CN101442391A - Method for processing receiving terminal signal and apparatus for receiving signal - Google Patents

Abstract

The invention discloses a method for processing a signal in the receiving end, which is applied to a closed-loop MIMO-OFDM system. The method comprises the following steps: receiving a signal sent by more than one subchannel on the same subcarrier of the sending side; filtering the received signal; and converting color noise in the received signal into white noise. The invention also discloses a signal receiving device which is applied to the closed-loop MIMO-OFDM system; the device comprises a signal receiving module for receiving the signal sent by more than one subchannel on the same subcarrier of the sending side, and a filtering module for filtering the received signal and converting the color noise of the received signal into the white noise. Through the proposal, the method filters the received signals sent by a plurality of subchannels of the same subcarrier, converts the color noise of the received signal into the white noise, eliminates interference between a plurality of the subchannels caused by time delay of feedback and increases system gain.

Description

Method and signal receiving device that receiving end signal is handled

Technical field

The present invention relates to the signal processing technology field, method and a kind of signal receiving device that particularly a kind of receiving end signal is handled.

Background technology

All adopt multiple-input, multiple-output (the Multiple Input MultipleOutput of many antennas at transmitting terminal and receiving terminal, MIMO) the theoretical channel capacity of system increases with smaller value approximately linear in number of transmit antennas and the reception antenna number, and the mode that mimo system improves capacity is called space division multiplexing.Mimo system and OFDM (Orthogonal Frequency Division Multiplexing, OFDM) the MIMO-OFDM system that combines of technology, can effectively resist the multipath fading of wireless channel, be acknowledged as the most competitive technology of the 4th third-generation mobile communication.

Usually, the transmitting terminal of mimo system is not known channel condition information, yet, if transmitting terminal can be known channel condition information, then we can necessarily handle transmitting at transmitting terminal, reduce and disturb, improve average received signal to noise ratio (Signal to Noise Ratio, SNR), thereby reduce average error sign ratio, we claim the precoding that is treated to of this transmitting terminal.

A symmetrical time division duplex (Time Division Duplex, TDD) in the system, the channel information of up-downgoing channel can be shared mutually, in this case, does not need system configuration is done any change, base station and travelling carriage can both adopt precoding.(receiving terminal need be passed to transmitting terminal so that it knows descending channel information with channel information by a feedback channel for Frequency Division Duplex, FDD) system for asymmetric TDD system or Frequency Division Duplexing (FDD).In a word, this mode by closed loop makes transmitting terminal obtain channel information, thereby the mode that improves the mimo system capacity is one of present MIMO Study on Technology focus.

Under the mimo system of closed loop, do not consider feedback delay and feedback error, the conventional wave beam forming is to choose channel matrix eigenvalue of maximum characteristic of correspondence vector as launching beam moulding vector.This closed-loop MIMO system as shown in Figure 1, wherein, at transmitting terminal, feedback adjusting module 101 receives the channel information that receiving terminal feeds back to, after being the beam forming vector, adjust sending sense, then by transmitting antenna 102 emissions according to this beam forming vector; At receiving terminal, after reception antenna 103 receives signal, in channel estimation module 104, receive whole OFDM symbol after, according to the calculated signals channel matrix H of this OFDM symbol correspondence that receives, H is reception antenna number (W _r) * number of transmit antennas (W _t) complex matrix, feedback module 105 calculates the beam forming vector according to above-mentioned channel matrix H again, feeds back to transmitting terminal then.Wherein, the beam forming vector that calculates of feedback module 105 is a matrix H ^HH has the characteristic vector of eigenvalue of maximum.By to matrix H ^HH carries out characteristic value decomposition, can obtain: w _t=eigvec _Max[H ^HH]=μ ^Max, μ wherein ^MaxExpression H ^HThe eigenvalue of maximum characteristic of correspondence vector of H, this characteristic vector are that mould is 1 unit vector.

Concrete feedback principle is as follows:

It is d that order sends signal, at the multidimensional additive white Gaussian noise (AWGN) of receiving terminal is

And the covariance matrix of noise $E\left[{\mathrm{nn}}^H\right]=N_0{I}_{M_r},$ Wherein

Be unit matrix, N ₀For original noise variance, make μ ^MaxExpression H ^HThe eigenvalue of maximum characteristic of correspondence vector of H, then receiving terminal carries out high specific to received signal and merges that (the maximum weight vector that merges is: w _r=(Hw _t) ^H=(H μ ^Max) ^H) after, the decision signal that obtains can be expressed as:

$\hat{d}={\left(H{\mathrm{\μ}}^{\max }\right)}^{H}H{\mathrm{\μ}}^{\max }d+{\left(H{\mathrm{\μ}}^{\max }\right)}^{H}n$

$={\mathrm{\λ}}^{\max }{\left({\mathrm{\μ}}^{\max }\right)}^H{\mathrm{\μ}}^{\max }d+{\left({\mathrm{\μ}}^{\max }\right)}^H{H}^{H}n$

$={\mathrm{\λ}}^{\max }d+s^{\max }{U}^{\max }n---\left(1\right)$

Wherein, λ ^MaxFor with μ ^MaxCorresponding eigenvalue of maximum, s ^MaxAnd U ^MaxRepresent H respectively ^HIn the maximum singular value of H and the singular value decomposition (SVD) with μ ^MaxThose row of corresponding U battle array.

