U.S. patent application number 14/620177 was filed with the patent office on 2015-08-27 for equalization method and equalizer.
The applicant listed for this patent is Panasonic Corporation. Invention is credited to HIROYUKI MOTOZUKA, NAGANORI SHIRAKATA, YOSHINORI SHIRAKAWA, KOICHIRO TANAKA.
Application Number | 20150244546 14/620177 |
Document ID | / |
Family ID | 53786150 |
Filed Date | 2015-08-27 |
United States Patent
Application |
20150244546 |
Kind Code |
A1 |
SHIRAKAWA; YOSHINORI ; et
al. |
August 27, 2015 |
EQUALIZATION METHOD AND EQUALIZER
Abstract
An equalization method includes carrying out frequency domain
conversion of M received signals into a 2M received vector having
2M elements, carrying out channel estimation and noise/interference
estimation based on the 2M vector, calculating a 2M channel vector
and a (2M).times.(2M) noise/interference matrix, selecting a 2M-1
or less channel vector from the calculated 2M channel vector,
selecting a (2M-1).times.(2M-1) or less noise/interference matrix
from the calculated (2M).times.(2M) noise/interference matrix,
calculating a 2M-1 or less equalization coefficient vector as
equalization coefficients based on the selected 2M-1 channel vector
and the selected (2M-1).times.(2M-1) noise/interference matrix,
selecting a 2M-1 or less received vector from the 2M received
vector, and equalizing the selected 2M-1 received vector by using
the calculated equalization coefficients.
Inventors: |
SHIRAKAWA; YOSHINORI;
(Kanagawa, JP) ; SHIRAKATA; NAGANORI; (Kanagawa,
JP) ; TANAKA; KOICHIRO; (Hyogo, JP) ;
MOTOZUKA; HIROYUKI; (Kanagawa, JP) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Panasonic Corporation |
Osaka |
|
JP |
|
|
Family ID: |
53786150 |
Appl. No.: |
14/620177 |
Filed: |
February 11, 2015 |
Current U.S.
Class: |
375/229 |
Current CPC
Class: |
H04L 25/03968 20130101;
H04L 27/265 20130101; H04L 25/03159 20130101; H04B 1/123 20130101;
H04L 25/022 20130101; H04L 25/0204 20130101; H04L 25/0224
20130101 |
International
Class: |
H04L 25/03 20060101
H04L025/03; H04B 1/12 20060101 H04B001/12; H04L 27/26 20060101
H04L027/26; H04L 25/02 20060101 H04L025/02 |
Foreign Application Data
Date |
Code |
Application Number |
Feb 21, 2014 |
JP |
2014-031965 |
Claims
1. An equalization method comprising: in a receiver with multiple
antennas, carrying out frequency domain conversion of M received
signals received by multiple antennas into a 2M received vector
having 2M elements; calculating a 2M channel vector and a
(2M).times.(2M) noise/interference matrix by carrying out channel
estimation and noise/interference estimation based on the 2M
received vector; selecting a 2M-1 or less channel vector from the
calculated 2M channel vector and selecting a (2M-1).times.(2M-1) or
less noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals; calculating a 2M-1 or less equalization coefficient vector
as equalization coefficients based on the 2M-1 or less channel
vector and the (2M-1).times.(2M-1) or less noise/interference
matrix; selecting a 2M-1 or less received vector from the 2M
received vector; and equalizing the 2M-1 or less received vector by
using the equalization coefficients.
2. The equalization method according to claim 1, wherein the
calculating of the equalization coefficients is carried out based
on MMSE by using one of matrix operation processing and scalar
operation processing, which are switched based on quality of the
received signals.
3. The equalization method according to claim 1, wherein the
selecting of the 2M-1 or less received vector, the 2M-1 or less
channel vector, or the (2M-1).times.(2M-1) or less
noise/interference matrix is carried out by using any one of the
following methods: (1) selection for each frequency bin based on
SINRs of respective frequencies of the respective received signals;
(2) selection for each frequency bin based on SINRs and received
signal powers of respective frequencies of the respective received
signals; (3) based on a mean value of SINRs in a predefined range
of frequency bins of the respective received signals, selection for
each of the predefined ranges of frequency bins; (4) based on a
mean values of SINRs and a mean value of received signal powers in
a predefined range of frequency bins of the respective received
signals, selection for each of the predefined ranges of frequency
bins; and (5) switching between selection for each frequency bin
based on SINRs of respective frequencies of the respective received
signals and, based on a mean value of SINRs in a predefined range
of frequency bins of the respective received signals, selection for
each of the predefined ranges of frequency bins.
4. The equalization method according to claim 1, the method further
comprising: replacing a portion of the 2M-1 or less channel vector
and a portion of the (2M-1).times.(2M-1) or less noise/interference
matrix with zeros; calculating the 2M-1 or less equalization
coefficient vector as the equalization coefficients based on both
the 2M-1 or less channel vector, the portion of which is replaced
with zeros, and the (2M-1).times.(2M-1) or less noise/interference
matrix, the portion of which is replaced with zeros; replacing a
portion of the 2M-1 or less received vector with zeros; and
equalizing the 2M-1 or less received vector, the portion of which
is replaced with zeros, by using the equalization coefficients.
5. The equalization method according to claim 1, the method further
comprising: selecting a 2M-2 or less channel vector from the
calculated 2M channel vector and selecting a (2M-2).times.(2M-2) or
less noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals; calculating a 2M-2 or less equalization coefficient vector
as second equalization coefficients based on the 2M-2 or less
channel vector and the (2M-2).times.(2M-2) or less
noise/interference matrix; selecting a 2M-2 or less received vector
from the 2M received vector; equalizing the 2M-2 or less received
vector by using the second equalization coefficients; and switching
between a first equalization method in which the 2M-1 or less
received vector is equalized by using the equalization coefficients
and a second equalization method in which the 2M-2 or less received
vector is equalized by using the second equalization coefficients,
based on a predefined criterion.
6. An equalizer comprising: a frequency domain converter which
carries out frequency domain conversion of M received signals
received by multiple antennas into a 2M received vector having 2M
elements; a channel and noise/interference estimator which carries
out channel estimation and noise/interference estimation based on
the 2M received vector to calculate a 2M channel vector and a
(2M).times.(2M) noise/interference matrix; a first selector which
selects a 2M-1 or less channel vector from the calculated 2M
channel vector and selects a (2M-1).times.(2M-1) or less
noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals; an equalization coefficient calculator which calculates a
2M-1 or less equalization coefficient vector as equalization
coefficients based on the 2M-1 channel vector and the
(2M-1).times.(2M-1) or less noise/interference matrix; a second
selector which selects a 2M-1 or less received vector from the 2M
received vector; and a frequency domain equalizer which equalizes
the 2M-1 or less received vector by using the equalization
coefficients.
7. The equalizer according to claim 6, wherein the equalization
coefficient calculator calculates the equalization coefficient
based on MMSE by using one of matrix operation processing and
scalar operation processing, which are switched based on quality of
the received signals.
8. The equalizer according to claim 6, wherein the 2M-1 or less
received vector, the 2M-1 or less channel vector, or the
(2M-1).times.(2M-1) or less noise/interference matrix by using any
one of the following methods: (1) selection for each frequency bin
based on SINRs of respective frequencies of the respective received
signals; (2) selection for each frequency bin based on SINRs and
received signal powers of respective frequencies of the respective
received signals; (3) based on a mean value of SINRs in a
predefined range of frequency bins of the respective received
signals, selection for each of the predefined ranges of frequency
bins; (4) based on a mean values of SINRs and a mean value of
received signal powers in a predefined range of frequency bins of
the respective received signals, selection for each of the
predefined ranges of frequency bins; and (5) switching between
selection for each frequency bin based on SINRs of respective
frequencies of the respective received signals and, based on a mean
value of SINRs in a predefined range of frequency bins of the
respective received signals, selection for each of the predefined
ranges of frequency bins.
9. The equalizer according to claim 6, further comprising: a first
zero replacer which replaces a portion of the 2M-1 or less channel
vector and a portion of the (2M-1).times.(2M-1) or less
noise/interference matrix with zeros; and a second zero replacer
which replaces a portion of the 2M-1 or less received vector with
zeros, wherein the equalization coefficient calculator, based on
both the 2M-1 or less channel vector, the portion of which is
replaced with zeros, and the (2M-1).times.(2M-1) or less
noise/interference matrix, the portion of which is replaced with
zeros, calculates the 2M-1 or less equalization coefficient vector
as the equalization coefficients, and the frequency domain
equalizer equalizes the 2M-1 or less received vector, the portion
of which is replaced with zeros, by using the equalization
coefficients.
10. The equalizer according to claim 6, further comprising: a third
selector which selects a 2M-2 or less channel vector from the
calculated 2M channel vector and selects (2M-2).times.(2M-2) or
less noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals; a second equalization coefficient calculator which
calculates a 2M-2 or less equalization coefficient vector as second
equalization coefficients based on the 2M-2 or less channel vector
and the (2M-2).times.(2M-2) or less noise/interference matrix; a
fourth selector which selects a 2M-2 or less received vector from
the 2M received vector; a second frequency domain equalizer which
equalizes the 2M-2 or less received vector by using the second
equalization coefficients; and a controller which switches between
equalization by the frequency domain equalizer and equalization by
the second frequency domain equalizer.
