CN101582742A - Method for detecting iteration of multiple input multiple output (MIMO) system, system thereof and device thereof - Google Patents

Abstract

The invention discloses a method for detecting the iteration of a multiple input multiple output (MIMO) system, a system thereof and a device thereof. The method comprises the following steps of: when initializing an iteration process, adopting a soft in soft out parallel soft interference cancellation MMSESIC MIMO detecting method with a signal which is received by a receiver and a channel fading matrix which is estimated by the channel; when continuing the iteration process, taking the sum of residual interference and noise which are deleted by the parallel soft interference as an equivalent noise with the signal which is received by the receiver, bit priori information which is outputted by a channel encoder in the former iteration and a channel fading matrix which is estimated by the channel, wherein, the each component of the equivalent noise is similar to be mutually independent gaussian random variables, and executing a soft in soft out MMSESIC MIMO detecting computation under the model; and hard deciding code bit posterior information which is outputted by the soft in soft out channel encoder, executing a CRC check, outputting the hard decision result of the information source code bit posterior information, and taking the result as the final result of interative detection.

Description

Mimo systems iteration detection method, system and equipment

Technical field

The present invention relates to the wireless communication system technologies field, be meant a kind of mimo systems iteration detection method, system and equipment especially.

Background technology

Along with the rapid increase of wireless communication user quantity and developing rapidly of broadband multimedia services, people have proposed more and more higher requirement to the transmission rate and the performance of wireless communication system.In the today of being becoming tight frequency spectrum resource day, multiple-input, multiple-output (MIMO:Multiple Input MultipleOutput) technology receives much concern because of transmission quality and the spectrum efficiency that it can significantly improve wireless communication system.Present research thinks that for mimo system, iterative receiver is the effective way of approaching the mimo channel capacity.In iterative receiver, the MIMO detector should be realized soft inputting and soft output.Soft under the optimum meaning of bit error rate performance goes into the soft (SISO of going out, Soft InputSoft Output) the MIMO detector is maximum likelihood (ML, Maximum Likelihood) detector, but its complexity increases along with the big or small exponentially of number of antennas and modulation constellation, can't practical application.Therefore, the MIMO detection algorithm of high-performance low complex degree becomes the important content in the iterative receiver research.

In the research at iterative receiver, extensively adopt serial interference delete (SIC, SerialInterference Cancellation) MIMO detection algorithm at present based on least mean-square error (MMSE, Minimum Mean Square Error).Yet, softly to go into the soft MMSESIC of going out MIMO and detect and relate to the complex matrix inversion operation, complexity is still higher.In order further to reduce complexity, Takumi ITO, Xiaodong Wang, people such as Yoshikazu Kakura are at Performance comparison of MF and MMSE combined iterative softinterference canceller and V-BLAST technique in MIMO/OFDM systems, VTC 2003-Fall, October 2003, (1): among the 488-492, proposition adopts the soft soft MMSE SIC MIMO of going out of going into to detect in primary iteration is handled, and in successive iterations is handled, adopt soft soft matched filter (MF, Match Filter) the SIC MIMO of going out of going into to detect.Be respectively N at the dual-mode antenna number _r, N _tMimo system in, this scheme under the less prerequisite of performance of BER loss, with the complexity of iterative receiver from O (N _r ³N _t) be reduced to O (N _r ²N _t).Yet this scheme does not consider that the MIMO detector can offset many interference between antennas preferably, and therefore, in fact the successive iterations stage has still obtained less performance gain with higher complexity.

Summary of the invention

In view of this, the objective of the invention is to propose a kind of mimo systems iteration detection method, system and equipment, under the prerequisite that iterative detection performance does not significantly descend, reduce it and handle complexity.

Based on above-mentioned purpose a kind of multiple-input, multiple-output mimo system iteration detection method provided by the invention, comprising:

When A. primary iteration is handled, utilize the signal that receiver receives, the channel fading matrix that channel estimating obtains, adopt the soft soft serial interference deletion in minimum mean square error multiple-input, multiple-output MMSE of the going out SIC MIMO detection method of going into;

When B. successive iterations is handled, utilize the bit prior information of channel decoder output in signal that receiver receives, the last iteration, and the channel fading matrix that obtains of channel estimating, and the residual interference behind the soft interference delete that will walk abreast and noise sum are as equivalent noise, each component of equivalent noise is approximately separate Gaussian random variable, carries out the soft soft parallel soft interference delete multiple-input, multiple-output MMSE PSIC MIMO detection computations of least mean-square error that goes out of going under this model;

C. go into the soft coded-bit posterior information that goes out channel decoder output and carry out hard decision soft, and make CRC check, if this encoding block of CRC check is incorrect, and the iterations that has carried out then returns step B less than total iterations N; Otherwise enter step D;

D. iterative detection finishes, and the hard decision result of output source bits posterior information is as the final result of iterative detection.