Then can draw, the signal to noise ratio of signal is:

$\mathrm{SNR}=\frac{E\left({\left({\mathrm{\λ}}^{\max }d\right)}^{*}\left({\mathrm{\λ}}^{\max }d\right)\right)}{E\left({\left(s^{\max }{U}^{\max }n\right)}^H\left(s^{\max }{U}^{\max }n\right)\right)}={\mathrm{\λ}}^{\max }\frac{E_s}{N_0}---\left(2\right)$

Because in the following formula, transmit signal energy and original noise variance N0 are definite value, so system performance gain is fully by the eigenvalue on the selected characteristic direction ^MaxDetermine.

Under the MIMO-OFDM system, for guaranteeing the system performance gain maximum, receiving terminal is chosen N optimum direction in all characteristic direction the insides of all subcarriers, it is the characteristic vector direction of a corresponding N eigenvalue of maximum, and feed back to transmitting terminal, transmitting terminal with it as sending direction, promptly can increase system gain, above-mentioned calculating process is also referred to as the optimal beam moulding algorithm under the MIMO-OFDM system, under the bigger channel of the frequency selectivity that does not have feedback delay and mistake, the performance phase divided ring of optimal beam moulding (promptly not adopting precoding) has very big gain.

But, have following shortcoming in the above-mentioned technology:

Receiving terminal at the closed-loop MIMO-ofdm system that has feedback processing, owing to just can calculate after channel matrix H need be received whole OFDM symbol, could calculate transmission beam forming vector and feed back to transmitting terminal according to channel matrix then, so have an OFDM symbol lengths time T at least _sFeedback delay.If add frame structure, receiver Processing Algorithm and the needed time of transmitting feedback information of certain length, time of delay will be longer, makes the precoding vectors of receiving terminal feedback or matrix have the possibility that lost efficacy.Therefore, in actual communication systems, must consider the influence of feedback delay.

Suppose to send with optimal direction on i the subcarrier, received signal can be expressed as so:

${\hat{d}}_i={\mathrm{\λ}}_i{d}_i+s_i{U}_{i}n---\left(3\right)$

If be the n moment this moment, corresponding channel matrix is H _i(n); When channel changes, become H _i(n+1) after, be equivalent to characteristic direction and characteristic value variation has all taken place.The matrix character base of this moment just is not And become

Equally, Ci Shi characteristic value also becomes

And system still uses H this moment _i(n) ^HH _i(n) characteristic vector

As sending vector.Because

Also be one group of base of this Linear Space, so we can be expressed as ${\mathrm{\&mu;}}_i^j={\mathrm{<figure-callout\; id="1"\; label="c\; i\; j"\; filenames="A200710187849E00201.png,A200710187849E00211.png"\; state="\{\{state\}\}">c</figure-callout>}}_{ij}^1{\mathrm{\&omega;}}_i^1+\&CenterDot;\&CenterDot;\&CenterDot;+c_i^{\mathrm{jn}}{\mathrm{\&omega;}}_i^n,$ That is, can regard as with

Projected to new characteristic vector

On.Wherein, All be plural number, expression

On projection, and have

$c_i^{\mathrm{jk}}=\frac{{\left({\mathrm{\&mu;}}_i^j\right)}^H{\mathrm{\&omega;}}_i^k}{\left|\right|{\mathrm{\&mu;}}_i^j\left|\right|\left|\right|{\mathrm{\&omega;}}_i^k\left|\right|}---\left(4\right)$

Wherein ‖ ‖ represents vector is asked two norms.Because

With All be orthonormal basis, so $\left|\right|{\mathrm{\&mu;}}_i^j\left|\right|=\left|\right|{\mathrm{\&omega;}}_i^k\left|\right|=1,$ Therefore

${\hat{d}}_i={\left({\mathrm{\&mu;}}_i^j\right)}^H{H}_i{(n+1)}^H{H}_i(n+1){\mathrm{\&mu;}}_i^j{d}_i+{\left({\mathrm{\&mu;}}_i^j\right)}^H{H}_i{(n+1)}^{H}n$

$={\left(\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{c}_i^{\mathrm{jk}}{\mathrm{\&omega;}}_i^k\right)}^H{H}_i{(n+1)}^H{H}_i(n+1)\underset{k=1}{\overset{n}{\mathrm{\&Sigma;}}}{c}_i^{\mathrm{jk}}{\mathrm{\&omega;}}_i^k{d}_i+{\left({\mathrm{\&mu;}}_i^j\right)}^H{H}_i{(n+1)}^{H}n---\left(5\right)$

Then signal to noise ratio becomes

${\mathrm{SNR}}_i^j=\left(\underset{k}{\mathrm{\&Sigma;}}{\left(c_i^{\mathrm{jk}}\right)}^H{c}_i^{\mathrm{jk}}{\mathrm{\&eta;}}_i^j\right)\frac{E_s}{N_0}---\left(6\right)$

As can be seen, gain is by λ _iBe reduced to

This is to cause because the launching beam forming direction no longer is the optimal energy direction of transfer of channel.If do not consider channel estimating, this gain reduction is unavoidable.Especially for optimal beam moulding algorithm, carry out beam forming owing to chosen two or more direction on some subcarrier, when not having feedback delay, the subchannel on these same subcarriers is orthogonal; As behind the feedback delay, they are orthogonality relations no longer just, must will have a strong impact on systematic function to interference effect is arranged each other.Therefore, when having feedback delay and mistake, optimum beam forming algorithm performance descends very fast, even may be worse than open-loop performance.

