CN100561992C - A Signal Estimation Method for Hierarchical Demodulation of MIMO System - Google Patents

Abstract

The invention discloses a signal estimation method used in multi-input multi-output system layered demodulation, comprising: 1. determining the amplitude factor; determining the first symbol of the estimated signal according to the first symbol of the current layer signal; 2. Demodulate the current symbol; 3. Re-modulate the current estimated signal symbol to obtain a new estimated signal symbol; 4. Calculate the difference between the new estimated signal symbol and the corresponding current layer signal symbol Phase error; 5. Low-pass filtering the phase error to update the phase error statistic; 6. Eliminating the phase error statistic updated through low-pass filtering from the new estimated signal symbol; 7. Judging whether the symbol of the current layer signal has been processed, If the processing is complete, the method ends, otherwise, go to step 2. The estimated signal obtained by the invention can be directly used in the layered demodulation algorithm, and has the obvious advantages of simple calculation, accurate estimation, and allowing the current layer signal to contain frequency offset.

Description

A kind of signal estimation method that is used for layered demodulation of multi-input and multi-output system

Technical field

The present invention relates to the signal estimation method in a kind of multi-input and multi-output wireless communication system hierarchical de-modulating, what be particularly related to is signal estimation method in multi-input and multi-output wireless communication system V-BLAST (Vertical Bell Laboratories Layered Space-Time, vertical-during dblast) the hierarchical de-modulating algorithm.

Background technology

Multiple-input and multiple-output (MIMO, Multiple-Input Multiple-Output) technology is the important breakthrough of wireless mobile communications field intelligent antenna technology, it is by launching separate signal simultaneously on different antennae, it is space multiplexing technique, make full use of the disperse characteristic of space channel, both can under the situation that does not increase bandwidth, improve the capability of communication system and the availability of frequency spectrum exponentially.

MIMO technique is of long duration, the seventies just the someone propose MIMO technique is used for communication system, but be that the nineties is by AT﹠amp to the work of laying a foundation that the Multiple Input Multiple Output of mobile radio system produces huge promotion; T Bell laboratory scholar finishes.Nineteen ninety-five Teladar has provided the MIMO capacity under the decline situation; Foshinia had provided a kind of multiple-input and multiple-output Processing Algorithm in 1996---(D-BLAST) algorithm during diagonal angle-dblast; Tarokh in 1998 etc. have discussed the empty time-code that is used for multiple-input and multiple-output; People such as Wolniansky in 1998 adopt vertically-and Bell Laboratory branch (V-BLAST) algorithm set up a MIMO experimental system, reached the availability of frequency spectrum more than the 20bit/s/Hz in laboratory test, and this availability of frequency spectrum is extremely difficult the realization in conventional system.These work are subjected to various countries scholar's very big attention, and make the research work of multiple-input and multiple-output obtain developing rapidly.

The hierarchical de-modulating algorithm, particularly the V-BLAST algorithm occupies critical role in the MIMO signal Processing Algorithm.Receiver separates the signal of an independent channel successively in the V-BLAST algorithm by a graded from mix received signal, from mix received signal the ideal signal of this independent channel is deleted then, realizes hierarchical de-modulating.

This part content can be seen in " Layered Space-time Architecture for Wireless Communication inFading Environment When Using Multiple Antennas ", Foschini G.J, Bell Labs TechnicalJournal, 1996, Autumn, pp 41-59, perhaps " V-BLAST:an architecture for realizing very high data ratesover the rich-scattering wireless channel ", WOLNIANSKY P W, FOSCHINI G.J, GOLDEN G D, etal.Proc IEEE ISSSE ' 98[C] .Pisa, Italy, 1998.295-300.The existing many pieces of papers of the analysis of V-BLAST algorithm performance are announced, for example " Performance Analysis of the V-BLAST Algorithm:An Analytical Approach ", SergeyLoyka, Francois Gagnon, IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL.3, NO.4, JULY 2004.

