeatable than analog filters. The digital filters are
advantageously used in the present invention for more effective
cancellation.
[0070] In FIG. 5A, the receiver direct converter 220 preferably
converts the coupled signal 217 using the same local oscillator
signal, VCO.sub.RX, as the reference direct converter 238, because
the canceller 230 focuses on the portion of the filtered transmit
signal spectrum that is within the receiver bandwidth, as
illustrated in FIGS. 2A, 2B and 4.
[0071] FIG. 5B illustrates an alternate embodiment of a transceiver
200' of FIG. 3 with separate receive and transmit antenna radiators
120, 122 and an adaptive transmit canceller 230. In FIG. 5B, the
transceiver 200' comprises a receiver bandpass filter 214, a low
noise amplifier (LNA) 216, a receiver direct converter 220, a
digital adaptive coherent spectral canceller 230, a matched or
channel filter 250, a directional coupler 235, a transmitter
bandpass filter 218 and a transmitter 280.
[0072] In FIG. 5B, the transmitter 280 receives data and control
signals from the digital processor interface. The transmitter 280
is coupled to the transmitter bandpass filter 218, which is coupled
to the directional coupler 235. The directional coupler 235 is
coupled to the transmitter antenna radiator 122 and the spectral
canceller 230. The receiver antenna radiator 120 is coupled to the
receiver bandpass filter 214, which is coupled to the LNA 216. The
LNA 216 is coupled to the receiver direct converter 220, which is
coupled to the digital processor interface and the spectral
canceller 230. The spectral canceller 230 is coupled to the
processor interface and the matched or channel filter 250, which is
also coupled to the digital processor interface.
[0073] Even though there is no circulator 212 in the transceiver
200' of FIG. 5B as in the transceiver 200 of FIG. 5A, the
transceiver 200' of FIG. 5B may still experience transmit spectral
spillage or leakage into the receive path as the receiving antenna
radiator 120 receives the signal energy radiated by the
transmitting antenna radiator 122.
[0074] In addition to canceling or attenuating leakage from the
subject transmitter frequency, the transceiver 200' of FIG. 5B can
also be configured to cancel interferences from other transmit
antennas located in the proximity of the receiver.
[0075] FIG. 6 illustrates one embodiment of a digital coherent
spectral canceller 230 within the transceivers 200, 200' of FIGS.
5A and 5B. As explained above, the canceller 230 comprises a
combiner 232, an adaptive digital transversal filter 234, a
reference bandpass filter 236, a reference direct converter 238 and
a cross correlation measurement module 240.
[0076] In FIG. 6, the reference bandpass filter 236 receives a
signal from the directional coupler 235 (FIG. 5A) and outputs a
signal to the reference direct converter 238. The reference
converter 238 is coupled to the adaptive digital transversal filter
234, the cross correlation measurement module 240 and the digital
processor interface (FIG. 5A). The adaptive digital transversal
filter 234 is coupled to the canceller adaptation module 910 and
the combiner 232, which is coupled to the matched or channel filter
250 (FIG. 5A) and the receiver direct converter 220 (FIG. 5A). As
shown in FIG. 5A, the combiner 232 may also be coupled to the cross
correlation measurement module 240. The cross correlation
measurement module 240 is coupled to the matched filter 250 (FIG.
5A) and the canceller adaptation module 910.
[0077] In FIG. 6, the reference direct converter 238 creates a
digitized reference of the spectral energy of the transmitter 280
within the bandwidth of the receiver to produce an output, Z(n).
The output of the reference direct converter 238, Z(n), comprises
an in-phase component, Z.sub.I(n), and a quadrature component,
Z.sub.Q(n), as shown in FIG. 6. The reference direct converter 238
outputs both the I and Q components of the signal Z(n) to the
adaptive digital transversal filter 234 and to the cross
correlation measurement module 240. The cross correlation
measurement module 240 also receives an I and a Q component of a
residue signal R(n) from the matched or channel filter 250. The
adaptive digital transversal filter 234 outputs an in-phase
component, U.sub.I(n), and a quadrature component, U.sub.Q(n) (an I
and a Q component) of an output signal, U(n), to the combiner
232.
[0078] In FIG. 6, the combiner 232 preferably uses a digital delay
.DELTA.T.sub.S 420, 422 to compensate for differences in time
delays between the reference path and the leakage receiver path.
The digital delay reduces time de-correlation effects and
instabilities due to different path lengths and processing times
between the receiver and reference paths.
[0079] In FIG. 6, the in-phase and quadrature component outputs of
the combiner 232 are given as:
Y.sub.I(n)=X.sub.I(n-.DELTA.)-.orgate..sub.I(n) (1)
Y.sub.Q(n)=X.sub.Q(n-.DELTA.)-.orgate..sub.Q(n) (2)
[0080] where the terms are defined as follows:
[0081] X.sub.I(n) Digitized output of I-path of direct converter
220,
[0082] X.sub.Q(n) Digitized output of Q-path of direct converter
220,
[0083] Y.sub.I(n) I-path output of combiner 232,
[0084] Y.sub.Q(n) Q-path output of combiner 232,
[0085] U.sub.I(n) I-path output of adaptive digital transversal
filter 234,
[0086] U.sub.Q(n) Q-path output of adaptive digital transversal
filter 234,
[0087] .DELTA. Integer number of sample periods, T.sub.S, for delay
added to approximately compensate for the difference in delay
between the reference and signal paths.