Description
CROSS REFERENCES TO RELATED APPLICATIONS
[0001] This application claims priority to Japanese Patent
Application No. 2014-031965, filed on Feb. 21, 2014, the contents
of which are hereby incorporated by reference.
BACKGROUND
[0002] 1. Technical Field
[0003] The present disclosure relates to an equalization method and
an equalizer applicable to a radio communication apparatus.
[0004] 2. Description of the Related Art
[0005] In a radio communication apparatus, a transmitter carries
out coding of a transmission signal, interleaving, modulation,
scrambling, and generation of a symbol signal. Modulation speed in
the modulation is referred to as a symbol rate. The transmitter
filters the symbol signal to generate a sampled signal, inputs the
sampled signal to a mixer with a carrier signal, generates a radio
frequency (RF) signal, and transmits the RF signal to a
receiver.
[0006] To the transmitted RF signal, channel characteristics of a
transmission channel and interference are added by the time when
the RF signal reaches the receiver, and noise is further added in
the receiver.
[0007] The receiver carries out mixing of the received RF signal,
and then carries out time sampling of a complex signal by an
analog-to-digital converter (ADC). The receiver filters the sampled
signal, and then converts the filtered signal to a frequency domain
signal by a discrete Fourier transformation (DFT) or a fast Fourier
transformation (FFT). The receiver equalizes the frequency domain
signal, and then converts the equalized signal into a time domain
signal by an inverse discrete Fourier transformation (IDFT) or an
inverse fast Fourier transformation (IFFT) to obtain a symbol
signal. The receiver carries out descrambling of the symbol signal,
demodulation, de-interleaving, decoding, and regeneration of the
transmitted signal. In the series of processing of the receiver,
the equalization eliminates channel characteristics and
interference between the transmitter and the receiver.
[0008] Examples of equalization according to the related art are
described in, for example, Japanese Unexamined Patent Application
Publication No. 2006-245810 and Japanese Patent No. 5166246.
[0009] Japanese Unexamined Patent Application Publication No.
2006-245810 describes a technique in which, in a receiver with a
single antenna system, when a sampling frequency of an ADC in the
receiver is higher than twice the symbol rate (for the purpose of
simplification, it is assumed that the sampling frequency is twice
the symbol rate), equalization is carried out by combining specific
frequencies to calculate equalization coefficients.
[0010] Japanese Patent No. 5166246 describes a technique in which,
in a receiver with two antenna systems (for the purpose of
simplification, the receiver is assumed to have two antenna
systems), when the sampling frequency of an ADC in the receiver is
twice the symbol rate, equalization is carried out by combining
specific frequencies through an inverse matrix operation for a
4.times.4 matrix, which has complex number elements, to calculate
equalization coefficients.
SUMMARY
[0011] Japanese Patent No. 5166246 described above assumes a radio
communication system using the code division multiple access
(CDMA), the time division multiple access (TDMA), the frequency
division multiple access (FDMA), the orthogonal frequency division
multiple access (OFDMA), or the single-carrier FDMA (SC-FDMA), and
with a carrier frequency of 1 to 5 GHz and a symbol rate in the
order of Mbps.
[0012] In a radio communication system with a symbol rate in the
order of Mbps, a receiver is desired to have an ADC which carries
out time sampling with a high precision and to calculate
equalization coefficients with high precision. Thus, the receiver
is also desired to have a processor which calculates a 4.times.4
inverse matrix, the calculation of which accompanies a large amount
of operations. However, because the system operates with a symbol
rate in the order of Mbps, a sufficient processing speed of the
processor has been achieved.
[0013] In the above-described radio communication system with a
symbol rate in the order of Mbps, the receiver is assumed to be a
mobile phone or a radio base station. Thus, although there is
interference from other channels, because a distance between an
interfering station and the receiver is longer than a distance
between a desired station and the receiver, influence of
interference from the interfering station has not been
significant.
[0014] However, in the radio communication standard WiGig.RTM. or
IEEE 802.11ad, which uses a carrier frequency of 60 GHz, because a
radio communication system uses a symbol rate in the order of Gbps,
a processing speed of a processor is not sufficient. Thus, it is
desirable to carry out an inverse matrix operation of a 4.times.4
matrix by an operational circuit. However, because of a large
amount of operations, implementation in circuitry is difficult.
Even if the implementation in circuitry is achieved, another
problem in that the circuitry consumes a lot of power is
caused.
[0015] In IEEE 802.11ad, short distance radio communication is
assumed. Thus, there may be a high possibility that a distance
between an interfering station and a receiver is shorter than a
distance between a desired station and the receiver, which causes a
problem of significant influence from the interfering station.
[0016] Thus, one non-limiting and exemplary embodiment provides an
equalization method and an equalizer which reduce an amount of
operations for performing equalization in a receiver with multiple
receiving systems.
[0017] Additional benefits and advantages of the disclosed
embodiments will be apparent from the specification and Figures.
The benefits and/or advantages may be individually provided by the
various embodiments and features of the specification and drawings
disclosure, and need not all be provided in order to obtain one or
more of the same.
[0018] In one general aspect, the techniques disclosed here feature
an equalization method, the method including, in a receiver with
multiple antennas, carrying out frequency domain conversion of M
received signals by multiple antennas into a 2M received vector
including 2M elements, carrying out channel estimation and
noise/interference estimation based on the 2M received vector,
calculating a 2M channel vector and a (2M).times.(2M)
noise/interference matrix, selecting a 2M-1 or less channel vector
from the calculated 2M channel vector and selecting a
(2M-1).times.(2M-1) or less noise/interference matrix from the
calculated (2M).times.(2M) noise/interference matrix, based on
quality of the received signals, calculating a 2M-1 or less
equalization coefficient vector as equalization coefficients based
on the 2M-1 or less channel vector and the (2M-1).times.(2M-1) or
less noise/interference matrix, selecting a 2M-1 or less received
vector from the 2M received vector, and equalizing the 2M-1 or less
received vector by using the equalization coefficients. In the
equalization method, the 2M received vector and the 2M channel
vector are vectors having 2M elements each of which is a complex
number. The (2M).times.(2M) noise/interference matrix is a matrix
having (2M).times.(2M) elements each of which is a complex number.
The 2M-1 or less equalization coefficient vector is a vector having
2M-1 or less elements each of which is a complex number.
[0019] In another general aspect, the techniques disclosed here
feature an equalizer including a frequency domain converter which
carries out frequency domain conversion of M systems of received
signals received by multiple antennas into a 2M received vector
having 2M elements, a channel and noise/interference estimator
which carries out channel estimation and noise/interference
estimation based on the 2M received vector to calculate a 2M
channel vector and a (2M).times.(2M) noise/interference matrix, a
first selector which selects a 2M-1 or less channel vector from the
calculated 2M channel vector and selects a (2M-1).times.(2M-1) or
less noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals, an equalization coefficient calculator which calculates a
2M-1 or less equalization coefficient vector as equalization
coefficients based on the 2M-1 or less channel vector and the
(2M-1).times.(2M-1) or less noise/interference matrix, a second
selector which selects a 2M-1 or less received vector from the 2M
received vector, and a frequency domain equalizer which equalizes
the 2M-1 or less received vector by using the equalization
coefficients.
[0020] These general and specific aspects may be implemented using
a system, a method, and a computer program, and any combination of
systems, methods, and computer programs.
[0021] According to the present disclosure, it is possible to
reduce an amount of operations for performing equalization in a
receiver with multiple receiving systems.
BRIEF DESCRIPTION OF THE DRAWINGS
[0022] FIG. 1 is a diagram illustrating a configuration of an
equalization coefficient calculator and equalizer usable in a
receiver of a first embodiment of the present disclosure;
[0023] FIG. 2 is a block diagram illustrating a configuration of a
receiver which has two antenna systems;
[0024] FIG. 3 is a diagram illustrating a frequency response of
received signals on multiple receiving antenna systems in a case in
which two antenna systems are used and two-times oversampling is
employed;
[0025] FIG. 4 is a diagram illustrating frequency response revised
from the frequency response illustrated in FIG. 3 by dividing
frequency bins into the lower side k.sub.L and the upper side
k.sub.U and combining portions of frequency response corresponding
to the divided frequency bins;
[0026] FIG. 5 is a diagram illustrating a configuration example of
an equalization coefficient calculator and equalizer in the
receiver illustrated in FIG. 2;
[0027] FIG. 6 is a diagram illustrating frequency response when
adjacent channel interference is produced on one of the reception
systems;
[0028] FIG. 7 is a diagram illustrating frequency response when
received signal power is low on one of the receiving antenna;
[0029] FIG. 8 is a diagram illustrating frequency response when
adjacent channel interference is produced on both receiving
systems;
[0030] FIG. 9 is a diagram illustrating frequency response when a
null point is produced on a received signal on one of the reception
systems;
[0031] FIG. 10 is a diagram illustrating frequency response when
received signal power on both reception systems is so high that
co-channel interference is produced;
[0032] FIG. 11 is a diagram illustrating frequency response when
received signal power on both reception systems is high, an
interference level is low, and no interference is produced;
[0033] FIG. 12 is a diagram illustrating a configuration of an
equalization coefficient calculator and equalizer usable in a
receiver of a second embodiment of the present disclosure;
[0034] FIG. 13 is a diagram illustrating a configuration of an
equalization coefficient calculator and equalizer usable in a
receiver of a third embodiment of the present disclosure; and
[0035] FIG. 14 is a diagram illustrating a configuration of an
equalization coefficient calculator usable in a receiver of a
fourth embodiment of the present disclosure.