Be that example is carried out comparatively detailed introduction to the hierarchical de-modulating algorithm with the V-BLAST algorithm below:

If the antenna number of multi-input multi-output system emission side is M, recipient's antenna number is N; Emission side is H to recipient's Channel Transmission matrix, and dimension equals N * M, and satisfies 2≤M≤N, its element h _IjIt is transmission factor from transmitting antenna j to reception antenna i; The mixed signal that receiver receives is r.

Then the V-BLAST algorithm can be explained with following several steps:

Initialization: H ₁=H (1)

For?m＝1:M (2)

Calculate $G_m=H_m^{+}={\left(H_m^H{H}_m\right)}^{-1}{H}_m^H---\left(3\right)$

Carry out sorting operation $\mathrm{km}=\underset{j\≠\{k1,k2,\·\·\·,k(j-1)\}}{\arg \min }{\left|\right|{\left(G_m\right)}_j\left|\right|}^2---\left(4\right)$

Choose $w_{\mathrm{km}}^T={\left(G_m\right)}_{\mathrm{km}}---\left(5\right)$

Single channel signal separates $y_{\mathrm{km}}=w_{\mathrm{km}}^T\·r---\left(6\right)$

Separation signal is quantized to obtain estimated signal $\hat{d}=Q\left(y_{\mathrm{km}}\right)---\left(7\right)$

Cancellation estimated signal from received signal $r=r-{\left[H_m\right]}_{\mathrm{km}}\·\hat{d}---\left(8\right)$

H _M+1The H of=deletion km row _m(9)

Next m (10)

Wherein, (G _m) _jRepresenting matrix G _mJ row vector, || (G _m) _j|| ²Represent its mould square; [H _m] k _mRepresenting matrix H _mKm row; In above-mentioned classical documents, (7) step $\hat{d}=Q\left(y_{\mathrm{km}}\right)$ Be called " the current layer signal y of reconstruct _KmEstimated signal

" or " quantizing (amplitude limit) operation, with the constellation that obtains being suitable for using ".

The above the hierarchical de-modulating algorithm particularly principle, step of V-BLAST algorithm is open, and is known by the scholar who is engaged in association area work, but for one of important step of hierarchical de-modulating algorithm---(7) step of V-BLAST algorithm particularly $\hat{d}=Q\left(y_{\mathrm{km}}\right)$ Adopt which kind of concrete grammar to realize, also do not find disclosed specific implementation method at present.

Summary of the invention

The objective of the invention is to disclose that a kind of multi-input and multi-output wireless communication system hierarchical de-modulating algorithm that is used for---particularly the V-BLAST algorithm is by current layer signal y _KmObtain estimated signal

Method, promptly $\hat{d}=Q\left(y_{\mathrm{km}}\right)$ A kind of implementation method of middle operator " Q () ", this method realizes simple, has higher performance, and is applicable to current layer signal y _KmIn have the situation of frequency shift (FS), therefore be fit to very much practical engineering application.

For reaching above goal of the invention, the present invention specifically is achieved in that

A kind of signal estimation method that is used for layered demodulation of multi-input and multi-output system comprises the steps:

Step 1, initialization:

Determine amplitude factor;

Determine first symbol of estimated signal according to first symbol of current layer signal;

The phase error statistic is resetted;

Step 2, current sign is carried out demodulation;

Step 3, current estimated signal symbol is modulated again, obtained new estimated signal symbol by demodulation result;

Step 4, calculate new estimated signal symbol and the phase error between the corresponding current layer signal symbol;

Step 5, with the phase error low-pass filtering, upgrade the phase error statistic;

Step 6, the phase error statistic that elimination is upgraded through low-pass filtering from new estimated signal symbol;

Step 7, judge whether that the symbol of current layer signal disposes, if dispose then method finishes; Otherwise to step 2.

The initialization of described step 1:

By current layer signal, the relation of Channel Transmission vector and weighted vector and ideal signal, the amplitude information of acquisition estimated signal is amplitude factor;

Amplitude factor is used for determining the amplitude size of estimated signal;

The phase error statistic resets, and makes usually that its value is 0.

