CN100388645C - Pre-coding method and device for improving V-BLAST detection performance - Google Patents

Abstract

The present invention provides a pre-encoding method in a multi-antenna system based on V-BLAST technology, which comprises the following steps: a channel matrix H fed by a reverse channel is received; the singular value of the matrix is decomposed according to a formula * H=U. D. V< + >* so as to obtain a unitary matrix V and a diagonal matrix D which is the diagonal matrix comprising singular values; + represents Hermitian transposition; the Hermitian transposition matrix of the matrix V and the matrix D is calculated so as to obtain a matrix V< + > and a matrix D< + >; the matrix multiplication operation is carried out to generate a code generation matrix X=V. D < + >. V< + >; the matrix X generated by the codes encodes the input modulation symbol vector a so as to obtain and output an encoding symbol vector b=X. a=V. D < + >. V< + >. a. After the pre-encoding is adopted, the equivalent space sub-channel has the same receiving power, and thereby, the capacity of the space relative channel is effectively increased. Compared with the existing pre-treatment and after-treatment proposal based on the channel singular value decomposition, the present invention obviously improves the performance of the traditional V-BLAST system, and the hardware cost of a receiver is eliminated.

Description

Precoding method and device for improving V-BLAST detection performance

Technical Field

The invention designs a pre-coding method and a pre-coding device for improving V-BLAST (Vertical-Bell laboratories layered Space-Time, Vertical-Bell laboratories layered Space-Time structure) detection performance. The precoding method and device for improving the V-BLAST detection performance can be applied to a multi-antenna system based on the V-BLAST technology.

Background

The V-BLAST technology-based multi-antenna system capacity is linearly related to the number of transmit antennas, and high spectral efficiency can be obtained without sacrificing additional bandwidth and transmit power, thus receiving much attention. The conventional V-BLAST technology uses the same modulation scheme and transmission power, transmits independent data at each transmitting antenna unit, and the receiving end uses a successive interference cancellation detection algorithm based on the optimal ordering ([1] g.d. golden, c.j.fosschii, r.a.valentlea and p.w.wolniansky.monitoring and initial laboratory using V-BLAST-time communication architecture. electronics Letters, vol.35, No.1, 7th January 1999.). Under the condition of space independent multipath fading channel, the traditional V-BLAST technology obtains good performance. However, in a spatially correlated multipath fading channel, strong spatial correlation reduces the signal-to-noise ratio of several equivalent spatial subchannels of the spatially correlated channel, thus reducing the channel capacity ([5] Da-Shan Shiu, Gerard j.fosschii, Michael j.gans, etc. fading correlation and bits effect on the capacity of multi-element antenna systems ieee Trans on com., vol.48, No.3, March 2000, pp.502-513.) if the transmitting end still multiplexes the individual spatial channels at a higher rate to transmit data, the receiving end will have difficulty in recovering the data transmitted by the equivalent spatial subchannels at a lower signal-to-noise ratio. These errors propagate from layer to layer in V-BLAST, resulting in a severe degradation of the error rate performance ([2] Cong Shen, Hairuo Zhang, Lin Dai and Shidong Zhou. detection algorithm V-BLAST performance over error propagation. electronics Letters, Vol.39, No.13, 26th June 2003.). Therefore, improving the performance of V-BLAST on the spatial correlation channel is an important issue to be faced with in order to increase the system capacity using the multi-antenna system.

In the published literature, the channel singular value decomposition method has been applied to rate and power control in multi-antenna systems. Document [3] makes the design of rate and power allocation more convenient by diagonalizing the spatial correlation fading channel matrix at the preprocessing unit at the transmitting end and the post-processing unit at the receiving end. However, this method does not change the snr of the equivalent spatial subchannel of the spatial channel, and thus cannot improve the performance of V-BLAST when the modulation scheme and the transmission power are constant. The channel singular value decomposition method is also widely used in adaptive modulation of multi-antenna systems ([4] Young-Doo Kim Inhyoung Kim, Jihoon Choi, etc. adaptive modulation for MIMO systems with V-BLAST detection. vehicular Technology Conference, 2003, VTC 2003-Spring, The57th IEEE semi annual, Vol.2, 22-25, April 2003.).

