1. An electronic device comprising: circuitry configured to: exchange capability information with another electronic device, the capability information including at least a first field, a second field and a third field; and perform wireless communication with the another electronic device based on the capability information, wherein the first field indicates a capability of performing a transmission beamforming, the second field indicates capabilities of a beamformee for receiving a sounding packet, and the third field indicates a maximum number of space time streams that is corresponding to a number of columns of a MIMO (multiple-input multiple-output) channel matrix for the wireless communication.

2. The electronic device of claim 1, wherein the capabilities of the beamformee includes dimension information for the sounding packet based on antenna information of the beamformee.

3. The electronic device of claim 2, wherein the sounding packet being generated based at least in part on information included in the third field.

4. The electronic device of claim 3, wherein a dimension of the sounding packet is equal to or less than a number indicated by the maximum number of space time streams.

5. The electronic device of claim 4, wherein the capability information is present in a management frame.

6. The electronic device of claim 5, wherein the management frame includes: a beacon signal, a measure pilot, an association response, a re-association response, and/or a probe response.

7. The electronic device of claim 6, wherein the capability information is defined as a beamforming capability field in IEEE802.11 standard.

8. The electronic device of claim 7, wherein the beamforming capability field is included in HT capability field defined by the IEEE802.11 standard.

9. The electronic device of claim 8, further comprising: a plurality of antennas configured to transmit or receive signals.

10. The electronic device of claim 9, wherein a number of the plurality of antennas is two.

11. The electronic device of claim 9, wherein a number of the plurality of antennas is three.

12. The electronic device of claim 4, the wireless communication is beamforming communication established based on channel estimation using the sounding packet.