In the described step 1,

Determine first symbol of estimated signal according to first symbol of current layer signal, specifically can realize like this:

With amplitude factor with when the product of first symbol of anterior layer divided by mould when first symbol of anterior layer, promptly obtain first symbol of estimated signal.

In the described step 2,, use existing estimated signal symbol and/or current layer signal symbol that current sign is carried out demodulation according to the modulation system that system adopts.

In the described step 3, current estimated signal symbol is modulated again, obtained new estimated signal symbol, when obtaining new estimated signal symbol, need the amplitude of control character simultaneously, make its amplitude equal amplitude factor all the time according to demodulation result.

In the described step 4, obtain the phase error between new estimated signal symbol and the corresponding current layer signal symbol, specifically be achieved in that

${\mathrm{Err}}_{\mathrm{Crt}}=\mathrm{angle}\{{\hat{d}}_k\·y_k^{\′}/|y_k\left|\right\},$ Wherein angle{} represents to ask the multiple angle of a plural number,

Represent new estimated signal symbol, y _kRepresent corresponding current layer signal symbol, y ' _kBe y _kConjugation.

In the described step 5, low-pass filtering specifically can realize like this:

The phase error statistic of upgrading is ρ Err+ (1-ρ) Err _Crt,

Wherein, Err is the phase error statistic, Err _CrtBe the phase error between new estimated signal symbol and the corresponding current layer signal symbol, the span of ρ is (0,1).

In the described step 6, from new estimated signal symbol, eliminate the phase error statistic, specifically be achieved in that

The new estimated signal symbol that upgrades is

Be new estimated signal symbol, its amplitude equals amplitude factor, and Err is the phase error statistic.

A kind of hierarchical de-modulating algorithm that is used for provided by the present invention---particularly the V-BLAST algorithm is obtained the method for estimated signal by current layer signal, because the method that has adopted amplitude control, phase error low pass to eliminate, therefore the estimated signal that obtains can be directly used in the hierarchical de-modulating algorithm, have simultaneously and calculate simply, estimate accurately to allow to contain in the current layer signal the obvious advantage of frequency shift (FS).

A kind of hierarchical de-modulating algorithm that is used for disclosed in this invention---particularly the V-BLAST algorithm obtains the method for estimated signal by current layer signal, has the following advantages:

The first, because the method that has adopted amplitude control, phase error low pass to eliminate, therefore the estimated signal that obtains can be directly used in the hierarchical de-modulating algorithm.

The second, because therefore the method that has adopted amplitude control, phase error low pass to eliminate uses the inventive method to allow to contain frequency shift (FS) in the current layer signal.

In a word, adopt the signal in the method realization hierarchical de-modulating algorithm provided by the invention to estimate to have and calculate simply, estimate remarkable advantage accurately.

Description of drawings

Fig. 1 is the schematic flow sheet of the inventive method.

Embodiment

Below in conjunction with accompanying drawing, will carry out comparatively detailed explanation to preferred embodiment of the present invention.

The method of the invention comprises:

The first step, initialization: determine amplitude factor; Determine first symbol of estimated signal according to first symbol of current layer signal; The phase error statistic is resetted;

In second step, current sign is carried out demodulation by existing estimated signal symbol (with current layer signal symbol);

The 3rd step, by demodulation result current estimated signal symbol is modulated again, obtain new estimated signal symbol;

In the 4th step, calculate the phase error between new estimated signal symbol and the corresponding current layer signal symbol;

In the 5th step,, upgrade the phase error statistic with the phase error low-pass filtering;

In the 6th step, from new estimated signal symbol, eliminate the phase error statistic of upgrading through low-pass filtering;

In the 7th step, judge whether the symbol of current layer signal disposes, if dispose then the method end; Otherwise to second step.