The technology adopts a channel singular value decomposition method, designs a new precoding method and a device according to the equal receiving power principle of an equivalent space subchannel, and well improves the performance of V-BLAST of a fixed modulation mode and transmitting power under a space correlated fading channel.

Under the spatial correlation fading channel, there are two V-BLAST data transmission schemes with constant transmission rate and power. One is that the transmitting end and the receiving end do not adopt any traditional V-BLAST [1] aiming at the correlation processing of spatial channels, and the other is a space-time processing V-BLAST scheme based on singular value decomposition of a channel matrix ([3] Jeongseok Ha, Apurva N.Mody, JoonHyun Sung, etc. LDPC-coded OFDM with Alamouti/SVD diversity technology. Wireless Personal Communications, vol.23, No.1, October 2002, pp.183-194 (12)).

A block diagram of a multi-antenna system based on the conventional V-BLAST technology is shown in fig. 1, which illustrates a 4-transmission and 4-reception system as an example.

Assume that a typical V-BLAST system has M transmit antennas Tx and N receive antennas Rx, where N ≧ M. The source bit stream 1 passes through a digital modulator 2 to produce a serial stream of modulated symbols. Then converted into parallel transmit symbol vector a ═ (α) by vector encoder 3₁，α₂，…，α_M)^T. In one symbol time interval, the transmitting antennas transmit symbol vectors to a spatial fading channel H with the same power of a single antenna. A V-BLAST detector 4 based on optimally ordered successive symbol interference cancellation detects the received symbol vector and produces an estimate of the transmitted symbol vector. Then, the signal is sent to a digital demodulator 5 for demodulation.

A block diagram of a space-time processing V-BLAST system based on singular value decomposition of a channel matrix is shown in fig. 2. Singular value decomposition of the channel matrix H into

H＝U·D·V⁺(formula-1)

Wherein, U and V are unitary matrixes, D is a diagonal matrix containing H singular values, and + represents Hermitian transpose. In fig. 2, a transmission symbol vector is transmitted to a spatial correlation channel after undergoing unitary V transform. The receiving end passes through a post-processing unit U⁺Then, the signal is sent to a V-BLAST detector 4 for detection and a digital demodulator 5 for demodulation.

According to the document [5], assuming that the channel is a flat, quasi-static spatially correlated rayleigh fading channel, the transmitting antennas transmit with equal power, and the total transmitting power is ρ, the capacity of the multi-antenna system is the sum of the capacities of all M equivalent single-antenna systems, i.e., the capacity of the multi-antenna system is the sum of the capacities of all M equivalent single-antenna systems

<math><mrow> <mi>C</mi> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>M</mi> </munderover> <msub> <mi>log</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <mn>1</mn> <mo>+</mo> <mfrac> <mi>&rho;</mi> <msub> <mi>N</mi> <mi>&alpha;</mi> </msub> </mfrac> <mo>&CenterDot;</mo> <msubsup> <mi>&epsiv;</mi> <mi>k</mi> <mn>2</mn> </msubsup> <mo>)</mo> </mrow> </mrow></math> (formula-2)

Wherein epsilon_kIs the singular value of H. Accordingly,. epsilon_k ²Is the channel gain of the equivalent spatial subchannel.

A statistical model of a typical spatially correlated channel H can be described by the following equation [5 ]:

H＝Ψ_R·H_w·Ψ_T(formula-3)

Therein, Ψ_RAnd Ψ_TSemi-correlation matrices, H, for the receiving and transmitting antennas, respectively_wIs a spatially independent rayleigh fading channel matrix.

The spatial fading channel has a certain spatial correlation under the influence of the distance between the antenna elements and the scattering environment. This spatial correlation will affect ε_k K 1.., the numerical distribution of M, thereby affecting the capacity of the equivalent spatial subchannels. When the channel has strong correlation, epsilon_kK is 1.. times.mThe value distribution is severely unbalanced, with the largest and smallest singular values differing by up to tens of decibels.

For the conventional V-BLAST system with constant modulation scheme and transmission power in [1], the extremely small channel gain will seriously reduce the received snr, thereby causing a large number of false detection symbols in the high-rate transmission symbols and causing error propagation in the continuous interference cancellation process of V-BLAST, which seriously deteriorates the performance.