[0088] The outputs of adaptive digital transversal filter 234 are:
1 U 1 ( n ) = k = o K { W k - I ( j ) Z I ( n - k ) - W k - Q ( j )
Z Q ( n - k s ) } ( 3 ) U Q ( n ) = k = o K { W k - I ( j ) Z Q ( n
- k ) + W k - Q ( j ) Z I ( n - k ) } ( 4 )
[0089] where the terms are defined as follows:
[0090] Z.sub.I(n) Digitized output of I-path of the reference
direct converter 238,
[0091] Z.sub.Q(n) Digitized output of Q-path of the reference
direct converter 238,
[0092] W.sub.K-I(j) I coefficients of the adaptive digital
transversal filter 234 at the j-th iteration of an adaptation
algorithm of the canceller adaptation module 910,
[0093] W.sub.K-Q(j) Q coefficients of the adaptive digital
transversal filter 234 at the j-th iteration of an adaptation
algorithm of the canceller adaptation module 910,
[0094] K Number of taps (complex coefficients) in the adaptive
digital transversal filter 234.
[0095] The adaptation module 910 adjusts the coefficient vector,
W(j), of the adaptive digital transversal filter 234 to minimize
the transmitter energy within the bandwidth determined by either
the bandwidth of the direct converters 220 and 238 or the bandpass
filters 214 and 236. Specifically, this may be expressed as
minimizing the error 2 ( ) = 1 2 [ X _ ( ) - j Ts - Z _ ( ) A _ ( )
] 2 F _ ( ) 2 ( 5 )
[0096] where the terms are defined as follows:
[0097] X(.omega.) The digitized and filtered output of the receiver
direct converter 220,
[0098] Z(.omega.) The digitized and filtered output of the
reference direct converter 238,
[0099] A(.omega.) The transfer function of the adaptive digital
transversal filter 234,
[0100] F(.omega.) The transfer function of the matched or channel
filter 250, where F(.omega.)=1 in the alternative embodiment.
[0101] .omega..sub.2 Upper frequency (in radians) over which
transmitter spectrum is to be cancelled.
[0102] .omega..sub.1 Lower frequency (in radians) over which
transmitter spectrum is to be cancelled.
[0103] T.sub.s Sampling period of output of receiver direct
converter 220 and reference direct converter 238 (reciprocal of
sampling frequency).
[0104] Where the coefficients, {[W.sub.k-I(j), W.sub.k-Q(j),], k=1
. . . K}, are selected to minimize E(.omega.).
[0105] The coefficients associated with the adaptation algorithm
used by the adaptive digital transversal filter 234 which minimize
E(.omega.) are:
W.sub.k-I(j+1)=.rho..circle-solid.W.sub.k-I(j)+.upsilon..circle-solid.C.su-
b.k-I(j) (6)
W.sub.k-Q(j+1)=.rho..circle-solid.W.sub.k-Q(j)+.upsilon..circle-solid.C.su-
b.k-Q(j) (7)
[0106] where, 3 C k - I ( j ) = m = n M + n { Z I [ m - k ] R I [ m
] - Z Q [ m - k ] R Q [ m ] } , ( 8 ) C k - Q ( j ) = m = n M + n {
Z Q [ m - k ] R I [ m ] - Z I [ m - k ] R Q [ m ] } , ( 9 )
[0107] for k=1 . . . K, with M representing the number of samples
in the measurement interval, .rho. representing a constant set so
that the integrated correlated noise component does not grow with
the iterations, and the step size at each iteration.
[0108] Together, FIGS. 9A, 9B, 10A and 10B are used to illustrate
examples of the cancellation obtainable as a function of the
complex coefficients (taps) of the adaptive digital transversal
filter 234, where the bandwidth of F (.omega.) is one quarter of
the sampling frequency. In FIGS. 9A-10B, the horizontal axis
represents frequency.times. 1000/sample rate. The vertical axis
represents amplitude in decibels (dB).
[0109] FIG. 9A illustrates exemplary effective baseband responses
(transfer functions) 430, 432 or one embodiment of the receive
filter 214 and the transmit reference filter 236 of FIG. 5A, where
the receiver and reference filters 214, 236, H.sub.RX((.omega.) and
H.sub.REF(.omega.), are Chebyshev bandpass filters designed for
equal stop and pass bandwidths with different pass band ripples,
such as 0.01 dB and 1.0 dB respectively. In alternative
embodiments, the baseband responses 430, 432 shown in FIG. 9A may
be switched. In other words, either baseband response 430, 432
shown in FIG. 9A may be associated with the receive filter 214
and/or the transmit reference filter 236.
[0110] FIG. 9B illustrates 10 log.sup.10
.vertline.Y(.omega.).sup.2/.vertl- ine.X(.omega.).sup.2,
cancellation ratios in dB of the spectral leakage from the
transmitter 280 due to the digital coherent spectral 230 of FIG. 6.
Curves 434-440 represent a different number of taps for the digital
coherent spectral 230 of FIG. 6, where the receiver and reference
filters 214, 236, H.sub.RX (.omega.) and H.sub.REF (.omega.), are
Chebyshev bandpass filters with baseband responses illustrated in
FIG. 9A. In FIG. 9B, the performance of the digital coherent
spectral canceller 230 reflects cancellation over the bandwidth of
.omega..sub.1=0 to .omega..sub.2=250. In FIG. 9B the line 434
represents the amount of cancellation achieved with an adaptive
digital transversal filter 234 of one tap, which does not
compensate (equalize) for the frequency dependent mismatch
characteristics between the two paths. As shown in FIG. 9B, if the
adaptive digital transversal filter 234 uses an increased number of
taps, the cancellation performance is significantly improved.
