Patent Grant
7486655
This application claims the priority benefit of Taiwan application serial no. 94136284, filed on Oct. 18, 2005. All disclosure of the Taiwan application is incorporated herein by reference.
1. Field of Invention
The present invention relates to an orthogonal-frequency-division-multiplexing wireless system (OFDM wireless system), and particularly to an antenna-array-based multiple-input multiple-output orthogonal-frequency-division-multiplexing (MIMO-OFDM) system. The MIMO-OFDM system uses QR decomposition of the MIMO channel matrix to parameterize the channel state information (CSI).
2. Description of the Related Art
The wireless mobile communication today is capable of carrying mega transmission data and this has become a standard requirement in the modern society. To more effectively increase the channel capacity thereof, today's wireless mobile communication employs a so-called antenna-array-based MIMO-OFDM technology. In a MIMO-OFDM system with a closed loop, a spatial vector-coding scheme is used at the transmitter thereof, which can largely increase the channel capacity. However, the effectiveness of the spatial vector-coding scheme depends on the correctness of the estimated MIMO channel state information (CSI) at the receiver of the system. Therefore, the parameters and information fed back from the receiver to the transmitter must be sufficient for the transmitter to reproduce the CSI, even more so when the number of the transmitting antenna is larger than the number of the receiving antenna.
Among the currently available spatial vector-coding schemes, a preferred solution provides the maximum mutual information of the MIMO channels by means of singular value decomposition (SVD) of the MIMO channel matrix H for each frequency band to parameterize the CSI. The SVD can be expressed by:
And, by implementing a Givens rotation on the V matrixes of all the frequency bands for producing, a plurality of Givens rotation matrixes Gp,q(θ,φ) can be generated. Such transformation can be expressed by:
Afterwards, the Givens rotation matrix information of all the frequency bands are fed back to the transmitter, where the Givens rotation matrixes are combined for re-obtaining the V matrixes of all the frequency bands. Then, the V matrixes of all the frequency bands are applied to the vector coding at the transmitter. In the above-described solution scheme, the right/left singular vectors of the MIMO channel matrixes are used as the transmit/receive weighting vectors, and a water-filing power allocation is used to establish the optimum spatial multiplexing system.
Obviously, the above-described scheme has the such disadvantages as too much computation, high complexity and excessive feedback data or feedback rate. It is because the receiver needs to conduct computations to obtain the V matrixes from the MIMO channel matrixes H and then to conduct computations to obtain the Givens rotation matrixes from the V matrixes for each frequency band, so that the computation and high complexity are inevitable. In addition, the Givens rotation matrix information of all the frequency bands is required to be fed back, therefore the feedback information is proportional to the number of frequency bands. For 512 frequency bands, a typical number of frequency bands, the disadvantage of excessive feedback data or feedback rate is also inevitable. For example, if the channel matrix H has a dimension of 2×4, 62 complex numbers are needed to conduct a multiplication for each frequency band and at least 10 real numbers are required to be fed back, then the total feedback information reaches as high as 5,120 real numbers.
The present invention provides an antenna-array-based multiple-input multiple-output orthogonal-frequency-division-multiplexing (MIMO-OFDM) system. The MIMO-OFDM system can avoid excessive computation, high complexity and excessive feedback data or feedback rate in the prior art.
The present invention provides a feedback method in the antenna-array-based MIMO-OFDM system. The method is capable of reducing the complexity, lowering the computation, freely controlling the feedback data amount and responding to the feedback rate requirement.
The present invention provides a pre-coding method in the antenna-array-based MIMO-OFDM system. The method is able to approach an ideal channel capacity and an acceptable mutual information loss while reducing the feedback data.
The present invention provides an antenna-array-based MIMO-OFDM system. The receiver of the system includes MR receiving antennas, MR frequency band demodulators, a channel estimator & feedback parameter generator and N detectors. The transmitter of the system includes a vector-coding matrix generator, N coders, MT frequency band modulators and MT transmitting antennas. Wherein, MR, MT. N and n are natural numbers and 0<n≦N.