Optionally, an iterative process in this method comprises: soft go into signal that the soft MIMO of going out detector utilizes receiver to receive, channel fading matrix that channel estimating obtains and last iteration handle in the prior information of channel decoder output, the external information of calculation code bit, this external information is through deinterleaver, becomes the soft soft input prior information that goes out channel decoder of going into.Channel decoder is according to prior information and code structure, and the coded-bit external information that calculates is used for the next iteration processing that iteration receives as prior information behind interleaver.

Optionally, this method is in first time iteration, and the prior information of channel decoder output was 0 during described last iteration was handled.

Optionally, when the described successive iterations of this method is handled, carry out soft calculating of going into its weighing vector in the soft MMSE of the going out PSICMIMO detection computations process and do not comprise matrix inversion.

Optionally, this method is described softly goes into the soft MMSE of going out PSIC MIMO detection computations process and may further comprise the steps:

In last iteration, softly go into the soft bit prior information that goes out channel decoder output, calculate the average E (s of all symbols to be detected _i) and variance var (s _i), (i=1,2, L, N _t); According to E (s _i), the channel fading matrix H that signal r that receiver receives and channel estimating obtain calculates receiving terminal after PSIC handles, symbol s _kEquivalent single-shot overcharge received signal r ^kAs follows:

$r^k=r-\underset{j\≠k}{\overset{N_t}{\mathrm{\Σ}}}{h}_{j}E\left(s_j\right)---\left(1\right);$

Signal calculated s _kEstimated value ${\hat{s}}_k=w^k{r}^k$ Wherein $w=[w_1^k,w_2^k,...,w_{N_r}^k]$ , its i element calculates by following formula (2):

$w_i^k=\frac{h_{\mathrm{ik}}^{*}}{(\underset{j=1}{\overset{N_r}{\mathrm{\Σ}}}{\left|h_{\mathrm{jk}}\right|}^2+{\mathrm{\σ}}_{n^k}^2\left(i\right))}---\left(2\right)$

Wherein, ${\mathrm{\σ}}_{n^k}^2\left(i\right)=[\underset{j=1}{\overset{N_t}{\mathrm{\Σ}}}\left({\left|h_{\mathrm{ij}}\right|}^2\mathrm{var}\left(s_j\right)\right)+{\mathrm{\σ}}_n^2]-{\left|h_{\mathrm{ik}}\right|}^2\mathrm{var}\left(s_k\right)--\left(3\right);$

Calculate according to formula (4) (5) and to be mapped to symbol s _kThe external information λ of coded-bit ₁[b _j ^k]:

$p\left\{{\hat{s}}_k\right|s_k=S_l\}=\frac{1}{\mathrm{\π}{\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2}{{\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)})---\left(4\right)$

μ _S-MMSE(k)＝w ^kh _k

Wherein, ${\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)={\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)-{\left[{\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)\right]}^2---\left(5\right)$

${\mathrm{\λ}}_1\left[b_j^k\right]=\log \frac{\underset{S_l\∈S_j^{+}}{\mathrm{\Σ}}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2\underset{m=1,m\≠j}{\overset{M}{\mathrm{\Π}}}P(b_m^k=B_m^l)}{{\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)})}{\underset{S_l\∈S_l^{-}}{\mathrm{\Σ}}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2\underset{m=1,m\≠j}{\overset{M}{\mathrm{\Π}}}P(b_m^k=B_m^l)}{{\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)})}---\left(6\right).$

Optionally, the described symbol s of this method _kEquivalent single-shot overcharge received signal r ^kComputational process in, symbol soft estimate E (s _i) obtain by following process:

Utilize the soft soft bit prior information λ that goes out channel decoder output that goes into ₂[b _{π (i)}], according to the principle of averaging, the soft estimate E (s of compute sign _i), the prior information that this estimated value detects as MMSE PSICMIMO:

$E\left(s_i\right)=\underset{S_j\∈{\mathrm{\Ω}}_S}{\mathrm{\Σ}}{S}_j\underset{m=1}{\overset{{\log }_{2}M}{\mathrm{\Π}}}\frac{\exp \left(B_m^j{\mathrm{\λ}}_2\right[b_m^i\left]\right)}{1+\exp \left(B_m^j{\mathrm{\λ}}_2\right[b_m^i\left]\right)}(i=\mathrm{1,2},...,N_t)---\left(8\right)$

Optionally, the variance of symbol soft estimate obtains by following process in this method formula (3):

Utilize prior information λ ₂[b _{π (i)}], according to the principle of averaging, the soft estimate of compute sign mould square:

$E\left({\left|s_i\right|}^2\right)=\underset{S_j\∈{\mathrm{\Ω}}_S}{\mathrm{\Σ}}{\left|S_j\right|}^2\underset{m=1}{\overset{{\log }_{2}M}{\mathrm{\Π}}}\frac{\exp \left(B_m^j{\mathrm{\λ}}_2\right[b_m^i\left]\right)}{1+\exp \left(B_m^j{\mathrm{\λ}}_2\right[b_m^i\left]\right)}(i=\mathrm{1,2},...,N_t)---\left(9\right)$

Utilize the soft estimate of symbol soft estimate that formula (8), (9) calculate and symbol mould square, according to the computing formula of variance, the variance of compute sign soft estimate:

var(s _i)＝E(|s _i| ²)-E(s _i) ²(i＝1，2，…，N _t) (10)

Optionally, the described total iterations of this method is the fixed numbers of presetting, and perhaps determines adaptively according to certain criterion in iterative process.

Based on above-mentioned purpose, the present invention also provides a kind of multiple-input, multiple-output mimo system iterative detection system, comprise: reception antenna, channel estimator, softly go into the soft MIMO of going out detector, deinterleaver, softly go into soft go out channel decoder and interleaver, above-mentioned iteration detection method is adopted in wherein said soft detection of going into the soft MIMO of going out detector.

Based on above-mentioned purpose, the present invention also provides a kind of multiple-input, multiple-output mimo system iterative detection equipment, and this equipment is arranged on the soft of mimo system receiver goes in the soft MIMO of the going out detector, adopts above-mentioned iteration detection method.

From above as can be seen, the present invention proposes a kind of low-complexity MIMO system iterative detection method based on MMSE, system and equipment, the soft external information that the soft MMSE of going out SIC MIMO detection algorithm obtains coded-bit of going into is adopted in iterative processing for the first time to mimo system, soft soft MMSE PSIC (the parallel soft interference delete that goes out to simplify of going into is adopted in follow-up iterative processing, Parallel Soft Interference Cancelation) the MIMO detection algorithm obtains the external information of coded-bit, and, determine the iterative processing number of times of iterative receiver reality jointly by CRC check result and total iterations N.Like this, can under the prerequisite that iterative detection performance does not significantly descend, significantly reduce it and handle complexity.By simulation analysis as can be known, use the present invention, the overall performance of system does not significantly descend, and complexity is from O (N _r ³N _t) be reduced to O (N _rN _t).And the present invention uses simply, and is very little to the existing system change, has compatible preferably with existing system.

Description of drawings

Fig. 1 is the structural representation of prior art mimo system transmitting terminal;

Fig. 2 is the structural representation of embodiment of the invention MIMO receiving system;

Fig. 3 is the schematic flow sheet of the embodiment of the invention based on the low-complexity MIMO system iterative detection method of MMSE;

Fig. 4 is traditional Iterative detection algorithm and the performance comparison schematic diagram of Iterative detection algorithm of the present invention under an instantiation.

Embodiment

For making the purpose, technical solutions and advantages of the present invention clearer, below in conjunction with specific embodiment, and with reference to accompanying drawing, the present invention is described in more detail.

Referring to shown in Figure 1, the mimo system transmitter comprises: information source maker 101, channel encoder 102, interleaver 103, modulator 104, deserializer 105 and transmitting antenna 106.

Information source { a that information source maker 101 generates _iGenerate symbol sebolic addressing { s after the modulation through channel encoder 102 coding, interleaver 103 Bit Interleaves and modulator 104 _i, symbol sebolic addressing { s _iThrough deserializer 105 serial to parallel conversion demultiplexing signal flow [s ₁, s ₂..., s _Nt], launch by transmitting antenna 106.

Referring to shown in Figure 2, the mimo system receiver comprises: reception antenna 201, channel estimator 202, softly go into the soft MIMO of going out detector 203, deinterleaver 204, softly go into soft go out (SISO) channel decoder 205 and interleaver 206.

Multiple signals that the mimo system transmitter sends stream is through mimo channel, received by the many reception antennas 201 of receiver, and any iterative process comprises: softly go into the signal that the soft MIMO of going out detector 203 utilizes reception antenna 201 to receive

SISO channel decoder 205 calculates and passes through the bit prior information λ of bit interleaver 204 gained in (pass of r and transmitter s emission signal s is r=Hs+n, and wherein H is a mimo channel decline matrix, and n is a white Gaussian noise), the last iterative processing ₂[b _{π (i)}] the channel fading matrix H that obtains of (for the first time iterative processing in, this information is 0) and 202 pairs of channel estimating of channel estimator, go into the soft MIMO detector 203 that goes out, the external information λ of calculation code bit by soft ₁[b _{π (i)}], this external information generates λ through deinterleaver 204 bit deinterleavings ₁[b _i], as the input prior information of SISO channel decoder 205.SISO channel decoder 205 is according to described prior information and coding structure, the coded-bit external information λ that calculates ₂[b _i], behind interleaver 206 Bit Interleaves, generate λ ₂[b _{π (i)}], be used for iterative receiver iterative processing next time as prior information.