13. The electronic device of claim 12, the beamforming communication is corresponding to an Explicit beamforming.

WIRELESS 60 ZCWTZBW SYSTEMld,W IRELESS C0WJ:CATSOM
APPARATUS AND WIRELESS CBmZTNPCATGON METHOD
CROSS REFERENCES TO RELATED APP1,ICATIONS
The present invention contains subject matter related
to Japanese Patent Applications JP 2006-124538 and JP
2007-056245 filed in the Japanese Patent Office onApril 27,
2006, and March 6, 2007, respectively, the entire contents of
which being incorporated herein by reference.
. . BACKGROUND
1. Technical Field
The invention relates to awireless communication system,
a wireless communication apparatus and a wireless
communication method using spatial multiplexing, and more
particularly, to a wireless communication system, a wireless
communicatioi~a pparatus and a wireless communicati~~mre thod,
in which a transmitter and a receiver share channel information
toperform closed loop type spatial multiplexing transmission.
In particular, the invention relates to a wireless
communication system, a wireless communication apparatus and
a wireless communication method, which perform beamforming on
the basis of inf ormation which is fed back from a receiver when
a transmitter transmits a packet, and more particularly, to
a wireless communication system, a wireless communication
apparatus and a wireless communication method, which perform
beamforming by- feeding back beamforming information between
abeamformerandbeamformee whicharedifferent fromeachother
in the number of antennas or the number of supported streams.
2. ~ackground Art
As a system for removing wire in an existing wired
communication method, a wireless network is attracting
attention. A standard of the wireless network may be the IEEE
(The instituteof ElectricalandElectronicsEngineers) 802 .I1
or the IEEE 802.15.
For example, in the IEEE 802.11a/g, as a standard of a
wireless LAN, an orthogonal frequency division multiplexing
(OFDM) modulation method which is one of a multi-carrier method
is employed. In the OFDM modulation method, since
transmission data is distributed to a plurality of carriers
having orthogonal frequenciesand is transmitted, the band of
each carrier becomes narrow, frequency use efficiency is very
high, and frequency-selective fa.d in. g interference is strong.
Inaddition, inthe IEEE 802 . llalgstandard, amodulation
method for accomplishing a communication speed of a maximum
of 54 Mbps is supported, but a next-generation wireless LAN
standard for realizing a new high bit rate is required.
As one of atechnology of realizing a high speed of
wireless communication, multi-input multi-output (MIMO)
communication is attracting attention. This is a
communication method in which both a transmitter side and a
receiver side respectively include a plurality of antennas to
realize spatially multiplexed streams. The transmitter side
performs spatial/temporal encoding and multiplexing of plural
pieces of transmission data and distributes and transmits the
plural pieces of transmission data to N transmission antennas
through channels. The receiver side performs
spatial/temporal decoding of reception signals received by M
reception antennas through the channels to obtain reception
data without crosstalk between the streams (for example, see
JP-8-2002-44051 (Patent pocument 1) ) . Ideally, spatial
streams corresponding to the smaller number (MZN [N, MI ) of the
transmission and reception antennas are formed.
According to the MIMO communication method, a
transmission capacity can increase according to the number of
antennas and a communication speed improvement can be realized,
without increasing a frequency band. Since the spatial
multiplexing is used, frequency use efficiency is high. The
MIMOmethoduses channel characteristics and is different from
a simpletransmission/reception adaptive array. For example,
in the IEEE 802 . llnwhich is the extension standard of the IEEE
802 . lla/g, anOFDM - MIMOmethodus ing OFDM inprimarymodulation
is employed. Currently, the IEEE 802.1122 is being
standardized in a task group n(TGn) and a specification
established therein is based on a specification established
inEnhancedwireless consortium (EWC) formedonOctober, 2005.
In the M"1IQ communication, in order to spatially divide
a spatially multiplexed reception signal y into the stream
signals x, a channel matrix H is acquired by any method and
the spatially multiplexed reception signal needs to be
spatially divided into a plurality of original streams using
the channel matrlx H by a predetermined algorithm.
The channel matrix H is obtained by allowing a
transmitter/receiver side to transmit/receive existing
training sequence, estimating the channels by a difference
between the actuallyreceived signal and the existing sequence
and arranging propagation channels of a combination of
transmission and reception antennas in a matrix form. When
the number of transmission antennas is N and the number of
receptionantennas isM, the channel matrix is MxN (rowxcolumn)
matrix. Accordingly, the transmitter side transmits N
training sequcl-cce and the receiver side acquires thi. channel
matrix H using the received training sequence.
A method of spatially dividing a reception signal is
largely classified into an open loop type method in which a
receiver independentlyperforms spatial division on the basis
of the channel matrix H and a closed loop type method in which
a transmitter side gives weights to the transmission antennas
on the basis of the channel matrix to perform adequate
beamforming toward a receiver to form an ideal spatial
orthogonal channel.
As an open loop type MIMB transmission method, there is
a zero force (for example, see A. Benjebbour, H. Murata and
S. Yoshida, "Performance of iterative successive detection
, Proc . IEEEVTCSpring,
vol. 2, pp. 1287-1291, Rhodes. Greece, May 2001 (Non-Patent
Document2) ) oraminimummeansquareerror (VD'ISE) (forexample,
see A. Benjebbour, H. Murata and S . Yoshida, " Perromance
comparison of ordered successive receivers for space-time
transmission", Pros. IEEE VTC Fall, ~ 0 % .4, pp.2053-2057,
Atlantic City, USA, sept.2001 (Non-Patent Document 3)) . The
open loop type MIMO transmission method is a relative simple
algorithm forobtaining receptionweightmatrixw for spatially
dividing the reception signal from the channel matrix H, in
which a feedback operation for sharing the channel information
between the transmitter and the receiver is omitted and the
transmitter and the receiver independently perform spatial
multiplexing transmission.
,As an ideal one of a closed,loop type MIMO transmission
method, a singularvalue decomposition (SVD) -MIMOmethodusing
SVD of the channel matrix H is known (for example, see
http://radio3.ee.uec.ac.jp/MIMO(IEICE-Ts). Pdf (October 24,
2003) (Non-Patent Document 1) ) . Inthe SVD-MIMOtransmission,
a numerical matrix having channel information corresponding
to antenna pairs as elements, that is, a channel information
matrix H, is subjected to the singular value decomposition to
obtain UDV'. A transmitter side uses V in a transmission
antenna weight matrix and transmits a beamformed packet to a
receiver and a receiver si.de typically gives (UD) "l as a
receptionantenna weight matrix. Here, D is a diagonal matrix
having square roots of singular values hi corresponding to
qualities of the spatial streams in diagonal elements (the
subscript i indicates an ith spatial stream). The singular
values hi are arranged in the diagonalelements of the diagonal
matrix D in ascending order and power ratio distribution or
modulation method allocation is performed according to
communication quality represented by the level of the singular
value with respect to the streams such that a plurality of
spatial orthogonal multiplexed propagation channels which are
logically independent are realized. The receiver side can
extract a plurality of original signal sequence without
crosstalk and theoretically accomplish maximum pertormance.
In the closed loop type MIMO conununication system,
adequate beamforming is performed when the transmitter
transmits the packet, but inionnation on the channel
information needs to be fed back from the receiver side for
receiving the packet.
For example, in the EWC HT (High Throughput) MAC (Media
Access Control) Specification Version V1.24, two kinds of
procedures, that is, "implicit feedback" and "explicit
feedbacls", are defined as the procedure for feeding back the
informationon the channel matrix between the transmitter and
the receiver.
In the "implicit feedback", the transmitter estimates
a backward channel matrix from the receiver to the transmitter
using training sequence transmitted from the receiver, and a
forward channel matrix from the transmitter to the receiver
is computed to perform beamforming on the assumption that
bidirectional channel characteristics between the transmitter
and the receiver are reciprocal.
In the "explicit feedbackN, the receiver estimates a
forward channel matrix from the transmitter to the receiver
using training sequence transmitted from the transmitter and
returns a packet including the channel matrix as data to the
transmitter, and transmitter performs the beamforming using
the received channel matrix.' Alternatively, the receiver
computes a transmission weight matrix for allowing the
transmitter to perform the beamforming from the estimation
channel matrix and returns a packet including the transmission
weight matrix as the data to the transmitter. In the explicit
feedback, since the weight matrix is computed on the basis of
the estimated forward channel matrix, it may not be assumed
that the channels are reciprocal.
In view of packet transmission, the transmitter is an
initiator and the receiver is a receiver. However, in view
of beamforming, the initiator for transmitting the packet is
a beamfarmer and the receiver for receiving the beamformed
packet is abeamformee. Communication fromthe beamformer to
the beamformee is referred to as "forward" and communication
from the beamformee to the beamformer is referred to as
"backward" . For example, when an access point (AP) transmits
a data frame to a client terminal (STA) as the beamformer, the
accesspointperformthebeamformingonthebasis of the channel
information transmitted from the client in the explicit
feedback .
Fig. 14 shows a state where the beamformee estimates the
channel matrix excited by a training signal transmitted from
the beamformer. In the same drawing, a STA-A having three
antennas is the beamformer and a STA-B having two antennas is
the beamformee and feedback is performed on the basis of a CSI
format. In the below-described description or equations, a
subscript AF3 indicates forward transmission from tht STA-A to
the STA-B. A numerical subscript corresponds to the antenna
number of the corresponding terminal.
The training sequence transmitted from the antennas of
the STA-A are (~AB~ZA,BZ ,t -3) and the signals received by the
antennas of the STA-A through a channel NAB are (rABl,r m2), the
following equation is obtained.
where, the channelmatrixn-is a 2x3 matrixandexpressed
bythe followingequation. But, hij is achannelcharacteristic
value of ath antenna of the STA-A to ith antenna of the STA-B.
When the channel matrix HAB is subjected to singular value
decomposition, the following equation is obtained. Here, ?JAB
is a matrix having an inherent normalized vector of kIRB~rnHrV RB
is an inherent normalized vector of HAB~HAaBn d DAB is a diagonal
matrix having a square root of an inherent vector of HABHAsHo r
HABHkIRaB s the diagonal elements. In addition, UAB and VAB are
unitary matrices and complex conjugate transposed matrices
thereof become inverse matrices.
The transmission weight matrix necessary for larmingthe
frame transmitted from the STA-A to the STA-B is the matrix
VAB obtained by performing the singular value decomposition
with respect to the forward channel matrix Hm. When the
beamformee receives a sounding packet, the beamforrnee divides
the sounding packet into spatial strearn trainings to construct
the estimation channel matrix HAB. The CSI composed of MTMO
channel coefficients hll, hlz, ... which are elements of the
channel matrix is collected and fed back to the STA-A.
If a transmission vector composed of transmission
signals of the antennas of the STA-A is x and a reception signal
of the STA-B is y, the reception signal becomes y=Hmx in a case
where the beamforming is not performed (un-steered) , but the
reception signal y becomes the following equation in a case
where the beamforming are performedby the transmission weight
matrix VAB (steered) .
Accordingly, the STA-B can perform spatial division to
the original streambymultiplying a receptionvector including
the reception signals of the antennas by D,-'URB~ as a reception
weight.
Fig. 15 shows a frame exchange procedure for transmitting
beamforming from the access point to the client terminal by
the explicit feedback.
This procedure is initiated by the access point which
sends the sounding packet including a CSI feedback request.
The ~ounding packet includes the training sequence
excitedby the channel matrix. Accordingly, when the sounding
packet is received, the client terminal divides the spatial
stream training to estimate the channel matrix H and collects
the CSI. The CSI data is included in the packet as a CSI
feedback (CFB) and returned to the access point
The access point computes the transmissionweight matrix
for beamforming from the received CFB and multiplies the
transmission signal by it to transmit the beamformed packet
tothe client terminal. Eveninaplacewherethe communication
was hard to be accomplished in the past, communication is
accomplished at a high transmission rate by the beamforming.
As described above, in the explicit feedback, the
beamformer can receive the explicit feedbackof the estimation
channel matrix fromthebeamformee. The format of the feedback
format of the estimationchannel matrix is largelyclassified
into a case where an MIMO channel coefficient is sent and a
case where a transmission weight matrix V for beamforming
computed by the beamformee.
The former format is called channel state information