DETAILED DESCRIPTION
Underlying Knowledge Forming Basis of the Present Disclosure
[0036] Before embodiments of an equalization method and an
equalizer according to the present disclosure are described, an
illustrative example of equalization on a receiver having multiple
antenna systems will be described first.
[0037] In the following description, a complex number that is an
element in the time domain is denoted by a lowercase letter with an
index n in a sample period, for example, h(n). A complex number
that is an element in the frequency domain is denoted by an
uppercase letter with an index k of a frequency bin, for example,
H(k). A vector is assumed to contain a plurality of elements, each
of which is an afore-mentioned complex number, and is denoted by a
lowercase letter with an underscore, for example, h. A matrix is
denoted by an uppercase letter with an underscore, for example,
H.
[0038] FIG. 2 is a block diagram illustrating a configuration of a
receiver which includes R (R=2) antenna systems. The receiver
includes antennas 210a and 210b, RF processors 211a and 211b, ADCs
212a and 212b, DFTs 213a and 213b, a channel and noise/interference
estimator 220, an equalization coefficient calculator and equalizer
230, an IDFT 240, a demodulator 241, and a decoder 242. In FIG. 2,
solid arrows denote complex numbers, each of which is an element in
a vector or a matrix, and dashed arrows denote matrices.
[0039] The antennas 210a and 210b receive RF signals from a
transmitter. For the purpose of simplicity, transmission power is
normalized to a mean value of 1.0. The RF signals, to which, while
being transmitted from the transmitter to the receiver, channel
characteristics between the transmitter and the receiver are added,
and interference waves from a source other than the desired
transmitter are also added, are received by the antennas 210a and
210b.
[0040] The RF processors 211a and 211b convert the received the RF
signals to baseband signals, which are complex signals. In the
conversion, noise such as a thermal noise from a circuit in the
receiver is added to the baseband signals.
[0041] In the ADCs 212a and 212b, the baseband signals, which are
complex signals, are sampled at C (C.gtoreq.1) times of a symbol
rate and converted to digital complex baseband signals r(n), which
is expressed by the formula (1). In the following description, for
the purpose of illustration, an oversampling multiple C is assumed
to be 2.
r(n)=[r.sub.a(n)r.sub.b(n)] (1)
where .sup.T denotes a transpose.
[0042] In the above formula, the complex baseband signal r.sub.a(n)
on the system of antenna a is expressed by the formula (2).
r.sub.a(n)=FIR(h.sub.a(n)t(n))+i.sub.a(n)+no.sub.a(n) (2)
where FIR( ) denotes a finite impulse response (FIR) filter, t(n)
denotes a sampled transmission signal, h.sub.a(n) denotes a channel
of the system of antenna a, i.sub.a(n) denotes an interference, and
no.sub.a(n) denotes a noise.
[0043] The channel, denoted by h.sub.a(n) in the formula (2), is a
frequency response to which channel characteristics between the
transmitter and the receiver and transmit and receive filter
characteristics are added.
[0044] The DFTs 213a and 213b work as a frequency domain converter,
which carries out frequency domain conversion of the digital
complex baseband signal r(n) to acquire a received vector r(k) in
the frequency domain, which is expressed by the formula (3).
r(k)=[R.sub.a(k)R.sub.b(k)].sup.T
R.sub.a(k)=H.sub.a(k)T(k)+I.sub.a(k)+NO.sub.a(k)
0.ltoreq.k.ltoreq.2K-1 (3)
[0045] In the formula (3), k denotes a frequency bin. K denotes the
number of points of DFT when the symbol rate C is 1.
[0046] The channel and noise/interference estimator 220, from the
received vector r(k) in the frequency domain, estimates a channel
h(k) as a channel vector and an undesired signal matrix U(k) as a
noise/interference matrix which represents noise and interference
collectively.
[0047] In FIG. 2, R.sub.a(k.sub.L), R.sub.a(k.sub.U),
R.sub.b(k.sub.L), R.sub.b(k.sub.U), H.sub.a(k.sub.L),
H.sub.a(k.sub.U), H.sub.b(k.sub.L), and H.sub.b(k.sub.U) denote
complex signals in individual frequency bins, respectively.
U(k.sub.LU) denotes a vector which has signal components in
individual frequency bins. The denotations apply to other
drawings.
[0048] Although various computation methods may be used for the
channel and noise/interference estimator 220, the channel h(k) is
computed by extracting reference signals for a certain period from
among the received vectors r(k) and multiplying the extracted
reference signals with an original reference signals. The undesired
signal matrix U(k) is computed as an expected value of a complex
conjugate covariance as expressed by the formulae (4) and (5).
u _ ( k ) = [ I a ( k ) + NO a ( k ) I b ( k ) + NO b ( k ) ] T = [
U a ( k ) U b ( k ) ] T ( 4 ) U _ ( k ) = E u _ ( k ) u _ H ( k ) (
5 ) ##EQU00001##
where E[ ] denotes an expectation operation, and .sup.H denotes a
complex conjugate transpose.
[0049] FIG. 3 is a diagram illustrating a frequency response of
received signals on a plurality of receiving antenna systems in a
case of R=2 and C=2, i.e. a case in which two antenna systems are
used and two-times oversampling is employed. In this case, filters
on the transmitter are assumed to be roll-off filters. In FIG. 3, a
frequency bin K corresponds to a symbol rate frequency
f.sub.sym.
[0050] FIG. 4 illustrates a case in which frequency bins in the
frequency response illustrated in FIG. 3 are partitioned into the
lower side k.sub.L and the upper side k.sub.U. In FIGS. 3 and 4,
the received vector r(k) in the frequency domain is expressed by
the formula (6).
r(k)=r(k.sub.LU)=[R.sub.a(k.sub.L)R.sub.a(k.sub.U)R.sub.b(k.sub.L)R.sub.-
b(k.sub.U)].sup.T
0.ltoreq.k.sub.LU.ltoreq.K-1
0.ltoreq.k.sub.L.ltoreq.K-1
K.ltoreq.k.sub.U.ltoreq.2K-1
k.sub.U=k.sub.L+K (6)
[0051] The channel h(k) and the undesired signal matrix U(k), which
includes noise and interference, are computed by the channel and
noise/interference estimator 220 by the formulae (7) and (8).
h _ ( k ) = h _ ( k LU ) = [ H a ( k L ) H a ( k U ) H b ( k L ) H
b ( k U ) ] T ( 7 ) U _ ( k ) = U _ ( k LU ) = E [ u a ( k L ) u a
* ( k L ) u a ( k L ) u a * ( k U ) u a ( k L ) u b * ( k L ) u a (
k L ) u b * ( k U ) u a ( k U ) u a * ( k L ) u a ( k U ) u a * ( k
U ) u a ( k U ) u b * ( k L ) u a ( k U ) u b * ( k U ) u b ( k L )
u a * ( k L ) u b ( k L ) u a * ( k U ) u b ( k L ) u b * ( k L ) u
b ( k L ) u b * ( k U ) u b ( k U ) u a * ( k L ) u b ( k U ) u a *
( k U ) u b ( k U ) u b * ( k L ) u b ( k U ) u a * ( k U ) ] ( 8 )
##EQU00002##
where * denotes a complex conjugate.
[0052] In the formula (8), diagonal elements of U(k) indicate power
and non-diagonal elements indicate correlations between
antennas.
[0053] FIG. 5 is a diagram illustrating a configuration example of
the equalization coefficient calculator and equalizer in the
receiver illustrated in FIG. 2. The equalization coefficient
calculator and equalizer includes a 4-vector equalization
coefficient calculator 510, a 4-vector frequency domain equalizer
520, a controller 530, selectors 540 to 544, a 2-vector
equalization coefficient calculator 550, a 2-vector frequency
domain equalizer 560, and a selector 570. In FIG. 5, solid arrows
other than the signals output from the controller indicate complex
numbers, each of which is an element of a vector or a matrix, and
dashed arrows indicate matrices.
[0054] The 4-vector equalization coefficient calculator 510
calculates a 1.times.4 equalization coefficient vector w(k.sub.LU),
which is expressed by the formula (9), by the minimum mean square
error (MMSE) criterion.
w(k.sub.LU)=h.sup.H(k.sub.LU)[h(k.sub.LU)h.sup.H(k.sub.LU)+U(k.sub.LU)].-
sup.-1 (9)
where [ ].sup.-1 denotes an inverse matrix operation.