For the space-time processing V-BLAST scheme based on singular value decomposition of channel matrix in [3], the channel matrix equivalence between the transmitted symbol vector and the post-processed received symbol vector is

H_*＝U⁺H.V (formula-4)

Wherein H_*Is the equivalent channel matrix of V-BLAST, which will directly affect the detection performance of V-BLAST. Bringing formula-1 into formula-4, and finishing to obtain

H_*Either as D (formula-5)

Obviously, the pre-processing at the transmitting end and the post-processing at the receiving end diagonalize the channel, facilitating the design of further rate and power control. However, the singular value distribution of the pre-processed and post-processed multi-antenna channel matrix is not changed, so that this method cannot improve the modulation scheme and the performance of V-BLAST with constant transmission power.

Disclosure of Invention

The singular value decomposition of the spatial channel matrix without preprocessing is shown in (equation-1).

Wherein,

$D=\left[\begin{array}{cccc}{d}_{11}& 0& ...& 0\\ 0& d_{22}& ...& ...\\ ...& ...& ...& 0\\ 0& ...& 0& d_{\mathrm{MN}}\end{array}\right]$ (formula-6)

In order to make the equivalent space related sub-channels have the same receiving power after passing through the sending end preprocessing unit x, i.e. H_*After singular value decomposition, the diagonal matrix is a unit matrix I, so that

H·X＝U·I·V⁺(formula-7)

Can be solved to obtain:

X＝V·D⁺·V⁺(formula-8)

The transmission code symbol a is coded by using (equation-2) as a code generating matrix, a code vector b can be obtained, and then the code vector b is transmitted to a spatial fading channel in parallel. Fig. 3 presents a block diagram of the V-BLAST system based on spatial precoding proposed herein.

The source bit stream 1 passes through a digital modulator 2 to produce a serial stream of modulated symbols. Then converted into parallel transmit symbol vector a ═ (α) by vector encoder 3₁，α₂，…，α_M)^TThe combined signal is transmitted to the space precoder 9, encoded according to the encoding generator matrix of (expression-8), and the encoded vector b is output as X · a as V · D⁺·V⁺A. Within one symbol interval, the transmit antennas transmit the code vector b to the spatially fading channel H with the same power for a single antenna. A V-BLAST detector 4 based on optimally ordered successive symbol interference cancellation detects the received symbol vector and produces an estimate of the transmitted encoded symbol vector. Then, the signal is sent to a digital demodulator 5 for demodulation.

After the precoding, the channel matrix between the transmitted symbol vector a and the V-BLAST received vector is equivalent to

H_*＝H·V·D⁺·V⁺(formula-9)

Putting (formula-1) into the above formula, and finishing to obtain

H_*＝U·I·V⁺(formula-10)

Obviously, after the spatial precoding, the equivalent spatial subchannels have the same received power, and the capacity of the spatial correlation channel is effectively improved. Compared with the existing preprocessing and post-processing scheme based on channel singular value decomposition, the precoding device designed by the scheme can obviously improve the performance of the traditional V-BLAST system with constant speed and power, and eliminates the hardware overhead brought to a receiver.

Drawings

The accompanying drawings, which are incorporated herein and constitute part of this specification, illustrate embodiments of the invention and, together with the description, explain the invention by way of example.

FIG. 1 shows a block diagram of a V-BLAST based multi-antenna system;

FIG. 2 is a block diagram of a space-time processing V-BLAST system based on singular value decomposition of a channel matrix;

FIG. 3 shows a system block diagram of spatial precoding V-BLAST in the present invention;

FIG. 4 shows a block diagram of a spatial precoding system in accordance with the present invention;

FIG. 5(a) shows a comparison of the performance of 4-way V-BLAST on spatially weakly correlated channels;

FIG. 5(b) shows the performance comparison of 4-way V-BLAST on the strongly spatially correlated channels;

FIG. 5(c) shows a comparison of performance of 2-way V-BLAST on spatially weakly correlated channels;

FIG. 5(d) shows the performance comparison of 2-way V-BLAST on strongly spatially correlated channels.

Detailed Description

The following detailed description of the embodiments refers to the accompanying drawings.