(CSI) . The beamformer needs to compute the transmission
weight matrix V for beamforming by constructing the channel
matrixli fsomthe receivedCSI andperforrningthe singularvalue
decomposition.
The latter is classified into a case where the
transmission weight matrix V for beamforming is sent in an
uncompressed format and a case where the transmission weight
matrix V for beamforming is sent in a compressed format.
According to the explicit feedback, a processing burden for
estimating the channel matrix in the beamformer side and a
processing burden for calculating the transmission weight
matrix from the channel matrix are reduced.
Fig. 16 shows a scheme of a HT coiltrol field of an MAC
frame defined in the EWC specification. The HTC field has 32
bits, but, among them, 22"d to 23rd CSI/steering fields can
specify a feedback type received from the beamformee in the
explicit feedback (see Fig. 17) .
As described above, the processing burden of the
beamformer which performs beamforming with respect to a
transmission frame is reduced by the explicit feedback.
However, when the beamformer and the beamformee are different
from each other in the number of antennas or the number of
supported streams, several problems are caused at the time of
beamforming.
In a spatial multiplexing type cornmication apparatus,
the dimension number which is allowed by the processing
capability including the estimation of the channel matrix H,
the computation of the transmission weight matrix for
beamforming, and themultiplicationof the transmissionvector
and the transmission weight matrix V for beamforming is
generallydesiynedaccordingto the number of antennas included
therein. Accordingly, the transmission weight matrix for
beamforming cannot be constructed by spatially dividing a
training signal transmitted from the beamformer having the
number of antennas which is larger than an allowable dimension,
the transmission weight matrix for beamforming cannot be
computed from the channel matrix which is fed back from the
beamformee, or the transmission weight matrix for beamforming
which is fed back f r ~tmhe beamformee cannot be multiplied with
the transmission vector.
First, consider a case where the explicit feedback is
performed with a CSI format.
In a case where the number N of antennas of the STA-A
is smaller than or equal to the number M oE antennr~s of the
STA-B, no problem is specially caused in the beamformee side.
Fig. 18 shows a state where the explicit feedback is performed
with a CSI format when N=2 and M=3. The STA-B includes a
processing capability of M streams, and can estimate an MxN
channel matrix excitedbya training signal includingN streams
and feed back the collected CSI information to the STA-A. The
STA-A side can suppress the fed-back MxN channel matrix to a
range of N rows and compute the transmission weight matrix for
beamforming by the singular value decomposition from the NxN
channel matrix.
However, in a case of N>M, problems are caused. This
isbecause, whenthe STA-B canprocess onlyMstreams, the STA-B
obtains only MxM estimation channel matrix using M packets
although the STA-A side transmits the sounding packet for
exciting N-dimensionab spatial channel matrix. Fig. 19 shows
a state where the explicit feedback is performed with the CSI
format when N=3 and M=2.
In the EWC specification, when the explicit feedback is
applied, a scheme of informing information on channel
estimation maximum dimension is defined as one of the
capabilities of the beamformee side. It is defined that the
HT terminal corresponding to high-speed transmission declares
that it itself is the HT terminal by including a HT capability
field in a predetermined management frame.
Fig. 20 shows a foxmat of a HT capability elemcnt. In
a TxBF (transmitbeamfoming) capability field, anyHT function
of the beamforming is specified. Fig. 21 shows the
configuration of the Tx beamforming capability field. The Tx
beamforming capability field has 32 bits, but, among them, lgth
to 2oth bits are allocated to the CSI number of beamformer
antennae, 2lSt to 22nd bits are allocated to the uncompressed
steering matrix of beamformer antennae, and 23rd to 24th bits
are allocated to the compressed steering matrix of beamformer
antennae. In these fields, the spatial dimension number of
the sounding packet which can be received from the beamfornier
when the beamformee performs the explicit feedback with each
t. format is described.
However, in the EWC specification, since it is n0.t
def inedwhich sounding packet is transmitted by the beamformer,
the STA-A may transmit the sounding packet for exciting more
than M channels even when the STA-B informs of its own maximum
dimension number by the above-described scheme and thus the
STA-B is forced to estimate MxN channel matrix.
As a method of solving such problems without
deteriorating the beamforming characteristics, it may be
considered that a channel estimation maximum dimension N,,,
corresponding to a rated maximum number of antennas is given
to the STA-B as the beamformee (for example, if it is based
on the IEEE specification, N,,,=4) .
For example, when the number of antennas of i-he STA-B
is M=2 and the rated maximum number of antennas is Nm,,=4, the
STA-B can compute only a 2x2 matrix in consideration of the
communication with the terminal having the same' number of
antenna, but needs tocompute a2x4 matrix. Inthis case, since
calculation or processing circuit needs to be doubled,
miniaturization or. low cost of the apparatus is hard to be
realized.
The same is also applied to the explicit feedback for
feeding back the transmission weight matrix V for beamforming,
instead of the CSI format.
In a case where the number N of antennas of the STA-A
is smaller than or equal to the number M of antennas of the
STA-B, no problem is specially caused in the beamformee side.
Fig. 22 shows a state where the transmission weight matrix V
for beamforming is fedback by the explicit feedback when N=2
and M=3. The STA-B includes a processing capability of M
streams, and can estimate an MxN channel matrix excited by a
training signal including N streams, compute an NxM
transmission weight matrix V for beamforming by the singular
value decomposition from the estimation channel matrix, and
feed backs the transmission weight matrix information to the
STA-A. The STA-A side can perform beamforming using the
fed-back transmission weight matrix for beamforming.
However, in a case of N>M, problems are caused. This
isbecause, whcnthe STA-B canprocess onlyM streams, ilie STA-B
obtains only an MxM estimation channel matrix using M packets
although the STA-A side transmits the sounding packet for
exciting N-dimensional spatial channel matrix. Fig. 23 shows
a state where the transmission weight matrix V is fed back by
the explicit feedback when N=3 and M=2.
In the EWC specification, when the explicit feedback is
applied, a scheme of informing information on channel
estimation maximum dimension is defined as one of the
capabilities of the beamfsrmee side (described above).
However, the STA-A may transmit the sounding packet for
exciting more than M channels even when the STA-B informs of
its ownmaximumdimensionnumber by the above-deeicribedscheme
and thus the STA-B is forced to estimate MxN channel matrix.
As a method of solving such a problem without
deteriorating the beamforming characteristics, it may be
considered that a channel estimation maximum dimension N,,,
corresponding to a rated maximum number of antennas is given
to the STA-H as the beamformee (for example, if it is based
on the IEEE specification, NmJ,,,=4a)n d a processing capability
which can compute the transmission weight matrix for
beamforming is given to the obtained N,,,xN estimation channel
matrix.
For example, when the number of antennas of the STA-B
is M=2 and the rated maximum number of antennas is N,,,,,=4, the
STA-B can compirte only a 2x2 matrix in consideration of the
communication with the terminal having the same number of
antenna, but must compute a 2x4 matrix. In this case, since
calculation or processing circuit needs to be doubled,
miniaturization, low cost and low power consumption of the
apparatus are hard to be realized.
SUMMARY OF THE INVENTION
It is desirable to provide an excellent wireless
communication system, wireless communj.cation apparatus and
wireless communicationmethod, whichare capable of performing
communication at a high transmission rate by a beamformed
packet by allowing a terminal which operates as a beamfermer
to suitably set a transmission weight matrix on the basis of
an estimation channel matrix fed back from a terminal which
operates as a beamforrnee.
It is also desirable to provide an excellent wireless
communication system, wireless communication apparatus and
wireless communication method, which are capable of suitably
performing beamforming by the explicit feedback without
deteriorating beamforming characteristics or increasing a
processing capability of channel estimation or a computing
capability of a matrix for beamforrning' in the beamformee even
when a beamformer and a beamf ormee are different f romeach other
in the number of antennas or the number of supported streams.
According to an embodiment of the invention, there is
provided a wireless communication system which performs data
transmission using spatially multiplexed streams froma first
terminal including N antennas to a second terminal including
M antennas (N is an integer of 2 or more and M is an integer
of 1 or more), the system including: notifying means for
"
notifying the first terminal of a maximum dimension N,,, at the
time of estimating a channel matrix of the second terminal (N,,,
is an integer of N or less); training means for transmitting
a sounding packet including training sequence for exciting a
channel corresponding to the maximum dimension N,,,,, from the
firstterminal tothe secondterminal; m a case ofM+M, channel
matrix estimation means for dividing the training sequence
received by the antennas of the second terminal into N, or
less streams and estimating the channel matrix having M rows
and N, or less columns; a beamforming information feedback
unit which prepares beamforming information necessary for
calculating a transmission weight matrix for beamforming in
the first terminal on the basis of the channel matrixestimated
in the second terminal and feeding back the beamforming
information from the second terminal to the first terminal;
transmission weight matrix setting means for setting the
transmission weight matrix for beamforming at the time of
transmitting data from the first terminal to the second
terminal on the basis of the beamforming information fed back
from the second terminal to the first terminal; and bearrrforming
means for performing beamforming in transmission signals of
the antennas of the first terminal using the transmission
weight matrix forbeamformingwhenadatapacket is transmitted
from the first terminal to the second terminal.
The term "system" described herein indicates a logical
set of apparatuses (or function modules for realizing specific
functions) and it is not specially considered whether the
apparatuses or the function modules are included in a single
casing (The same is true in the below description).
As a technology for realizing a high speed of wireless
communication, there is an MIMO communication method which
includes a plurality of antenna elements in a transmitter side
andareceiver side andrealizes spatiallymultiplexed streams .
In particular, in a closed loop type MIMO communication system,
a terminal of a data packet transmission side performs
beamforming on the basis of feedback of information on an
estimation channel matrix from a terminal of a reception side
such that a plurality of spatial orthogonal multiplexed
propagation channels which are logically independent are
realized and the receiver side can extract a plurality of
original signal sequence without crosstalk, thereby
theoretically accomplishing maximum performance.
As a procedure of performing feedback of the channel
matrix from the terminal of the reception side to the terminal
of thetransmisnionside, for example, twokinds of proa:edures,
that is, "implicit feedback" and "explicit feedback", are
defined in the EWC HT MAC specification. Among them, in the
explicit feedback, the first terminal which operates as a
beamformer performs the beamforming of a transmission packet
to perform communication using the transmission weight matrix
forbeamforming based on the channel information fedback from
the second terminal which operates as a beamformee. It is
possible to reduce aprocessingburdennecessaryforperforming
the beamforming in the beamformer.
However, when the beamformer and the beamformee are
different from each other in the number of antennas or the
number of supported streams, there are several problems at the
time of the beamforming. This is because the terminal having
a smaller number of antennas needs to perform channel
estimation, calculationoE the transmission weight matrix and
the multiplication of the transmission weigh matrix with the
dimension number larger than or equal to the number which is
considered at the time of designing.
In particular, in a case where the number N of antennas
of the beamformer is larger than the number M of antennas of
thebeamformee and the STA-B cancorrespondtoatmostMatreams,
the STA-B cannot obtain the estimation channel matrix in spite
that the sounding packet for exciting the spatial channel
matrix of N dimensions is transmitted fromthe STA-A, because
the STA-B can correspond to at most MxM estimatio~c~h annel
matrix.
Accordingly, in the wireless communication system
according to the embodiment of the invention, when the
beamforming is performed according to the explicit feedback,
the maximum dimension N,,, at the time of estimating the channel
matrixof the second terminal is notifiedto the first terminal
and the first terminal transmits the soundingpacket including
the training sequence for exciting the channel corresponding
to the maximum dimension &,. Here, "corresponding to the
maximum dimension N," does not indicate that the excited