Summary of the invention

In view of this, the invention provides method and a kind of receiving system that a kind of receiving end signal is handled, can improve systematic function.

The method that a kind of receiving end signal provided by the invention is handled is applied in the closed loop MIMO-OFDM MIMO-OFDM system, and this method comprises:

Reception is from the signal that sends more than a subchannel on the same subcarrier of transmitter side;

The signal that receives is carried out filtering, the coloured noise in the received signal is converted to white noise.

A kind of receiving system provided by the invention is applied in closed-loop MIMO-ofdm system, and this device comprises:

Signal receiving module is used to receive the signal that sends more than a subchannel from the same subcarrier of transmitter side;

Filtration module is used for the signal that receives is carried out filtering, and the coloured noise in the received signal is converted to white noise.

By such scheme as can be seen, the signal that among the present invention a plurality of subchannels on the same subcarrier that receives is sent carries out filtering, coloured noise in the received signal is converted to white noise, eliminated in closed-loop MIMO-ofdm system because the interference between above-mentioned a plurality of subchannels that feedback delay causes, thereby increased system gain, improved systematic function.

Description of drawings

Fig. 1 is a closed-loop MIMO system schematic diagram of the prior art;

Fig. 2 is the flow chart of the method specific embodiment of receiving end signal processing of the present invention;

Fig. 3 is concrete calculating and the filtering flow chart in the step 202 of Fig. 2;

Fig. 4 is a kind of preferable composition schematic diagram of receiving system specific embodiment of the present invention;

Fig. 5 to Figure 13 is for to carry out the result schematic diagram that emulation obtains to prior art and the present invention program.

Embodiment

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

The flow process of method first embodiment that receiving end signal of the present invention is handled as shown in Figure 2, this embodiment is applied in the receiver side of closed-loop MIMO-ofdm system, be used for solving because the signal interference problem that closed-loop system feedback delay and other delays cause comprises the steps:

Step 201, reception are from the signal more than a subchannel transmission on the same subcarrier of transmitting terminal.

From this step as can be seen, present embodiment is primarily aimed on same subcarrier, sends the situation of signal on a plurality of subchannels of different directions respectively, solves the interference problem between a plurality of subchannels on the above-mentioned same subcarrier.

Step 202, the signal that receives is carried out filtering, the coloured noise in the received signal is converted to white noise.

In this step, can obtain equivalent MIMO vertical demixing time space (V-BLAST) model, according to the coloured noise factor calculation of filtered parameter in this model according to received signal.

In the present embodiment, can not need from all characteristic directions of all subcarriers, to choose N optimum subchannel, and only use N subchannel choosing optimum 2 2N that direction constituted subchannels of and suboptimum optimum separately from each subcarrier, can send data by two subchannels at most on the subcarrier of selecting like this, this shortcut calculation complexity is very low and performance loss is also very little.

Suppose in step 201 signal that receives be transmitting terminal on a subcarrier at μ ₁And μ ₂The signal that the subchannel of two different directions sends comprises d ₁And d ₂Wherein, μ ₁And μ ₂For to transmitting terminal to H (n) ^HH (n) carries out maximum and time big characteristic value characteristic of correspondence vector that characteristic value decomposition obtains, and H (n) is at n channel matrix constantly.Then concrete calculating and filtering comprise the steps: as shown in Figure 3 in the above-mentioned steps 202

Step 2021, carry out channel estimating, obtain current channel matrix H (n+1) at receiving terminal.

Because it is the n+1 moment that the moment of channel estimating is carried out in the influence of time delay, receiving terminal, so obtain current channel matrix H (n+1).

Step 2022, to H (n+1) ^HH (n+1) carries out characteristic value decomposition, and the matrix character base that obtains is { ω ₁, ω ₂..., ω _n.

The matrix character basis representation that step 2023, usefulness solve sends the direction of signal:

${\mathrm{\&mu;}}_1=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{c}_{1j}{\mathrm{\&omega;}}_j,$ ${\mathrm{\&mu;}}_2=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{c}_{2j}{\mathrm{\&omega;}}_j,$ Wherein, c _1jAnd c _2jRepresent μ respectively ₁And μ ₂At ω _jOn projection, its method for solving is seen formula (4).