In the receiver provided by the present invention, the above-described MR frequency band demodulators are coupled to the MR receiving antennas. The frequency band demodulators demodulate the signals received by the receiving antennas, thus each frequency band demodulator obtains N tones of N different frequency bands. The above-described channel estimator & feedback parameter generator is coupled to the frequency band demodulators and according to the outputs from the frequency band demodulators, estimates the obtained channel state information (CSI) of the N different frequency bands. Further, the channel estimator & feedback parameter generator produces feedback parameters according to the CSI and feeds back the feedback parameters to the transmitter. The above-described N detectors are coupled to the frequency band demodulators and the channel estimator & feedback parameter generator, and according to the CSI, decode the outputs from the frequency band demodulators to obtain N receiving data vectors, wherein each single detector receives the tones with a same frequency band output from all the frequency band demodulators.
In the transmitter of the system provided by the present invention, the above-described vector-coding matrix generator, according to the feedback parameters, produces N spatial vector-coding matrixes W(n). That is, H(n) is a MIMO channel matrix of the n-th frequency band, and the QR decomposition of H(n) can be expressed by:
H(n)=R(n)Q*(n),
wherein R(n) is a lower triangular matrix with a dimension of MR×MT, Q(n) is an orthonormal matrix with a dimension of MT×MT and Q*(n) represents the complex conjugate transpose matrix of Q(n). When Q(n)=[q1(n) q2(n) . . . qMT(n)], qm(n) represents the m-th column vector in the matrix Q(n), thus:
W(n)=[q1(n)q2(n) . . . qM(n)],
wherein M and m are natural numbers and 0<M≦MT. The above-described N coders are coupled to the vector-coding matrix generator and receive N transmission data vectors. Each transmission data vector has a dimension of M×1. According to the vector-coding matrixes, the coders conduct coding on the transmission data vectors and each coder outputs MT tones with a same frequency band but different from the other coders. The above-described MT frequency band modulators are coupled to the coders and the transmitting antennas and conduct modulating on the outputs from the coders, then output the modulated signals at the transmitting antennas, wherein each frequency band modulator receives the tones of N different frequency bands output from all the coders.
According to the MIMO-OFDM system described in the embodiment of the present invention, the above-described channel estimator & feedback parameter generator, according to the CSI, implements Givens rotations on the MIMO channel matrixes H(n) to generate a plurality of Givens rotation matrixes Gp,q(θ,φ) and produces feedback parameters according to the θ and φ of the Givens rotation matrixes. In the embodiment, the channel estimator & feedback parameter generator according to the Givens rotation matrixes Gp,q(θ,φ), uses a preferred method, i.e., a delta modulation on θ and φ, to produce the feedback parameters. In particular, the channel estimator & feedback parameter generator uses a delta modulation with a one-bit quantizer or a multi-bit quantizer on θ and φ to produce the feedback parameters.
According to the MIMO-OFDM system described in the embodiment of the present invention, the channel estimator & feedback parameter generator at the receiver, according to the MIMO channel matrixes of a part of frequency bands, only produces the Givens rotation matrixes Gp,q(θ,φ) of the part of frequency bands. The channel estimator & feedback parameter generator, according to θ and φ of the Givens rotation matrixes of the part of frequency bands, further produces the feedback parameters. That is, only the information of a part of the frequency bands is contained in the produced feedback parameters. The vector-coding matrix generator at the transmitter, according to the produced feedback parameters, would generate θ(n),φ(n) of the part of the frequency bands. Afterwards, an interpolation method is used to generate θ(n),φ(n) of all the frequency bands, and further to generate all spatial vector-coding matrixes.
The MIMO-OFDM system provided by the present invention is applicable to a system with a receiving antenna number smaller than the transmitting antenna number; that is, MR<MT. In particular, when MR<<MT, the effect of the present invention can be best proven.