spatial dimension is limited to N,,," . Generally, it indicates
that the second terminal estimates the channel in a format that
an estimation pro.cess of a dimension larger than MxN,,, is not
performed. As an example, the below-described staggered
sounding packet may be considered. In this case, the entire
exciting channel dimension is N ( >Nmax) , but the training
sequence are divided into a part for exciting the channel of
Nma, dimensions and a part for exciting the channel of N-ha,
dimensions such that the second terminal can perform channel
estimationusingonlythepartN,,,whichcanbeprocessed. That
is, even when the exciting channel space dimension is larger
than N,,,, the sounding packet which is considered such that
the channel estimation is possible in the . . processing capability
in consideration of the maximum dimension N,,, at the time of
estimating the channel matrix of the second terminal
corresponds to "corresponding to the maximum dimension N,".
That is, since the first terminal suppresses the number
of streams of the channel for exciting the sounding packet
according to the processing capability OF the second terminal
for estimating the channel matrix, the second terminal surely
receives the sounding packet in a range of its own maximum
dimension number. In this case, it is possible to reduce the
size of the channel matrix estimation circuit of the second
terminal as the beamf ormee and to realize low cost or low power
consumption of the apparatus. . '
The first terminal is designed to include the processing
capability corresponding to the number of its own streams and
includes the processing capability such as computation for
requestingthe transmissionweight matrix forbeamforming from
, .
the channel matrix of NxN dimensions or less or multiplication
of the transmissionvector and the transmission weight matrix
forbeamforming of NxNdimensions or less. Accordingly, when
the CSI information, that is, MxN,,, channel matrix, is fedback
fromthe second terminal, the first terminal can compute the
transmission weight matrix for beamforming. Alternatively,
even when the MxN,,, transmission weight matrix for beamforming
computed from the channel matrix lin the compressed or
uncompressed format) is fed back from the second terminal, the
transmission weight matrix for beamforming can be multiplied
with the trans~uissionv ector in the range of its own p:rocessing
capability and thus no problem is caused.
On a protocol according to the EWC specification, in a
predetermined management frame, a capability description
field fordescribing a possible maximum spatial dimensionwhen
abeamformee of explicit feedback receives apacket including
training sequence is defined. Accordingly, the notifying
means can notify the first terminal of the maximum dimension
N,,, at the time of estimating the channel matrix of the second
terminal using the management frame for describing the
capability description field. The management frame is, for
example, a type of transmission frame of the beacon which is
notified in a frame period, a measure pilot, an association
response and a re-association response which respond to the
request of association from the client terminal, a probe
response which responds to the request of BBS information from
the client terminal, or an association request and
re-association request for requesting network association by
the client terminal (or a communication station other than the
access point) and a probe request for requesting BBS
information to the access point. Accordingly, even when the
second terminal operates as any one of an access point and a
client terminal, the notifyingmeans canperformnotification.
The beamformer may include a signal for requesting the
CSI information in the sounding packet including the training
sequence for cxciting the channel. In particular, the CSI
informationmaybe requestedby specifyingthe feedbackmethod
received from the beamformee in the explicit feedback, in the
CSI/Steering f ieldprovided in the HT control field of the MAC
frame. Accordingly, the training means may include a request
signal forrequesting feedback of the channel information from
the first terminal tothe secondterminal inthe soundingpacket
for exciting the channel.
~ntheEWCspecification,a zero length frame (ZLF) (also
called a null data packet (NDP) and hereinafter referred to
as "ZLF") dedicatedtothe soundingpacket, whichincludes only
a PHY header part including the training sequence for exciting
the channel and does not include an MAC frame, is defined.
Since the ZLF does not have the MAC header, the CSI information
cannot be requested by the HT control field. In such a case,
the training means does not include the signal for requesting
the CSI information in the sounding packet and requests the
CSI information in the HT control field of a general packet
transmitted prior thereto.
It is possible to reduce an operation amount necessary
for estimating the channel matrix by transmitting the sounding
packet in a staggered format for temporally dividing a training
signal part used for a space division process of a data part
and a training signal for exciting a channel of a spatial
dimension larger than or equal to the number of streams of data
from the first terminal to the second terminal.
In particular, the first terminal excites the channel
of N,,, spatial dimensions in a training signal part used for
the space division process of the data part and allows a
training signal for exciting the channel of 69-Nm,, remaining
spatial dimensions to be not related to the space division of
the sibnal, with respect to the sounding packet.
In this case, when the second terminal receives the
sounding packet, the channel of N, spatial dimensions is
excited to estimateMxNma,channelmatrixinthe training signal
part used for the space division process of the data part, but
the training signal forexciting the channelof N-N,,,remaining
spatial dimensions does not need to be processed. Although
a part attached to the end of the training is not processed
in order to excite remaining N-Nu, channels, no problem is
caused in channel estimation or data symbol demodulation.
If the direct mapping for mapping one antenna branch to
each transmission stream is performed when the first terminal
transmits the sounding packet with the stream suppressed to
themaximumdimensionN,,,or less which is allowedby the second
terminal, the beamforming effect deteriorates. Since the
first terminal includes N antennas but does not use all the
antennas and the beamforming must be originally performed with
respect to the MxN channel matrixbut the dimension number is
suppressed to MxN,,,, a beam gain is reduced and thus a
transmissiondiversitygain is also reduced. The trcn~smission
power of N,,, antenna branches used for transmission increases
and distortion of the signal increases in a transmission end.
When the dimension number of the sounding packet is
suppressed, for example, the first terminal may perform
conversion for mapping N,,, spatial streams to all N
transmission antenna branches by a spatial expansion and
compensate the deterioration of the characteristics by the
transmission diversity. For example, it is possible to
perform the mapping to the transmission signals to all N
transmission antenna branches by multiplying the sounding
packet oE N,,, dimensions specified from the second terminal
by an NxN,,, mapping matrix.
Correlation between the transmission antenna branches
may not be sufficiently reduced by the multiplication of the
mapping matrix. Accordingly, the first terminal may give
different cyclic shift delay amounts to the transmission
antennabranches after themultiplicationof the mappingmatrix.
The undesired beamforming may be performed when identical or
similar signals are transmitted through different spatial
streams. However, it is possible to reduce the correlation
between the transmission antenna branches to reduce the
undesired directional characteristics. The cyclic shift
delay described herein is an operation for cutting out a part
of the time axis waveform of the OFDM symbol and fitting the
part to the opposite end (corresponding to phase rotation on
a frequency axis) (see Fig. 10) , which is different from the
simple delay of the transmission timing between the
transmission antenna branches.
When the first terminal performs the conversion for
mapping the N,,, spatial streams to all the N transmission
antenna branches, other mapping to the transmission antenna
branches may be performed in the subcarrier unit, as shown in
Fig. 11. In this case, since the correlation between the
transmission antenna branches is high in the subcarrier unit,
the cyclic shiftdelaymaybeusedtogether, as describedabove.
According to the embodiment of the invention, it is
possible to an excellent wireless communication system,
wireless communication apparatus and wireless communication
method, whichare capableofperformingcommunicationatahigh
transmissionratebyabeamformedpacketbyallowingaterminal
which operates a beamformer to suitably set a transmission
weight matrix on the basis of an estimation channel matrix fed
back from a terminal which operates as a beamformee.
According to the embodiment of the invention, it is
possibletoprovide anexcellent wireless communication system,
wireless communication apparatus and wireless communication
method, which are capable of suitably performing beamforming
by the explicit feedback without deteriorating beamforming
characteristics or increasing a processing capability of
channel estimation or a computing capability of a m,~trix for
beamforming in the beamformee even when a beamformer and a
beamformee are different from each other in the number of
antennas or the number of supported streams .
According to the wireless communication system of the
embodiment of the invention, it is possible to reduce
complexity of the circuit or power consumption of a
communication terminal which is the beamformee by suppressing
the dimension number of the sounding packet transmitted from
the beamformer according to the processing capability of the
beamformee when the explicit feedback for- Eeeding back CSI
information or transmission weight matrix for beamforming is
performed, even if the number of antennas of the beamformer
is larger than the number of antennas of the beamformee.
According to the embodiment of the invention, it is
possible to perform the beamforming while maintaining the
transmission diversity effect to some extent by mapping
transmission streams to all transmission antennabranches when
suppressing the dimension number of the sounding packet
transmitted from the beamformer.
The other objects, features and advantages of the
invention will become apparent and more readily appreciated
from the following description of the embodiments, taken in
conjunction with the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
Fig. 1Ais a schematic diagramof anoperationprocedure
(in a case where CSI information is fed back) of explicit
feedback according to an embodiment of the invention.
Fig. 1Bis a schematic diagramof anoperationprocedure
(in a case where a transmission weight matrix for beamforming
is fedback) ofthe explicit feedbackaccordingtoanembodiment
of the invention.
Fig. 2 is a view showing the configuration of a
transmitter side of a wireless communication apparatus which
can operate as a STA-A (or STA-B) in a. wireless communication
system shown in Figs. 1A and 1B.
Fig. 3 is aview showing the configurationof a receiver
side of thewireless communicationapparatuswhichcanoperate
as the STA-A (or STA-B) in the wireless communication system
shown in Figs. 1A and IB.
Fig. 4A is a view showing an example of a transmission
operation of a ZLPpacket.
Fig. 4B is a view showing an example of the transmission
operation of the ZLF packet.
Fig. 5 is a view showing a format example of a staggered
sounding packet.
Fig. 6 is a view showing a format example of the staggered
sounding packet.
Fig. 7 is a view showing a fonnat example of the staggered
sounding packet.
Fig. 8 is a view showing a fonnat example of the staggered
sounding packet.
Fig. 9 is aviewshowinga format exampleofthe staggered
sounding packet.
Fig. 10 is a view showing a state where a cyclic shift
delay is applied to an OFDM symbol.
Fig. 11isaviewshowinganexampleofpeb-farmingmapping
to transmission antennas in a subcarrier unit when the number
of antennas of the STA-A is N=3 and the number of antennas of
the STA-B is M-2.
Fig. 12 is a flowchart illustrating a pr. ocess when an
apparatus operates as a beamformer on the basis of the explicit
feedback procedure.
Fig. 13 is a flowchart illustrating a process when the
apparatus operates as the beamformer on the basis of the
explicit feedback procedure.
Fig. 14 is a view showing a state where a beamformee
estimates a channel matrix excited by a training signal
transmitted from a beamfosmer.
Fig. 15 is aview showing a frame exchange procedure for
transmitting beamforming from an access point to a client
terminal with the explicit feedback.
Fig. 16 is a view showing a scheme of a HT control field
of an MAC frame defined in the EWC specification.
Fig. 17 is aviewshowinga scheme of aCSIlsteeringfield
included in the HT control field.
Fig. 18 is a view showing a state where the explicit
feedback is performed with a CSI format.
Fig. 19 is a view showing a state where the explicit
feedback is performed with the CSI format.
Fig. 20 is a view showing a format of a HT capability
element
Fig. 21 is a view showing the configuration of a Tx
beamfarming capability field included in the I+?' capability
element.
Fig. 22 is a view showing a state where a transmission
weight matrix V for beamforming is fed back by the explicit
feedback.
Fig. 23 is a view showing the state where the transmission
weight matrix V for beamforming is fed back by the explicit
feedback.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
Hereinafter, embodiments of the invention will be
described in detail with reference to the accompanying
drawings.
A wireless communication system according to the
embodiment of the invention performs closed loop type MIMO