In the method for the invention, establishing current layer signal is y, and y={y ₁, y ₂..., y _n, wherein the subscript of each element of y is represented the time sequencing of symbol; If estimated signal is

, and $\hat{d}=\{{\hat{d}}_1,{\hat{d}}_2,...,{\hat{d}}_n\},$ Wherein

The subscript of each element is represented the time sequencing of symbol;

If the Channel Transmission matrix of multi-input multi-output system is H, for M * N multi-input multi-output system (having M transmitting antenna, N reception antenna), the dimension of H equals N * M, can be expressed as H=[h ₁, h ₂..., h _M], h wherein _k, k=1,2 ..., M represents k the column vector of H, promptly k transmitting antenna of multi-input multi-output system transmitter is to the Channel Transmission vector of a receiver N reception antenna.

The initial work of the described first step is at first determined amplitude factor A, and amplitude factor A is used for determining that estimated signal is

The amplitude size.

If the current layer signal y={y that multi-input multi-output receiver obtains ₁, y ₂..., y _nBe transmitter k (k ∈ 1,2 ..., M}) the ideal signal s={s of transmission antennas transmit ₁, s ₂..., s _nChannel used for vector transfer h _k(k ∈ 1,2 ..., M}), be received machine N reception antenna and receive, and by weighted vector w ^TReceived signal vector after the merging, i.e. y=w ^T(h _kS), this relation is easily the personnel that are engaged in association area and understands.

The inventive method will obtain the estimated signal of ideal signal s

, promptly expect estimated signal

Approach ideal signal s, then as far as possible Ideal signal s be need on amplitude, phase place, all approach, and for receiver, anterior layer received signal y, weight w worked as ^T, Channel Transmission vector h _kBe known quantity, therefore can pass through known relationship formula y=w ^T(h _kS)=(w ^TH _k) s obtains the amplitude information of ideal signal s, the amplitude information of s equals the amplitude of y divided by (w ^TH _k), this amplitude information promptly equals amplitude factor A.

In the operating process of the inventive method, as long as guarantee estimated signal

The amplitude size equal amplitude factor A all the time, promptly guaranteed estimated signal

Amplitude equal the amplitude of ideal signal s, this is " an estimated signal

Approach ideal signal s as far as possible " one of necessary condition.

The described first step is according to first symbol of current layer signal y ₁Determine first symbol of estimated signal

, the method for determining can be, but be not limited to ${\hat{d}}_1=A\·y_1/\left|y_1\right|.$

The described first step resets error statistics amount Err, makes Err=0 usually.

By existing estimated signal symbol (with current layer signal symbol) current sign is carried out demodulation in described second step.According to different modulation systems, the required input signal of demodulation is difference to some extent, and the part modulation system only uses existing estimated signal symbol to carry out demodulation to current sign, for example the QPSK modulation system; Also have some modulation systems need use existing estimated signal symbol and current layer signal symbol that current sign is carried out demodulation simultaneously, for example the DQPSK modulation system similarly also has Pi/4DQPSK modulation system etc.。

Described the 3rd the step in by demodulation result to current estimated signal symbol Modulate, obtain new estimated signal symbol When obtaining new valuation signal code, need control character simultaneously

Amplitude, make its amplitude equal amplitude factor A all the time.

By second step and the 3rd step, receiver can obtain the desirable symbol s that sends _kAn approximation , estimated value

Amplitude equal the desirable symbol s that sends _k, but in estimated value Also there is phase noise, needing to reduce the influence of phase noise by the 4th step, the 5th step and the 6th step.

Obtain new estimated signal symbol in described the 4th step

With corresponding current layer signal symbol y _kBetween phase error, its method can be, but be not limited to ${\mathrm{Err}}_{\mathrm{Crt}}=\mathrm{angle}\{{\hat{d}}_k\·y_k^{\′}/|y_k\left|\right\},$ Wherein angle{} represents to ask the multiple angle of a plural number.