[0111] The filtering of the reference transmitter output signal
with a reference filter 236 having substantially the same design as
the receiver filter 214 (FIG. 9A) (i.e., the transfer functions
H.sub.RX (.omega.) and H.sub.REF (.omega.) in FIG. 5A are
substantially similar) presents at least two advantages. First, it
may reduce dynamic range requirements by providing high attenuation
to signals outside the receiver band so that the reference direct
converter 238 (FIG. 5A) only digitizes the signal within the
spectral sidelobes of the transmit signal that fall within the
receiver bandwidth. This attenuation may significantly reduce the
resolution and linearity requirements of the reference direct
converter 238 because the sidelobes within the receiver filter's
bandwidth will normally be suppressed more than 40 dB below the
peak of the transmitter signal.
[0112] Second, using a reference bandpass filter 236 with the same
design (i.e., transfer function) as the receiver filter 214
associated with the duplexer 210 may also increase the amount of
suppression because the filters 214, 236 provide the same amplitude
and phase variations on both signals before the signals are
inputted to the direct converters 220, 238, respectively. The
adaptive transversal equalizer 234 may advantageously compensate
for any small differences in the transfer functions using a smaller
number of taps than required if H.sub.RX (.omega.) and H.sub.REF
(.omega.) in FIG. 5A were not substantially similar.
[0113] FIG. 10A illustrates exemplary cancellation ratio responses
442, 444 of another embodiment of the receiver filter 214 and the
transmit reference filter 236 of FIG. 5A, where the receiver and
reference filters, H.sub.RX (.omega.) and H.sub.REF (.omega.), are
Chebyshev bandpass filters designed for different bandwidths and
with different pass band ripples, such as 0.5 dB and 1.5 dB
respectively. The transfer functions 442, 440, H.sub.RX (.omega.)
and H.sub.REF (.omega.), of FIG. 10A are less similar than the
transfer functions 430, 432 of FIG. 9A. In alternative embodiments,
the baseband responses 442, 444 shown in FIG. 10A may be switched.
In other words, either baseband response 442, 444 shown in FIG. 10A
may be associated with the receive filter 214 and/or the transmit
reference filter 236.
[0114] FIG. 10B illustrates 10
log.sub.10.vertline.Y(.omega.).vertline..su-
p.2/X(.omega.).vertline..sup.2, cancellation ratios in dB of the
spectral leakage from the transmitter 280 due to the digital
coherent spectral 230 of FIG. 6. Curves 446-452 represent a
different number of taps for the digital spectral canceller 230 of
FIG. 6, where the receiver and reference filters 214, 236,
H.sub.RX(.omega.) and H.sub.REF(.omega.), are Chebyshev bandpass
filters with baseband responses illustrated in FIG. 10A. In this
example, because the transfer functions 442, 440, H.sub.RX(.omega.)
and H.sub.REF(.omega.), of FIG. 10A are less similar than the
transfer functions 430, 432 of FIG. 9A, more complexity (taps) may
be needed for the adaptive digital transversal filter 234 to
achieve substantially the same level of cancellation performance as
the performance shown in FIG. 9B.
[0115] FIG. 7 illustrates one embodiment of a direct converter 220
within the transceiver 200, 200' of FIGS. 5A or 5B. The structure
of the reference direct converter 238 is substantially similar to
the structure of the receiver direct converter 220. In FIG. 7, the
direct converter 220 comprises an LNA 202, a divider 205, a
translating delta-sigma modulator 213 for the in-phase path, a
translating delta-sigma modulator 211 for the quadrature path, a
receiver frequency synthesizer and clock generator 500, and I &
Q decimation filters 300. In one embodiment, the LNA 202 and the
LNA 216 of FIG. 5A are two separate components. In another
embodiment, the LNA 202 and the LNA 216 represent a single LNA
component. The divider 205 may be implemented with a variety of
active elements and/or passive elements. The divider 205 may be
implemented as a splitter. Alternatively, the output of the LNA 202
can be coupled directly to both of the translating delta-sigma
modulators 211, 213.
[0116] The LNA 202 of the direct converter 220 and/or the LNA 216
receives an RF signal 201 from the receiver bandpass filter 214
(FIG. 5A). The LNA 202 is coupled to the divider 205, which is
coupled to both translating delta-sigma modulators 213, 211. The
translating delta-sigma modulators 213, 211 are coupled to the I
& Q decimation filters 300 and the receiver synthesizer and
clock generator 500. The receiver synthesizer and clock generator
500 is coupled to the reference direct converter 238 (FIG. 5A) and
the digital processor interface (FIG. 5A). The decimation filters
300 are coupled to the combiner 232 (FIG. 5A).
[0117] In FIG. 7, the LNA 202 amplifies the incoming signal 201
while avoiding the addition of excessive noise and distortion. The
divider 205 separates the amplified signal into in-phase and the
quadrature components and outputs the in-phase and quadrature
components to the in-phase and quadrature modulators 213, 211,
respectively. The modulators 213, 211 translate the received I
& Q signal components by f.sub.RX to the baseband of the
received signal and digitize the I & Q signal components. The
modulators 213, 211 produce outputs to the decimation filter 300.
In one embodiment, the digital data output of each of the
modulators 213, 211 is a one-bit data stream at the same rate equal
to f.sub.RX.
[0118] The decimation filter 300 performs digital filtering and
decimation to produce digital words at the rate of a clock, CLK. In
FIG. 7, the decimation filter 300 converts the stream of 1-bit
digital words into a stream of N-bit digital words that comprise
conventional binary representations of the signals. In one
embodiment, the decimation filter 300 produces a clock signal,
CLK_M, which indicates the rate at which the binary representations
are created and is used to transfer the filtered output. The rate
at which the words produced by the decimation filter 300 is
determined by the oversampling ratio. For an oversampling ratio of
M, the output clock rate is 1/M times the rate of the incoming
data, or alternatively stated, for the embodiment of FIG. 13, the
output clock rate is equal to f.sub.RX/M where f.sub.RX is the rate
of the translation and conversion clock, CLK.