communication and more particularly, a terminal of a
transmitter side performs beamforming in order of p~rforming
feedback for a channel matrix, for example, in order of the
"explicit feedback" defined in the EWC HT MAC specification.
Intheexplicit feedback, abeamformerbeamforms atransmission
packet using a transmission weight matrix for beamforming
obtained on the basis of an estimation channel matrix fedback
from a beamformee so as to perform communication.
However, a processing capability for performing channel
estimation or a processing capability for computing a matrix
for beamforming is given to a terminal in consideration of the
number of its own . . antennas. ~ccordin~li~f, the number of
antennas of the beamformer is large, the bsa.mformee may not
divide the packet into spatial stream trainings to estimate
the channel matrix or obtain the matrix for beamforming from
the estimation channel matrix because the spatial dimension
number is large even when the beamformer transmits a sounding
packet to excite the channel.
Accordingly, in the wireless communication system
according to the embodiment of the invention, when the
beamforming is performed by the explicit feedback, a maximum
dimension N,,, of the beamformee at the time of the channel
matrix estimation is notified to the beamformer, and the
soundingpacket transmitted fromthebearnformeris excited such
that the spatial dimension number of the channel to be estimated
by the beamformee is suppressed to the maximum N,,, or less.
Accordingly, since the beamformee surely receives the sound
packet in a range of its own capability, the size of a channel
matrix estimation circuit can be reduced and low cost or low
power consumption of an apparatus ca6 be realized.
Fig. 1A is a schematic diagram of an operation procedure
of the explicit feedback according to the embodiment of the
invention. Here, the number of antennas of a STA-A as the
beamformer is 3 and the number of antennas of a STA-B as the
beamformee or the maximum spatial dimension number at the time
of computing the transmission weight matrix for beamforming
is 2 . The procedure is performed on the basis of the EWC MAC
specification.
The STA-B previously notifies the STA-A that the maximum
dimension number at the time of estimating the channel matrix
is 2. The STA-A excites the channel to a format in which the
spatial dimension of the channel to be estimated by the STA-B
is 2x2 and transmits the sounding packet to the STA-B using
two streams.
Since the STA-B receives the sounding packet in the range
of its owncapability, a 2x2 forward estimation channel matrix
can be easily generated. CSI infonrlation composed of
coefficients of the estimation channel matrix is prepared and
fed back to the STA-A using two streams.
Since the STA-A receives the fed-back CSI information
using three antennas, the information is received in a 3x2
spatial dimension, but is propertyprocessedbecause the STA-A
is designed to include a processing capability corresponding
to the number of its own streams. When the 2x2 channel matrix
is extracted from the CSI information, computation such as
singular value decomposition is performed such that a 2x2
transmission weight matrix V for beamforming can be easily
obtained.
The STA-A multiplies a two-dimensional transmission
vector by the 2x2 transmissionweight matrixV Eorbeamforming
to perform beamforming and transmits two transmission streams
in order to transmit a data packet. Alternatively, two
transmission streams are mapped to three antennas by space
expansion to perform the beamforming in a state where
transmission diversity effect is maintained to some extent.
Thereafter, a request of the sounding packet and the
channel estimation and the computation of the transmission
weight matrix for beamforming 'due to the reception of the
sounding packet are repeatedly performed whenever the STA-A
performs the beamforming.
Since the STA-B surely receives the sounding packet in
the range of its own capability, the size of the channel matrix
estimation circuit can be reduced and low cost or low power
. ,
consumption of the apparatus can be :realized.
Fig. 1B is a schematic diagram of an operation procedure
of the explicit feedback when the transmissionweight matrix
V for beamforming is fed back, instead of the CSI in.formation.
Here, the number of antennas of th.e STA-A is 3 and the number
of antennas of the STA-B and the maximum spatial dimension
number at the time of computing the transmission weight matrix
for beamf orming are 2.
The STA-B previously informs the STA-A that the maximum
dimensionnumber at the time of estimating the channel matrix
is 2. The STA-A excites the channel to a format in which the
spatial dimension of the channel to be estimated by the STA-B
is 2x2 and transmits the sounding packet to the STA-B using
two streams.
Since the STA-B receives the sounding paclcet in the range
of its own capability, a 2x2 forwardestimation channel matrix
can be easily generated. The 2x2 transmission weight matrix
V for beamforming is obtained by performing the computation
such as the singular value decomposition fromthe 2x2 channel
matrix and fed back to the STA-A using two streams.
Since the STA-A receives the f ed-back information using
three antennas, the information is received in a 3x2 spatial
dimension, but is property processed to extract the
transmission weight matrixV forbeamforming because the STA-A
is designed to include a processing capability corresponding
to the number of its own streams.
Since the spatial dimension number 2x2 of the
transmission weight matrix V for beamforming is in the range
of the capability of the STA-A having three antennas, the 2x2
transmission weight matrix V for beamforming can be easily
multiplied with the two-dimensional transmission vector, the
beamforming is performed and two transmission streams are
transmittedinordertotransmit adatapacket. ~lternatively,
two transmission streams are mapped to three antennas by space
expansion to perform the beamforming in a state where
transmission diversity effect is maintained to some extent.
Thereafter , the request of the sounding packet, and the
channel estimation and the computation of the transmission
weight matrix for beamforming due to the reception of the
sounding packet are repeatedly performed whenever the STA-A
performs the beamforming.
Since the STA-E! surely receives the sounding packet in
the range of its own capability, the size of the channel matrix
estimation circuit can be reduced and low cost or low power
consumption of the apparatus can be realized.
In the operation procedure shown in Figs. 1A and lB, the
STA-B needs to notify the STA-A of the maximum dimension number
2 at the time of estimating the channel matrix. In the EWC
specification, whenthe explicit feedback is applied, a scheme
of informing of information on the channel estimation maximum
dimension as one of the capability of the beamformee is
determined and can be used.
In the EWC specification, it is defined that the HT
terminal corresponding to the high-speed tr;\r~smission
transmits a HT capability element to declare that it is the
HT terminal. The HT terminal may include a HT capability field
in a predetermined management frame and declare any element
of a HT function by the HT capability element.
In the TxBF (transmit beamforming) capability field (see
Fig. 21) included in the format of the HT capability element,
any HT function of the beamforming (see Fig. 20) is specif ied.
The Tx beamforming capability field has 32 bits, but,
among them, lgth to 2oth bits are allocated to the CSI number
of beamformer antennae, 21St to 22M' bits are allocated to the
uncompressed steering matrix of beamformer antennae, and ~ 3 ' ~
to 24th bits are allocated to the compressed steering matrix
of beamformer antennae. In these fields, the spatial
dimension nunher of the sounding packet which can be received
fromthe beamformer when the beamformee performs the explicit
feedback with each format is described.
The HT capability element may be included in the
predetermined management frame. For example, when the STA-B
operates as the access point, the HT capability field maybe
included in a type of transmission frame of the beacon which
is notified in each frame period, a measure pilot, an
association response and a re-association response which
respond tothe request of association fromthe client terminal,
or a probe response which responds to the request of BBS
information from the client terminal such that the climension
number of the CSI information is notified to the STA-A which
participates in the network operated by the STA-B. When the
STA-B operates as the client terminal (or a communication
station other than the access point) , the HT capability field
may be included in a type of transmission frame of an
association request and re-association request for requesting
network association to the STA-Awhich operates as the access
rjoint and a probe request for requesting BBS information to
the access point. Accordingly, when the STA-B operates as the
access point or the client terminal, the maximum dimension
which is allowed in the CSI information may be notified to the
STA-B, by transmitting the HT capability element.
In the beamforming procedure shown in Figs. 1A and 1B,
the STA-A as the beamformer includes a signal for requesting
the CSI information in the sounding packet including training
sequence for exciting the channel. In particular, in a
CSI/Steering field provided in the HT control field (see Fig.
16) of the MAC frame, a feedback method received from the
beamformee can be specified in the explicit feedback (see Fig.
17) .
In the EWC specification, a zero length frame (ZLF)
dedicated to the sounding packet, which includes only a PHY
header part including the training sequence for excitihg the
channel and does not include anMAC frame, is defined. Since
the ZLF does nut have the MAC header, the CSI informat ion cannot
berequestedbytheHTcontrolfield. Insuchacase, the signal
for requesting the CSI information is not included in the
sounding packet and the CSI information is requested in the
HT control fieldof a general packet transmittedpriorthereto.
Fig. 4A shows an example of a transmission operation of
the ZLF packet. As shown, the ZLF packet is transmitted when
a short interframe space (SIFS) or a reduced inter frame space
(RIFS) elapses after a general data packet is transmitted. In
the NT control fisLd in the MAC header included in the general
data packet, the CSI request for the subsequent ZLF packet is
performed by specifying the GISI/Steesing field.
In an example shown in Fig. 4B, the STA-A requests of
the feedback of the CSI information in the data frame for
requesting an immediate response, but declares that the ZLF
is continuously transmitted therein. When the STA-B returns
anACKaccording to the immediate response, the STA-Atransmits
the ZLF when the SIFS elapses after the ACK is received.
Up to now, the STA-A which suppresses the spatial
dimension number of the channel excited according to the
processing capability of the STA-B and transmits the sounding
packet was described. As a packet transmission method for
suppressing the spatial dimension of the channel, there is a
staggered format.
The staggered packet has a training signal part used for
a space division process of a data part and a packet structure
for temporally dividing a training signal for exciting a
channel of the spatial dimension larger than or equal to the
number of streams of data and the receiver side reduces an
operation amount necessary for estimating the channel matrix.
The STA-A excites the channel of N,,, spatial dimensions
corresponding to the processing capability of the STA-B in a
training signal part used for the space division process of
the data part and allows a training signal for exciting the
channel of N-Nmax remaining spatial dimensions to be not related
to the space division of the signal, with respect to the
sounding packet. In this case, when the STA-E receives the
soundingpacket, the channelofN spatial dimensions is excited
to estimate MxN,,, channel matrix in the training signal part
used for the space division process of the data part, but the
training signal for exciting the channel of N-N,,, remaining
spatial dimensions does not need to be processed. Although
a part attached to the end of the training is not processed
in order to excite remaining N-N,, channels, no problem is
caused in channel estimation or data symbol demodulation.
Now, the procedure of the explicit feedback when the
sounding packet of the staggered format is used wi1.l be
described. For simplification of the description, although
a case where each stream is directly mapped to each antenna
branch will be described, the invention is not limited to the
case.
Fig. 5 shows a format example of the staggered sounding
packetwhenthebeamformeehavingthree antennas transmits data
of one stream.
A HT-STF (short training field) is a training symbol for
improving automatic gain control (AGC) in the MIMO system,
which includes QPSK-modulated OFDM signals of 52 tones. A
HT-LTF (long training field) is a training symbol for
performing the channel estimation for each input signal which
is spatially modulated in the receiver side, which includes
BPSK-modulate4 OFDM signals of 56 tones. These are training
signals defined in a HT mode of the EWC specification. A value
of -400nsec which is described in the HT-LTF simultaneously
transmitted froma third antenna is a cyclic shift delay amount
which is provided in order to avoid unintended beamf orming when
identical or similar signals are transmittedthroughdifferent
spatial streams, which shifts and connects a time axis
wavelength of an OFDM symbol sent from the third antenna by
-400 nanoseconds.
In the example shown in Fig. 5, one stream is transmitted
with a format having a data stream, but, with a temporal
separation therefrom, training signals for exciting the
channel of the remaining spatial dimensionare transmitted from
the other two antennas which are not used for the space division
proeess of the data part.
Fig. 6 shows a format example of the staggered sounding