Step 2024, according to above-mentioned transmit direction μ ₁And μ ₂The signal that receives is carried out high specific merge, obtain decision signal

With

${\hat{d}}_1={\left({\mathrm{\&mu;}}_1\right)}^H{\left(H(n+1)\right)}^{H}H(n+1)({\mathrm{\&mu;}}_1{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="A200710187849E00201.png,A200710187849E00211.png"\; state="\{\{state\}\}">d</figure-callout>}}_1+{\mathrm{\&mu;}}_2{d}_2)+{\left({\mathrm{\&mu;}}_1\right)}^H{\left(H(n+1)\right)}^{H}n$

(7)

$=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_j{\left(c_{1j}\right)}^H{c}_{1\mathrm{<figure-callout\; id="1"\; label="j\; d"\; filenames="A200710187849E00201.png,A200710187849E00211.png"\; state="\{\{state\}\}">j</figure-callout>}}{d}_{}{}_1+\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_j{\left(c_{1j}\right)}^H{c}_{2\mathrm{<figure-callout\; id="2"\; label="j\; d"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">j</figure-callout>}}{d}_{}{}_2+{\left(H(n+1){\mathrm{\&mu;}}_1\right)}^{H}n$

${\hat{d}}_2={\left({\mathrm{\&mu;}}_2\right)}^H{\left(H(n+1)\right)}^{H}H(n+1)({\mathrm{\&mu;}}_1{\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="A200710187849E00201.png,A200710187849E00211.png"\; state="\{\{state\}\}">d</figure-callout>}}_1+{\mathrm{\&mu;}}_2{d}_2)+{\left({\mathrm{\&mu;}}_2\right)}^H{\left(H(n+1)\right)}^{H}n$

(8)

$=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_j{\left(c_{2j}\right)}^H{c}_{2\mathrm{<figure-callout\; id="2"\; label="j\; d"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">j</figure-callout>}}{d}_{}{}_2+\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_j{\left(c_{2j}\right)}^H{c}_{1\mathrm{<figure-callout\; id="1"\; label="j\; d"\; filenames="A200710187849E00201.png,A200710187849E00211.png"\; state="\{\{state\}\}">j</figure-callout>}}{d}_{}{}_1+{\left(H(n+1){\mathrm{\&mu;}}_2\right)}^{H}n$

Wherein,

For The characteristic of correspondence value, n is the initial noise of receiving terminal.

Step 2025, obtain the equivalent MIMO V-BLAST model of above-mentioned decision signal.

It is as follows to obtain 22 MIMO V-BLAST models of receiving of equivalence according to above-mentioned decision signal:

$y=\left(\begin{array}{c}{\hat{d}}_1\\ {\hat{d}}_2\end{array}\right)=H_{\mathrm{eq}}\left(\begin{array}{c}{d}_1\\ {\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">d</figure-callout>}}_2\end{array}\right)+\left(\begin{array}{c}{\left({\mathrm{\&mu;}}_1\right)}^H{\left(H(n+1)\right)}^H\\ {\left({\mathrm{\&mu;}}_2\right)}^H{\left(H(n+1)\right)}^H\end{array}\right)n=H_{\mathrm{eq}}s+\left(\begin{array}{c}a\\ b\end{array}\right)n=H_{\mathrm{eq}}s+\mathrm{An}---\left(9\right)$

Wherein, H _EqBe equivalent channel matrix, s is for sending signal, and a, b are intermediate parameters, and A is the coloured noise factor, and n is the initial noise of receiving terminal.Wherein:

$H_{\mathrm{eq}}=\left(\begin{array}{cc}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left|\right|c_{1j}\left|\right|}^2{\mathrm{\&eta;}}_j& \underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left(c_{1j}\right)}^{*}{c}_{2j}{\mathrm{\&eta;}}_j\\ \underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left(c_{2j}\right)}^{*}{c}_{1j}{\mathrm{\&eta;}}_j& \underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left|\right|c_{1j}\left|\right|}^2{\mathrm{\&eta;}}_j\end{array}\right)---\left(10\right)$

$A=\left(\begin{array}{c}{\left({\mathrm{\&mu;}}_1\right)}^H{\left(H(n+1)\right)}^H\\ {\left({\mathrm{\&mu;}}_2\right)}^H{\left(H(n+1)\right)}^H\end{array}\right)={\left(\begin{array}{cc}\left(H(n+1)\right){\mathrm{\&mu;}}_1& \left(H(n+1)\right){\mathrm{\&mu;}}_2\end{array}\right)}^H$

$={\left(H(n+1)\left(\begin{array}{cc}{\mathrm{\&mu;}}_1& {\mathrm{\&mu;}}_2\end{array}\right)\right)}^H---\left(11\right)$

Unite above-mentioned formula (7) to formula (11) and can get H _Eq=AA ^H

An asks covariance matrix to noise:

$E\left[{\mathrm{Ann}}^H{A}^H\right]=H_{\mathrm{eq}}{\mathrm{\&sigma;}}_n^2---\left(12\right)$

Wherein,

Be original noise variance.Because H _EqBe not diagonal matrix, so this moment, the noise An in the equivalent model was a coloured noise, this is not a quadrature just because of two subchannels on the above-mentioned subcarrier each other, causes interference effect is arranged each other.

Step 2026, the coloured noise factors A in the above-mentioned MIMO V-BLAST model is decomposed, obtain filtering parameter.

In this step, make B=A ^H, so just can obtain H _Eq=B ^HB does singular value decomposition (SVD) with B, obtains

$B=U_{\mathrm{eq}}{S}_{\mathrm{eq}}{V}_{\mathrm{eq}}^H$

$A=B^H=V_{\mathrm{eq}}{S}_{\mathrm{eq}}{U}_{\mathrm{eq}}^H---\left(13\right)$

Wherein, S _EqBe A ^HThe singular value diagonal matrix,

Be filtering parameter, U _EqBe A ^HLeft singular matrix.