With the phase error low-pass filtering, upgrade the phase error statistic in described the 5th step.The method of low-pass filtering has a variety of, can be, but be not limited to Err=ρ Err+ (1-ρ) Err _Crt, wherein ρ is a low-pass filter coefficients, and possible span is (0,1), and typical value is 0.8.

Described the 6th the step in from new estimated signal symbol

The middle phase error statistic Err that eliminates, method is ${\hat{d}}_k={\hat{d}}_k\·e^{-\mathrm{jErr}},$ This moment estimated value

Amplitude equal amplitude factor A, and phase place through low-pass filtering to ideal symbol s _kApproach.

Judge whether that the symbol of current layer signal disposes in described the 7th step, if dispose then the method end; Otherwise to described second step.

π/4DQPSK modulation system is adopted in preferred embodiment of the present invention, and establishing current layer signal is y, and y={y ₁, y ₂..., y _n, wherein the subscript of each element of y is represented the time sequencing of symbol; Estimated signal is

, and $\hat{d}=\{{\hat{d}}_1,{\hat{d}}_2,...,{\hat{d}}_n\},$ Wherein

The subscript of each element is represented the time sequencing of symbol.

In the above-mentioned first step, determine that amplitude factor A equals 1; And according to first symbol of current layer signal y ₁Determine first symbol of estimated signal

, the method for determining is ${\hat{d}}_1=A\·y_1/\left|y_1\right|;$ Simultaneously phase error statistic Err is resetted, i.e. Err=0.

In above-mentioned second step, by existing estimated signal symbol and current layer signal symbol current sign is carried out differential ference spiral, calculate $z=y_k\·{\hat{d}}_{k-1}^{\′}.$

Above-mentioned the 3rd the step in, according to π/4DQPSK modulation system, by demodulation result to current estimated signal symbol

Modulate, obtain new estimated signal symbol

, the while is control character when obtaining new valuation signal code

Amplitude, make it equal amplitude factor 1 all the time, concrete grammar is ${\hat{d}}_k={\hat{d}}_{k-1}\·e^{\mathrm{j\δ}},$ When wherein the multiple angle of z lays respectively at 1,2,3,4 quadrants, e ^{J δ}Equal e respectively ^{J π/4}, e ^{J3 π/4}, e ^{-j3 π/4}, e ^{-j π/4}

In above-mentioned the 4th step, obtain new estimated signal symbol

With corresponding current layer signal symbol y _kBetween error, its method is ${\mathrm{Err}}_{\mathrm{Crt}}=\mathrm{angle}\{{\hat{d}}_k\·y_k^{\′}/|y_k\left|\right\}.$

In above-mentioned the 5th step,, upgrade the error statistics amount with the phase error low-pass filtering.The method of low-pass filtering is Err=ρ Err+ (1-ρ) Err _Crt, wherein the ρ value is 0.8.

In above-mentioned the 6th step, from new estimated signal symbol The middle phase error statistic Err that eliminates, method is ${\hat{d}}_k={\hat{d}}_k\·e^{-\mathrm{jErr}}.$

In above-mentioned the 7th step, judge whether that the symbol of current layer signal disposes, if dispose then the method end; Otherwise in described second step.

In the preferred embodiment of the present invention, above-described processing procedure can be represented with following unification:

${\hat{d}}_1=A\·y_1/\left|y_1\right|$

Err＝0

for?k＝2:n

$z=y_k\·{\hat{d}}_{k-1}^{\′}$

${\hat{d}}_k={\hat{d}}_{k-1}\·e^{\mathrm{j\δ}}$

${\mathrm{Err}}_{\mathrm{Crt}}=\mathrm{angle}\{{\hat{d}}_k\·y_k^{\′}/|y_k\left|\right\}$

Err＝ρ·Err+(1-ρ)·Err _Crt

${\hat{d}}_k={\hat{d}}_k\·e^{-\mathrm{jErr}}$

end

According to π/4DQPSK modulation system, when the multiple angle of z lays respectively at 1,2,3,4 quadrants, e ^{J δ}Equal e respectively ^{J π/4}, e ^{J3 π/4}, e ^{-j3 π/4}, e ^{-j π/4}ρ gets 0.8.