The low-complexity MIMO system iterative detection method that the present invention proposes based on MMSE,

Specifically comprise step:

Step 1: primary iteration is handled.

Utilize the signal r that receiver receives, the channel fading matrix H that channel estimating obtains, adopt the soft soft MMSE SIC of the going out MIMO detection method of going into, the external information of calculation code bit, this external information be through deinterleaver, becomes the soft soft input prior information λ that goes out channel decoder that goes into ₁[b _i].The SISO channel decoder is according to prior information and code structure, and the coded-bit external information that calculates is used for iterative receiver iterative processing next time as prior information behind interleaver.

Step 2: successive iterations is handled.

Utilize the bit prior information λ of channel decoder output in signal r that receiver receives, the last iteration ₂[b _{π (i)}], and the channel fading matrix H that obtains of channel estimating adopts the soft soft iterative processing that MMSE PSIC MIMO detects that goes out to simplify of going into.

The iterative processing that the MMSE PSIC MIMO that simplifies in the present embodiment detects mainly comprises: residual interference behind the soft interference delete that will walk abreast and noise sum are as equivalent noise, each component of equivalent noise is approximately separate Gaussian random variable, carries out the soft soft MMSE of the going out PSIC MIMO detection computations of going under this model.Wherein, the calculating of weighing vector can not comprise matrix inversion.

At sending signal s _kDetection, specifically may further comprise the steps:

In last iteration, softly go into the soft bit prior information λ that goes out channel decoder output ₂[b _{π (i)}], calculate the average E (s of all symbols to be detected _i) and variance var (s _i), (i=1,2, L, N _t), then according to E (s _i), r and H calculate receiving terminal after PSIC handles, symbol s _kEquivalent single-shot overcharge (SIMO, Single Input Multiple Output) received signal r ^k

$r^k=r-\underset{j\≠k}{\overset{N_t}{\mathrm{\Σ}}}{h}_{j}E\left(s_j\right)---\left(1\right);$

Signal calculated s _kEstimated value ${\hat{s}}_k=w^k{r}^k$ , wherein w is the weighing vector of MMSE filtering, $w=[w_1^k,w_2^k,...,w_{N_r}^k]$ , its i element calculates by following formula (2) (3):

$w_i^k=\frac{h_{\mathrm{ik}}^{*}}{(\underset{j=1}{\overset{N_r}{\mathrm{\Σ}}}{\left|h_{\mathrm{jk}}\right|}^2+{\mathrm{\σ}}_{n^k}^2\left(i\right))}---\left(2\right)$

${\mathrm{\σ}}_{n^k}^2\left(i\right)=[\underset{j=1}{\overset{N_t}{\mathrm{\Σ}}}\left({\left|h_{\mathrm{ij}}\right|}^2\mathrm{var}\left(s_j\right)\right)+{\mathrm{\σ}}_n^2]-{\left|h_{\mathrm{ik}}\right|}^2\mathrm{var}\left(s_k\right)--\left(3\right);$

Wherein, || ^*The expression complex conjugate, σ _n ²The noise variance of expression receiver, h _IjExpression is positioned at the element that channel matrix H i is capable, j is listed as.

According to obtaining the sign estimation value , according to formula (4) (5) compute sign prior probability $p\left\{{\hat{s}}_k\right|s_k=S_l\}:$

$p\left\{{\hat{s}}_k\right|s_k=S_l\}=\frac{1}{\mathrm{\π}{\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2}{{\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)})---\left(4\right)$

μ _S-MMSE(k)＝w ^kh _k

${\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)={\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)-{\left[{\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)\right]}^2---\left(5\right)$

H wherein _kThe k row of expression channel matrix H.According to prior information $p\left\{{\hat{s}}_k\right|s_k=S_l\},$ Calculate and be mapped to symbol s _kThe outer log-likelihood ratio λ of coded-bit ₁[b _j ^k] (being external information).