Step 2027, according to above-mentioned filtering parameter

To the signal d that receives ₁And d ₂Carry out filtering.

According to the computational process of step 2026, can obtain:

$H_{\mathrm{eq}}=B^{H}B=V_{\mathrm{eq}}{S}_{\mathrm{eq}}{U}_{\mathrm{eq}}^H{U}_{\mathrm{eq}}{S}_{\mathrm{eq}}{V}_{\mathrm{eq}}^G=V_{\mathrm{eq}}{D}_{\mathrm{eq}}{V}_{\mathrm{eq}}^H---\left(14\right)$

Wherein, D _Eq=S _EqS _Eq, be H _EqThe characteristic value diagonal matrix, bring the signal model that formula (9) obtains into;

$y=H_{\mathrm{eq}}s+\mathrm{An}$

$=V_{\mathrm{eq}}{D}_{\mathrm{eq}}{V}_{\mathrm{eq}}^{H}s+V_{\mathrm{eq}}{S}_{\mathrm{eq}}{U}_{\mathrm{eq}}^{H}n---\left(15\right)$

After this, signal is carried out filtering, promptly be multiplied by filtering parameter simultaneously on the both sides of following formula

Obtain:

$V_{\mathrm{eq}}^{H}y=V_{\mathrm{eq}}^H{H}_{\mathrm{eq}}s+V_{\mathrm{eq}}^H\mathrm{An}$

$=V_{\mathrm{eq}}^H{V}_{\mathrm{eq}}{D}_{\mathrm{eq}}+V_{\mathrm{eq}}^{H}s+V_{\mathrm{eq}}^H{V}_{\mathrm{eq}}{S}_{\mathrm{eq}}{U}_{\mathrm{eq}}^{H}n$

$=D_{\mathrm{eq}}{V}_{\mathrm{eq}}^{H}s+S_{\mathrm{eq}}{U}_{\mathrm{eq}}^{H}n---\left(16\right)$

Noise for this moment just has:

$E\left[V_{\mathrm{eq}}^H{\mathrm{Ann}}^H{A}^H{V}_{\mathrm{eq}}\right]$

$=E\left[V_{\mathrm{eq}}^H{\mathrm{AA}}^H{V}_{\mathrm{eq}}\right]{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">n</figure-callout>}}^2=E\left[V_{\mathrm{eq}}^H{H}_{\mathrm{eq}}{V}_{\mathrm{eq}}\right]{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">n</figure-callout>}}^2=E\left[V_{\mathrm{eq}}^H{V}_{\mathrm{eq}}{D}_{\mathrm{eq}}{V}_{\mathrm{eq}}^H{V}_{\mathrm{eq}}\right]{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">n</figure-callout>}}^2$

$=D_{\mathrm{eq}}{\mathrm{\&sigma;}}_n^2---\left(17\right)$

Because D _EqBe diagonal matrix, then Ci Shi noise has become white noise, has eliminated two subchannel interferences, has offset because the system gain that feedback delay causes descends.

The signal model of this moment becomes:

$r=V_{\mathrm{eq}}^{H}y$

$=V_{\mathrm{eq}}^H{H}_{\mathrm{eq}}s+V_{\mathrm{eq}}^H\mathrm{An}$

$=H_{\mathrm{eqa}}s+W---\left(18\right)$

Wherein, H _EqaBe filtered equivalent channel matrix, W is filtered noise.

In the present embodiment, by the processing of step 202, the coloured noise in the received signal become white noise after, can also carry out soft output to signal and detect, specifically comprise:

At first, the soft output of V-BLAST system can be expressed as:

${\mathrm{\&Lambda;}}_k^{\left(i\right)}\&ap;\frac{1}{{\mathrm{\&sigma;}}_w^2}[\underset{d:d_k\&Element;D_0^i}{\min }{\left|\right|\hat{s}-H_{\mathrm{eqa}}d\left|\right|}^2-\underset{d:d_k\&Element;D_1^i}{\min }{\left|\right|\hat{s}-H_{\mathrm{eqa}}d\left|\right|}^2]---\left(19\right)$

Wherein, E _SBe transmit signal energy, the variance of the noise W in the formula (18) ${\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">n</figure-callout>}}^2=D_{\mathrm{eq}}{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">n</figure-callout>}}^2.$

In the formula (19), under the situation that adopts least mean-square error (MMSE) decoding to detect, maximum singular value $\hat{s}=G_{\mathrm{MMSE}}r,$ Wherein MMSE detects matrix $G_{\mathrm{MMSE}}=E_S{H}_{\mathrm{eqa}}^H{(E_S{H}_{\mathrm{eqa}}{H}_{\mathrm{eqa}}^H+{\mathrm{\&sigma;}}_n^2{D}_{\mathrm{eq}})}^{-1}.$

List of references is adopted in then above-mentioned soft output: Dominik Seethaler, Gerald Matz, " AnEfficient MMSE-Based Demodulator for MIMO Bit-Interleaved CodedModulation " IEEE Communications Society Globecom 2004, the algorithm that provides, for:

${\mathrm{\&Lambda;}}_{k,\mathrm{MMSE}}^{\left(i\right)}=\frac{W_{k,k}}{1-W_{k,k}}[\underset{d\&Element;D_0^i}{\min }{\mathrm{\&psi;}}_k^2\left(d\right)-\underset{d\&Element;D_1^i}{\min }{\mathrm{\&psi;}}_k^2\left(d\right)]---\left(20\right)$

Wherein, $W_{k,k}={(I+D_{\mathrm{eq}}{\mathrm{\&sigma;}}_n^2{\left(H_{\mathrm{eq}}^{H}H\right)}^{-1})}^{-1},$ ${\mathrm{\&psi;}}_k^2\left(d\right)=|\frac{{\hat{s}}_k}{W_{k,k}}-d|,$ Subscript k represents subcarrier, the soft output of i bit in subscript (i) the expression estimate symbol,

Transmit i bit in the set of expression is 0 assemble of symbol,

Transmit i bit in the set of expression is 1 assemble of symbol.

In the formula (19), under the situation that adopts maximum likelihood (ML) decoding to detect, soft output can be expressed as:

${\mathrm{\&Lambda;}}_k^{\left(i\right)}\&ap;\frac{1}{{\mathrm{\&sigma;}}_{w,k}^2}[\underset{d:d_k\&Element;D_0^i}{\min }{\left|\right|\hat{s}-H_{\mathrm{eqa}}d\left|\right|}^2-\underset{d:d_k\&Element;D_1^i}{\min }{\left|\right|\hat{s}-H_{\mathrm{eqa}}d\left|\right|}^2]---\left(21\right)$

Wherein, ${\mathrm{\&sigma;}}_{w,\mathrm{<figure-callout\; id="2"\; label="k"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">k</figure-callout>}}^2={\left(D_{\mathrm{eq}}{\mathrm{\&sigma;}}_n^2\right)}_{k,k}.$

Soft output in the formula (19) detects also can adopt ZF (ZF) algorithm to carry out, because this algorithm is conventionally known to one of skill in the art, no longer illustrates here.

Present embodiment with top, the received signal of the subcarrier that uses two subchannels is calculated, and for the subcarrier that has only used a subchannel, received signal can be expressed as:

$\hat{d}={\left(\mathrm{\&mu;}\right)}^H{\left(H(n+1)\right)}^{H}H(n+1)\&prime;\mathrm{\&mu;d}+{\left(\mathrm{\&mu;}\right)}^H{\left(H(n+1)\right)}^{H}n$

$=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_j{\left(c_{1j}\right)}^{*}{c}_{1j}d+{\left(\mathrm{\&mu;}\right)}^H{\left(H(n+1)\right)}^{H}n$

$=h_{\mathrm{eq}}d+w---\left(22\right)$

After the log-likelihood ratio of i position was used approximate formula, can obtain soft being output as:

${\mathrm{\&Lambda;}}^{\left(i\right)}\&ap;\frac{1}{{\mathrm{\&sigma;}}_w^2}[\underset{d\&Element;D_0^i}{\min }{\left|\right|r-h_{\mathrm{eq}}d\left|\right|}^2-\underset{d\&Element;D_1^i}{\min }{\left|\right|r-h_{\mathrm{eq}}d\left|\right|}^2]---\left(23\right)$

Wherein, r is a received signal, h _EqBe the equivalent channel coefficient, d transmits, in addition, in the following formula

Draw by following formula:

${\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">w</figure-callout>}}^2=E\left[{\mathrm{ww}}^H\right]=E\left[{\left(\mathrm{\&mu;}\right)}^H{\left(H(n+1)\right)}^H{\mathrm{nn}}^{H}H(n+1)\mathrm{\&mu;}\right]$

$={\left(\mathrm{\&mu;}\right)}^H{\left(H(n+1)\right)}^{H}H(n+1)\mathrm{\&mu;}{\mathrm{\&sigma;}}_n^2---\left(24\right)$

$=\mathrm{\&alpha;}{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A200710187849E00192.png,A200710187849E00201.png"\; state="\{\{state\}\}">n</figure-callout>}}^2$

In the present embodiment, after above-mentioned steps 2021, because the MIMO-OFDM system is a closed-loop system, the channel matrix that obtains according to channel estimating at receiving terminal sends the calculating of beam forming vector, and the transmission beam forming vector that will calculate feeds back to transmitting terminal.In this case, though system still exists feedback delay and other delays, but because in the signal processing method of the receiving terminal that the embodiment of the invention provides, by filtering the interference signal that is caused by feedback delay and other delays is eliminated, eliminate the influence of above-mentioned delay, improved system performance gain.

Below again the specific embodiment of signal receiving device of the present invention is described.

The signal receiving device specific embodiment is applied in the MIMO-OFDM system of closed loop, is used for solving the signal interference problem that causes owing to closed-loop system feedback delay and other delays, and this signal receiving device specifically comprises:

Signal receiving module is used to receive the signal that sends more than a subchannel from the same subcarrier of transmitter side.

Filtration module is used for the signal that receives is carried out filtering, and the coloured noise in the received signal is converted to white noise.