The present invention provides a feedback method in the MIMO-OFDM system, wherein the system includes MR receiving antennas and MT transmitting antennas. The feedback method includes the following steps. First, the signals received by the MR receiving antennas are demodulated, where the signals received by each receiving antenna can be demodulated to obtain tones of N different frequency bands. Next, according to the tones of all the different frequency bands, the obtained channel state information (CSI) of N different frequency bands is estimated. Afterwards, all the tones are divided into N sets and each set sequentially includes MR tones with a same frequency band obtained by different receiving antennas. According to the CSI, the tones of every set are detected and decoded, thus N receiving data vectors are obtained. Meanwhile, according to the CSI, the feedback parameters are produced and output to the transmitter, wherein H(n) is the MIMO channel matrix of the n-th frequency band and the QR decomposition of H(n) is expressed by:
H(n)=R(n)Q*(n),
wherein R(n) represents a lower triangular matrix with a dimension of MR×MT, Q(n) represents an orthonormal matrix with a dimension of MT×MT and Q*(n) represents the complex conjugate transpose matrix of Q(n). When Q(n)=[q1(n) q2(n) . . . qMT(n)], qm(n) represents the m-th column vector in the matrix Q(n), and when the spatial vector-coding matrix W(n)=[q1(n) q2(n) . . . qM(n)], the tone of every same frequency band detected and decoded to obtain all the receiving data vectors corresponds to coding the tones by using the spatial vector-coding matrixes W(n). The above-described MT, N, M, m and n are natural numbers and 0<M≦MT, 0<n≦N.
According to the feedback method in the MIMO-OFDM system described in the embodiment of the present invention, the Givens rotations are used on the MIMO channel matrix to generate a plurality of Givens rotation matrixes Gp,q(θ,φ), and the feedback parameters are produced according to the θ and φ of the Givens rotation matrixes. In the present embodiment, a delta modulation is, as the preferred method, conducted on θ and φ in the Givens rotation matrixes Gp,q(θ,φ) to produce feedback parameters. In particular, a delta modulation with a one-bit quantizer or a multi-bit quantizer is conducted on θ and φ to produce feedback parameters.
According to the feedback method in the MIMO-OFDM system described in the embodiment of the present invention, according to the MIMO channel matrixes of a part of frequency bands, only the Givens rotation matrixes Gp,q(θ,φ) of the part of frequency bands are generated. Further, according to θ and φ of the Givens rotation matrixes of the part of frequency bands, the feedback parameters are produced.
The present invention provides a pre-coding method in the MIMO-OFDM system, wherein the system includes MR receiving antennas and MT transmitting antennas. The pre-coding method includes the following steps. First, the feedback parameters output from the receiver are received. Next, according to the received feedback parameters, N spatial vector-coding matrixes W(n) are produced. In more detail, when H(n) is a MIMO channel matrix of the n-th frequency band, the QR decomposition of H(n) can be expressed by:
H(n)=R(n)Q*(n),
wherein R(n) is a lower triangular matrix with a dimension of MR×MT, Q(n) is an orthonormal matrix with a dimension of MT×MT and Q*(n) represents the complex conjugate transpose matrix of Q(n). When Q(n)=[q1(n) q2(n) . . . qMT(n)], qm(n) represents the m-th column vector in the matrix Q(n), thus W(n)=[q1(n) q2(n) . . . qM(n)]. Afterwards, N transmission data vectors are received, wherein each transmission data vector has a dimension of M×1. According to the vector-coding matrixes, the transmission data vectors are coded and each transmission data vector is coded to obtain MT tones with a same frequency band, wherein the tones obtained from different transmission data vector have different frequency bands from each other. Further, all the tones are divided into MT sets and each set sequentially includes N tones with different frequency bands obtained from the different transmission data vectors. After conducting modulations on the tones of each set, the modulated signals are output by the transmitting antennas. The above-described MR, MT, M, N, m and n are natural numbers and 0<M≦MT, 0<n≦N.
According to the pre-coding method in the MIMO-OFDM system described in the embodiment of the present invention, a plurality of Givens rotation matrixes Gp,q(θ,φ) are generated according to the feedback parameters and then N spatial vector-coding matrixes W(n) are produced according to the Givens rotation matrixes Gp,q(θ,φ). In the present embodiment, a delta modulation is, as the preferred method, conducted on θ and φ in the Givens rotation matrixes Gp,q(θ,φ) to produce feedback parameters. In particular, a delta modulation with a one-bit quantizer or a multi-bit quantizer is conducted on θ and φ to produce feedback parameters.
According to the pre-coding method in the MIMO-OFDM system described in the embodiment of the present invention, θ(n) and φ(n) of only partial frequency bands are generated according to the feedback parameters, then an interpolation method is used to generate θ(n) and φ(n) of all the frequency bands and further the spatial vector-coding matrixes are produced.