packet when data of one stream is transmitted from the
beamformee having four antennas. In the shown example, one
stream is transmitted with the format having the data stream
and, witha temporal separation therefrom, the training signals
for exciting the channel of the remaining spatial dimension
are transmitted from the other three antennas which are not
used for the space division process of the data part. In the
current EWC specification, it is defined that four I-IT-LTFs are
used in the training of three streams.
Fig. 7 shows a format example of the staggered sounding
packet when data of two streams is transmitted from the
beamformee having three antennas. In the shown example, two
streams are transmittedwith the formathaving the data stream
and, witha temporal ~e~arationtherefromt:h e training signals
for exciting the channel of the remaining spatial dimension
are transmitted from the other one antenna which is not used
for the space division process of the data part.
Fig. 8 shows a format example of the staggered sounding
packet when data of two streams is transmitted from the
beamformee having four antennas. In the shown example, two
streams are transmitted with the format having the data stream
and, withatemporal separationtherefrom, the trainingsignals
for exciting the channel of the remaining spatial dimension
are transmitted from the other two antennas which are not used
for the space division process of the data part.
Fig. 9 shows a format example of the staggered sounding
packet when data of three streams is transmitted from the
beamformee having four antennas. In the shown example, three
streams are transmittedwith the format having the data stream
and, witha temporal separationtherefrom, the training signals
for exciting the channel of the remaining spatial dimension
are transmitted from the other one antema which is not used
forthe space divisionprocess of the datapart. Inthe current
EWC specification, it is defined that four HT-LTFs are used
in the training of three streams.
As can be seen from Figs. 5 to 9, in a wireless
communication apparatus inwhich the number of antennas is two
and the maximum number of estimatable streams is two, the
reception of the data part (payload) of the packet and the
estimation of a necessary channel matrix are in the processing
capability range which is considered upon designing, when the
staggered sounding packet has the structure shown in Fig. 5,
7 or 8 . Pig. 6 shows the staggeredsoundingpacketofone stream,
which is not suitably applied to the invention.
Inawireless communicationapparatus inwhich the number
of antennas is three and the maximum number of estimatable
streams is three, the reception of -the staggered sounding
packet shown in Figs. 5 to 9 and the estimation of a necessary
channel matrix are in the processing capability range which
is considered upon designing. In a wireless comnvunication
apparatus in which the maximum number of streams is three, the
speciEication in which four HT-LTFs are received and the
channel estimation of three streams is performed therefrom is
originally requested and no problem is caused in the structure
of the apparatus.
As can be seen from Fig. 5 to 9, when the number N of
antennas in a initiator of the sounding packet (that is, the
terminal which operates the beamformer in the explicit
feedback) is larger than the number M of antennas in a receiver
of the sounding packet (that is, the terminal which operates
as the beamformee in the explicit feedback), the beamformee
can selectively estimate only necessary M streams without
performing the channel estimation of N streams (that is,
without preparing an MxN channel matrix), by suitably using
the staggered format.
If the beamformer includes four antennas and the
beamformee includes two antennas, a circuit burden of the
beamformer may not be reduced although the staggered sounding
packet of the frame format shown in Fig. 6 is used. No problem
is caused in the channel estimation from training (HTLTF) of
a first stream, but, in order to estimate the channel with
respect to one stream, four HT-LTFs of the other three streams
which are not used for the space division process of the data
part transmitted with a temporal separation therefrom needs
tobe computed. Thus, the sizeof the circuitofthebeamformee
which can support at most two streams increases.
When the beamformer includes three antennas and the
beamformee includes two antennas, the staggered sounding
packet of the frame format shown in Fig. 5 or 7 is used.
When the sounding packet shown in Fig. 7 is transmitted
from the beamformer, the beamformee can estimate the channel
of two spatial dimensions necessary for the beamforming using
the training signal part in a reception stream of the first
to second antermas used for the space division of the data part.
Since the reception streams of the third to fourth antennas
which are not used for the space division of the data part
transmitted with temporal separation do not need to be
processed, the problem that the size of the circuit increases
in the beamformee which can support at most two streams is not
caused. No problem is caused in the channel estimation or the
data symbol demodulation although the part attached to the end
of the training is not processed in order to excite the third
to fourth channels.
When the sounding packet shown in Fig. 5 is transmitted
fromthebeamforrner, thebeamformee firstestimates the channel
using the training signal part in a reception stream of the
first antennas used for the space division of the data part.
However, two HT-LTF in each of the other two streams, which
are not used forthe space divisionof the datapart transmitted
with the temporal separation, needs to be processed, in order
toestimate the channel withrespect toone stream. The channel
estimation of a 2x2 matrix may be performed from two HT-LTFs
and the channel estimation of the other one stream may be
performed. However, in this case, since the channel
estimation result of a first stream needs to be buffered in
another place, the size of the circuit slightly increases
compared with the case shown in Fig. 7 which the buffer is not
needed.
As another example, when the beamformer includes four
antennas and the beamformee includes three antennas, the
staggered sounding packet of the frame format shown in Fig.
6, 8 or 9 is used.
When the sounding packet of the frame format shown in
Figs. 6 and 8 is used, as described above, the charmel
estimation is performed without a problem, but there is a
problem that the channel estimation result of first one or two
streamneeds tobe bufferedinanotherplace . When the sounding
packet of the frame format shown in Fig. 9 is used, the
beamformee can estimate the channel of two spatial dimensions
necessary for the beamforming using the training signal part
in a reception,streamof the first to third antennas used for
the space division of the data part. Since the reception stream
of the fourth antenna which is not used for the space division
of the datapart transmittedwiththe temporal separation does
not need to be processed, the problem that the size of the
circuit increases in the beamformee which can support at most
three streams is not caused. No problem is caused in the
channel estimation or the data symbol demodulation although
the part attached to the end of the training is not processed
in order to excite the fourth channel.
However, if the direct mapping for mapping one antenna
branch to each transmission stream is performed when the STA-A
transmits the sounding packet with the stream suppressed to
the rnaximum dimension N,,, or less which is allowed by the STA-B,
thebearnformingeffectdeteriorates . Since the STA-Aincludes
N antennas but does not use all the antennas and the beamf orming
must be originally performed with respect to the MxN channel
matrix but the dimension number is suppressed to MxN,,,, a beam
gain is reduced and thus a transmission diversity gain is also
reduced. The transmission power of N,,, antenna branches used
for transmission incrpases and distortion of the signal
increases in a transmission end.
Withrespect to suchaproblem, whenthe dimensionnumber
of the sounding packet is suppressed, for example, the STA-A
may perform conversion for mapping N,,, spatial streams to all
N transmission antenna branches by a spatial expansion and
compensate the deterioration of the characteristics by the
transmission diversity. For example, it is possible to
perform the mapping of to allN transmission antenna branches
bymultiplyingthe soundingpacket of Nm,,dimensions specified
from the STA-B by an NxN,,, mapping matrix.
For example, when the number of antenna of the STA-A is
N=3 andthenumberof antennasofthe STA-BisM=2, thedimension
number is suppressed to the two-dimensional transmission
stream (sL, sz) in order to transmit the sounding packet from
the STA-A, but the transmission signal can be, for example,
mapped to three transmission antennas by a mapping matrix E
described below.
That is, in the STA-A, a three-dimensional transmission
vector is obtainedby the two-dimensional transmission stream
(sl, s2) by the 3x2 mapping matrix E. The STA-B receives the
two-dimensional stream (rl, r2) by propagating the channel
composed of 2x3 dimension shown Equation 2.
When the stream conversion is performed at the time of
transmitting the sounding packet, the same mapping matrix E
needs to be multiplied even when the beamforming is performed
to transmit the data stream.
When the transmission streams are mapped to the
transmission branches using the mapping matrix, there is a
typical problein that undesired directional characteristics
occur due to strong correlation between the signals when the
signals are transmitted from the antennas. In order to avoid
this problem, it is preferable that an orthogonal matrix is
used such that the correlationbetween the antelma transmission
signals is reduced, if possible.
For example, when two streams are mapped to four
transmission antenna branches, the following mapping matrix
is multiplied.
As a non-orthogonal example, when two streams are mapped to
three transmission antenna branches, the following mapping
matrix is multiplied.
The correlation between the transmission antenna
branches may not be sufficiently reduced by only the
multiplicationof themappingmatrix. Inthe STA-A, different
cyclic shift delay amounts may be given to the transmission
antennabranches afterthemultiplicationof themappingmatrix.
When identical or similar signals are transmitted through
different spatial streams, an unintended beam may be formed,
but the correlation between the transmission antenna branches
can be reduced by giving the different cyclic shift delay
amounts , thereby reducing undesired directional
characteristics.
For example, when a valid symbol length of an OFDM symbol
is about 3.2 microseconds and a guard interval is about 800
microseconds, the cyclic shift delay amounts, that is, 0
nanoseconds, 50 nanoseconds, 100 nanoseconds and 150
nanoseconds, are given to four transmission antenna branches
such that the correlationbetween the transmission signals from
the antennas can be reduced to reduce occurrence of the
directional characteristics. In this case, the cyclic shift
delay amounts may be described as described below.
0 0 0
exp(- j2zkAF50nsec) 0 0
0 exp(- j2rrk~, 100n sec) 0
0 0 exp(- j 2 d F ISOnsec
In the above equation, AF is a subcarrier interval and
kis a serial numberof the subcarrier. It is possible to simply
realize mapping from the streams to the transmission antenna
branches by multiplying the above-described mapping matrix E
by such a matrix from the left (corresponding to the procedure
of giving the cyclic shift delay after the multiplication of
the mapping matrix) , and it is possible to reduce the
correlation between the transmission signals of the antennas
to reduce the undesired directional characteristics. Such a
conversionscheme is called "spatial expansion" definedinthe
EWC specification.
The cyclic shift delay described herein is an operation
for cutting out a part of the time axis waveform of the OE'DM
symbol and fitting the part to the opposite end (corresponding
to phase rotation on a frequency axis) (see Fig. lo), which
is different fromthe simple delay of the transmission timing
between the transmission antenna branches.
In the STA-A, when the conversion for mapping the N,
spatial streams to all the N transmission antenna branches,
other mapping to the transmission antenna branches may be
performed in the subcarrier unit. Fig. 11 is a view showing
an example of performing mapping to transmission ant.ennas in
a subcarrier unit when the number of antennas of the STA-A is
N=3 and the number of antennas of the STA-B is M=2. In this
case, since the correlationbetween the transmissionantenna
branches is highinthe subcarrierunit, the cyclic shiftdelay
may be used together, as described above,
Figs. 2 and 3 show the configurations of the transmitter
and the receiver of awireless communicationapparatus which
can operate as the STA-A (or the STA-B) in the wireless
communication system shown in Fig. I, respectively. The
number of antennas of the STA-A is N (the number of antennas
of the STA-B is M) and N (or M) is at most four, for example,
on the basis of the IEEE specification, but only two antennas
are shown in the figures in order to avoid conflict of the
figures.
Transmission data supplied to a data generator 100 is
scrambledby a scrambler 102 . Subsequently, error correction
encoding is performed by an encoder 104. For example, in the
EWC HT PHY specification, the scrambling and encoding methods
are defined according to the definition of the IEEE 802.11a.
The encoded signal is input to a data division unit 106 to be
divided into the transmission streams.
In a case where the apparatus operates as the beamformer,
the data generator 100 generates anMAC frame for describing
the request of CSI information when performing the explicit
feedback. In a case where the apparatus operate:; as the
beamformee, a channel matrix estimation unit 216a of the
receiver constructs a data frame including the CSI information
on the basis of the estimated channel matrix, in response to