${\mathrm{\λ}}_1\left[b_j^k\right]=\log \frac{\underset{S_l\∈S_j^{+}}{\mathrm{\Σ}}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2\underset{m=1,m\≠j}{\overset{M}{\mathrm{\Π}}}P(b_m^k=B_m^l)}{{\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)})}{\underset{S_l\∈S_l^{-}}{\mathrm{\Σ}}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\μ}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2\underset{m=1,m\≠j}{\overset{M}{\mathrm{\Π}}}P(b_m^k=B_m^l)}{{\mathrm{\σ}}_{S-\mathrm{MMSE}}^2\left(k\right)})}---\left(6\right)$

Above-mentioned MIMO detects the external information of output through after the deinterleaving, goes into the soft prior information that goes out channel decoder as soft.Channel decoder is according to prior information and code structure, the posterior information of calculation code bit and external information, and wherein, external information is gone into the soft prior information that goes out the MIMO detector as soft in the next iterative processing behind interleaver.

Step 3: iteration stopping is judged.

Coded-bit posterior information to the output of SISO channel decoder is carried out hard decision, and makes CRC check.If this encoding block of CRC check is incorrect, and the iterations that has carried out then returns step 2 less than total iterations N; Otherwise enter step 4.Wherein, total iterations N may be the fixed numbers of presetting, and also may determine adaptively according to certain criterion in iterative process.

Step 4: iterative detection finishes.

Output source bits posterior information declare the result firmly, as the final result of iterative detection.Iterative detection finishes.

With an instantiation technical solution of the present invention is explained in further detail below, wherein parameter setting does not influence generality.

Set up departments the system in dual-mode antenna count N _r=N _t=4, the Turbo code generator polynomial, in interweave and hole knockout with reference to 3GPP standard 3GPP TS 25.222, Multiplexing and ChannelCoding (TDD), 2004, the coding block length is 5112, code check 1/2, the SISO decoder adopts the MAX-LOG-MAP algorithm, and the decoding iterations is 8 times, the QPSK modulation, default maximum iteration time N is 4, and channel is smooth uncorrelated Rayleigh channel, and is the ideal communication channel estimation.

As shown in Figure 1, source bits { a _iThrough coding, interweave and 2 ^MThe M-QAM modulation back on rank generates symbol sebolic addressing { s _i, { s _iBe divided into N through serial to parallel conversion _tThe road signal flow is launched by transmitting antenna, through mimo channel H, by receiving terminal N _rThe root reception antenna receives.Ignore time parameter, the signal phasor of establishing the transmitting antenna transmission is $s={[{\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="A2009100866500018H2.png,A2009100866500020H1.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="A2009100866500018H2.png,A2009100866500020H1.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,...,s_{N_t}]}^T$ , the mimo channel matrix $H={\left(h_{\mathrm{ik}}\right)}_{N_r\&times;N_t}$ , then the received signal vector representation is:

$r=\mathrm{Hs}+n=\underset{k=1}{\overset{N_t}{\mathrm{\&Sigma;}}}{h}_k{s}_k+n---\left(7\right)$

Referring to shown in Figure 3, use the low-complexity MIMO system iterative detection method embodiment that the present invention proposes based on MMSE, step is as follows:

Step 301, primary iteration are handled, and the channel fading matrix H of utilizing received signal r and channel estimating to obtain is softly gone into the soft MIMO of going out detector and adopted the soft soft MMSE of the going out SIC MIMO of going into to detect the external information λ that obtains coded-bit ₁[b _{π (i)}], this algorithm particular content sees MathiniSellathurai and Simon Haykin for details, Turbo-BLAST for wirelesscommunications:theory and experiments.IEEE Transactions on SignalProcessing, 2002,50 (10): 2538-2546.This external information is through deinterleaver, becomes the soft soft input prior information that goes out channel decoder of going into.Channel decoder is according to prior information and code structure, the coded-bit external information λ that calculates ₂[b _i], behind interleaver, be used for iterative receiver iterative processing next time as prior information.

Step 302 is utilized the bit prior information λ that channel decoder is exported and interweaved in received signal r, the last iteration ₂[b _{π (i)}], and the channel fading matrix H that obtains of channel estimating adopts the soft soft iterative processing that MMSE PSIC MIMO detects that goes out to simplify of going into.Wherein at sending symbol s _kDetection, may further comprise the steps:

1) utilizes prior information λ ₂[b _{π (i)}], according to the principle of averaging, the soft estimate E (s of compute sign _i), the prior information that this estimated value detects as MMSE PSIC MIMO:

$E\left(s_i\right)=\underset{S_j\&Element;{\mathrm{\&Omega;}}_S}{\mathrm{\&Sigma;}}{S}_j\underset{m=1}{\overset{{\log }_{2}M}{\mathrm{\&Pi;}}}\frac{\exp \left(B_m^j{\mathrm{\&lambda;}}_2\right[b_m^i\left]\right)}{1+\exp \left(B_m^j{\mathrm{\&lambda;}}_2\right[b_m^i\left]\right)}(i=\mathrm{1,2},...,N_t)---\left(8\right)$

Wherein, B _m ^jBe constellation point S _jM bit, S _j∈ Ω _S, Ω _SIt is the modulation constellation points set.