[0055] The equalization is in general accomplished by carrying out
vector multiplication of a frequency domain received vector
r(k.sub.LU) by an equalization coefficient vector w(k.sub.LU) and
is expressed by the formula (10). The 4-vector frequency domain
equalizer 520 calculates an equalized output T.sub.r(k.sub.LU) by
the formula (10).
T.sub.r(k.sub.LU)=w(k.sub.LU)r(k.sub.LU) (10)
where T.sub.r(k.sub.LU) is a frequency domain converted signal of a
transmitted symbol signal, estimated from the received signal.
[0056] As illustrated in FIG. 4, the frequency response of the
received signal includes, due to the roll-off filter of the
transmitter, frequency bins at which signals become substantially
zero and frequency bins at which signals form a slope. The range of
the slope is determined by a roll-off rate .alpha. by the formula
(11).
K - .alpha. K 2 .ltoreq. k .ltoreq. K + .alpha. K 2 - 1 3 2 K -
.alpha. K 2 .ltoreq. k .ltoreq. 3 2 K + .alpha. K 2 - 1 ( 11 )
##EQU00003##
[0057] Frequency bins K.sub.2 and K.sub.4, located in the range of
the slope, form a pair, and an equalized signal
T.sub.abLU(k.sub.LU) is calculated by the above-described formula
(10).
[0058] In the frequency response of the received signal illustrated
in FIG. 4, a frequency bin K.sub.3 at which the signal is
substantially zero exists. The frequency bin K.sub.3 and a
frequency bin K.sub.1 form a pair, and, because replacing the
signal at K.sub.3 with zero causes terms related to K.sub.3 in the
formulae (7) and (8) to be zero, the terms related to K.sub.3 do
not have to be calculated. Thus, frequency bins at which the
signals are substantially zero are excluded from calculation
objects by the selectors 540 to 544 in FIG. 5. The selectors 540 to
544 select signals by following commands of control signals from
the controller 530. The 2-vector equalization coefficient
calculator 550 calculates a 1.times.2 equalization coefficient
vector w.sub.ab(k.sub.LU). The 2-vector frequency domain equalizer
560 carries out equalization to calculate an equalized output
T.sub.ab(k.sub.LU).
[0059] The selector 570 selects T.sub.ab(k.sub.LU) and
T.sub.abLU(k.sub.LU) to obtain a frequency domain converted signal
T.sub.r(k.sub.LU) of the transmitted symbol signal estimated from
the received signals. The selector 570, by following commands in
control signals from the controller 530, selects either of the
output signals.
[0060] Although, in the example of the receiver illustrated in
FIGS. 2 and 5, the calculation is simplified by selection of
frequency bins as described above, it is necessary to calculate a
4.times.4 inverse matrix in a specific range in the formula
(9).
[0061] There is a problem in that circuit implementation of the
inverse matrix operation of a 4.times.4 matrix is difficult, and,
even if the circuit implementation is possible, an implemented
circuit consumes a significant amount of power.
[0062] Accordingly, in the following embodiments, examples of
equalization methods and equalizers by which an amount of
operations needed for equalization on a receiver having multiple
receiving systems is reduced and calculation of equalization
coefficients is simplified, the circuit implementation of which is
easy, and which consume a low power will be described.
First Embodiment
[0063] FIG. 1 illustrates a configuration of an equalization
coefficient calculator and equalizer usable for a receiver of a
first embodiment of the present disclosure, which is equivalent to
a configuration of the above-described equalization coefficient
calculator and equalizer of the receiver illustrated in FIG. 2.
[0064] The equalization coefficient calculator and equalizer in the
receiver includes a controller 110, selectors 120 and 130, a
3-vector equalization coefficient calculator 140, and a 3-vector
frequency domain equalizer 150. In FIG. 1, solid arrows other than
the signals output from the controller indicate complex numbers,
each of which is an element of a vector or a matrix, and dashed
arrows indicate matrices.
[0065] In FIG. 1, the controller 110, based on a channel
h(k.sub.LU) and an undesired signal matrix U(k.sub.LU), calculates
signal power values |H(k.sub.L)|.sup.2, |H.sub.a(k.sub.U)|.sup.2,
|H.sub.b(k.sub.L)|.sup.2, and |H.sub.b(k.sub.U)|.sup.2 of each
vector at each frequency bin, and selects diagonal elements in
U(k.sub.LU), which indicates power values of undesired signals.
Based on the signal power values and the diagonal elements, the
controller 110 calculates ratios of signal power values to power
values of undesired signals (signal to interference and noise ratio
(SINR)) SINR.sub.a(k.sub.L), SINR.sub.a(k.sub.U),
SINR.sub.b(k.sub.L), and SINR.sub.b(k.sub.U).
[0066] The selector 120 selects a predefined number of signals from
a frequency domain received vector r(k.sub.LU) based on quality of
the received signals by following commands of control signals from
the controller 110. As an example, the selector 120 selects three
signals from the frequency domain received vector r(k.sub.LU),
which is a 4-vector, in descending order of the calculated SINRs to
generate a 3-vector r.sub.sel(k.sub.LU).
[0067] The selector 130 selects a predefined number of signals from
the channel h(k.sub.LU) as a channel vector based on quality of the
received signals by following commands of control signals from the
controller 110. As an example, the selector 130 selects three
signals from the channel h(k.sub.LU), which is a 4-vector, in
descending order of the calculated SINRs to generate a 3-vector
h.sub.sel(k.sub.LU). The selector 130 selects a predefined number
of signals from the undesired signal matrix U(k.sub.LU) as a
noise/interference matrix based on quality of the received signals
by following commands of control signals from the controller 110.
As an example, the selector 130 selects 3.times.3 signals from the
4.times.4 undesired signal matrix U(k.sub.LU) to generate
U.sub.sel(k.sub.LU).
[0068] With the above-described processing, it becomes possible for
the 3-vector equalization coefficient calculator 140 and the
3-vector frequency domain equalizer 150 to carry out processing by
vector operations with 3 or less dimension.
[0069] In the above-described processing example, the 3-vector
equalization coefficient calculator 140 calculates an equalization
coefficient vector, which is a 3-vector, as equalization
coefficients based on the generated channel h.sub.sel(k.sub.LU) and
undesired signal matrix U.sub.sel(k.sub.LU). The 3-vector frequency
domain equalizer 150 carries out equalization of the received
vector r.sub.sel(k.sub.LU), which is a 3-vector, by using the
calculated equalization coefficient vector, which is a
3-vector.
[0070] FIGS. 6 to 11 are diagrams illustrating frequency responses
which indicate received signal power and noise/interference power
on a plurality of receiving antenna systems in various conditions.
FIG. 6 is a diagram illustrating a frequency response when adjacent
channel interference is produced on one of the reception systems.
FIG. 7 is a diagram illustrating a frequency response when received
signal power is low on one of the reception systems. FIG. 8 is a
diagram illustrating a frequency response when adjacent channel
interference is produced on both reception systems. FIG. 9 is a
diagram illustrating a frequency response when a null point is
produced on a received signal on one of the reception system. FIG.
10 is a diagram illustrating a frequency response when received
signal power is so high on both reception systems that co-channel
interference is produced. FIG. 11 is a diagram illustrating a
frequency response when received signal power on both reception
systems is high, an interfering noise level is low, and thus no
interference is produced.
[0071] When adjacent channel interference is produced on one of the
reception systems, as illustrated in FIG. 6, a signal around the
frequency of K/2 on the system of receiving antenna a, on which the
interference influences significantly, is not selected. In this
case, the selectors 120 and 130 select a signal around the
frequency of K/2 on the system of receiving antenna b, a signal
around the frequency of 3K/2-1 on the system of receiving antenna
a, and a signal around the frequency of 3K/2-1 on the system of
receiving antenna b.
[0072] Although an example in which the equalization coefficients
are calculated by selecting a 3-vector signal is described in the
first embodiment, a method in which a 2-vector is selected may also
be employed. A signal selection criterion by which, as an example,
signals are selected, based on quality of the received signals, in
descending order from a signal with as good quality as possible,
e.g. from a signal with a high SINR, may be employed. When
interference is produced, signals are selected from signals with
high SINRs and low interference levels.
[0073] When the received signal power on one of the reception
systems is low as illustrated in FIG. 7, a 2-vector on the other
system of receiving antenna a, on which the received signal power
is high, is selected. In this case, the selectors 120 and 130
select a signal around the frequency of K/2 on the system of
receiving antenna a and a signal around the frequency of 3K/2-1 on
the system of receiving antenna a.
[0074] When adjacent channel interference is produced on both
reception systems as illustrated in FIG. 8, no signal around the
frequency of K/2 on both the system of receiving antenna a and the
system of receiving antenna b, which are significantly influenced,
is selected. In this case, the selectors 120 and 130 select a
signal around the frequency of 3K/2-1 on the system of receiving
antenna a and a signal around the frequency of 3K/2-1 on the system
of receiving antenna b.