the reception of the CSI information request. Alternatively,
compressedor uncompressed data frame composed of transmission
weighmatrix coefficients for beamforming calculated fromthe
estimation channel matrix may be constructed.
In eachtransrnission stream, the transmission signal is
punctured by a puncture 108 according to a data rate applied
to each stream, interleaved by an interleaver 110, mapped to
an IQ signal space by a mapper 112 , thereby becoming a conjugate
baseband signal. In the EWC HT PHY specification, an
interleaving scheme expands the definitionof the IEEE 802.1la
such that the same interleaving is not performed among a
pluralityof streams. As themapping scheme, BPSK, QPSK, 16QAM
or 64QAM is applied according to the IEEE 802.11a.
A selector 111 inserts the training sequence into the
transmission signal of each interleaved spatial stream at an
adequate timingandsupplies it tothemapper112 . The training
sequence include the HT-STF for improving the AGC in the MIMO
system and the HT-LTF for performing the channel estimation
for each input signal which is spatially modulated in the
receiver side. For example, in the HT-LTF, the training
sequence of each the transmission stream is inserted with the
staggered format.
When the beamforming is performed with respect to the
transmission signal, in a spatial multiplexer 114, a
beamforming transmission weight matrix computation unit 114a
calculates the transmission weight matrix V for beamforming
from the channel matrix H using a computation method such as
the singular value decomposition and a transmission weight
matrix multiplication unit 114b multiplies the transmission
vector having the transmission streams as the element by the
transmission weight matrix V set by the transmission weight
matrix setting unit 114a, thereby performing the beamf arming.
In order to transmit the sounding packet, the beamforming is
not performed with respect to the transmission signal.
Whenthe CSI information is fedback fromthe beamformee,
the transmission weight matrix setting unit 114a calculates
the transmission weight matrix V for beamforming on the basis
of the CSI information and sets it to the transmission weight
matrix multiplication unit 114b. When the compressed or
uncompressed transmission weight matrix V for beamforming is
fed back from the beamformee, it is set to the transmission
weight matrix multiplication unit 114b without change.
An inverse fast Fourier transform unit (IFFT) 115
converts the subcarriers arranged in a frequency region into
a time axis signal.
A stream number adjustment unit 116 adjusts the number
of transmission streams to the maximum dimension N,,,, or less
which is received from the STA-B as the beamformee. When the
direct mapping 4s performed, from the problem that the
beamforming effect deteriorates, the transmission signals may
be mapped to all the N transmission antenna branches by
multiplication of the NxN,,, mapping matrix and the
deterioration of the characteristics maybe compensatedby the
transmission diversity. Occurrence of the undesired
directional characteristics may be reduced by giving the
different cyclic shift delay amounts to the transmission
branches. The stream number adjustment unit 116 may be
realized by first multiplying the transmission vector by the
mapping matrix and then multiplying it by a matrix for cyclic
shift delay.
A guard insertion unit 118 adds a guard interval. A
digital filter 120 performs band limitation, a DA converter
(DAC) 122 converts it into an analog signal, and an RF unit
134 up-converts the analog signal to an adequate Eseguencyband
and transmits it to the channel through each transmission . .
antenna.
Meanwhile, the data which reaches the receiver through
the channel is analog-processed in an RF unit 228, converted
into a digital signal by an AD converter (ADC) 226, and input
to a digital filtex 224, in each reception antenna branch.
Subsequently, a synchronization circuit 222 performs
processes including packet detection, timing detection and
frequency offset correction and a guard removing unit 220
removes the guard interval added to the top of the data
transmission section. The fast Fourier transform unit (FFT)
, .
218 transforms a time axis signal to a frequency axis signal.
A space division unit 216 performs a space division
process of the spatially multiplexed reception signal. In
particular, a channel matrix estimation unit 216a divides the
spatial stream training included in the PHY header of the
sounding packet and constructs an estimation channel matrix
H from the training sequence.
An antenna reception weight matrix computationunit 216b
calculates an antenna reception weight matrix W on the basis
of the channel matrix H obtained by the channel matrix
estimation unit 216a. In a case where the beamforming is
performed with respect to the reception packet and the
estimation channel matrix is subjected to the singular value
decomposition (see Equation 3) , the estimation channel matrix
becomes equal to an UD and the antenna reception weight W is
calculated therefrom. A method of calculating the antenna
reception weight W is not limited to the singular value
decomposition and a calculation method such as zero forcing
or MMSE may be used. An antenna reception weight matrix
multiplication unit 216c multiplies the reception vector
having the reception streams as the element by the antenna
reception weight matrix W to perform spatial decoding of the
spatial multiplexed signal, thereby obtaining independent
signal sequence for each stream.
In the explicit feedback, when the apparatus operates
as the beamformee, the CSI information is constructed from the
estimation channel matrix H obtained by the channel matrix
estimation unit 216a and fed back from the transmitter side
to the beamf ormer as the transmission data. When the
transmissionweight matrixV forbeamforming is requested from
the beamformer instead of the CSI information, the matrix V
obtained by performing the singular value decomposition with
respect to the estimation channel matrix H in the antenna
reception weight matrix computation unit 216b is fed back.
A channel equalization circuit 214 performs remaining
frequency offset correction and channel tracking with respect
to the signal sequence of each stream. A demapper 212 demaps
the reception signal on the IQ signal space, a deinterleaver
210 performs deinterleaving, and a depuncture 208 performs
depuncturing at a predetermined data rate.
A data synthesis unit 206 synthesizes a plurality of
reception streams to one stream. This data synthesis process
performs an operation which is opposed to the data division
performed in the transmitter side. A decoder 204 performs
error correction decoding, a descrambler 202 performs
descrarnbling, and a data acquiring unit 200 acquires the
reception data.
When the apparatus operates as the beamformer, the CSI
information acquired by the data acquiring unit 200 or the
compressed/uncompressed transmission weight matrix V for
beamforming is sent to the transmission weight matrix setting
unit 114a of the transmitter side when the explicit feedback
is performed.
When the wireless communication apparatus operates as
the terminal of the data transmission side, that is, the
beamformee, in the closed loop type MZMQ communication, the
beamformeenatifi,esthebeamformerof its ownallowablemaximum
dimension N,,,,,, and the spatial dimension number of the channel
excitedby the soundingpacket transmitted fromthe beamformer
is suppressed to the maximum dimension N,,,, or less . Accordingly,
since the beamformee surely receives the sounding packet in
the range of its own capability, it is possible to reduce the
size of the circuit of the channel matrix estimation unit 216a
and to realize low cost or low power consumption of the
apparatus.
Fig. 12 is a flowchart illustrating a process when the
wireless communication apparatus shown in Figs. 2 and 3
operates as the initiator, that is, thebeamformer, onthebasis
of the explicit feedback procedure. Here, it is assumed that
the number of antennas of the beamformer is N and the number
of antennas of the beamformee is M.
First, iiotification of the maximum spatial dimension
number N, is received from the receiver which operates as the
beamformee (step S1) . Hereinafter, it is assumed that N,,,=M.
~henotificationisperformedbyreceivingthemanagementf rame
such as the message of network association or the signal
including the KT capability field.
Subsequently, the spatial dimension number of the
excited channel is suppressed to MxM and the sounding packet
is transmitted to the STA-B using two streams to obtain the
feedback of the transmission weight matrix V for beamforming
or the CSZ information (step 52) .
Slnce the channel of M spatial dimensions is excited in
the training signal part of the sounding packet such that the
beamformee receives it using M antennas, it is possible to
estimate an MxM channel matrix. According to the CSI
information request, the CSI information is prepared on the
basis of the estimation channel matrix and a packet in which
the information is carried in a data part returns to the
beamformer. Alternatively, when the feedback of the
transmissionweight matrixV forbeamforming is requested, the
estimation channel matrix is subjected to the singular value
decomposition and a packet in which the coefficient data of
the transmission weight matrix V for beamforming is included
in the compressed or uncompressed format returns to the
beamformer.
When the CSI information is received, the hcamformer
constructs the channel matrix (step 53) and obtains the
transmission weight matrix for beamforming at the time of
forward data transmission (step 54) . Alternatively, the
transmission weight matrix V for beamfosming may be received
in the step S3 and the step S4 may be skipped.
The beamforming is performed using the transmission
weight matrix forbeamforming inthe transmissionvector having
the transmission signal from the antennas as the element and
the data packet is transmitted to the receiver (step S5) . It
is possible to make an ideal spatial orthogonal channel by
applying the transmission antenna weight on the basis oE the
channel matrix and performing adequate beamforming which is
directed to the receiver.
Fig. 13 a flowchart illustrating a process when the
wireless communication apparatus shown in Figs. 2 and 3
operatesasthereceiver, that is, thebeamformee, on thebasis
of the explicit feedbackprocedure. Here, it is assumedthat
the number of antennas of the beamformer is N and the number
of antennas of the beamformee is M.
First, the maximum spatial dimension number N, of the
sounding packet is notified to the initiator which operates
as thebeamformer (step $11) . Hereinafter, it is assumedthat
N,,,=M. The notification is performed by receiving the
management frame such as the message of network association
or the beacon including the HT capability field.
Subsequently, when the sounding packet is transmitted
from the beamformer, the channel of M spatial dimensions is
excited. The beamformee receives it using M antennas (step .
512) and estimates the MxM channel matrix (step 513)
The CSI information is prepared from the estimation
channel matrix and the packet in which it is included in the
data part returns to the beamformer (step 514).
Alternatively, when the feedback of . . the transmission
weight matrix V for beamforming is requested as the sounding
packet., in the step S13, the estimation channel matrix is
subjected to the singular value decomposition to obtain the
transmission weight matrix V for beamf orming . In the step 514 ,
the packet in which the coefficient data of the transmission
weight matrix V for beamforming is included in the compressed
or uncompressed format instead of the CSI information returns
to the beamformer.
Since thebeamformee surelyreceives the soundingpacket
in the range of its own capability, it is possible to reduce
the size of the circuit of the channel matrix estimation unit
216a and to realize the low cost or low power consumption of
the apparatus.
Although the invention will be described in detail with
reference to the specific embodiment, it is apparent to those
skilled in the art that the embodiment may be modified or
substitutedwithout departing fromthe scope of the invention.
Although the embodiment in which the invention applies
to the MIMO communication system according to the EWC
specification in the IEEE 802.11n is described in the present
specification, the scope of the invention is not limited to
the embodiment. As the MIMO communication system using the
stream which is spatially multiplexed from a first terminal
includingN antennas to a second terminal includingMantennas,
it is possible to suitably apply the invention to various types
of communication systems in which the beamEormer performs the
beamforming using the channel information fed back from the
beanlf ormee .
Although the embodiment which is applied to the IEEE
802.11n which is extension standard of the IEEE 802.11 is
described in the present specification, the invention is not
limited to the embodiment. The invention is applicable to a
variety of wireless communication systems using an MIMO
communication method such as a mobile WiMax (Worldwide
interoperability forMicrowave) basedon the IEEE 802 .16e, the
IEEE 802.20 which is a high-speed wireless communication
standard for a mobile object, the IEEE 802.15.3~w hich is a
high-speed wireless PAN (Personal Area Network) using 60 M z
(milliwave) band, a wireless HD (High Definition) which
transmitting an uncompressed HD image using wireless
transmission of 60 GHz (milliwave) band, and a fourth
generat ion (4G) mobile telephone.
The invention is disclosed as an exemplary aspect and
the contents described in the present specification are
restrictivelyanalyzed. The scope of the invention is defined
by claims.
It should be understood by those skilled in the art that
various modifications, combinations, sub-combinations and
alterations may occur depending on design requirements and
other factors insofar as they are within the scope of the
appended claims or the equivalents thereof,

