The present invention relates to a method and arrangement for reducing feedback data in a communication system. In particular, the present invention relates to a method and arrangement for reducing data feedback, especially channel response, in a Multiple-input Multiple-Output (MIMO) system.
The existing wireless mobile communication systems provide several types of services and mostly depend on channel coding to overcome any inferiority of channels. However, due to the increasing demands, for example for a high-quality multimedia services, in which users can communicate with anyone regardless of time and place, the existing services have evolving into data-oriented services. Accordingly, there is a high demand for next generation wireless transmission technology for transmitting the larger amount of data at a lower error rate. In particular, it is very important to transmit data at a high rate in a link in which the amount of required data is large.
For the next generation wireless communication, various antenna systems have been proposed. For example, a Multiple-input multiple-output (MIMO) system increases spectrum efficiency through all of transmission antennas without excessive use of a frequency bandwidth. Generally, MIMO is classified into Space-Time Coding (STC), Diversity, Beam Forming (BF), and Spatial Multiplexing (SM) according to the transmission structure and scheme of a transmitter, all of which provide high data rate and reliability.
A MIMO system adopts multiple antennas or array antenna to transmit/receive data in the transmitter and receiver. Multiple antennas are provided in different spatial positions, with different fading features, thus the received signals of adjacent antennas can be approximated as uncorrelated entirely as long as the spacing between adjacent antennas for transmitting/receiving signals in the MIMO system is large enough. The MIMO system takes full advantage of the spatial characteristics of multipath for implementing space diversity transmission and reception.
At the receiver site N Rx 104 antennas receive the broadcasted signals, which are transformed by multiple access inverse transform unit 105, performing multiple access demultiplexing processing, and provided to a MIMO detection unit 106.
Usually, a MIMO antenna system with nR receive and nT transmit antennas operating in a frequency non-selective channel is described by the following matrix representation:
y=Gx+z (1)
Wherein y is the nR×1 received signal vector, G is the nR×nT MIMO channel response, z is the independent and identically distributed elements Additive White Gaussian Noise (AWGN) at the receiver with individual variance of σ2z and x is the nT×1 transmitted signal vector with a certain power constraint.
The best performance of such a system is achieved when the channel response is known to the transmitter so that the transmit signals can be designed accordingly. This is disclosed, for example in G. G Raleigh & J. M. Cioffi, “Spatio-temporal Coding For Wireless Communication” IEEE trans. On Comm. Vol. 46, no. 3 Mar. 1998, pp. 357-366, and K. C. Zangi and L. G Krasny “Capacity achieving Transmitter and Receiver Pairs for MISO Channels” IEEE Transaction on Wireless Communications Vol. 2 No. 6 Nov. 2003, pp 1204-1216.
In many cases, the channel response is only known to the receiver through the reference signals sent by the transmitter on a forward link and therefore requires being explicitly fed back to the transmitter on a reverse link. Such a feedback may sometimes be a significant overhead especially for a configuration with a large number of antennas. It may also require a substantial amount of computational power.
Conventionally, the entire channel matrix G is fed back from the receiver. For the case of two transmit antennas and one receiver antenna, for example, the close-loop mode in 3rd Generation Project Plan (3GPP) feeds back one phase factor optimized for the frequency selective channel to adjust one of the transmitter antennas.
It is therefore an object of the present invention to provide a method and an arrangement in a MIMO antenna system that reduces feedback overhead. Thus, one problem solved by the present invention is to extend a simple solution to the complex MIMO.
Moreover, the present invention extends and generalises the close-loop mode concept in 3GPP and similar concepts to any number of transmit and receive antennas and also to channels with much greater frequency selectivity.
These objects are achieved using a method for reducing amount of data in a transmission in a communication system. The communication system comprises a number of transmitter antennas, nT, and a number of receiver antennas, nR, for parallel spatially independent transmission and reception of signals. A channel response is represented by a matrix containing nR×nT complex variables. The invention comprises substituting the channel response matrix with a representative subset of elements in the decomposed expression of the channel response matrix, in the data transmission.