[0075] Although an example in which a 3-vector of equalization
coefficients is calculated is described in the first embodiment, by
combining SINRs and amounts of signal power
|H.sub.a(k.sub.L)|.sup.2, |H.sub.a(k.sub.U)|.sup.2,
|H.sub.b(k.sub.L)|.sup.2, and |H.sub.b(k.sub.U)|.sup.2 of each
vector at each frequency bin, a vector with a low power among
3-vectors may be replaced with zero.
[0076] When a null point is produced on the received signals on one
of the reception systems as illustrated in FIG. 9, use of signals
on the system of receiving antenna a on which the null point is
produced is avoided. For example, at a frequency bin at which a
null point is produced, signals on the other system of receiving
antenna b are selected.
[0077] As described above, in the first embodiment, signals are
selected by using SINRs at individual frequency bins, and, when a
null point is produced in a range defined by the formula (12) due
to multipath communication or the like, use of signals at the null
point is avoided.
0 .ltoreq. k .ltoreq. K - .alpha. K 2 3 2 K + .alpha. K 2 - 1
.ltoreq. k .ltoreq. 2 K - 1 ( 12 ) ##EQU00004##
[0078] When received signal power values on both reception systems
are high and interfering noise is so low that no interference is
detected as illustrated in FIG. 11, from among signals around the
frequency of K/2 on the system of receiving antenna a, around the
frequency of K/2 on the system of receiving antenna b, around the
frequency of 3K/2-1 on the system of receiving antenna a, and
around the frequency of 3K/2-1 on the system of receiving antenna
b, two or three signals are selected in descending order of
SINRs.
[0079] According to the above-described first embodiment, the
selectors 120 and 130 select three or two signals and an
equalization coefficients vector with three or less dimensions is
calculated. With this configuration, it is possible to reduce an
amount of operations of MMSE which is used by the 3-vector
equalization coefficient calculator 140 and an amount of operations
by the 3-vector frequency domain equalizer 150.
[0080] According to the first embodiment, because signals are
selected based on an SINR at each frequency bin, when a null point
is produced due to, for example, multipath communication, it
becomes possible to avoid selection of a signal at the null
point.
Second Embodiment
[0081] FIG. 12 is a diagram illustrating a configuration of an
equalization coefficient calculator and equalizer which is used for
a receiver in a second embodiment of the present disclosure. In the
second embodiment, another configuration of the equalization
coefficient calculator and equalizer of the receiver will be
described. In FIG. 12, components that are the same as those in
FIG. 1 are denoted by the same reference characters and description
thereof will be omitted.
[0082] The equalization coefficient calculator and equalizer in the
receiver includes a controller 910, selectors 120 and 130, zero
replacers 920 and 930, a 3-vector equalization coefficient
calculator 140, and a 3-vector frequency domain equalizer 150. In
FIG. 12, solid arrows other than the signals output from the
controller indicate complex numbers, each of which is an element of
a vector or a matrix, and dashed arrows indicate matrices.
[0083] In FIG. 12, the controller 910, based on the channel
h(k.sub.LU) and the undesired signal matrix U(k.sub.LU), calculates
signal power values |H.sub.a(k.sub.L)|.sup.2,
|H.sub.a(k.sub.U)|.sup.2, |H.sub.b(k.sub.L)|.sup.2, and
|H.sub.b(k.sub.U)|.sup.2 of individual vectors at individual
frequency bins, and selects diagonal elements of U(k.sub.LU), which
indicates power values of undesired signals. Based on the signal
power values and the diagonal elements, the controller 910
calculates ratios of signal power values to power values of
undesired signals SINR.sub.a(k.sub.L), SINR.sub.a(k.sub.U),
SINR.sub.b(k.sub.L), and SINR.sub.b(k.sub.U).
[0084] The selector 120, by following commands of control signals
from the controller 910, selects three signals from the frequency
domain received vector r(k.sub.LU), which is a 4-vector, in
descending order of the calculated SINRs and generates a 3-vector
r.sub.sel(k.sub.LU).
[0085] The selector 130, by following commands of control signals
from the controller 910, selects three signals from the channel
h(k.sub.LU), which is a 4-vector, in descending order of the
calculated SINRs and generates a 3-vector h.sub.sel(k.sub.LU). The
selector 130 also selects 3.times.3 signals from the 4.times.4
undesired signal matrix U(k.sub.LU) and generates
U.sub.sel(k.sub.LU).
[0086] The zero replacer 920, by following commands of control
signals from the controller 910, replaces a portion of the received
vector r.sub.sel(k.sub.LU) with zeros. The zero replacer 930, by
following commands of control signals from the controller 910,
replaces individual portions of the channel h.sub.sel(k.sub.LU) and
the undesired signal matrix U.sub.sel(k.sub.LU) with zeros. As an
example, the zero replacers 920 and 930 replace complex numbers of
elements corresponding to low signal power values
|H.sub.a(k.sub.L)|.sup.2, |H.sub.a(k.sub.U)|.sup.2,
|H.sub.b(k.sub.L)|.sup.2, and |H.sub.b(k.sub.U)|.sup.2 with zeros.
The zero replacers 920 and 930 change r.sub.sel(k.sub.LU) and
h.sub.sel(k.sub.LU), which are selected by the selectors 120 and
130, to r.sub.zero(k.sub.LU) and h.sub.zero(k.sub.LU),
respectively.
[0087] For example, r.sub.sel(k.sub.LU) is expressed by the formula
(13), and, when the signal power on the system of receiving antenna
b is low, r.sub.zero(k.sub.LU) is expressed by the formula
(14).
r.sub.sel(k.sub.LU)=[R.sub.a(k.sub.L)R.sub.b(k.sub.L)R.sub.b(k.sub.U)].s-
up.T (13)
r.sub.zero(k.sub.LU)=[R.sub.a(k.sub.L)0R.sub.b(k.sub.U)].sup.T
(14)
[0088] In a similar manner, terms of h.sub.sel(k.sub.LU) and
U.sub.sel(k.sub.LU) corresponding to H.sub.b(k.sub.L) are replaced
with zeros.
[0089] With this processing by the zero replacers 920 and 930, it
becomes possible for the 3-vector equalization coefficient
calculator 140 and the 3-vector frequency domain equalizer 150 to
carry out virtually 2-vector processing.
[0090] When the received signal power on both reception systems is
so high that co-channel interference takes place as illustrated in
FIG. 10, the zero replacement is carried out in a range expressed
by the formula (15).
K 2 .ltoreq. k .ltoreq. 3 2 K - 1 ( 15 ) ##EQU00005##
[0091] In the second embodiment, by the zero replacers 920 and 930
carrying out zero replacement when co-channel interference takes
place or the like, it is possible to further reduce an amount of
operations of MMSE, which is used by the 3-vector equalization
coefficient calculator 140, and an amount of operations by the
3-vector frequency domain equalizer 150.
Third Embodiment
[0092] FIG. 13 is a diagram illustrating a configuration of an
equalization coefficient calculator and equalizer which is used in
a receiver of a third embodiment of the present disclosure. In the
third embodiment, still another configuration of the equalization
coefficient calculator and equalizer in the receiver will be
described. In FIG. 13, components that are the same as those in
FIG. 1 are denoted by the same reference characters and description
thereof will be omitted.
[0093] The equalization coefficient calculator and equalizer in the
receiver includes a controller 610, selectors 120, 130, 620, and
630, a 3-vector equalization coefficient calculator 140, a 3-vector
frequency domain equalizer 150, a 2-vector equalization coefficient
calculator 640, a 2-vector frequency domain equalizer 650, and a
selector 660. In FIG. 13, solid arrows other than the signals
output from the controller indicate complex numbers, each of which
is an element of a vector or a matrix, and dashed arrows indicate
matrices.
[0094] In FIG. 13, the controller 610, based on the channel
h(k.sub.LU) and the undesired signal matrix U(k.sub.LU), calculates
signal power values |H.sub.a(k.sub.L)|.sup.2, |H.sub.a
(k.sub.U)|.sup.2, |H.sub.b(k.sub.L)|.sup.2, and
|H.sub.b(k.sub.U)|.sup.2 of individual vectors at individual
frequency bins, and selects diagonal elements of U(k.sub.LU), which
indicate the power values of undesired signals. Based on the signal
power values and the diagonal elements, the controller 610
calculates ratios of the signal power values to power values of
undesired signals SINR.sub.a(k.sub.L), SINR.sub.a(k.sub.U),
SINR.sub.b(k.sub.L), and SINR.sub.b(k.sub.U).
[0095] The selector 120, by following commands of control signals
from the controller 610, selects three signals from the frequency
domain received vector r(k.sub.LU), which is a 4-vector, in
descending order of the calculated SINRs and generates a 3-vector
r.sub.sel3(k.sub.LU).
[0096] The selector 130, by following commands of control signals
from the controller 610, selects three signals from the channel
vector h(k.sub.LU), which is a 4-vector, in descending order of the
calculated SINRs and generates 3-vector h.sub.sel3(k.sub.LU). The
selector 130 also selects 3.times.3 signals from the 4.times.4
undesired signal matrix U(k.sub.LU) to generate
U.sub.sel3(k.sub.LU).
[0097] With the above-described processing, as with the first
embodiment, it becomes possible for the 3-vector equalization
coefficient calculator 140 and the 3-vector frequency domain
equalizer 150 to carry out processing by 3-vector operations.