Claims:
We Claim:

1. An electronic device comprising:
circuitry configured to:
exchange capability information with another electronic device, the capability information including at least a first field, a second field and a third field; and
perform wireless communication with the another electronic device based on the capability information,
wherein the first field indicates a capability of performing a transmission beamforming, the second field indicates capabilities of a beamformee for receiving a sounding packet, and the third field indicates a maximum number of space time streams that is corresponding to a number of columns of a MIMO (multiple-input multiple-output) channel matrix for the wireless communication.

2. The electronic device of claim 1, wherein the capabilities of the beamformee includes dimension information for the sounding packet based on antenna information of the beamformee.

3. The electronic device of claim 2, wherein the sounding packet being generated based at least in part on information included in the third field.

4. The electronic device of claim 3, wherein a dimension of the sounding packet is equal to or less than a number indicated by the maximum number of space time streams.

5. The electronic device of claim 4, wherein the capability information is present in a management frame.

6. The electronic device of claim 5, wherein the management frame includes: a beacon signal, a measure pilot, an association response, a re-association response, and/or a probe response.

7. The electronic device of claim 6, wherein the capability information is defined as a beamforming capability field in IEEE802.11 standard.

8. The electronic device of claim 7, wherein the beamforming capability field is included in HT capability field defined by the IEEE802.11 standard.

9. The electronic device of claim 8, further comprising: a plurality of antennas configured to transmit or receive signals.

10. The electronic device of claim 9, wherein a number of the plurality of antennas is two.

11. The electronic device of claim 9, wherein a number of the plurality of antennas is three.

12. The electronic device of claim 4, the wireless communication is beamforming communication established based on channel estimation using the sounding packet.

13. The electronic device of claim 12, the beamforming communication is corresponding to an Explicit beamforming.