CN100544327C - A Low Complexity Minimum Mean Square Error Serial Interference Removal Detector - Google Patents

Abstract

The present invention provides a Minimum Mean Square Error Serial Interference Cancellation (MMSE-SIC) detector suitable for multiple-input multiple-output (MIMO) systems, the main features of which are: minimum mean square error (MMSE) for each detection The weighted vector can be obtained through a simple recursive method, and a sorting scheme based on the MMSE sorting criterion is directly provided while the recursive calculation is performed. Compared with the traditional MMSE-SIC detector, the detector provided by the invention greatly reduces the computational complexity without any performance loss. In addition, the present invention is also applicable to other communication systems that can be modeled as MIMO systems, such as Code Division Multiple Access (CDMA) systems.

Description

A kind of detector for serial interference deletion in minimum mean square error of low complex degree

Technical field

The invention belongs to wireless communication technology field, relate to a kind of detector that is applied to multiple-input, multiple-output (MIMO) system, this detector is applicable to that also other can be modeled as the communication system of mimo system simultaneously, for example code division multiple access (CDMA) system.

Background technology

Development along with cellular mobile communication, multimedia service, the capacity requirement of radio communication is increasing rapidly in the world wide, and available radio frequency resources is very limited, how to provide bigger channel capacity to become the main challenge of following high-speed radiocommunication system development in limited frequency band.Mimo system is a kind of wireless communication system that utilizes many transmit antennas and Duo Gen reception antenna to carry out transfer of data, and very large channel capacity can be provided, and its availability of frequency spectrum and number of antennas are linear under desirable propagation conditions.Because mimo system has the high availability of frequency spectrum, therefore be considered to one of Main physical layer technology of following high-speed radiocommunication system.

Serial interference deletion in minimum mean square error (MMSE-SIC) detector is a kind of effective detector that is applicable to mimo system, it detects the data symbol of transmission successively from a plurality of receiving data streams according to least mean-square error (MMSE) criterion, after detecting a certain transmission symbol, the caused interference of this symbol is deleted from received signal, and then detect the next symbol that sends.Its detection order can be determined according to certain ranking criteria (as: least mean-square error ranking criteria, maximum Signal to Interference plus Noise Ratio ranking criteria etc.).In traditional MMSE-SIC detector owing to will repeatedly carry out complicated matrix inversion and sort operation when calculating the MMSE matrix, when the dual-mode antenna number more for a long time, its computation complexity is very high.

How under the prerequisite that guaranteed performance does not descend, the computational complexity that reduces detector is the key that this detector carries out practical application.

Summary of the invention

The object of the present invention is to provide a kind of MMSE-SIC detector, it is guaranteeing to reduce computational complexity significantly under the constant prerequisite of detection performance.

The technical scheme of MMSE-SIC detector provided by the invention is: this detector comprises N continuous time detection (N represents number of transmit antennas) in same sending time slots, when detecting for the i time, by expansion weighting matrix W _iAcquisition is used for the weighing vector ω of this detection _iWith its corresponding transmitting antenna sequence number k _i, then to the emission symbol

Detect and obtain its decision value

And from received signal, delete

To other not interference of detection signal.Its key point is to expand weighting matrix W _iAdopt the recursion Calculation Method to obtain weighing vector ω _iWith its corresponding transmitting antenna sequence number k _iCan be from W _iIn directly obtain.

W _iRecursive algorithm as follows:

1, calculates expansion weighting matrix W ₁=R _N, R _NCan be by R ₁, R ₂..., R _N-1Progressively recursion obtains:

When j=1, calculate $R_j={({\left|\right|{\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="C200510080117E00101.png,C200510080117E00111.png"\; state="\{\{state\}\}">h</figure-callout>}}_1\left|\right|}^2+{\mathrm{\&sigma;}}^2)}^{-1}\left[\begin{array}{cc}{h}_1^H& \mathrm{\&sigma;}\end{array}\right].$ Here, h _jThe j row of expression channel matrix H; σ represents the standard deviation of noise; () ^HThe expression conjugate transpose; ‖ ‖ represents the Frobenius norm.