Preferably, further can comprise in this device:

Soft demodulation module is used for that described filtered signal is carried out soft output and detects.

Preferably, further also can comprise in this device:

The filtering parameter determination module is used for that the signal that receives is carried out high specific and merges, and obtains decision signal, and determines filtering parameter according to this decision signal;

Above-mentioned filtration module carries out filtering according to described definite filtering parameter.

Preferably, comprise in the above-mentioned filtering parameter determination module:

High specific merges submodule, is used for that the signal that receives is carried out high specific and merges, and obtains decision signal;

Equivalence V-BLAST models treated submodule is used for obtaining equivalent MIMO V-BLAST model according to described judgement channel, and the noise in this model is An, and wherein A is the coloured noise factor, and n is the initial noise of receiving terminal;

The singular value decomposition submodule is used for carrying out singular value decomposition to behind the described coloured noise factors A transposition, obtains $A^H=U_{\mathrm{eq}}{S}_{\mathrm{eq}}{V}_{\mathrm{eq}}^H,$ Wherein, S _EqBe A ^HThe singular value diagonal matrix, Be described definite filtering parameter, U _EqBe A ^HLeft singular matrix.

Preferably, further comprise in the above-mentioned described filtering parameter determination module:

The channel estimating submodule is used for carrying out channel estimating at receiving terminal, obtains current channel matrix H (n+1);

The characteristic value decomposition submodule is used for to H (n+1) ^HH (n+1) carries out characteristic value decomposition, obtains matrix character base { ω ₁, ω ₂..., ω _n;

Transmit direction is represented submodule, is used to adopt described matrix character base { ω ₁, ω ₂..., ω _nThe expression described received signal transmit direction;

Then above-mentioned high specific merges submodule and carries out the high specific merging to received signal according to described transmit direction.

Further comprise in the present embodiment:

Feedback module, the channel matrix H (n+1) that is used for obtaining according to above-mentioned channel estimating is calculated and is sent the beam forming vector, and should send the beam forming vector and feed back to transmitting terminal, so that can sending the beam forming vector according to this, transmitting terminal adjusts the signal transmit direction.

A kind of preferable implementation of signal receiving device specific embodiment as shown in Figure 4, wherein the annexation between each module repeats no more here in above-mentioned existing explanation.

In order to prove that the present invention improves the technique effect of system gain, the inventor has carried out emulation experiment, and the environment of this experiment and the parameter that relates to are:

Under the channel of the bad city, 6 footpaths of COST207, carry out, wherein the concrete setting for the 6 footpath bad city channels of COST207 please see this reference paper for details: G.L.St ü ber, Priciples of MobileCommunication, 2nd ed.Norwell, MA:Kluwer, 2001.; The sub-carrier number of ofdm signal chooses 1024, i.e. N _IFFT=1024, cyclic prefix type during is 1/4, and promptly CP=256 sends signal and takes the BPSK modulation, and signal bandwidth is 10M, and has taked ten OFDM symbol time length feedback delays; Chnnel coding adopts the convolution code of 3GPP LTE (1,2,9), and the octal system generator polynomial is (561,753); Weaving length is 10 OFDM symbols; Receiving terminal viterbi decoder decoding depth is 45; Antenna configurations is 22 receipts.

As shown in drawings, Fig. 5, Fig. 6 and Fig. 7 are the performances that receiving terminal adopts BPSK, QPSK and three kinds of different modulating modes of 16-QAM under the translational speed of 3km/h; Fig. 8, Fig. 9 and Figure 10 are the performance map of above-mentioned three kinds of modulation systems under 20km/h; Figure 11, Figure 12 and Figure 13 then are the performances of above-mentioned three kinds of modulation systems under 40km/h.Adopted MMSE and ML to carry out soft output emulation.

In each accompanying drawing, traditional scheme is meant that each subcarrier selects the optimal direction of oneself to send separately.This scheme is simple in structure at receiver end, but when postponing, can accurately not calculate the attenuation coefficient of signal, this BPSK and not influence of QPSK for hard decision the time, but that the 16-QAM of hard decision will demodulation occur is inaccurate.And after we add chnnel coding, no matter be 16-QAM or the modulation of simple phase modulation, soft output all can be affected.For the present invention program, then be to have selected in all subcarriers, a best N direction sends, and simultaneously, because receiving terminal has increased the consideration to time delay, has avoided self-interference on the same subcarrier.Scheme 1 is to have the subcarrier of having chosen two subchannels to take the scheme of MMSE filtering to those, and scheme 2 then is these subcarriers to be taked the scheme of maximum likelihood filtering.

From Fig. 5 to simulation result shown in Figure 13, we in low speed (3km/h, 20km/h), adopt the solution of the present invention as can be seen, and there is comparatively obvious gain in system, and particularly after modulation system uprised, it was more obvious to gain.This be because, according to traditional algorithm of not considering feedback delay, when feedback delay occurring, can't accurately estimate the equivalent model of signal, this error is when the modulation of high-order more, influence can be serious more.And the present invention program can estimate the model of signal accurately, and performance just can not occur quick decline with the rising of order of modulation.