[0098] The selector 620, by following commands of control signals
from the controller 610, selects a predefined number of signals
from the frequency domain received vector r(k.sub.LU) based on
quality of the received signals. As an example, the selector 620
selects two signals from the frequency domain received vector
r(k.sub.LU), which is a 4-vector, in descending order of the
calculated SINRs to generate a 2-vector r.sub.sel2(k.sub.LU).
[0099] The selector 630, by following commands of control signals
from the controller 610, selects a predefined number of signals
from the channel h(k.sub.LU) as a channel vector based on quality
of the received signals. As an example, the selector 630 selects
two signals from the channel h(k.sub.LU), which is a 4-vector, in
descending order of the calculated SINRs to generate a 2-vector
h.sub.sel2(k.sub.LU). The selector 630, by following commands of
control signals from the controller 610, also selects a predefined
number of signals from the undesired signal matrix U(k.sub.LU) as a
noise/interference matrix based on quality of the received signals.
As an example, the selector 630 selects 2.times.2 signals from the
4.times.4 undesired signal matrix U(k.sub.LU) to generate
U.sub.sel2(k.sub.LU).
[0100] With the above-described processing, it becomes possible for
the 2-vector equalization coefficient calculator 640 and the
2-vector frequency domain equalizer 650 to carry out processing by
2-vector operations.
[0101] An output of the 3-vector frequency domain equalizer 150 is
denoted by T.sub.sel3(k.sub.LU), and an output of the 2-vector
frequency domain equalizer 650 is denoted by
T.sub.sel2(k.sub.LU).
[0102] The selector 660, by following commands of control signals
from the controller 610, selects either of T.sub.sel3(k.sub.LU) or
T.sub.sel2(k.sub.LU) based on a predefined selection criterion and
outputs the selection as T.sub.r(k.sub.LU). The output selection
criterion is changeable by setting. As an example of the selection
criterion, when SINRs of two signals with the best quality and the
second best quality are sufficiently high based on quality of the
received signals, T.sub.sel2(k.sub.LU) is selected. For example, in
a range where frequency response of the received signal power forms
a flat shape, it is possible to carry out equalization by 2-vector
operations.
[0103] According to the third embodiment, the selectors 120 and 130
selecting three signals makes it possible to reduce an amount of
operations of MMSE, which is used by the 3-vector equalization
coefficient calculator 140, and an amount of operations of the
3-vector frequency domain equalizer 150.
[0104] The selectors 620 and 630 select two signals, the 2-vector
equalization coefficient calculator 640 calculates equalization
coefficients, and the 2-vector frequency domain equalizer 650
carries out equalization. When the SINRs of two signals with the
best quality and the second best quality are sufficiently high, it
is possible to carry out calculation of equalization coefficients
and equalization by 2-vector operations. Accordingly, at frequency
bins where 2-vector operations do not influence performance, it is
possible to further reduce an amount of operations compared with
3-vector operations.
Fourth Embodiment
[0105] FIG. 14 is a diagram illustrating a configuration of an
equalization coefficient calculator usable in a receiver of a
fourth embodiment of the present disclosure. In the fourth
embodiment, another configuration of the 3-vector equalization
coefficient calculator usable in the equalization coefficient
calculators and equalizers which were described in the first to
third embodiments.
[0106] The 3-vector equalization coefficient calculator includes a
controller 710, a vector-vector multiplier 721, a matrix adder 722,
an inverse matrix calculator 723, a vector-matrix multiplier 730, a
power calculator 741, a diagonal element extractor 742, a scalar
adder 743, dividers 751, 752, and 753, and a selector 760. In FIG.
14, solid arrows other than the signals output from the controller
indicate complex numbers, each of which is an element of a vector
or a matrix, and dashed arrows indicate matrices.
[0107] The vector-vector multiplier 721, the matrix adder 722, the
inverse matrix calculator 723, and the vector-matrix multiplier 730
function as an MMSE calculator in matrix operation processing. The
power calculator 741, the diagonal element extractor 742, the
scalar adder 743, and the dividers 751 to 753 function as an MMSE
calculator in scalar operation processing.
[0108] A signal h.sub.sel(k.sub.LU), which is a 3-vector channel
selected by the selector 130, is expressed by the formula (16).
h.sub.sel(k.sub.LU)=[H.sub.a(k.sub.L)H.sub.a(k.sub.U)H.sub.b(k.sub.L)].s-
up.T (16)
[0109] A 3-vector signal U.sub.sel(k.sub.LU), which is an undesired
signal matrix selected by the selector 130, is expressed by the
formula (17).
U _ sel ( k LU ) = [ u a ( k L ) u a * ( k L ) u a ( k L ) u a * (
k U ) u a ( k L ) u b * ( k L ) u a ( k U ) u a * ( k L ) u a ( k U
) u a * ( k U ) u a ( k U ) u b * ( k L ) u b ( k L ) u a * ( k L )
u b ( k L ) u a * ( k U ) u b ( k L ) u b * ( k L ) ] ( 17 )
##EQU00006##
[0110] In this case, it is possible to calculate the equalization
coefficients w.sub.selm(k.sub.LU) based on MMSE by the formula
(18).
w.sub.selm(k.sub.LU)=h.sup.H.sub.sel(k.sub.LU)[h.sub.sel(k.sub.LU)h.sup.-
H.sub.sel(k.sub.LU)+U.sub.sel(k.sub.LU)].sup.-1 (18)
[0111] In the formula (18), the inverse matrix term is calculated
through processing by the vector-vector multiplier 721, the matrix
adder 722, and the inverse matrix calculator 723.
[0112] The vector-matrix multiplier 730 multiplies the inverse
matrix calculated in the above-described processing by the selected
h.sub.sel(k.sub.LU) to obtain the equalization coefficient
w.sub.sel(k.sub.LU)
[0113] When there is no correlation between antennas, it is
possible to express U.sub.sel(k.sub.LU) in the formula (17) by the
formula (19).
U _ sel ( k LU ) = [ u a ( k L ) u a * ( k L ) 0 0 0 u a ( k U ) u
a * ( k U ) 0 0 0 u b ( k L ) u b * ( k L ) ] ( 19 )
##EQU00007##
[0114] In the formula (19), a diagonal element indicates undesired
signal power at each frequency bin, and, when there is no
correlation between frequency bins, U.sub.sel(k.sub.LU) is also
expressed by the formula (20).
U _ sel ( k LU ) = NO 2 ( k LU ) [ 1 0 0 0 1 0 0 0 1 ] = NO 2 ( k
LU ) I ( 20 ) ##EQU00008##
where NO.sup.2(k.sub.LU) denotes noise power existing between
antennas and frequencies without correlation, and I denotes an
identity matrix.
[0115] The equalization coefficient w.sub.sels(k.sub.LU) based on
MMSE is expressed by the formula (21).
w _ sels ( k LU ) = h _ sel H ( k LU ) [ h _ sel ( k LU ) h _ sel H
( k LU ) + U _ sel ( k LU ) ] - 1 = h _ sel H ( k LU ) [ h _ sel (
k LU ) h _ sel H ( k LU ) + NO 2 ( k LU ) ] - 1 = [ I + h _ sel H (
k LU ) [ NO 2 ( k LU ) I ] - 1 h _ sel ( k LU ) ] - 1 h _ sel H ( k
LU ) [ NO 2 ( k LU ) I ] - 1 = [ I + NO - 2 ( k LU ) h _ sel H ( k
LU ) h _ sel ( k LU ) ] - 1 h _ sel H ( k LU ) [ NO 2 ( k LU ) I ]
- 1 = [ no 2 ( I + NO - 2 ( k LU ) h _ sesl H ( k LU ) h _ sel ( k
LU ) ) ] - 1 h _ sel H ( k LU ) = h _ sel H ( k LU ) [ h _ sel H (
k LU ) h _ sel ( k LU ) + NO 2 ( k LU ) I ] - 1 ( 21 )
##EQU00009##
[0116] The inverse matrix term in the formula (21) may be expressed
by a scalar which indicates power, not a matrix, as by the formula
(22).
h.sup.H.sub.sel(k.sub.LU)h.sub.sel(k.sub.LU)+NO.sup.2I(k.sub.LU)=|H.sub.-
a(k.sub.L)|.sup.2+|H.sub.a(k.sub.U)|.sup.2+|H.sub.b(k.sub.L)|.sup.2+NO.sup-
.2(k.sub.LU) (22)
[0117] The inverse matrix term expressed by the formula (22) is
calculated through processing by the power calculator 741, the
diagonal element extractor 742, and the scalar adder 743. The
diagonal element extractor 742 extracts an element in diagonal
elements or calculates a mean value of the diagonal elements. The
equalization coefficient w.sub.sel(k.sub.LU) is obtained by the
dividers 751, 752, and 753 dividing the selected
h.sub.sel(k.sub.LU) by the scalar representing the inverse matrix,
which is obtained through the above-described processing.