When 2≤j≤N, $R_j=\left[\begin{array}{cc}{R}_{j-1}-G_j& -{\mathrm{\&beta;}}_j{d}_j\\ g_j^H& {\mathrm{\&beta;}}_j\end{array}\right].$ Here, $d_j={\tilde{R}}_{j-1}{h}_j,$

Expression is by R _J-1The matrix that constitutes of preceding M row; M represents the reception antenna number; β _j=σ α _j, α _j=(σ ²+ ‖ f _j‖ ²+ σ ²‖ d _j‖ ²) ^-1, f _j=h _j-H _J-1d _j, H _J-1The matrix that the preceding j-1 row of expression channel matrix H constitute; $g_j={\left[\begin{array}{cc}{\mathrm{\&alpha;}}_j{f}_j^T& -{\mathrm{\&beta;}}_j{d}_j^T\end{array}\right]}^T,$ () ^TThe expression transposition; $G_j=d_j{g}_j^H.$

2, calculate expansion weighting matrix W _i(i=2,3 ..., N).

${\stackrel{\&OverBar;}{W}}_i=V_{i-1}-{\left|\right|p_{i-1}\left|\right|}^{-2}{\mathrm{\&Lambda;}}_{i-1}{p}_{i-1}^H{q}_{i-1}$

Here, Λ _I-1Expression deletion W _I-1L _I-1The matrix that obtains behind the row; V _I-1Expression deletion Λ _I-1M+l _I-1The matrix that obtains behind the row; p _I-1Expression W _I-1L _I-1OK; q _I-1Expression deletion p _I-1M+l _I-1The vector that obtains behind the individual element.

Obtain ω _iAnd k _iMethod as follows:

1, extracts expansion weighting matrix W _iPreceding M row obtain MMSE weighting matrix W _i, extract expansion weighting matrix W _iThe back N-i+1 row matrix Z that obtains sorting _i

2, take out Z _iThe capable sequence number l of minimum diagonal entry correspondence _i, from W _iExtract l _iRow obtains the weighing vector ω of 1 * M dimension _i, ω _iPairing transmitting antenna sequence number is k _i

Implementing beneficial effect of the present invention is: compare with traditional MMSE-SIC detector, MMSE-SIC detector provided by the invention adopted a kind of simply, recursive algorithm calculates and to detect required MMSE weighing vector each time efficiently, sequencing schemes based on the MMSE ranking criteria when calculating, recursion directly is provided, this processing method has greatly reduced the computational complexity of detector, has guaranteed that simultaneously the performance of detector is not subjected to any loss.

Description of drawings

Fig. 1 is the basic principle block diagram of mimo system.

Fig. 2 is the flow chart that the MMSE-SIC detector detects for the i time.Here, i=1,2 ..., N.

Fig. 3 calculates weighing vector ω _iWith its corresponding transmitting antenna sequence number k _iFlow chart.Here, i=1,2 ..., N.

Fig. 4 calculates expansion weighting matrix W ₁Flow chart.

Fig. 5 calculates expansion weighting matrix W _iFlow chart.Here, i=2,3 ..., N.

Fig. 6 is the performance comparison diagram (QPSK) of MMSE-SIC detector provided by the invention and traditional MMSE-SIC detector.

Fig. 7 is the performance comparison diagram (16QAM) of MMSE-SIC detector provided by the invention and traditional MMSE-SIC detector.

Embodiment

The present invention will be described in detail below by drawings and Examples.

Detector provided by the invention is applicable to mimo system, or can be modeled as other communication system of mimo system.For example, the present invention can directly be used as the multi-user detector of cdma system.Be that example is described below with the mimo system.