The present invention uses a QR decomposition of the MIMO channel matrixes to parameterize the channel state information (CSI), instead of using a complicate singular value decomposition (SVD) to parameterize the CSI, therefore the complexity is reduced and the computation is reduced. In addition, in the present invention, only θ and φ in the Givens rotation matrixes of a part of frequency bands are fed back and an interpolation method is used to generate θ and φ in the Givens rotation matrixes of all the frequency bands and to further produce the spatial vector-coding matrixes of all the frequency bands, therefore the feedback data amount can be freely controlled and the feedback rate requirement can be well responded to.
The accompanying drawings are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the invention and, together with the description, serve for explaining the principles of the invention.
The related prior art is characterized in that a singular value decomposition (SVD) of the MIMO channel matrix of each frequency band is used to parameterize the channel state information (CSI); further a plurality of Givens rotation matrixes Gp,q(θ,φ) are generated by using the Givens rotation on the V matrixes of all the frequency bands; furthermore. the information of the Givens rotation matrixes of all the frequency bands is fed back to the transmitter and is combined (re-produced) at the transmitter to re-obtain the V matrixes of all the frequency bands; finally, the re-obtained V matrixes are used in the vector coding at the transmitter. However, it is questionable whether or not the SVD is the only approach to provide the solution. Is there a simpler computation approach? Is the information of all the frequency bands required to be fed back. Accordingly, the present invention provides a solution characterized in that QR decompositions of the MIMO channel matrixes are used to parameterize the CSI. Hence, the Givens rotation can be directly applied to the MIMO channel matrixes for generating the Givens rotation matrixes Gp,q(θ,φ). Another major feature of the present invention is that only θ and φ of the Givens rotation matrixes of partial frequency bands are fed back and an interpolation method is used to generate θ and φ of the Givens rotation matrixes of all the frequency bands and to further produce spatial vector-coding matrixes of all the frequency bands.
In the transmitter of the system provided by the present invention, the N coders 128 receive N transmission data vectors s(n), each coder 128 corresponds to a different frequency band and each transmission data vector has a dimension of M×1. The coders 128 conduct spatial vector-coding on the transmission data vectors s(n) according to the channel state information (CSI) of each different frequency band. That is, the coders 128 conduct coding on the Ntransmission data vectors s(n) according to N vector-coding matrixes W(n). Since each different frequency band has a different vector-coding matrix W(n) with a dimension of MT×M, each coder 128 outputs MT tones with a same frequency band but different from the other coders, wherein M and n are natural numbers and 0<M≦MT, 0<n≦N. Referring to
In the receiver provided by the present invention, MR frequency band demodulators 104 are coupled to the MR receiving antennas. The frequency band demodulators 104 demodulate the radio signals received by the receiving antennas 102, thus each frequency band demodulator 104 obtains N tones of N different frequency bands. In particular, the demodulation conducted by the frequency band demodulators 104 is a fast Fourier transform (FFT). The channel estimator & feedback parameter generator 106 is coupled to the frequency band demodulators 104 and according to the outputs from the frequency band demodulators 104, estimates the obtained channel state information (CSI) of the N different frequency bands. The N detectors 108 are coupled to the frequency band demodulators 104 and the channel estimator & feedback parameter generator 106. Each detector 108 corresponds to a different frequency band. A single detector 108 receives the tones with a same frequency band output from all the frequency band demodulators 104 and according to the CSI of the different frequency bands, conducts decoding on the outputs from the frequency band demodulators 104 to obtain N receiving data vectors ŝ(n).