[0118] The controller 710, based on quality of the received
signals, calculates the equalization coefficient by switching two
MMSE calculators, a calculator carrying out matrix operation
processing and a calculator carrying out scalar operation
processing. As an example, the controller 710 decides whether or
not correlations between antennas and between frequency bins exist
and controls each block so as to make either of the processing
systems operate. For example, when there is no interference in the
received signal but a noise is added to the received signal,
correlations between antennas and between frequency bins are
small.
[0119] When the controller 710 decides that there is neither
correlation between antennas nor correlation between frequencies
(the correlation values are less than predefined values), the
controller 710 calculates the equalization coefficient
w.sub.sels(k.sub.LU) by making the power calculator 741, the
diagonal element extractor 742, the scalar adder 743, and the
dividers 751 to 753 carry out scalar operation processing. When the
controller 710 decides that there is a correlation either between
antennas or between frequencies (either of the correlation values
is equal to or greater than the predefined value), the controller
710 calculates the equalization coefficient w.sub.selm(k.sub.LU) by
making the vector-vector multiplier 721, the matrix adder 722, the
inverse matrix calculator 723, and the vector-matrix multiplier 730
carry out matrix operation processing. The controller 710 makes the
selector 760 select an equalization coefficient and outputs
w.sub.sel(k.sub.LU).
[0120] In the fourth embodiment, although a configuration example
in a case of 3-vectors is described as an example of the
equalization coefficient calculator, because the configuration does
not depend on the number of vectors, the configuration may also be
applied to cases of 2-vectors and 4-vectors.
[0121] According to the fourth embodiment, because the controller
710 can select an equalization coefficient calculation method with
a low amount of MMSE calculation, it is possible to reduce the
amount of operations.
Fifth Embodiment
[0122] A fifth embodiment is a variation of the equalization
coefficient calculator and equalizer of the third embodiment
illustrated in FIG. 13. In the fifth embodiment, an equalization
coefficient calculator with a lower amount of operations is
selected on the 2-vector equalization coefficient calculator 640 of
the third embodiment, in accordance with the existence of a
correlation between antennas or between frequency bins, as with the
fourth embodiment. For example, when the correlation between
antennas or between frequency bins is small, it is possible to
carry out equalization by selecting an equalization coefficient
calculator with a lower amount of operations because there is no
interference in the received signal. With this configuration, the
amount of operations may be further reduced.
Sixth Embodiment
[0123] A sixth embodiment is a variation of the equalization
coefficient calculator and equalizer of the first embodiment
illustrated in FIG. 1. In the first embodiment, because the
3-vector equalization coefficient calculator 140 is used, inverse
matrix operation needs to be carried out when interference
resistance is required. On the other hand, when there is no
interference, as described in the fourth embodiment, the 3-vector
equalization coefficient calculator can carry out the scalar
operation expressed by the formulae (21) and (22). Therefore, in
the sixth embodiment, an equalization coefficient calculator with a
lower amount of operations is selected on the 3-vector equalization
coefficient calculator 140 of the first embodiment, in accordance
with the existence of a correlation between antennas or between
frequency bins, as with the fourth embodiment.
[0124] The amount of operations necessary for the equalization
coefficient calculation through scalar operations for 4-vectors is
less than the amount for the calculation through 3.times.3 or
2.times.2 inverse matrix operations, and lower power is consumed
for the calculation thorough scalar operations. In other words,
though depending on a trade-off between interference resistance and
SINR, it is possible to reduce the amount of operations by
switching 3.times.3 or 2.times.2 inverse matrix operations and
scalar operations of 4-vectors.
Seventh Embodiment
[0125] A seventh embodiment is a variation of the selectors 120 and
130 of the first embodiment illustrated in FIG. 1 and the selectors
620 and 630 of the third embodiment illustrated in FIG. 13. The
afore-mentioned selectors 120, 130, 620, and 630 carry out
selection based on SINRs at individual frequency bins. In the
seventh embodiment, a selector, based on a mean value of SINRs at
individual frequency bins in each predefined range of frequency
bins, carries out selection for each of the ranges of frequency
bins. In other words, the equalizer carries out selection of a
signal based on a mean value of quality for each predefined range
of frequency bins, and carries out equalization with a small amount
of operations. With this configuration, in an operation circuit,
the number of switching operations to switch signals may be reduced
compared with a case of carrying out control for each frequency
bin, which leads to a reduction in power consumption.
Eighth Embodiment
[0126] An eighth embodiment is a variation of the controller 110 of
the first embodiment illustrated in FIG. 1, the controller 910 of
the second embodiment illustrated in FIG. 12, and the controller
610 of the third embodiment illustrated in FIG. 13. The
afore-mentioned controller 110, 610, and 910 control each selector
based on an SINR at each frequency bin of a received signal. In the
eighth embodiment, a controller makes a decision by using not only
an SINR but also an amount of power of a received signal and
carries out control in accordance with the SINR and the received
signal power.
[0127] For example, not only a case in which the signal power S is
1.0 and the power of interference and noise I+N is 0.1 but also a
case in which the signal power S is 0.1 and the power of
interference and noise I+N is 0.01 may be included in cases in
which the SINR is 10. In the case in which the power S is 0.1 and
the power I+N is 0.01, because the signal power S is sufficiently
low, it is possible to carry out processing in which no signal is
selected or the signal is replaced with zero. Carrying out control
by taking into consideration the received signal power as described
above makes it possible to reduce an amount of operations.
Ninth Embodiment
[0128] A ninth embodiment is exemplified by a case in which a lot
of receiving antenna systems are used. In the first to third
embodiments, cases in which there are two receiving antenna
systems, and a complex baseband signal is sampled at twice the
symbol rate by ADCs 212a and 212b were described. In the ninth
embodiment, a case in which there are M receiving antenna systems,
and a complex baseband signal, which is received by the M systems
of antenna, is sampled at C-times the symbol rate will be
described.
[0129] Even though the received signal is sampled at C-times the
symbol rate by the ADCs, valid signal components, due to a
transmission filter and a reception filter, exist within a range of
signals sampled at twice the symbol rate. Therefore, as with the
first to third embodiments, even in a case of C>2, the number of
elements of a channel vector required for the calculation of an
equalization coefficient becomes 2M. A noise/interference matrix
becomes a (2M).times.(2M) matrix.
[0130] 2M-1 or less signals are selected as elements of the channel
vector, (2M-1).times.(2M-1) or less signals are selected as
elements of the noise/interference matrix, and, based on the
selected channel vector and noise/interference matrix, 2M-1 or less
equalization coefficient vector is calculated.
[0131] Accordingly, when the ninth embodiment is applied to the
first and second embodiments, it is possible to carry out
equalization by 2M-1 vector operations. When the ninth embodiment
is applied to the third embodiment, the equalization is carried out
by switching an equalization by 2M-1 vector operations and an
equalization by 2M-2 vector operations.
[0132] According to the ninth embodiment, it is possible to carry
out equalization with 2M-1 or less vector operations and to reduce
an amount of operations.
[0133] As described above, according to the embodiments, in the
equalization on a receiver having multiple antenna systems, when
the equalization is carried out by combining different antenna
systems and frequency signals, it is possible to reduce an amount
of operations required for the equalization by selecting a
frequency signal based on an SINR or the like. Therefore, it is
possible to carry out equalization the circuit implementation of
which is easy and the power consumption of which is low.
Summary of an Aspect of the Disclosure
[0134] Various aspects of the embodiments according to the present
disclosure include the followings.
[0135] An equalization method of the present disclosure includes,
in a receiver having multiple antennas, carrying out frequency
domain conversion of M received signals, which are received by the
plurality of antennas, into a 2M received vector, which includes 2M
elements, carrying out channel estimation and noise/interference
estimation based on the 2M received vector, calculating a 2M
channel vector and a (2M).times.(2M) noise/interference matrix,
selecting a 2M-1 or less channel vector from the calculated 2M
channel vector and selecting a (2M-1).times.(2M-1) or less
noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals, calculating 2M-1 or less equalization coefficient vector
based on the 2M-1 or less channel vector and the
(2M-1).times.(2M-1) noise/interference matrix, selecting a 2M-1 or
less received vector from the 2M received vector, and equalizing
the 2M-1 or less received vector by using the equalization
coefficient.
[0136] An equalization method of the present disclosure includes,
in a receiver having multiple antennas, carrying out frequency
domain conversion of M received signals, which are received by the
plurality of antennas, into a 2M received vector, which includes 2M
elements, carrying out channel estimation and noise/interference
estimation based on the 2M received vector, calculating a 2M
channel vector and a (2M).times.(2M) noise/interference matrix,
selecting a 2M-1 or less channel vector from the calculated 2M
channel vector and selecting a (2M-1).times.(2M-1) or less
noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals, replacing a portion of the 2M-1 channel vector with zeros,
replacing a portion of the (2M-1).times.(2M-1) or less
noise/interference matrix with zeros, calculating a 2M-1 or less
equalization coefficient vector as equalization coefficients based
on both the channel vector, the portion of which is replaced with
zeros, and the noise/interference matrix, the portion of which is
replaced with zeros, selecting a 2M-1 or less received vector from
the 2M received vector, replacing a portion of the 2M-1 received
vector with zeros, and equalizing the 2M-1 or less received vector,
the portion of which is replaced with zeros based on the
equalization coefficient.