Fig. 4 shows a complete system block diagram of spatial pre-coding V-BLAST. At the transmitting end, the information bit stream output by the source encoder 41 is subjected to the digital modulator 2 and the serial-parallel conversion 42 to form a modulation symbol vector a, and is sent to the precoder 9 for encoding. The precoder 9 encodes the modulation symbol vector a according to the channel matrix fed back by the reverse channel, outputs a coded vector b, and forms a data frame, which multiplexes the same transmit antennas as the pilot channel. Each transmit antenna unit transmits data and pilot on forward channel 53 with the same transmit power. At the receiving end, the received data of the pilot channel is used to estimate the channel matrix, and the auxiliary V-BLAST detector 4 performs detection of the next data frame and feeds back the channel matrix to the transmitting end through the reverse channel for precoding of the next data frame.

Meanwhile, the received data of the data channel is sent to a V-BLAST detector 4 to detect the transmitted symbol according to the pre-estimated channel matrix, and an estimated transmitted symbol vector is generated and sent to a digital modulator 5 for demodulation and a source decoder 44 for decoding after parallel-serial conversion.

The inventive device described herein works as follows:

step 1: receiving a channel matrix H fed back by a reverse channel, and performing matrix singular value decomposition to obtain a unitary matrix V and a diagonal matrix D shown in a formula-1;

step 2: calculating Hermitian transpose matrix of the matrixes V and D to obtain a matrix V⁺And D⁺；

And step 3: performing matrix multiplication to generate code generation matrix X ═ V.D⁺·V⁺；

And 4, step 4: coding an input modulation symbol vector a by using a code generation matrix to obtain a coded symbol vector b ═ X · a ═ V · D⁺·V⁺A, and output.

The precoder 9 proposed herein can ensure that equivalent spatial subchannels of a multi-antenna system have the same received power under a spatial correlation fading channel, and effectively improve the capacity of the spatial correlation channel. When the method is combined with the V-BLAST technology, the V-BLAST can transmit data at a fixed speed and transmission power, and the high frequency spectrum efficiency is obtained and the good performance is improved.

The performance improvements described above are verified herein by computer simulations.

Simulation conditions are as follows:

modulator/demodulator: 16QAM

Reverse channel: error-free, delay-free feedback channel

Forward channel: flat fading, quasi-static spatially correlated channels. The spatial correlation matrix is generated according to document [5], in which the angular spread of the signals received by the transmitter is 5 degrees, the average arrival angle is 0 degree, the distance between the antenna elements in weak correlation is 4 wavelengths, and the distance between the antenna elements in strong correlation is half wavelength, without cross correlation.

Channel estimation: ideal channel estimation

Number of transmitting/receiving antennas: 2-transmission 2-reception and 4-transmission 4-reception

V-BLAST Detector: MMSE detection

Signal-to-noise ratio: the total received power to noise power ratio for each receive antenna.

Comparison scheme: the spatial pre-coding V-BLAST scheme proposed herein, the conventional V-BLAST detection scheme, the scheme combining the spatial pre-processing method based on the channel SVD decomposition with V-BLAST.

FIG. 5(a) shows a comparison of the performance of 4-way V-BLAST on spatially weakly correlated channels. FIG. 5(b) shows the performance comparison of 4-way V-BLAST on the strongly spatially correlated channels. Figure 5(c) shows a comparison of the performance of 2-way V-BLAST in spatially weakly correlated channels. FIG. 5(d) shows the performance comparison of 2-way V-BLAST on strongly spatially correlated channels.

In the figure, the solid line shows the performance of the spatial precoding V-BLAST proposed herein, and the dashed line with a "+" sign is [1]]Performance of V-BLAST of (1) "^*The dotted line of the symbol is [3]]Performance of the space-time processing V-BLAST scheme based on singular value decomposition of the channel matrix in (1). As is apparent from the performance comparison of fig. 5, the pre-coding V-BLAST proposed herein achieves significant performance improvement at the same spectral efficiency, whether in a weakly correlated channel or a strongly correlated channel. Further, the performance improvement will be more pronounced if the actual channel coding is used.