2) utilize prior information λ ₂[b _{π (i)}], according to the principle of averaging, the soft estimate of compute sign mould square:

$E\left({\left|s_i\right|}^2\right)=\underset{S_j\&Element;{\mathrm{\&Omega;}}_S}{\mathrm{\&Sigma;}}{\left|S_j\right|}^2\underset{m=1}{\overset{{\log }_{2}M}{\mathrm{\&Pi;}}}\frac{\exp \left(B_m^j{\mathrm{\&lambda;}}_2\right[b_m^i\left]\right)}{1+\exp \left(B_m^j{\mathrm{\&lambda;}}_2\right[b_m^i\left]\right)}(i=\mathrm{1,2},...,N_t)---\left(9\right)$

3) utilize the soft estimate of symbol soft estimate that formula (8), (9) calculate and symbol mould square, according to the computing formula of variance, the variance of compute sign soft estimate:

var(s _i)＝E(|s _i| ²)-E(s _i) ²(i＝1，2，…，N _t) (10)

4) the symbol soft estimate E (s that utilizes formula (8) to calculate _i), the r soft interference delete that walks abreast to received signal:

$r^k=r-\underset{j\&NotEqual;k}{\overset{N_t}{\mathrm{\&Sigma;}}}{h}_{j}E\left(s_j\right)---\left(1\right)$

Utilize the variance of the symbol soft estimate that formula (10) calculates, calculate by variance residual interference and white Gaussian noise and equivalent noise approximate gained:

${\mathrm{\&sigma;}}_{n^k}^2\left(i\right)=[\underset{j=1}{\overset{N_t}{\mathrm{\&Sigma;}}}\left({\left|h_{\mathrm{ij}}\right|}^2\mathrm{var}\left(s_j\right)\right)+{\mathrm{\&sigma;}}_n^2]-{\left|h_{\mathrm{ik}}\right|}^2\mathrm{var}\left(s_k\right)$

(i＝1，2，…，N _r) (3)

The equivalent noise variance of utilizing formula (3) to calculate, and channel matrix H, each element of calculating MMSE weighing vector

:

$w_i^k=\frac{h_{\mathrm{ik}}^{*}}{(\underset{j=1}{\overset{N_r}{\mathrm{\&Sigma;}}}{\left|h_{\mathrm{jk}}\right|}^2+{\mathrm{\&sigma;}}_{n^k}^2\left(i\right))}---\left(2\right)$

The MMSE weighing vector that utilizes formula (2) to calculate, and the received signal behind the interference delete that obtains of formula (1), estimate sending symbol:

${\hat{s}}_k=w^k{r}^k---\left(11\right)$

The average and the variance of the sign estimation value of calculating formula (11) gained.

μ _S-MMSE(k)＝w ^kh _k

${\mathrm{\&sigma;}}_{S-\mathrm{MMSE}}^2\left(k\right)={\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)-{\left[{\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)\right]}^2---\left(5\right)$

The average and the variance of the sign estimation value of calculating according to formula (5) meet the characteristics of Gaussian Profile in conjunction with the output of MMSE filtering, calculate the external information λ of each bit that is mapped to symbol ₁[b _j ^k].

${\mathrm{\&lambda;}}_1\left[b_j^k\right]=\log \frac{\underset{S_l\&Element;S_j^{+}}{\mathrm{\&Sigma;}}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2\underset{m=1,m\&NotEqual;j}{\overset{M}{\mathrm{\&Pi;}}}P(b_m^k=B_m^l)}{{\mathrm{\&sigma;}}_{S-\mathrm{MMSE}}^2\left(k\right)})}{\underset{S_l\&Element;S_l^{-}}{\mathrm{\&Sigma;}}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2\underset{m=1,m\&NotEqual;j}{\overset{M}{\mathrm{\&Pi;}}}P(b_m^k=B_m^l)}{{\mathrm{\&sigma;}}_{S-\mathrm{MMSE}}^2\left(k\right)})}---\left(6\right)$

Wherein, b _j ^kExpression symbol s _kJ bit, S _l∈ Ω _S, S _j ⁺, S _j ^-Represent that respectively j bit is+1 ,-1 constellation point, and B _m ^lBe constellation point S _lM bit.λ ₁[b _j ^k] detect the external information of output for MIMO.The external information sequence is through after the deinterleaving, as the prior information of SISO channel decoder.The SISO channel decoder is according to prior information and code structure, the posterior information of calculation code bit and external information, and wherein, external information is behind interleaver, as the prior information of MIMO detector in the next iterative processing.