From the above described it can be seen that either the transmitter or the receiver uses the CSI of every different frequency band to conduct coding and decoding. One major feature of the present invention is that QR decompositions of the MIMO channel matrixes are used to parameterize the CSI. Hence, the channel estimator & feedback parameter generator 106 estimates the obtained CSI of N different frequency bands according to the outputs from all the frequency band demodulators 104 and according to the CSI produces the feedback parameters related to the QR decompositions of the MIMO channel matrixes, followed by feeding back the produced feedback parameters to the transmitter. The vector-coding matrix generator 126, according to the feedback parameters sent from the receiver, produces the CSI of all the different frequency bands, i.e. the above-described N spatial vector-coding matrixes W(n). When H(n) is a MIMO channel matrix of the n-th frequency band, the QR decomposition of H(n) can be expressed by:
H(n)=R(n)Q*(n),
wherein R(n) is a lower triangular matrix with a dimension of MR×MT, Q(n) is an orthonormal matrix with a dimension of MT×MT and Q*(n) represents the complex conjugate transpose matrix of Q(n). When Q(n)=[q1(n) q2(n) . . . qMT(n)], qm(n) represents the m-th column vector in the matrix Q(n), thus:
W(n)=[q1(n)q2(n) . . . qM(n)],
wherein M and m are natural numbers and 0<M≦MT.
As a matter of facts, the channel estimator & feedback parameter generator 106 is not required to really produce Q(n); instead, according to the CSI of the different frequency bands and directly applying the Givens rotation on the MIMO channel matrixes H(n), the generator 106 generates a plurality of Givens rotation matrixes Gp,q(θ,φ). Afterwards, according to θ and φ of the Givens rotation matrixes, the generator 106 produces the feedback parameters. Further, the vector-coding matrix generator 126 would directly combine (re-produce) the feedback parameters related to θ and φ of the Givens rotation matrixes and obtain Q(n). Furthermore, the vector-coding matrix generator 126 would produce the above-described N the spatial vector-coding matrixes W(n) according to a plurality of column vectors in the matrixes Q(n).
In the present embodiment, the channel estimator & feedback parameter generator 106 according to the Givens rotation matrixes Gp,q(θ,φ) uses a delta modulation on θ and φ to produce the feedback parameters. In particular, the channel estimator & feedback parameter generator 106 uses a delta modulation with a one-bit quantizer or a multi-bit quantizer on θ and φ to produce the feedback parameters.
According to the above-described operations at the receiver of the system, a feedback method in the MIMO-OFDM system provided by the present invention can be summarized into the following steps. First, the signals received by the MR receiving antennas are demodulated, wherein the signals received by each receiving antenna can be demodulated to obtain tones of N different frequency bands. Next, according to the tones of all the different frequency bands, the obtained channel state information (CSI) of N different frequency bands is estimated. Afterwards, all the tones are divided into N sets and each set sequentially includes MR tones with a same frequency band but obtained by different receiving antennas. According to the CSI, the tones of every set are detected and decoded, thus N receiving data vectors are obtained. Meanwhile, according to the CSI, the feedback parameters are produced and output to the transmitter, wherein when H(n) is the MIMO channel matrix of the n-th frequency band, the QR decomposition of H(n) is expressed by:
H(n)=R(n)Q*(n),
wherein R(n) represents a lower triangular matrix with a dimension of MR×MT, Q(n) represents an orthonormal matrix with a dimension of MT×MT and Q*(n) represents the complex conjugate transpose matrix of Q(n). When Q(n)=[q1(n) q2(n) . . . qMT(n)], qm(n) represents the m-th column vector in the matrix Q(n), and when the spatial vector-coding matrix W(n)=[q1(n) q2(n) . . . qM(n)], the tone of every same frequency band detected and decoded to obtain all the receiving data vectors corresponds to coding the tones by using the spatial vector-coding matrixes W(n). The above-described MR, MT, N, M, m and n are natural numbers and 0<M≦MT, 0<n≦N.
In the feedback method, the Givens rotations are directly used on the MIMO channel matrix H(n) to generate a plurality of Givens rotation matrixes Gp,q(θ,φ), and the feedback parameters are produced according to the θ and φ of the Givens rotation matrixes. A delta modulation is, as the preferred method, conducted on θ and φ in the Givens rotation matrixes Gp,q(θ,φ) to produce feedback parameters. In particular, a delta modulation with a one-bit quantizer or a multi-bit quantizer is conducted on θ and φ to produce feedback parameters.