A method of reducing feedback data in a communication system, the communication In more detail, the method of the invention comprises the steps of: decomposing an expression of the channel response matrix into products of a unitary transform, nR×nT diagonal matrix and a conjugate transpose of a unitary matrix, choosing the unitary matrix such that its diagonal elements are real, substituting the channel response matrix with a representative subset of elements in the decomposed expression of the channel response matrix. The representative subset comprising diagonal of a first matrix and lower triangle of a second matrix excluding the diagonal. Preferably, the first matrix is nR×nR non-trivial truncation of the nR×nT diagonal matrix, and the second matrix is nT×nR non-trivial truncation of the corresponding nT×nT unitary matrix. The second matrix is represented by:
According to one embodiment of the invention, the second matrix is solved by choosing a solution and rotating the phase of each column in the matrix such that the diagonal elements become real. Thus, the second matrix is restored from its lower triangle by:
Preferably, <w1, w1>=1 for w1,1 is determined by solving
Preferably, <w1, w2>=0 and <w2, w2>=0 is determined by solving:w1,w2
=0
w2,w2
=1
The method implies continuing restoration process until all columns are restored by solving:
wherein l is an integer ranging from 1 to nR.
Most preferably, one advantage of the invention is that the number of real coefficients required for specifying the channel response matrix is 2nRnT−n2R.
According to one aspect of the invention, the system comprises frequency selective channels and a frequency range is divided into a number of consecutive segments and the number of segments are chosen such that a channel response matrix being a function of a frequency, is substantially constant and the channel response matrix is then approximated in each segment using its mean and substitution of each segment. Thus, the system comprises frequency selective channels and a continuous channel response in a frequency domain is approximated by a finite number of frequency non-selective channels.
Preferably, a receiver end determines and feeds back to a transmit end the representative subset.
According to another aspect of the invention a method is provided for reducing feedback data in a Multiple-input Multiple-Output (MIMO) communication system. The system comprises at least two transmitter antennas, nT, and at least two receiver antennas, nR, for parallel spatially independent transmission and reception of signals. A channel response is represented by a matrix containing nR×nT complex variables. The method comprising the steps of: decomposing an expression of the channel response matrix into products of a unitary transform, nR×nT diagonal matrix and conjugate transpose of a unitary matrix, where the unitary matrix is chosen such that its diagonal elements are real, substituting the channel response matrix with a representative subset of elements in the decomposed expression of the channel response matrix, the representative subset comprising diagonal of a first matrix being nR×nR non-trivial truncation of the nR×nT diagonal matrix and lower triangle of a second matrix being nT×nR non-trivial truncation of the corresponding nT×nT unitary matrix excluding the diagonal.
According to another aspect of the invention a method is provided for a Single-Input Single-Output (SISO) communication system, the system comprising one transmitter antenna and a receiver antenna for transmission and reception of signals, wherein a channel response is represented by a matrix. The method comprises the steps of: decomposing an expression of the channel response matrix into products of a unitary transform, a diagonal matrix and Hermitian transpose of a unitary matrix, where the unitary matrix is chosen such that its diagonal elements are real, substituting the channel response matrix with a representative subset of elements in the decomposed expression of the channel response matrix, the representative subset comprising diagonal of a first matrix being non-trivial truncation of the diagonal matrix and lower triangle of a second matrix being non-trivial truncation of the corresponding unitary matrix excluding the diagonal.
The invention also relates to an arrangement in a communication network. The communication network comprises a number of transmitter antennas, nT, and a number of receiver antennas, nR. Each of the antennas being arranged for substantially parallel and spatially independent transmission and reception of signals, wherein a channel response is realized as a matrix containing nR×nT complex variables. The arrangement further comprises a data processing unit for decomposing an expression of the channel response matrix into products of a unitary transform, a diagonal matrix and a conjugate transpose of a unitary matrix, where the unitary matrix is chosen such that its diagonal elements are real and means for substituting and transmitting the channel response matrix with a representative subset of elements in the decomposed expression of the channel response matrix, wherein the representative subset comprises diagonal of a first matrix and lower triangle of a second matrix excluding the diagonal. The arrangement further comprises means for generating the first matrix as a nR×nR non-trivial truncation of the nR×nT diagonal matrix. Preferably, the arrangement further comprises means for generating the second matrix as a nT×nR non-trivial truncation of the corresponding nT×nT unitary matrix.