Fig. 1 is the basic principle block diagram of mimo system.At transmitting terminal, data bit at first is mapped to and is the signal in the signal constellation (in digital modulation), through forming a plurality of parallel baseband signals behind the serial to parallel conversion, launches simultaneously from a plurality of different antennas respectively after ovennodulation then; After the wireless channel decline, signal and noise stack back from different transmit antennas are received simultaneously by a plurality of antennas, through generating a plurality of parallel baseband signals after the demodulation, the channel condition information that the MIMO detector utilizes channel estimator to produce recovers initial data from baseband signal.In the real system, data bit can will pass through deinterleaving and decoding accordingly earlier through encoding and interweaving before the receiver dateout before mapping.The mathematic(al) representation of this system's baseband signal input/output relation can be expressed as:

y＝Hx+ε (1)

In the following formula, x=[x ₁x ₂X _N] ^TThe expression emission signal vector, N represents number of transmit antennas, () ^TThe expression transposition, x _nExpression is from the signal of n transmit antennas emission; ε=[ε ₁ε ₂ε _M] ^TThe expression noise vector, M represents reception antenna number, ε _mRepresent the noise that m root reception antenna receives; Y=[y ₁y ₂Y _M] ^TThe expression received signal vector, y _mRepresent the signal that m root reception antenna receives; H is the matrix of M * N dimension, the equivalent baseband channel matrix of expression mimo system; Before carrying out MIMO detection processing, at first to obtain the estimated value of channel matrix by channel estimator, suppose that here receiver can free from errorly estimate channel matrix, for convenience of description, still is designated as H to the estimated value of channel matrix in the literary composition.

Fig. 2 be the flow chart that detects of MMSE-SIC detector the i time (i=1,2 ..., N).The step of this flow process is as follows:

Step 1: calculate weighing vector ω _iWith its corresponding transmitting antenna sequence number k _i(concrete grammar is seen Fig. 3).

Step 2: calculate

Decision value ${\hat{x}}_{k_i}=Q({\mathrm{\&omega;}}_i,y_i).$ Here, symbol Q () expression hard decision; When i=1, y ₁=y, when 2≤i≤N, y _iStep 3 by the i-1 time detection obtains.

Step 3: if i＜N, with signal

Interference from received signal y _iMiddle deletion obtains y _I+1, that is: $y_{i+1}=y_i-h_{k_i}{\hat{x}}_{k_i};$ Otherwise the output decision value finishes algorithm.Here, The k of expression channel matrix H _iRow.

Fig. 3 calculates weighing vector ω _iWith its corresponding transmitting antenna sequence number k _iFlow chart (i=1,2 ..., N).The step of this flow process is as follows:

Step 1: make i=1, recursion is calculated expansion weighting matrix W ₁(concrete grammar is seen Fig. 4).

Step 2: extract expansion weighting matrix W _iPreceding M row obtain weighting matrix W _i, extract expansion weighting matrix W _iThe back N-i+1 row matrix Z that obtains sorting _i

Step 3: take out Z _iThe capable sequence number l of minimum diagonal entry correspondence _i, from W _iThe middle l that extracts _iRow obtains the weighing vector ω of 1 * M dimension of the i time detection _i, the transmitting antenna sequence number k that it is corresponding _iBe vectorial L _iIn l _iThe value of individual element.Here, L _iExpression deletion vector [1 2 ... N] intermediate value equals k ₁, k ₂..., k _I-1Element after the vector that obtains.

Step 4: make i=i+1, if i≤N, by W _I-lRecursion is calculated expansion weighting matrix W _i(concrete grammar is seen Fig. 5) turns to step 2; Otherwise, finish algorithm.

Annotate: the weighing vector ω of the N time detection _NCalculating can obtain by following shortcut calculation:

${\mathrm{\&omega;}}_N={({\left|\right|h_{k_N}\left|\right|}^2+{\mathrm{\&sigma;}}^2)}^{-1}{h}_{k_N}^H$

Here, σ represents the standard deviation of noise; () ^HThe expression conjugate transpose; ‖ ‖ represents the Frobenius norm.