Step 303, iteration stopping are judged, the coded-bit posterior information of SISO channel decoder output is carried out hard decision, and make CRC check.If this encoding block of CRC check is incorrect, and the iterations that has carried out then returns step 302 less than total iterations N=4; Otherwise promptly this encoding block of CRC check is correct, and perhaps the iterations that has carried out enters step 304 more than or equal to total iterations N=4.

Step 304, iterative detection finishes, and the hard decision result of output source bits posterior information is as the final result of iterative detection.Iterative detection finishes.

Can know by inference by the aforementioned calculation step, the MMSE PSIC MIMO detection algorithm of simplification, its complexity is O (N _rN _t) magnitude, and the complexity that traditional MMSE PSIC MIMO detects will reach O (N _r ³N _t) magnitude, hence one can see that, and the low-complexity MIMO system iterative detection method based on MMSE proposed by the invention has significantly reduced the computation complexity that iteration receives.

Fig. 4 is for using the simulation result of the inventive method, wherein dotted line is for using the result of the inventive method the 4th iteration, compare with traditional iteration detection method, the low-complexity MIMO system iterative detection method based on MMSE proposed by the invention, performance are not significant to descend.

Above-described specific embodiment is specific embodiments of the invention only, is not limited to the present invention, and is within the spirit and principles in the present invention all, any modification of being made, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. a multiple-input, multiple-output mimo system iteration detection method is characterized in that,

When A. primary iteration is handled, utilize the signal that receiver receives, the channel fading matrix that channel estimating obtains, adopt the soft soft serial interference deletion in minimum mean square error multiple-input, multiple-output MMSE of the going out SIC MIMO detection method of going into;

When B. successive iterations is handled, utilize the bit prior information of channel decoder output in signal that receiver receives, the last iteration, and the channel fading matrix that obtains of channel estimating, and the residual interference behind the soft interference delete that will walk abreast and noise sum are as equivalent noise, each component of equivalent noise is approximately separate Gaussian random variable, carries out the soft soft parallel soft interference delete multiple-input, multiple-output MMSE PSIC MIMO detection computations of least mean-square error that goes out of going under this model;

C. go into the soft coded-bit posterior information that goes out channel decoder output and carry out hard decision soft, and make CRC check, if this encoding block of CRC check is incorrect, and the iterations that has carried out then returns step B less than total iterations N; Otherwise enter step D;

D. iterative detection finishes, and the hard decision result of output source bits posterior information is as the final result of iterative detection.

2. method according to claim 1, it is characterized in that, described iterative process comprises: soft go into signal that the soft MIMO of going out detector utilizes receiver to receive, channel fading matrix that channel estimating obtains and last iteration handle in the prior information of channel decoder output, the external information of calculation code bit, this external information is through deinterleaver, becomes the soft soft input prior information that goes out channel decoder of going into.Channel decoder is according to prior information and code structure, and the coded-bit external information that calculates is used for the next iteration processing that iteration receives as prior information behind interleaver.

3. method according to claim 2 is characterized in that, in the iteration, the prior information of channel decoder output was 0 during described last iteration was handled for the first time.

4. method according to claim 1 is characterized in that, when described successive iterations is handled, carries out soft calculating of going into its weighing vector in the soft MMSE of the going out PSIC MIMO detection computations process and does not comprise matrix inversion.

5. method according to claim 4 is characterized in that, describedly softly goes into the soft MMSE of going out PSIC MIMO detection computations process and may further comprise the steps:

In last iteration, softly go into the soft bit prior information that goes out channel decoder output, calculate the average E (s of all symbols to be detected _i) and variance var (s _i), (i=1,2, L, N _t); According to E (s _i), the channel fading matrix H that signal r that receiver receives and channel estimating obtain calculates receiving terminal after PSIC handles, symbol s _kEquivalent single-shot overcharge received signal r ^kAs follows:

$r^k=r-\underset{j\&NotEqual;k}{\overset{N_t}{\mathrm{\&Sigma;}}}{h}_{j}E\left(s_j\right)---\left(1\right);$

Signal calculated s _kEstimated value ${\hat{s}}_k=w^k{r}^k,$ Wherein $w=[w_1^k,w_2^k,...,w_{N_r}^k],$ Its i element calculates by following formula (2):

$w_i^k=\frac{h_{\mathrm{ik}}^{*}}{(\underset{j=1}{\overset{N_r}{\mathrm{\&Sigma;}}}{\left|h_{\mathrm{jk}}\right|}^2+{\mathrm{\&sigma;}}_{n^k}^2\left(i\right))}---\left(2\right)$