[0119] The decimator filter 300 also attenuates unwanted signals
and noise sources outside of the bandwidth of interest while
preserving the signals of interest. The desired filter
characteristics of the decimation filter 300 are selected on the
basis of the characteristics of the receive signal as well as the
conversion clock rate, f.sub.RX. In one embodiment, the decimation
filter 300 has a programmable characteristic which can be modified
to accommodate a variety of different waveforms. The decimation
filer 300 is preferably implemented with a finite impulse response
(FIR) filter whose characteristics are modified by changing the
value of the filter coefficients. The value of the filter
coefficients of a digital filter can be readily changed via
software making such modifications practical.
[0120] FIG. 8 illustrates another embodiment of a direct converter
800 within the transceiver 200, 200' of FIGS. 5A or 5B. In FIG. 8,
the direct converter 800 comprises an LNA 802, a pair of amps 808,
809, a pair of balanced mixers 810, 811, a pair of low pass
filter/amps 812, 813, a pair of analog-digital converters 814, 815,
a quadrature hybrid module 805 and a receiver frequency synthesizer
and VCO 816. In one embodiment, the LNA 802 and the LNA 216 of FIG.
5A are two separate components In another embodiment, the LNA 802
and the LNA 216 represent a single LNA component.
[0121] The LNA 802 and/or the LNA 216 receives the RF signal 201
from the receiver bandpass filter 214 (FIG. 5A). The LNA 202 is
coupled to the divider 804, which is coupled to both amps 808, 809.
The amp 808 is coupled to the balanced mixer 810, which is coupled
to the quadrature hybrid module 805 and the low pass filter/amp
812. The low pass filter/amp 812 is coupled to the analog-digital
converter 814, which is coupled to the analog digital convert 815,
the combiner 232 (FIG. 5A) and the digital processor interface
(FIG. 5A). The amp 809 is coupled to the balanced mixer 811, which
is coupled to the quadrature hybrid module 805 and the low pass
filter/amp 813. The low pass filter/amp 813 is coupled to the
analog-digital converter 815, which is coupled to the analog
digital converter 814, the combiner 232 (FIG. 5A) and the digital
processor interface (FIG. 5A).
[0122] FIG. 11 illustrates another embodiment of a direct converter
220' within the transceiver 200, 200' of FIGS. 5A or 5B. In FIG.
11, the direct converter 220' comprises a divider 205, a
translating delta-sigma modulator 213 for the in-phase path, a
translating delta-sigma modulator 211 for the quadrature path, a
clock generator 504, an I & Q decimation filters 300 and an I
& Q gain quadrature and offset correction module 410. In one
embodiment, the direct converter 220' of FIG. 11 may include an
input LNA coupled to the RF input of the divider 205.
[0123] In FIG. 11, the divider 205 receives an RF input from the
LNA 216 (FIG. 5A). The divider 205 is coupled to both translating
delta-sigma modulators 213, 211. The translating delta-sigma
modulators 213, 211 are coupled to the I & Q decimation filters
300 and the clock generator 504. The clock generator 504 is coupled
to the reference direct converter 238 (FIG. 5A) and the digital
processor interface (FIG. 5A). The decimation filters 300 are
coupled to the I & Q gain quadrature and offset correction
module 410, which is coupled to the combiner 232 (FIG. 5A).
[0124] FIG. 12 illustrates another embodiment of a direct converter
800' within the transceiver 200, 200' of FIGS. 5A or 5B. In FIG.
12, the direct converter 800' comprises a divider 804, a pair of
amps 808, 809, a pair of balanced mixers 810, 811, a pair of low
pass filter/amps 812, 813, a pair of analog-digital converters 814,
815, a quadrature hybrid module 805 and an I & Q gain,
quadrature and offset correction module 850. In one embodiment, the
direct converter 800' of FIG. 12 may include an input LNA coupled
to the RF input of the divider 205.
[0125] In FIG. 12, the divider 804 receives an RF input from the
LNA 216 (FIG. 5A). The divider 804 is coupled to both amps 808,
809. The amp 808 is coupled to the balanced mixer 810, which is
coupled to the quadrature hybrid module 805 and the low pass
filter/amp 812. The loss pass filter/amp 812 is coupled to the
analog-digital converter 814, which is coupled to the analog
digital converter 815, the I & Q gain, quadrature and offset
correction module 850 and the digital processor interface (FIG.
5A). The amp 809 is coupled to the balanced mixer 811, which is
coupled to the quadrature hybrid module 805 and the low pass
filter/amp 813. The low pass filter/amp 813 is coupled to the
analog-digital converter 815, which is coupled to the analog
digital converter 814, the I & Q gain, quadrature and offset
correction module 850 and the digital processor interface (FIG.
5A).
[0126] FIG. 13 illustrates one embodiment of a translating
delta-sigma modulator 211 within the direct converters 220, 220' of
FIGS. 7 and 11. Because the modulator 213 is substantially similar
to the modulator 211, only the modulator 211 will be described
herein. In FIG. 13, the modulator 211, 213 comprises a
complementary amplifier 310, a switch 312, a loop amplifier 314, a
one-bit digital-to-analog converter 316, a loop filter 318 and an
edge-triggered comparator 320.
[0127] In FIG. 13, the output 206 of the divider 205 (FIGS. 7 or
11) is input into the complementary amplifier 310, which is coupled
to the switch 312. The switch 312 is coupled to the loop amplifier
314. The loop amplifier 314 is coupled to the digital-to-analog
converter 316 and the loop filter 318, which is coupled to the
comparator 320. The CLK signal is coupled to the switch 312 and the
edge-triggered comparator 320.