Fig. 4 calculates expansion weighting matrix W _iFlow chart.The step of this flow process is as follows:

Step 1: make j=1, calculate $R_j={({\left|\right|h_1\left|\right|}^2+{\mathrm{\&sigma;}}^2)}^{-1}\left[\begin{array}{cc}{h}_1^H& \mathrm{\&sigma;}\end{array}\right].$

Step 2: make j=j+1, take out R _J-1Preceding M row constitute matrix

Calculate $d_j={\tilde{R}}_{j-1}{h}_j,$

f _j＝h _j-H _j-1d _j，α _j＝(σ ²+‖f _j‖ ²+σ ²‖d _j‖ ²) ^-1，β _j＝σα _j， $g_j={\left[\begin{array}{cc}{\mathrm{\&alpha;}}_j{f}_j^T& -{\mathrm{\&beta;}}_j{d}_j^T\end{array}\right]}^T,$ $G_j=d_j{g}_j^H.$ Here, H _J-1The matrix that the preceding j-1 row of expression channel matrix H constitute.

Step 3: calculate $R_j=\left[\begin{array}{cc}{R}_{j-1}-G_j& -{\mathrm{\&beta;}}_j{d}_j\\ g_j^H& {\mathrm{\&beta;}}_j\end{array}\right].$

Step 4: when j＜N, return step 2; When j=N, make W ₁=R _N, finish algorithm.

Fig. 5 calculates expansion weighting matrix W _i(i=2,3 ..., flow chart N).The step of this flow process is as follows:

Step 1: with W _I-1Be split as vectorial p by row _I-1With matrix Λ _I-1Wherein, p _I-1Be W _I-1L _I-1OK, Λ _I-1Be deletion W _I-1L _I-1The matrix that obtains behind the row.

Step 2: deletion Λ _I-1M+l _I-1Row obtain matrix v _I-1

Step 3: deletion p _I-1M+l _I-1Individual element obtains vectorial q _I-1

Step 4: calculate ${\stackrel{\&OverBar;}{W}}_i=V_{i-1}-{\left|\right|p_{i-1}\left|\right|}^{-2}{\mathrm{\&Lambda;}}_{i-1}{p}_{i-1}^H{q}_{i-1},$ Finish algorithm.

Fig. 6 and Fig. 7 show two groups of performance comparison result of detector provided by the invention and traditional MMSE-SIC detector.That abscissa is represented among the figure is the energy per bit of emission data and the ratio (E of noise power spectral density _b/ N ₀), that ordinate is represented is bit error rate (BER).The dual-mode antenna number of system is 4, and channel is independent identically distributed MIMO flat Rayleigh fading channel, and supposes that receiver can free from errorly estimate channel, and detector adopts the sequencing schemes based on the least mean-square error ranking criteria.The system of Fig. 6 institute emulation adopts the QPSK modulation, and spectrum efficiency is 8bit/s/Hz; The system of Fig. 7 institute emulation adopts the 16QAM modulation, and spectrum efficiency is 16bit/s/Hz.Detector performance provided by the invention as can be seen from Figure and traditional MMSE-SIC detector performance are in full accord.

The computational complexity of following surface analysis MMSE-SIC detector provided by the invention.Here the operand with a complex multiplication is the unit of algorithm complex, ignores addition and subtraction, comparison, selection etc. and relatively simply handles, and only calculates the complexity of multiplication and division.When equating with number of transmit antennas N and reception antenna number M is example, and through calculating as can be known, the complexity of detector provided by the invention is roughly N ³Level, the complexity of traditional MMSE-SIC detector then reaches N ⁴Level.

In sum, detector provided by the invention has reduced computational complexity significantly under the prerequisite of not losing performance.