Should be understood that protection range of the present invention is illustrated in the appended claims, and can not be with the foregoing description of specification as restriction, every conspicuous modification within aim of the present invention is also due within protection scope of the present invention.

Claims (8)

1, a kind of signal estimation method in the multi-input multi-output system layered demodulation, is characterized in that, comprises the steps: Step 1. Initialization: Determine the magnitude factor; Determine the first symbol of the estimated signal according to the first symbol of the current layer signal; Reset the phase error statistics; Step 2, demodulating the current symbol; Step 3, remodulating the current estimated signal symbol by the demodulation result to obtain a new estimated signal symbol; Step 4, calculating the phase error between the new estimated signal symbol and the corresponding current layer signal symbol; Step 5, low-pass filtering the phase error, and updating the phase error statistics; Step 6, eliminating the phase error statistics updated by low-pass filtering from the new estimated signal symbols; Step 7, judging whether the symbol of the current layer signal has been processed, and if the processing is completed, the method ends; otherwise, go to step 2. 2. The signal estimation method used in multi-input multi-output system layered demodulation as claimed in claim 1, characterized in that: Initialization of the step 1: Through the current layer signal, the channel transmission vector and the relationship between the weight vector and the ideal signal, the amplitude information of the estimated signal is obtained, which is the amplitude factor; the amplitude factor is used to determine the amplitude of the estimated signal; Phase error statistics are reset. 3. The signal estimation method used in multi-input multi-output system layered demodulation as claimed in claim 2, characterized in that: In the step 1, Determine the first symbol of the estimated signal according to the first symbol of the current layer signal, which can be specifically implemented as follows: Divide the product of the magnitude factor and the first symbol of the current layer by the modulus of the first symbol of the current layer to obtain the first symbol of the estimated signal. 4. The signal estimation method used in multi-input multi-output system layered demodulation as claimed in claim 1, characterized in that: In the step 2, According to the modulation mode adopted by the system, the current symbol is demodulated by using the existing estimated signal symbol and/or the current layer signal symbol. 5. The signal estimation method used in multi-input multi-output system layered demodulation as claimed in claim 1, characterized in that: In the step 3, the current estimated signal symbol is remodulated according to the demodulation result to obtain a new estimated signal symbol, and at the same time, when obtaining the new estimated signal symbol, the amplitude of the symbol needs to be controlled so that its amplitude is always equal to the amplitude factor. 6. The signal estimation method used in multi-input multi-output system layered demodulation as claimed in claim 1, characterized in that: In the step 4, the phase error between the new estimated signal symbol and the corresponding current layer signal symbol is obtained, which is specifically implemented as follows: ${\mathrm{Err}}_{\mathrm{Crt}}=\mathrm{the\; angle}\{{\hat{d}}_k\&Center\; Dot;{\mathrm{the\; y}}_k^{\′}/|{\mathrm{the\; y}}_k\left|\right\},$ Where angle{} means to find the complex angle of a complex number,

represents the new estimated signal symbol, y _k represents the corresponding current layer signal symbol, and y′ _k is the conjugate of y _k . 7. The signal estimation method used in multi-input multi-output system layered demodulation as claimed in claim 1, characterized in that: In said step 5, Low-pass filtering can be implemented as follows: The updated phase error statistic is ρ Err+(1-ρ) Err _Crt , Wherein, Err is the phase error statistic, Err _Crt is the phase error between the new estimated signal symbol and the corresponding current layer signal symbol, and the value range of ρ is (0, 1). 8. The signal estimation method used in multi-input multi-output system layered demodulation as claimed in claim 1, characterized in that: In the step 6, the phase error statistic is eliminated from the new estimated signal symbol, which is specifically implemented as follows: The updated new estimated signal sign is

For the new estimated signal symbol, its magnitude is equal to the magnitude factor, and Err is the phase error statistic.