[0128] The modulator 211 translates the received I & Q signal
components by an amount equal to the frequency of the CLK signal,
f.sub.RX, to the baseband of the received signal and digitizes the
I & Q signal components. In FIG. 13, the complementary
amplifier 310 receives the modulated RF carrier signal. At a
non-inverting output, the complementary amplifier 310 produces a
voltage that is G times the voltage at the input to the
complementary amplifier 310. At an inverting output, the
complementary amplifier 310 produces a voltage that is -G times the
voltage at the input to the complementary amplifier 310. The
inverting and non-inverting outputs of the complementary amplifier
310 are coupled to two input ports of a switch 312. The control
port of the switch 312 determines which input port of the switch
312 is coupled to the output port of the amplifier 310 and is
driven by the conversion clock, CLK, such that the output port of
the switch 312 is alternately coupled to the inverting and
non-inverting outputs of the complementary amplifier 310.
[0129] Together, the complementary amplifier 310 and the switch 312
perform the functions of a commutator which inverts the polarity of
the modulated RF carrier signal on every half cycle of the
conversion clock, CLK. If the frequency of the conversion clock CLK
is chosen to be approximately equal to the carrier frequency of the
modulated RF carrier signal, effectively, the commutator translates
the modulation of the carrier signal down to D.C. centered or
frequency offset baseband. In addition to the low frequency signal
components, high frequency signal components are also generated by
the commutator. However, the high frequency components are
attenuated by the delta-sigma modulator 211 and further filtering.
In one embodiment, the frequency of the conversion clock CLK is
programmable to permit the translation of a variety of waveforms
over a range of center frequencies.
[0130] In one embodiment, the commutator comprised of the
complementary amplifier 310 and the switch 312 is not a
conventional downconverter. The mathematical paradigm for a
conventional downconverter is multiplication by a sinusoidal
signal. Practical implementations of conventional downconverters
(such as circuits employing using diode rings or Gilbert multiplier
circuits) are incapable of realizing this mathematical paradigm
without introduction of distortion and feed-through effects that
result in the creation of undesired spurious signals.
[0131] In contrast, the mathematical paradigm of the commutator is
that of alternately multiplying the input signal by +1 or -1 on
opposite half cycles of a clock signal. Practical implementations
of the commutator employing a fast switch behave more closely to
this mathematical paradigm, thus avoiding the production of
non-linear components of the signal in the baseband signal in
comparison to a conventional down converter.
[0132] The output of the switch 312 is coupled to the input of the
core delta-sigma modulator, which comprises a loop amplifier 314, a
loop filter 318, an edge-triggered comparator 320 and a one-bit
digital-to-analog (D/A) converter 316. In the preferred embodiment,
the core delta-sigma modulator is operated at the same frequency as
the commutator. Use of a conversion clock operating at or near the
carrier frequency provides a significant oversampling ratio in
typical embodiments and, hence, leads to high resolution, high
dynamic range performance according to well-known principles of
delta-sigma conversion.
[0133] In FIG. 13, the output of the loop amplifier 314 is the
difference between the voltage coupled to its non-inverting input
port and its inverting input port times a voltage gain, A.sub.V,
where the voltage gain is typically a large positive constant. The
loop filter 318 is typically an analog low pass filter but can be
embodied in other forms. In one embodiment, the loop amplifier 314
and loop filter 318 act as an integrator. When the voltage value at
the signal input to the edge-triggered comparator 320 is greater
than a predetermined threshold value at the time the conversion
clock transitions, the output is a logic value 1. When the voltage
value at the signal input to the edge-triggered comparator 320 is
less than the predetermined threshold value at the time the
conversion clock transitions, the output is a logic value 0. The
output of the edge-triggered comparator 320 is coupled to the input
of the one-bit digital-to-analog converter 316. The one-bit
digital-to-analog converter 316 produces one of two analog levels
at its output depending upon the digital logic value applied to its
input. The output of the one-bit digital-to-analog converter 316 is
coupled to the inverting input of the loop amplified 314.
[0134] The core delta-sigma modulator 211, 213 shown in FIG. 13 and
described above is a standard, one-bit digital-sigma modulator.
However, a variety of delta-sigma modulators and delta-sigma
modulation techniques can be combined with the teachings of the
present invention. For example, additional information concerning
delta-sigma modulators is found in Delta-Sigma Data Converters:
Theory, Design, and Simulation by Steven R. Norsworthy, published
by IEEE Press in 1996.
[0135] Because metal-oxide semiconductor (MOS) technology lends
itself inherently to implementing discrete-time filters based on
capacitor ratios, prior art systems use switched-capacitor
technology to implement delta-sigma modulators. Inherently,
switched-capacitor filters cause aliasing and hence, additional
interference to the system. In addition, because MOS
switched-capacitor circuits must be operated at a much lower
oversampling ratio, they do not have as much resolution for any
given order of the delta-sigma modulator, compared to one
embodiment of the modulator 211 of the present invention. In order
to gain resolution, prior art systems typically use higher order
loop filters which are only conditionally stable. As the order of
the delta-sigma modulator is increased, the implementation of a
stable loop that is capable of operating at high clock frequencies
becomes more difficult.
[0136] In contrast, one embodiment of the modulator 211 comprises a
continuous-time filter for the loop filter 318. As noted above, one
embodiment of the delta-sigma modulator 211 operates at or near the
carrier frequency. Due to the use of a high frequency clock, the
use of higher order filtering is not needed to achieve a high
degree of resolution. Therefore, the use of a lower order,
continuous-time filter is practical in conjunction with one
embodiment of the modulators 211, 213. Continuous-time filters are
less difficult and bulky to implement than switched-capacitor
filters. Furthermore, continuous-time filters can be operated at
much higher frequencies than switched-capacitor circuits in a given
semiconductor technology. Finally, the use of a continuous-time
filter has the added advantage of eliminating aliasing that is
potentially produced by switched-capacitor filters.