[0137] An equalization method of the present disclosure includes,
in a receiver having multiple antennas, carrying out frequency
domain conversion of M received signals, which are received by the
plurality of antennas, into a 2M received vectors, which includes
2M elements, carrying out channel estimation and noise/interference
estimation based on the 2M received vector, and calculating a 2M
channel vector and a (2M).times.(2M) noise/interference matrix,
includes a first method including selecting a 2M-1 or less channel
vector from the calculated 2M channel vector and selecting a
(2M-1).times.(2M-1) or less noise/interference matrix from the
calculated (2M).times.(2M) noise/interference matrix, based on
quality of the received signals, calculating a 2M-1 or less
equalization coefficient vector as first equalization coefficients
based on the 2M-1 or less channel vector and the
(2M-1).times.(2M-1) or less noise/interference matrix, selecting a
2M-1 or less received vector from the 2M received vector, and
equalizing the 2M-1 or less received vector based on the first
equalization coefficients, and a second method including selecting
a 2M-2 or less channel vector from the calculated 2M channel vector
and selecting a (2M-2).times.(2M-2) or less noise/interference
matrix from the calculated (2M).times.(2M) noise/interference
matrix, based on quality of the received signals, calculating a
2M-2 or less equalization coefficient vector as second equalization
coefficients based on the 2M-2 or less channel vector and the
(2M-2).times.(2M-2) or less noise/interference matrix, selecting a
2M-2 or less received vector from the 2M received vector, and
equalizing the 2M-2 or less received vector based on the second
equalization coefficients, and in accordance with a predefined
criterion, switches the first method and the second method.
[0138] Any one of the above-described equalization methods may,
calculates the equalization coefficient based on an MMSE by using
one of matrix operation processing and scalar operation processing,
which are switched based on quality of the received signal.
[0139] Any one of the above-described equalization methods may
select the received vector, the channel vector, or the
noise/interference matrix by using any one of the following
selection methods:
(1) selection for each frequency bin based on SINRs of respective
frequencies of the respective received signals; (2) selection for
each frequency bin based on SINRs and received signal powers of
respective frequencies of the respective received signals; (3)
based on a mean value of SINRs in a predefined range of frequency
bins of the respective received signals, selection for each of the
predefined ranges of frequency bins; (4) based on a mean values of
SINRs and a mean value of received signal powers in a predefined
range of frequency bins of the respective received signals,
selection for each of the predefined ranges of frequency bins; and
(5) switching between selection for each frequency bin based on
SINRs of respective frequencies of the respective received signals
and, based on a mean value of SINRs in a predefined range of
frequency bins of the respective received signals, selection for
each of the predefined ranges of frequency bins.
[0140] An equalizer of the present disclosure includes a frequency
domain converter which carries out frequency domain conversion of M
systems of received signals, which are received by multiple
antennas, into a 2M received vector, which includes 2M elements, a
channel and noise/interference estimator which carries out channel
estimation and noise/interference estimation based on the 2M
received vector to calculate a 2M channel vector and a
(2M).times.(2M) noise/interference matrix, a first selector which
selects a 2M-1 or less channel vector from the calculated 2M
channel vector and selects a (2M-1).times.(2M-1) or less
noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals, an equalization coefficient calculator which calculates a
2M-1 or less equalization coefficient vector as equalization
coefficients based on the 2M-1 or less channel vector and the
(2M-1).times.(2M-1) or less noise/interference matrix, a second
selector which selects a 2M-1 or less received vector from the 2M
received vector, and a frequency domain equalizer which equalizes
the 2M-1 or less received vector by using the equalization
coefficients.
[0141] An equalizer of the present disclosure includes a frequency
domain converter which carries out frequency domain conversion of M
received signals, which are received by multiple antennas, into a
2M received vector, which includes 2M elements, a channel and
noise/interference estimator which carries out channel estimation
and noise/interference estimation based on the 2M received vector
to calculate a 2M channel vector and a (2M).times.(2M)
noise/interference matrix, a first selector which selects a 2M-1 or
less channel vector from the calculated 2M channel vector and
selects a (2M-1).times.(2M-1) or less noise/interference matrix
from the calculated (2M).times.(2M) noise/interference matrix,
based on quality of the received signals, a first zero replacer
which replaces a portion of the 2M-1 or less channel vector and a
portion of the (2M-1).times.(2M-1) or less noise/interference
matrix with zeros, an equalization coefficient calculator which
calculates a 2M-1 or less equalization coefficient vector as
equalization coefficients based on both the 2M-1 or less channel
vector, the portion of which is replaced with zeros, and the
(2M-1).times.(2M-1) or less noise/interference matrix, the portion
of which is replaced with zeros, a second selector which selects a
2M-1 or less received vector from the 2M received vector, a second
zero replacer which replaces a portion of the 2M-1 or less received
vector with zeros, and a frequency domain equalizer which equalizes
the 2M-1 or less received vector, the portion of which is replaced
with zeros, by using the equalization coefficients.
[0142] An equalizer of the present disclosure includes a frequency
domain converter which carries out frequency domain conversion of M
received signals, which are received by multiple antennas, into a
2M received vector, which includes 2M elements, a channel and
noise/interference estimator which carries out channel estimation
and noise/interference estimation based on the 2M received vector
to calculate a 2M channel vector and a (2M).times.(2M)
noise/interference matrix, a first equalization processing unit
including a first selector which selects a 2M-1 or less channel
vector from the calculated 2M channel vector and selects a
(2M-1).times.(2M-1) or less noise/interference matrix from the
calculated (2M).times.(2M) noise/interference matrix, based on
quality of the received signals, a first equalization coefficient
calculator which calculates a 2M-1 or less equalization coefficient
vector as first equalization coefficients based on the 2M-1 or less
channel vector and the (2M-1).times.(2M-1) or less
noise/interference matrix, a second selector which selects a 2M-1
or less received vector from the 2M received vector, and a first
frequency domain equalizer which equalizes the 2M-1 or less
received vector by using the first equalization coefficients,
a second equalization processing unit including a third selector
which selects a 2M-2 or less channel vector from the calculated 2M
channel vector and selects a (2M-2).times.(2M-2) or less
noise/interference matrix from the calculated (2M).times.(2M)
noise/interference matrix, based on quality of the received
signals, a second equalization coefficient calculator which
calculates a 2M-2 or less equalization coefficient vector as second
equalization coefficients based on the 2M-2 or less channel vector
and the (2M-2).times.(2M-2) or less noise/interference matrix, a
fourth selector which selects a 2M-2 or less received vector from
the 2M received vector, and a second frequency domain equalizer
which equalizes the 2M-2 or less received vector by using the
second equalization coefficients, and a controller which switches
the first equalization processing unit and the second equalization
processing unit.
[0143] Although various embodiments have been described with
reference to drawings, it is indisputable that the present
disclosure is not limited to such embodiments. It is apparent that
those skilled in the art can devise numerous other variations and
modifications within the scope of the foregoing disclosure, and it
should be understood that such variations and modifications
naturally belong to the scope of the disclosure. Components in the
above-described embodiments may be combined in any manner without
departing from the scope of the disclosure.
[0144] Although, in the above embodiments, the present disclosure
was described by using a case in which the present disclosure is
configured with hardware as an example, it is also possible to
implement the present disclosure by software in cooperation with
hardware.
[0145] Each functional block, used in the description of each of
the above-described embodiments, is typically implemented by LSIs,
which are integrated circuits. The functional blocks may be
individually integrated into a single chip or may be collectively
integrated into a single chip with a portion or the whole of every
functional block. Although an integrated circuit is referred to as
an LSI above, an integrated circuit may also be referred to as an
IC, system LSI, super LSI, or ultra LSI in accordance with a degree
of integration.
[0146] A method of integrated circuit implementation is not limited
to developing an LSI, and the functional blocks may be implemented
by a dedicated circuit or a general-purpose processor. A field
programmable gate array (FPGA), which is programmable after an LSI
is fabricated, or a re-configurable processor, which makes it
possible to reconfigure connections and setting of circuit cells in
an LSI, may also be used.
[0147] Furthermore, if a technology of integrated circuit
implementation which substitutes for an LSI emerges due to progress
of the semiconductor technology or derivation of new technologies,
it goes without saying that integrated circuit implementation of
the functional blocks may be accomplished by using the new
technologies. Application of biotechnology or the like may be a
possible candidate.
[0148] It is possible to express the present disclosure as an
equalization method which is carried out in a radio communication
apparatus. It is also possible to express the present disclosure as
an equalizer which is an apparatus having a function to carry out
an equalization method or a program which makes a computer carry
out an equalization method or function as an equalizer. In other
words, it is possible to express the present disclosure in any
category of an apparatus, a method, and a program.
[0149] The present disclosure makes it possible to reduce an amount
of operations needed for equalization in a receiver with multiple
receiving antennas, and has an advantageous effect as an
equalization method and an equalizer usable in a radio
communication apparatus in, for example, a millimeter wave radio
communication.
* * * * *