Similarly, the above-described operations at the transmitter of the system a pre-coding method in the MIMO-OFDM system provided by the present invention can be summarized into the following steps. First, the feedback parameters output from the receiver are received. Next, according to the received feedback parameters, N spatial vector-coding matrixes W(n) are produced. In more detail, when H(n) is a MIMO channel matrix of the n-th frequency band, the QR decomposition of H(n) can be expressed by:
H(n)=R(n)Q*(n),
wherein R(n) is a lower triangular matrix with a dimension of MR×MT, Q(n) is an orthonormal matrix with a dimension of MT×MT and Q*(n) represents the complex conjugate transpose matrix of Q(n). When Q(n)=[q1(n) q2(n) . . . qMT(n)], qm(n) represents the m-th column vector in the matrix Q(n), thus W(n)=[q1(n) q2(n) . . . qM(n)]. Afterwards, N transmission data vectors are received, wherein each transmission data vector has a dimension of M×1. According to the vector-coding matrixes. the transmission data vectors are coded and after the coding each transmission data vector, the MT tones with a same frequency band are obtained, wherein the tones obtained from different transmission data vector have different frequency bands from each other. Further, all the tones are divided into MT sets and each set sequentially includes N tones with different frequency bands obtained from the different transmission data vectors. After conducting modulations on the tones of each set, the modulated signals are transmitted by the transmitting antennas. The above-described MT, MR, M, N, m and n are natural numbers and 0<M≦MT, 0<n≦N.
In the pre-coding method, a plurality of Givens rotation matrixes Gp,q(θ,φ) are generated according to the feedback parameters and then N spatial vector-coding matrixes W(n) are produced according to the Givens rotation matrixes Gp,q(θ,φ).
The vector-coding matrix generator at the transmitter only generates θ(n) and φ(n) of partial frequency bands according to the feedback parameters, then a clustering method or an interpolation method is used to generate θ(n) and φ(n) of all the frequency bands, and further all the spatial vector-coding matrixes are produced. Since θ(n) and φ(n) are the parameter data in the Givens rotation matrixes, the spatial vector-coding matrixes finally obtained by using the interpolation computation still have an orthogonal characteristic.
According to the above-described operations at the receiver and transmitter of the system, the feedback method and the pre-coding method in the MIMO-OFDM system of the present invention can be summarized as follows. In the feedback method, the Givens rotation matrixes Gp,q(θ,φ) of the partial frequency bands are generated only based on the MIMO channel matrixes of the partial frequency bands and then according to θ and φ of the generated Givens rotation matrixes of the partial frequency bands, the feedback parameters are produced. In the pre-coding method, according to the feedback parameters, θ(n) and φ(n) of the partial frequency bands are generated, and then the interpolation method is used to generate θ(n) and φ(n) of all the frequency bands to further produce the spatial vector-coding matrixes.
According to the results in
In summary, since the present invention uses the QR decomposition of the MIMO channel matrixes, instead of using the complicate singular value decomposition (SVD), to parameterize the channel state information (CSI), the complexity is reduced and the computation is lowered. In addition, in the present invention, only θ and φ in the Givens rotation matrixes of a part of frequency bands are fed back and an interpolation method is used to generate θ and φ in the Givens rotation matrixes of all the frequency bands and to further produce the spatial vector-coding matrixes of all the frequency bands, therefore the feedback data amount can be freely controlled and the feedback rate requirement can be well responded to.
It will be apparent to those skilled in the art that various modifications and variations can be made to the structure of the present invention without departing from the scope or spirit of the invention. In view of the foregoing, it is intended that the specification and examples to be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims and their equivalents.
| Number | Date | Country | Kind |
|---|---|---|---|
| 94136284 A | Oct 2005 | TW | national |
| Number | Name | Date | Kind |
|---|---|---|---|
| 5475793 | Broomhead et al. | Dec 1995 | A |
| 6956907 | Ketchum | Oct 2005 | B2 |
| 7158563 | Ginis et al. | Jan 2007 | B2 |
| 7218906 | Sayeed et al. | May 2007 | B2 |
| 7356017 | Codreanu et al. | Apr 2008 | B2 |
| 7424063 | Yee | Sep 2008 | B2 |
| 20040181419 | Davis et al. | Sep 2004 | A1 |
| 20050152484 | Sandhu et al. | Jul 2005 | A1 |
| 20050237971 | Skraparlis | Oct 2005 | A1 |
| 20060056531 | Li et al. | Mar 2006 | A1 |
| 20060285531 | Howard et al. | Dec 2006 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20070115799 A1 | May 2007 | US |