[0137] Many modem delta-sigma converters are currently available
that are implemented in silicon metal oxide semiconductor (MOS)
technology. Typically such designs use switched capacitor
techniques to sample the incoming signal for conversion. However,
circuits capable of processing high frequency input signals, such
as those formed from silicon bipolar, silicon germanium (SiGe), or
gallium arsenide (GaAs) technologies, can use current steering
architectures in order to increase system efficiencies.
[0138] FIGS. 14A-14E are spectral plots used to illustrate the
operation of various embodiments of the transceiver 200 of FIG. 1.
An understanding of the desired characteristics of the decimation
filter 300 can be understood with reference to FIG. 14A where the
vertical axis represents energy such as in units of decibels and
the horizontal axis represents frequency such as in units of
Gigahertz. FIG. 14A is a spectral plot showing received signal
energies 330, 332 and 334 centered about three different carrier
frequencies, F.sub.c1, F.sub.c2 and f.sub.c3, respectively. In one
embodiment, we shall assume that signal energies 330, 332 and 334
each comprise an incoming waveform comprising a digitally modulated
RF signal of interest. We shall also assume that the conversion
clock, CLK, operates at a frequency f.sub.RX which is between
frequencies f.sub.c1 and f.sub.c2. In FIG. 14A, the horizontal axis
has been segmented so that more of the signal energy can be
shown.
[0139] FIG. 14B represents the corresponding output of the switch
313 (excluding noise) of FIG. 13, when the spectrum shown in FIG.
14 is applied thereto. For example, in a typical embodiment, the
frequency f.sub.RX is equal to 1851.4 MHz and the frequencies
f.sub.c1, f.sub.c2 and f.sub.c3 are 1851, 1851.6 and 1852.2 MHz,
respectively. Each of the signal energies 330, 332 and 334 have a
bandwidth of approximately 100 kHz. Thus, in FIG. 14B, the signal
energies 336, 338 and 340 correspond to signal energies 330, 332
and 334, respectively, and are centered about -400 kHz, 200 kHz and
800 kHz, respectively. Note that the signal energy 336 has been
translated to the negative portion of the frequency axis.
[0140] A dashed line 342 of FIG. 14C represents the transfer curve
of the decimation filter 300 in one embodiment. In the embodiment
of FIG. 14C, the low pass decimation filter 300 passes all three
signal energies 336, 338 and 340. (For example, each of the signal
energies 336, 338 and 340 could be produced by a different
transmitting unit.) In this example, none of the signal energies
are centered about D.C. Thus, the effects of any DC offset in the
system and 1/f noise (denoted by the increase in the spectral noise
density curve 343 around zero frequency) can be reduced by
follow-on filtering, for example, matched filtering. The spectral
noise density curve 343 of FIG. 14C shows that the spectral noise
density level increases (primarily due to quantization noise of the
delta-sigma converter) as the frequency increases, expect for 1/f
increase near zero frequency. As a result, the noise level within
the bandwidth of signal energy 340 is greater than that for signal
energies 336 or 338. In one embodiment, the decimation filter 300
is implemented with low pass filtering and equivalent bandpass
filtering is implemented in the following matched filter.
[0141] In an alternative embodiment, the decimation filter 300 is
more frequency selective such that only one of the signal energies
(such as might be produced by a single transmitting unit) is passed
without substantial attenuation. For example, in FIG. 14D, the
dashed line 344 shows such a decimation filter 300 transfer
characteristic. As shown in FIG. 14D, in the alternative
embodiment, only the signal energy 338 is efficiently passed
through the decimation filter 300.
[0142] In yet another embodiment, the down-converted waveform is
centered about the D.C., i.e., the waveform has zero frequency
offset, as shown in FIG. 14E. Conversion to D.C. centered baseband
has the benefit of achieving higher resolution for a given clock
rate which can be a particular benefit for wide band signals where
the effects of quantization noise should be minimized. The effects
of 1/f noise are less pronounced in a wide band system and can be
filtered with a notch filter at zero frequency without
significantly degrading the performance. A dashed line 346 of FIG.
14E represents the transfer curve of the decimation filter 300 in
one such embodiment. More information concerning the design of
decimation filters can be found in Multi-Rate Digital Signal
Processing, Prentice-Hall, Inc., Englewood Cliffs, N.J., 1983 by R.
E. Crochiere and L. R. Rabiner.
[0143] It is advantageous for the receiver 106 (FIG. 1) to operate
in accordance with more than one communication protocol. For
example, the receiver can operate in a narrow band time division
multiple access (TDMA) system such as Global System for Mobile
Communication (GSM) or a wide band code division multiple access
(CDMA) system such as defined in the Telephone Industry
Association, Electronic Industry Association (TIA/EIA) interim
standard entitled "Mobile Station--Base Station Capability Standard
for Dual-Mode Wide band Spread Spectrum Cellular System,"
TIA/EIA/IS-95.
[0144] During TDMA operation, the decimation filter 300 can take on
a narrow band transfer characteristic, shown by dashed line 344.
Alternatively, during CDMA operation, the offset and bandwidth of
this filter, shown by line 342, may be increased according to well
known principles of digital filtering and signal reception.
Alternatively, a single wide band, low-pass decimation filter could
be utilized and the programmable bandwidth implemented in the
following matched filtering.
[0145] I&Q Gain Quadrature and Offset Correction
[0146] In FIG. 11, the output of the decimation filter 300 is the
input into the I & Q gain quadrature and offset correction
component 410. FIG. 15 is a block diagram showing one embodiment of
the component 410. In FIG. 15, the clock and data output of the
decimation filters 300 are coupled to a calibration circuit 350.