In at a high speed (40km/h), at this moment the channel condition information owing to the closed loop that obtains at transmitting terminal is very inaccurate, and systematic function descends, but the present invention program still has certain gain than traditional scheme, and also be that order of modulation is high more, it is big more to gain.

In addition, we it can also be seen that as a result from above, when carrying out soft output and detect, adopt the MMSE algorithm to compare with the ML algorithm, performance difference is also little, this is because have only subcarrier few in number to be to use 2 subchannels to transmit, and only for these subcarriers, just can have the performance difference of ML algorithm and MMSE algorithm.So when embodying the whole system performance at last, it is very little that difference becomes.That is to say, when we use the MMSE algorithm to carry out soft output, just can reach good performance, and needn't adopt the ML algorithm of high complexity.

More than be explanation, in concrete implementation process, can carry out suitable improvement, to adapt to the concrete needs of concrete condition method of the present invention to the specific embodiment of the invention.Therefore be appreciated that according to the specific embodiment of the present invention just to play an exemplary role, not in order to restriction protection scope of the present invention.

Claims (11)

1, a kind of method of receiving end signal processing is applied to it is characterized in that in the closed loop MIMO-OFDM MIMO-OFDM system, comprising:

Reception is from the signal that sends more than a subchannel on the same subcarrier of transmitting terminal;

Described received signal is carried out filtering, the coloured noise in the described received signal is converted to white noise.

2, method according to claim 1 is characterized in that, the coloured noise in the described received signal is converted to white noise after, this method further comprises: described filtered signal is carried out soft output detect.

3, method according to claim 2 is characterized in that, described soft output detects adopts ZF ZF algorithm, least mean-square error MMSE algorithm or maximum likelihood ML algorithm.

4, method according to claim 1 is characterized in that,

Described described received signal is carried out comprising before the filtering:

Described received signal is carried out high specific merge, obtain decision signal;

Determine filtering parameter according to described decision signal;

Describedly described received signal is carried out filtering comprise: multiply by described filtering parameter with described decision signal.

5, method according to claim 4 is characterized in that, describedly determines filtering parameter according to decision signal, further comprises:

Obtain equivalent MIMO vertical demixing time space V-BLAST model according to described decision signal, the noise in this model is An, and wherein A is the coloured noise factor, and n is the initial noise of receiving terminal;

Behind described coloured noise factors A transposition, carry out singular value decomposition, obtain: $A^H=U_{\mathrm{eq}}{S}_{\mathrm{eq}}{V}_{\mathrm{eq}}^H,$ Wherein, S _EqBe A ^HThe singular value diagonal matrix,

Be described filtering parameter, U _EqBe A ^HLeft singular matrix.

6, method according to claim 4 is characterized in that, before described received signal being carried out the high specific merging, further comprises:

Carry out channel estimating, obtain current channel matrix H (n+1);

To H (n+1) ^HH (n+1) carries out characteristic value decomposition, obtains matrix character base { ω ₁, ω ₂..., ω _n;

With described matrix character base { ω ₁, ω ₂..., ω _nThe expression described received signal transmit direction;

According to described transmit direction described received signal being carried out high specific merges.

7, a kind of signal receiving device is applied to it is characterized in that in closed-loop MIMO-ofdm system that this device comprises:

Signal receiving module is used to receive the signal that sends more than a subchannel from the same subcarrier of transmitting terminal;

Filtration module is used for described received signal is carried out filtering, and the coloured noise in the described received signal is converted to white noise.

8, device according to claim 7 is characterized in that, further comprises in this device:

Soft demodulation module is used for that described filtered signal is carried out soft output and detects.

9, device according to claim 7 is characterized in that, further comprises in this device:

The filtering parameter determination module is used for that described received signal is carried out high specific and merges, and obtains decision signal, and determines filtering parameter according to this decision signal;

Described filtration module carries out filtering according to described definite filtering parameter.

10, device according to claim 9 is characterized in that, comprises in the described filtering parameter determination module:

High specific merges submodule, is used for that described received signal is carried out high specific and merges, and obtains decision signal;

Equivalence V-BLAST models treated submodule is used for obtaining equivalent MIMO V-BLAST model according to described decision signal, and the noise in this model is An, and wherein A is the coloured noise factor, and n is the initial noise of receiving terminal;

The singular value decomposition submodule is used for carrying out singular value decomposition to behind the described coloured noise factors A transposition, obtains $A^H=U_{\mathrm{eq}}{S}_{\mathrm{eq}}{V}_{\mathrm{eq}}^H,$ Wherein, S _EqBe A ^HThe singular value diagonal matrix, Be described filtering parameter, U _EqBe A ^HLeft singular matrix.

11, device according to claim 10 is characterized in that, further comprises in the described filtering parameter determination module:

The channel estimating submodule is used to carry out channel estimating, obtains current channel matrix H (n+1);

The characteristic value decomposition submodule is used for to H (n+1) ^HH (n+1) carries out characteristic value decomposition, obtains matrix character base { ω ₁, ω ₂..., ω _n;

Transmit direction is represented submodule, is used to adopt described matrix character base { ω ₁, ω ₂..., ω _nThe expression described received signal transmit direction;

Then described high specific merges the transmit direction of submodule according to the described matrix character basis representation of described employing, the signal of described reception is carried out high specific merge.

Images