The invention also relates to a Multiple-Input Multiple-Output (MIMO) architecture unit in a MIMO system comprising a number of transmitter antennas, nT, and a number of receiver antennas, nR, each of the antennas being arranged for substantially parallel and spatially independent transmission and reception of signals, wherein a channel response is realized as a matrix containing nR×nT complex variables. The architecture unit further comprises a data processing unit for decomposing an expression of the channel response matrix into products of a unitary transform, a diagonal matrix and a conjugate transpose of a unitary matrix, where the unitary matrix is chosen such that its diagonal elements are real and means for substituting and transmitting the channel response matrix with a representative subset of elements in the decomposed expression of the channel response matrix, wherein the representative subset comprises diagonal of a first matrix and lower triangle of a second matrix excluding the diagonal.
The invention also relates to a Single-input Single-Output (SISO) receiver unit in a SISO system comprising a transmitter antenna and a receiver antenna, each of the antennas being arranged for independent transmission and reception of signals, wherein a channel response is realized as a matrix containing complex variables. The architecture unit further comprises a data processing unit for decomposing an expression of the channel response matrix into products of a unitary transform, a diagonal matrix and a conjugate transpose of a unitary matrix, where the unitary matrix is chosen such that its diagonal elements are real and means for substituting and transmitting the channel response matrix with a representative subset of elements in the decomposed expression of the channel response matrix, wherein the representative subset comprises diagonal of a first matrix and lower triangle of a second matrix excluding the diagonal.
The invention may be realized as a computer program product for reducing feedback data in a communication system, the communication system comprising a number of transmitter antennas, nT, and a number of receiver antennas, nR, for parallel spatially independent transmission and reception of signals, wherein a channel response is represented by a matrix containing nR×nT complex variables. Thus, the computer program product comprises: an instruction set for receiving, decomposing and storing an expression of the channel response matrix, the decomposing comprising a function for producing products of the unitary transform, nR×nT diagonal matrix, and a Hermitian transpose of the unitary matrix, an instruction set for choosing the unitary matrix in such a way that its diagonal elements are real, an instruction set for producing a representative subset of the elements in the decomposed expression, which comprises diagonal of nR×nR non-trivial truncation of the nR×nT diagonal matrix and lower triangle of nT×nR non-trivial truncation of the nT×nT unitary matrix excluding diagonals, and an instruction set for substituting the channel response matrix with the representative subset of the elements in the decomposed expression.
wherein W is the corresponding nT×nT unitary matrix multiplied from the right, Λ is an nR×nT diagonal matrix whose diagonal elements {√{square root over (λ1)}, . . . , √{square root over (λn
and note that since W is unitary, its column vectors are orthogonal:
Without loss of generality, it is assumed that all the elements in w1 are known except w1,1. Since the columns of {tilde over (W)} is uniquely determined up to a phase rotation, consequently w1,1 can be chosen as real and non-negative such that its solution to <w1, w1>=1, and
is unique. Assuming further, without loss of generality, that w2 is known except w1,2 and w2,2 and that w2,2 is chosen to be real and non-negative. Consequently, the two unknowns may be uniquely determined by solving the following equations:w1,w2
=0 (6)
w2,w2
=1
By continuing this process, it can be shown that the l unknowns {w1,l, . . . , wl,l} in wl can be uniquely determined by solving the equations
Given that wl,l is real and non-negative. Therefore, the number of real numbers required to completely specify {tilde over (W)} is
Together with the nR real eigenvalues in {tilde over (Λ)}, this reduces the number of real coefficients required to specify G from 2nRnT to 2nRnT−n2R, i.e. for example in the feedback signal from the receiver to the transmitter.
Thus, the invention according to one preferred embodiment comprises sending the diagonal of {tilde over (Λ)} and an abbreviated or simplified {tilde over (W)}, as feedback data from the receiver instead of sending G, which has nR×nT complex variables.