The calibration circuit 350 adjusts the relative gain and phase so
that the in-phase and quadrature signal paths are balanced with
respect to each other. In order to avoid introduction of distortion
into the signals, it is important that the relative gain and phase
of the in-phase and quadrature signal paths are the same.
[0147] One advantage of a digital signal processing architecture
shown in FIG. 15 is that the parameters can be controlled in the
digital circuit elements more easily than in analog circuit
elements. Typically, unbalances originate from the differences in
gain between the I and Q channels and errors in the relative
90.degree. phase shift between the I and Q channels. Additionally,
any differences in the DC offsets can be calibrated out. Additional
information concerning accomplishment of calibration can be found
in U.S. Pat. No. 5,422,889 entitled "OFFSET CORRECTION CIRCUIT,"
and in U.S. Pat. No. 5,604,929 entitled "SYSTEM FOR CORRECTING
QUADRATURE GAIN IN-PHASE ERROR IN A DIRECT CONVERSION SINGLE-SIDE
BAND RECEIVER INDEPENDENT OF THE CHARACTERISTICS OF THE MODULATED
SIGNAL."
[0148] In FIG. 15, the output of the calibration circuit 350 is
coupled to the input of a sampling rate converter 352. The sampling
rate converter 352 converts and synchronizes the data rate of the
signal to the rate of an external clock, CLK.sub.waveform. In one
embodiment, this function is accomplished with a linear or higher
order interpolation method such as the one described in "Advanced
Digital Signal Processing" by J. G. Proakis, et al., and McMillan
Publishing Co.
[0149] In FIG. 15, the output of the sampling rate converter 352 is
coupled to the input of a frequency translator 354. In one
embodiment, the frequency translator 354 is used to translate the
center frequency of the signal of interest to a D.C. centered
baseband. The frequency translator 354 multiplies the signal at the
output of the sampling rate converter 352 with a digital
representation of a sinusoidal signal having a frequency equal to
the center of the frequency of signal of interest. The advantage of
frequency translation is that it allows the matched filter 356 for
the signal to be implemented as a low pass filter and provides the
baseband I and Q inputs required for the digital demodulator input.
For the situation shown in FIG. 14E where there is only one signal
of interest and it has zero offset, the frequency translator 354 is
preferably not used.
[0150] When the frequency translator 354 is utilized, and the
output of the matched or channel filter 250 is utilized as an input
to the cross correlation measurement module 240 as shown in FIG.
5A, then the input Z(n) to the cross correlation measurement module
240 must also undergo the same frequency translation. The
translation can be accomplished by including a frequency translator
in the cross correlation measurement module 240.
[0151] In FIG. 15, the output of the frequency translator 354 is
coupled to a low pass filter 356 which can operate as a signal
matched filter. The low pass filter 356 is also used to reject
interference outside the bandwidth of interest. The output of the
low pass filter 356 provides a digital I and Q signal input to the
digital demodulator that is synchronized with the digital
demodulator clock--CLK.sub.WAVEFORM.
Clock Generator
[0152] FIG. 16 is a block diagram showing one embodiment of the
clock generator 500 of FIG. 7. In FIG. 16, a frequency synthesizer
360 produces an analog waveform at twice the rate of the conversion
clock, CLK. The output of the frequency synthesizer 360 is coupled
to the input of a limiting amplifier 362. In this embodiment, the
positive going zero crossing of the signal output by the frequency
synthesizer 360 is compared to a threshold by the limiting
amplifier 362. When the threshold is chosen appropriately, the
limiting amplifier 362 produces a waveform with digital logic
values at the same frequency as that of the output from the
frequency synthesizer 360 and having a 50% duty cycle (i.e., the
duration of the logic "1" pulse is the same as the duration of the
logic "0" pulse).
[0153] In FIG. 16, the limiting amplifier 362 drives a master slave
flip-flop 376 comprising a master latch 364 and a slave latch 368.
The master-slave flip-flop 376 is configured in a divide-by-two
configuration. In this configuration, a Q output 366 and a
{overscore (Q)} output 372 of the flip-flop 364 are connected to
the D and {overscore (D)} inputs of the flip-flop 368,
respectively, and a Q output 370 and a {overscore (Q)} output 374
of the flip-flop 368 are connected to the {overscore (D)} and D
inputs of the flip-flop 364, respectively. When the master-slave
flip-flop is connected in this manner, the four latch outputs 366,
372, 370 and 374, have clock phases of 0.degree., 90.degree.,
380.degree., and 270.degree., with respect to one another. Two of
these outputs (for example, output 366 and output 370) can be used
as I_CLK and Q_CLK, respectively. Although the implementation of
FIG. 16 is included explicitly herein for illustration purposes, a
variety of other means (such as a ring oscillator) can be used to
generate a clock signal in accordance with the present
invention.
Another Embodiment of the Modulator
[0154] FIG. 17 illustrates another embodiment of a translating
delta-sigma modulator 380 which employs double-sampling (i.e.,
samples on both edges of a clock signal). The delta-sigma modulator
380 of FIG. 17 operates under some of the same principles as the
single sampled architecture shown in FIG. 13 while doubling the
sample rate, thereby relaxing the speed requirements for circuitry
by a factor of two. The delta-sigma modulator 380 can be used as
within the architecture shown in FIG. 7 as the transmitting
delta-sigma modulators 211 and 213.
[0155] In FIG. 17, complementary amplifier 382 receives the
digitally modulated RF signal centered about the carrier frequency.