The technical scheme proposed by the method is suitable for time division multiplexing systems and slow fading channels. At present, in the HSDPA (High-Speed Downlink Packet Access) proposal regarding closed-loop MIMO (Multiple Input Multiple Output), the overhead of increasing reverse channel resources is almost the cost of all proposals. Therefore, the technical scheme provided by the invention obtains obvious performance improvement under the condition of increasing tolerable reverse channel overhead and transmitter implementation complexity, and has good application prospect.

Claims (12)

1. A precoding method in a multiple antenna system based on V-BLAST technology, comprising the steps of:

receiving a channel matrix H of the reverse channel feedback, and determining the feedback channel according to H-U.D.V⁺Matrix singular value decomposition is carried out to obtain a unitary matrix V, U and a diagonal matrix D, wherein D is the diagonal matrix containing H singular values, + represents the Hermite conjugate transpose,

calculating Hermite conjugate transpose matrixes of the matrixes V and D to obtain a matrix V⁺And D⁺；

Performing matrix multiplication operation to obtainGenerating code generating matrix X ═ V.D⁺·V⁺；

Encoding an input modulation symbol vector a by using a code generation matrix X to obtain an encoded symbol vector b ═ X · a ═ V · D⁺·V⁺A, and output.

2. A process according to claim 1, characterized in that

Before the modulation symbol vector a is coded according to the code generating matrix, the modulation symbol vector a is formed by digital modulation and serial-parallel conversion of a source bit stream.

3. A method as claimed in claim 1, characterized in that the data frame is formed from a vector b of code symbols, the data channel multiplexing the same transmit antennas as the pilot channel.

4. The method of claim 1, wherein each transmit antenna unit transmits data and pilot on a forward channel at the same transmit power.

5. The method of claim 1, wherein the received data of the pilot channel is used to estimate a channel matrix to aid V-BLAST detection of a next data frame, and the estimated channel matrix is fed back to the transmitter via a reverse channel for precoding of the next data frame.

6. The method of claim 5, wherein the received data of the data channel is V-BLAST detected based on a pre-estimated channel matrix to produce estimated transmitted symbol vectors, and said transmitted symbol vectors are parallel-to-serial converted, digitally demodulated and decoded.

7. A precoding system for a V-BLAST technique multi-antenna system, characterized in that

The system comprises a source encoder, a digital modulator, a serial-parallel conversion device, a pre-coding device and a transmitting antenna at a transmitting end, comprises a receiving antenna, a channel estimator, a V-BLAST detector, a parallel-serial conversion device, a digital demodulator and a source decoder at a receiving end,

the precoding device precodes the modulation symbol vector from the serial-parallel conversion device, receives the channel matrix H fed back by the reverse channel, and based on H ═ U.D.V⁺Matrix singular value decomposition is carried out to obtain a unitary matrix V, U and a diagonal matrix D, wherein D is the diagonal matrix containing H singular values, and + represents hermitian conjugate transpose; calculating Hermite conjugate transpose matrixes of the matrixes V and D to obtain a matrix V⁺And D⁺(ii) a Performing matrix multiplication to generate code generation matrix X ═ V.D⁺·V⁺(ii) a Encoding an input modulation symbol vector a by using a code generation matrix X to obtain an encoded symbol vector b ═ X · a ═ V · D⁺·V⁺A, and output by the transmitting antenna.

8. Precoding system according to claim 7, characterized in that

At the transmitting end, the information bit stream output by the information source coder forms a modulation symbol vector a after passing through a digital modulator and a serial-parallel conversion device.

9. Precoding system according to claim 7, characterized in that the coded symbol vectors b output by the precoding means form data frames, and that the data channels multiplex the same transmit antennas as the pilot channels.

10. Precoding system as claimed in claim 7, characterized in that each transmit antenna unit transmits data and pilot on the forward channel with the same transmit power.

11. Precoding system according to claim 9, characterized in that at the receiving end the channel estimator estimates the channel matrix using the received data of the pilot channel to assist the V-BLAST detector in the detection of the next data frame and feeds back the estimated channel matrix to the precoding means for precoding of the next data frame via the back channel.

12. The precoding system of claim 10 wherein for received data for the data channel, the V-BLAST detector V-BLAST detects the received data based on a pre-estimated channel matrix to produce an estimated transmitted symbol vector, the transmitted symbol vector being converted by the parallel to serial conversion device, demodulated by the digital demodulator, and decoded by the source decoder.

Images