Decomposing G using Eq. 2 provides a number of solutions for {tilde over (W)}. Accordingly, a solution with real diagonal elements is selected. This is obtained by choosing any solution and rotating the phase of each column in the matrix such that the diagonal elements become real. Then only the lower triangle elements, i.e. elements found only in the lower triangle of the matrix, excluding the main diagonal (Iij=0 if j<j) are fed back, which are sufficient to restore {tilde over (W)}.
{tilde over (W)} is restored by:
Thus, for the common case of MIMO spatial multiplexing where nR≈nT, the reduction in feedback overhead can be as much as approx. 50%.
A frequency selective channel is one whose channel response matrix G is a function of the frequency f, or G(f). According to another aspect of the present invention, for frequency selective channel equation (1) will be represented by
y(f)=G(f)×(f)+z(f), (9)
and the same approach as mentioned earlier can be applied to across the frequency. Instead of 2nRnT real functions of f in G(f), the feedback now consists of 2nRnT−n2R real functions of f. For example each element in the MIMO channel frequency response G(f) is Fourier transform of a digital filter such as an Finite Impulse Response (FIR) filter, e.g. having L taps.
The present invention is applied to frequency selective channel case is by assuming each entry in the concise representation, as descried earlier, as a function of frequency instead of scalar, using Transmit Diversity with Constrained Feedback (TDFC) to convert each function to a finite number of taps.
One problem which can occur is to quantify the number of taps required to accurately represent the eigenvalues {tilde over (Λ)}(f) and their corresponding eigenvectors {tilde over (W)}(f). For large number of transmit antennas, the components in {tilde over (Λ)}(f) are almost flat, and therefore require much less L taps to represent total number of FIR coefficients, it is possible to apply Transmit Diversity with Constrained Feedback (TDCF) techniques to the elements in {tilde over (Λ)}(f) and {tilde over (W)}(f), see for example L. Krasny and J. Guey, “Transmit Diversity with Constrained Feedback”, 14th IST Mobile & Wireless Communications Summit, Dresden, June 2005. Feedback overhead reduction is still possible as long as the Signal-Noise Ratio (SNR) loss due to the mismatch is smaller than feeding back the original channel response with equal amount of compression.
Alternatively, it is possible to approximate the continuous channel response in the frequency domain by a finite number of frequency non-selective channels and then apply the same principle to each sub-band. From sample point of view, an impulse response with L-tap can be represented by L equally spaced sampling points in the frequency domain. Therefore, dividing the frequency response of the channels into L sub-bands should be reasonable approximation and the proportion of the reduction in feed-back overhead is the same as the flat fading channel.
Thus, for frequency selective channel, the frequency range is divided into L consecutive segments. The number of segments are chosen such that G(f) is substantially constant. G(f) is then approximated in each segment using its mean and the aforementioned frequency non-selective technique is applied to each segment.
The system is assumed to operate in for example 3GPP Typical Urban (TU) channel, scaled four times to 20 MHz bandwidth with L=40 taps. Thus the frequency response of the channel is divided into 40 sub-bands or “chunks” each with a bandwidth of 500 KHz.
Although, the invention is described in an exemplary way to systems having nR≦nT, it will be appreciated by a skilled person that the teachings of the invention may equally be applied to systems having nR>nT.
The method of the invention may be implemented as hardware and/or software. An instruction set 600, as illustrated in
The invention is not limited to the described and illustrated embodiments; especially the number of transmitters, receivers and antennas may vary.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/SE2006/000296 | 3/7/2006 | WO | 00 | 1/22/2009 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO2007/102760 | 9/13/2007 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
4408332 | Sari | Oct 1983 | A |
20020159415 | Pan et al. | Oct 2002 | A1 |
20060093067 | Jalali et al. | May 2006 | A1 |
20060115030 | Erving et al. | Jun 2006 | A1 |
20060251198 | Ma et al. | Nov 2006 | A1 |
20090154587 | Cheong et al. | Jun 2009 | A1 |
Number | Date | Country | |
---|---|---|---|
20090233559 A1 | Sep 2009 | US |