Claims (6)

1. A low-complexity minimum mean square error serial interference cancellation MMSE-SIC detector suitable for multiple-input multiple-output MIMO systems. The detector contains N consecutive detections in the same transmission time slot, where N represents the transmitting antenna number, it is characterized in that the ith detection of the detector comprises the following steps: a) Calculate the extended weighting matrix W _i of the i-th detection according to the following recursive algorithm: When i=1, W ₁ =R _N , R _N is obtained from R ₁ , R ₂ ,..., R _N-1 step by step recursively; where, when j=1, $R_j={({||h_1||}^2+{\mathrm{\&sigma;}}^2)}^{-1}\left[\begin{array}{cc}{h}_1^h& \mathrm{\&sigma;}\end{array}\right];$ When j=2, 3, ..., N, $R_j=\left[\begin{array}{cc}{R}_{j-1}{-G}_j& {-\mathrm{\&beta;}}_j{d}_j\\ g_j^h& {\mathrm{\&beta;}}_j\end{array}\right];$ Here, _hj denotes the jth column of the channel matrix H, σ denotes the standard deviation of the noise, ‖·‖ denotes the Frobenius norm, ( ) ^H denotes the conjugate transpose, $d_j={\tilde{R}}_{j-1}{h}_j,$

Represents a matrix composed of the first M columns of R _j-1 , M represents the number of receiving antennas, β _j =σα _j , α _j =(σ ² +‖f _j ‖ ² +σ ² ‖d _j ‖ ² ) ^-1 , f _j = h _j -H _j-1 d _j , H _j-1 represents the matrix formed by the first j-1 columns of the channel matrix H, $g_j={\left[\begin{array}{cc}{\mathrm{\&alpha;}}_j{f}_j^T& {-\mathrm{\&beta;}}_j{d}_j^T\end{array}\right]}^T,$ (·) ^T means transpose, $G_j=d_j{g}_j^h;$ When i=2, 3, ..., N, the recursive algorithm of W _i is: first, split W _i-1 into vector p _i-1 and matrix Λ _i-1 by row, here, p _i-1 is the l i _-1 row of W _i-1 , Λ _i-1 is the matrix obtained after deleting the l _i-1 row of W _i-1 ; then, delete the M+l i _{-1 of Λ i- 1} column to get the matrix V _i-1 ; then, delete the M+l _i-1th element of p _i-1 to get the vector q _i-1 ; finally, calculate ${\stackrel{\&OverBar;}{W}}_i=V_{i-1}-{\left|\right|p_{i-1}\left|\right|}^{-2}{\mathrm{\&Lambda;}}_{i-1}{p}_{i-1}^h{q}_{i-1};$ b) Obtain the minimum mean square error MMSE weighting matrix W _i and sorting matrix Z _i of the ith detection according to W _i ; c) Take out the row number l _i corresponding to the smallest diagonal element of Z _i , and take out the l _{ith row} of W _i to obtain the 1×M-dimensional weighted vector ω _i of the ith detection, where M represents the number of receiving antennas; d) Use the weight vector ω _i to recover the data symbols sent by the k _ith transmit antenna corresponding to it

If i ≤ N-1, remove from received data

Interference to other undetected signals, where _ki is the serial number of the transmitting antenna corresponding to ω _i . 2. The detector according to claim 1, wherein the first M columns of the extended weight matrix W _i are the MMSE weight matrix W _i detected for the ith time. 3. The detector according to claim 1, wherein the last N-i+1 columns of the extended weighting matrix W _i are the sorting matrix Z _i for the ith detection. 4. The detector according to claim 1, wherein the formula for calculating the weighting vector ω _N is: ${\mathrm{\&omega;}}_N={({\left|\right|h_{k_N}\left|\right|}^2+{\mathrm{\&sigma;}}^2)}^{-1}{h}_{k_N}^h.$ 5. The detector according to claim 1, characterized in that the detector can be sorted according to other sorting criteria. 6. The detector according to claim 1, characterized in that the detector is adapted for a communication system which can be modeled as a MIMO system.

Images