Wherein, ${\mathrm{\&sigma;}}_{n^k}^2\left(i\right)=[\underset{j=1}{\overset{N_t}{\mathrm{\&Sigma;}}}\left({\left|h_{\mathrm{ij}}\right|}^2\mathrm{var}\left(s_j\right)\right)+{\mathrm{\&sigma;}}_n^2]-{\left|h_{\mathrm{ik}}\right|}^2\mathrm{var}\left(s_k\right)---\left(3\right);$

Calculate according to formula (4) (5) and to be mapped to symbol s _kThe external information λ of coded-bit ₁[b _j ^k]:

$p\left\{{\hat{s}}_k\right|s_k=S_l\}=\frac{1}{\mathrm{\&pi;}{\mathrm{\&sigma;}}_{S-\mathrm{MMSE}}^2\left(k\right)}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2}{{\mathrm{\&sigma;}}_{S-\mathrm{MMSE}}^2\left(k\right)})---\left(4\right)$

μ _S-MMSE(k)＝w ^kh _k

Wherein, ${\mathrm{\&sigma;}}_{S-\mathrm{MMSE}}^2\left(k\right)={\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)-{\left[{\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)\right]}^2---\left(5\right)$

${\mathrm{\&lambda;}}_1\left[b_j^k\right]=\log \frac{\underset{S_l\&Element;S_j^{+}}{\mathrm{\&Sigma;}}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2\underset{m=1,m\&NotEqual;j}{\overset{M}{\mathrm{\&Pi;}}}P(b_m^k=B_m^l)}{{\mathrm{\&sigma;}}_{S-\mathrm{MMSE}}^2\left(k\right)})}{\underset{S_j\&Element;S_j^{-}}{\mathrm{\&Sigma;}}\exp (-\frac{{|{\hat{s}}_k-{\mathrm{\&mu;}}_{S-\mathrm{MMSE}}\left(k\right)S_l|}^2\underset{m=1,m\&NotEqual;j}{\overset{M}{\mathrm{\&Pi;}}}P(b_m^k=B_m^l)}{{\mathrm{\&sigma;}}_{S-\mathrm{MMSE}}^2\left(k\right)})}---\left(6\right).$

6. method according to claim 5 is characterized in that, described symbol s _kEquivalent single-shot overcharge received signal r ^kComputational process in, symbol soft estimate E (s _i) obtain by following process:

Utilize the soft soft bit prior information λ that goes out channel decoder output that goes into ₂[b _{π (i)}], according to the principle of averaging, the soft estimate E (s of compute sign _i), the prior information that this estimated value detects as MMSE PSICMIMO:

$E\left(s_i\right)=\underset{S_j\&Element;{\mathrm{\&Omega;}}_S}{\mathrm{\&Sigma;}}{S}_j\underset{m=1}{\overset{{\log }_{2}M}{\mathrm{\&Pi;}}}\frac{\exp \left(B_m^j{\mathrm{\&lambda;}}_2\right[b_m^i\left]\right)}{1+\exp \left(B_m^j{\mathrm{\&lambda;}}_2\right[b_m^i\left]\right)}$ (i＝1，2，...，N _t) (8)

7. method according to claim 6 is characterized in that, the variance of symbol soft estimate obtains by following process in the formula (3):

Utilize prior information λ ₂[b _{π (i)}], according to the principle of averaging, the soft estimate of compute sign mould square:

$E\left({\left|s_i\right|}^2\right)=\underset{S_j\&Element;{\mathrm{\&Omega;}}_S}{\mathrm{\&Sigma;}}{\left|S_j\right|}^2\underset{m=1}{\overset{{\log }_{2}M}{\mathrm{\&Pi;}}}\frac{\exp \left(B_m^j{\mathrm{\&lambda;}}_2\right[b_m^i\left]\right)}{1+\exp \left(B_m^j{\mathrm{\&lambda;}}_2\right[b_m^i\left]\right)}$ (i＝1，2，…，N _t)

(9)

Utilize the soft estimate of symbol soft estimate that formula (8), (9) calculate and symbol mould square, according to the computing formula of variance, the variance of compute sign soft estimate:

var(s _t)＝E(|s _i| ²)-E(s _i) ²(i＝1，2，…，N _t) (10)

8. method according to claim 1 is characterized in that, described total iterations is the fixed numbers of presetting, and perhaps determines adaptively according to certain criterion in iterative process.

9. multiple-input, multiple-output mimo system iterative detection system, comprise: reception antenna, channel estimator, softly go into the soft MIMO of going out detector, deinterleaver, softly go into soft go out channel decoder and interleaver, it is characterized in that, describedly softly go into the soft detection that goes out the MIMO detector and adopt any described method of claim 1-8.

10. a multiple-input, multiple-output mimo system iterative detection equipment is characterized in that, this equipment is arranged on the soft of mimo system receiver goes in the soft MIMO of the going out detector, adopts any described method of claim 1-8.

Images