At a non-inverting output, the complementary amplifier 382 produces
a voltage that is G times the voltage at the input to the
complementary amplifier 382. At an inverting output, the
complementary amplifier 382 produces a voltage that is -G times the
voltage at the input to the complementary amplifier 382. The
inverting and non-inverting outputs of the complementary amplifier
382 are coupled to two input ports of a switch 384. The control
port of the switch 384 determines which input port is coupled to
the output port and is driven by the conversion clock, CLK, such
that the output port of the switch 384 is alternately coupled to
the inverting and non-inverting outputs of the complementary
amplifier 382.
[0156] Together, the complementary amplifier 382 and the switch 384
perform the functions of a commutator as explained more fully above
with reference to FIG. 13. The output of the switch 384 is coupled
to the input of the core double-sampling delta-sigma modulator. The
core double-sampling delta-sigma modulator is comprised of a
combiner 388, a loop amplifier 390, a loop filter 392, an
even-phase edge-triggered comparator 394A, an odd-phase
edge-triggered comparator 394B, an even-phase digital-to-analog
converter 396A and an odd-phase digital-to-analog converter
396B.
[0157] The output of the switch 384 is coupled to the non-inverting
input of the loop amplifier 390. The output of the loop amplifier
390 is the difference between the voltage coupled to its
non-inverting input port and its inverting input port times a
voltage gain A.sub.v where the voltage gain is typically a large
positive constant. The output of the loop amplifier 390 is coupled
to the input of the loop filter 392. In a preferred embodiment, the
loop filter 392 is an analog low pass filter but can be embodied in
other forms. In one embodiment, the loop amplifier 390 and loop
filter 392 act as an integrator.
[0158] The output of the loop filter 392 is coupled to the input of
the even-phase edge-triggered comparator 394A and also to the input
of the odd-phase edge-triggered comparator 394B. The clock inputs
of the even-phase edge-triggered comparator 394A and the odd-phase
edge-triggered comparator 394B are coupled to the conversion clock,
CLK. The even-phase edge-triggered comparator 394A and odd-phase
edge-triggered comparator 394B are clocked using opposite edges of
the comparison clock, CLK. For example, in one embodiment, the
even-phase edge-triggered comparator 394A performs a comparison on
the rising edge of the comparison clock, CLK, and the odd-phase
edge-triggered comparator 394B performs a comparison on the falling
edge of the comparison clock, CLK.
[0159] The logic values output by the even-phase edge-triggered
comparator 394A and the odd-phase edge triggered comparator 394B
are coupled to the input of the digital-to-analog converter 396A
and the digital-to-analog converter 396B, respectively. The outputs
of the digital-to-analog converter 396A and digital-to-analog
converter 396B are combined through the combiner 388 and drive the
inverting input of the loop amplifier 390. In one embodiment, the
combiner 388 simply adds the two values together. In another
embodiment, the combiner 388 time-division multiplexes the values
into the loop. One useful attribute of the first embodiment of the
combiner 388 is that linearity can be achieved without tight
matching between the tight-to-analog converter 396A and the
digital-to-analog converter 396B since their respective outputs are
effectively averaged before being presented to the loop
amplifier.
[0160] In one embodiment, the outputs of edge-triggered comparator
394A and edge-triggered comparator 394B are also coupled to the
decimation filter 300 in a similar manner to the single-sampled
case. In such an embodiment, typically the architecture of the
decimation filter 300 is appropriately modified to accommodate
processing the samples in the form of two bit serial words instead
of a single high speed serial bit stream.
[0161] Due to the continuous time nature of the direct converter
embodiments shown in FIGS. 7, 13 and 17, the direct converter
embodiments are not limited in dynamic range in contrast to
conventional multistage down converters. Thus, there is no need to
incorporate automatic gain control into the front end of the
receiver. For example, in FIG. 5A, the amplitude of the incoming
waveform applied to the receiver direct converter 220 is in fixed
proportion of an amplitude of a signal received by the receive
antenna radiator 102 because no automatic gain control mechanism is
included. As noted above, automatic gain control used to extend the
dynamic range of a receiver may be undesirable for spectrally
crowded applications such as cellular communications because the
automatic gain control may make the receiver sensitivity dependent
upon signals and interference that are outside the signal channel.
For example, it is possible for a strong signal in an adjacent
channel to capture the receiver front end and desensitize the
receiver such that a weak signal in the channel of interest is
undetectable. This is particularly harmful in a base station
receiver where the receiver receives incoming signals from multiple
remote units. Furthermore, the use of automatic gain control will
likely require the digital coherent canceller to track the gain
changes which will introduce errors and noise.
[0162] The digital decimation filtering of the receiver direct
converter 220 (FIGS. 5A, 7 and 8), the reference direct converter
238 (FIG. 5A), and the matched or channel filter 250 (FIG. 5A) with
low side lobes may significantly suppress the transmit signal
outside of the desired signal bandwidth. In a preferred
implementation, these digital decimation filters 220, 238, 250 have
approximately 90 dB attenuation. This attenuation advantageously
reduces the amount of filtering provided by the duplexer receiver
filter 214.
[0163] In one embodiment, the reference direct converter 238 and/or
the receiver direct converter 220 have a sampling rate
approximately equal to that of the carrier frequency of interest.
This may significantly reduce the requirements on the duplexer
receiver filter 214 and the reference bandpass filter 236 to
provide attenuation at the aliasing frequency (i.e., at half the RF
frequency).
[0164] The invention may be embodied in other specific forms
without departing from its spirit or essential characteristics. The
described embodiment is to be considered in all respects only as
illustrative and not restrictive and the scope of the invention is,
therefore, indcated by the appended claims rather than the
foregoing description. All changes which come within the meaning
and range of equivalency of the claims are to be embraced within
their scope.
* * * * *