CN101047688B - Method and device for estimating signal noise ratio - Google Patents

Abstract

This invention discloses a method for estimating S/N ratio including: selecting demodulation signal constellation points with the absolute value of the signal component greater than that of the edge constellation point of the modulation signal to compute noise power utilizing the selected demodulated signal constellation points, computing emission power and the ratio with noise power to get an estimated value of the S/N ratio. This invention also discloses a device estimating S/N ratio including: a constellation scheme mapping unit, a constellation point selection unit, a noise power computation unit, an emission signal power computation unit and S/N ratio computation unit.

Description

A kind of method of estimated snr and device

Technical field

The present invention relates to digital communication technology, be meant a kind of method and device of estimated snr especially.

Background technology

Digital communication system comprises transmitting terminal, receiving terminal and channel, and Fig. 1 is the simple structure block diagram of digital communication system, and wherein, transmitting terminal comprises information source, coding, planisphere mapping, and receiving terminal comprises constellation diagram mapping, decoding, the stay of two nights.Transmitting terminal with the bit stream that information source produces encode, planisphere mapping back emission; The signal process Channel Transmission of transmitting terminal emission is to receiving terminal; Receiving terminal carries out received signal to pass to the stay of two nights after constellation diagram mapping, the decoding.Wherein, coding adopts the error correcting code coding usually, and error correcting code decoding is adopted in corresponding decoding.The planisphere mapping can be modulated (QPSK) by quarternary phase-shift keying (QPSK), quadrature amplitude modulation modulation systems such as (QAM) is mapped to constellation point with bit stream.The constellation point that transmits that bit stream is carried out obtaining after the described modulation is called the modulation signal constellation point, and modulation signal constellation point x is made of two signal components, is respectively in-phase component x _RWith quadrature component x _I, by in-phase component x _RWith quadrature component x _IExpression modulation signal constellation point x is: x=x _R+ jx _IThe modulation signal constellation point is called the restituted signal constellation point through the received signal constellation point that channel arrives receiving terminal, and same, the restituted signal constellation point also is made of in-phase component and quadrature component.Channel can be white Gaussian noise channel (AWGN), Rayleigh channel channels such as (Rayleigh).

In digital communication system, signal to noise ratio (snr) is a parameter relevant with channel, in decoding of error correcting code for example or constellation diagram mapping process, need utilize SNR to obtain more excellent decoding performance or symbol de-maps performance more accurately.And SNR changes along with the variable power of surrounding environment and transmitter, therefore need SNR be estimated, obtains the estimated value of SNR more accurately, improves the performance of whole system.

Present stage, the method for estimation of SNR has the fitting of a polynomial method of estimation, and the training sequence method of estimation also has blind estimating method.Wherein, the signal that blind estimating method directly utilizes receiving terminal to receive is estimated SNR, the advantage of blind estimating method is an applied range, be easy to optimized Algorithm and without any need for ancillary method, but the estimated value of the SNR that obtains by blind estimation is compared not accurate enough with actual value, therefore, need to propose the less SNR method of estimation of evaluated error.

In the prior art, the method for estimating SNR by blind estimation approach comprises following step as shown in Figure 2:

Step 201: all restituted signal constellation point calculating noise power σ that the receiving terminal utilization receives ², noise power σ ²For: the average of the difference square of distance between all restituted signal constellation point and its nearest modulation signal constellation point.

Step 202: receiving terminal utilizes all restituted signal constellation point to calculate transmit signal power P _s, transmit signal power P _sComputational methods be: after utilizing all restituted signal constellation point to calculate received signal powers, calculate described received signal power and noise power σ ²Difference obtain transmit signal power P _sPerhaps, if transmit signal power is normalized to 1, then directly obtain transmit signal power P at transmitting terminal _s=1.

Wherein, described received signal power is the in-phase component of each restituted signal constellation point and the average of quadrature component square sum, and transmit signal power is the in-phase component of each modulation signal constellation point and the average of quadrature component square sum.

Step 203: calculate transmit signal power P _sWith noise power σ ²Ratio obtain the estimated value of SNR, i.e. SNR=P _s/ σ ²

In the above-described estimation SNR process, need during calculating noise power to determine exactly the restituted signal constellation point by which modulation signal constellation point is sent, and can obtain correct noise power like this.Think that in the prior art all restituted signal constellation point are to be sent by nearest modulation signal constellation point respectively, and utilize all restituted signal constellation point according to described nearest principle calculating noise power, but resulting noise power exists than mistake, its reason is: because the influence of channel, the restituted signal constellation point overlaps on together easily, be difficult to judge the modulation signal constellation point of restituted signal constellation point correspondence, therefore the noise power that calculates with nearest principle at all restituted signal constellation point certainly exists bigger error, thereby the estimated value of SNR and actual value differ bigger.Below in conjunction with the modulated signal constellation of transmitting terminal and the restituted signal planisphere of receiving terminal, further describe the reason that there is big evaluated error in the SNR estimated value that obtains according to prior art.

Fig. 3 is for carrying out the qpsk modulation signal planisphere that the QPSK modulation obtains to transmitting at transmitting terminal, wherein, constellation point power has been normalized to 1.Fig. 4 has the transmitting through behind the awgn channel of qpsk modulation signal planisphere shown in Figure 3, the QPSK restituted signal planisphere of the received signal that obtains at receiving terminal.As can see from Figure 4, the QPSK restituted signal constellation point that obtains behind the process awgn channel is owing to the influence of white Gaussian noise, and the QPSK restituted signal constellation point overwhelming majority has all departed from the qpsk modulation signal constellation point.Fig. 5 is after having qpsk modulation signal constellation point process Rayleigh channel arrival receiving terminal shown in Figure 3, received signal is carried out the QPSK restituted signal planisphere that obtains after channel estimating and the equilibrium at receiving terminal.As can see from Figure 5, owing to have influences such as multidiameter, phase deviation and decline in the Rayleigh channel, though therefore by channel estimating and balanced influence of removing channel, but the restituted signal constellation point still departs from the modulation signal constellation point, and the amplitude that some restituted signal constellation point departs from the modulation signal constellation point is very big.

Fig. 6 is for to carry out the 16QAM modulated signal constellation that 16 ary quadrature Modulation and Amplitude Modulation (16QAM) obtain at transmitting terminal to transmitting, wherein, constellation point power has been normalized to 1.Fig. 7 has the transmitting through behind the awgn channel of 16QAM modulated signal constellation shown in Figure 6, the 16QAM restituted signal planisphere of the received signal that obtains at receiving terminal.As can see from Figure 7, the 16QAM restituted signal constellation point that obtains behind the process awgn channel is owing to The noise, and the 16QAM restituted signal constellation point overwhelming majority has all departed from 16QAM modulation signal constellation point.

From Fig. 4, Fig. 5 and Fig. 7, find out easily, the restituted signal constellation point be distributed in the modulation signal constellation point around, even be distributed in from the farther position of modulation signal constellation point, and the restituted signal constellation point is easier to overlap on together, is difficult to judge the restituted signal constellation point by which modulation signal constellation point is sent.For example, at shown in Figure 4 near near the restituted signal constellation point two reference axis, at the restituted signal constellation point near near restituted signal constellation point two reference axis and modulation signal constellation point centre position shown in Figure 7, at shown in Figure 5 near near the restituted signal constellation point two reference axis and away from the restituted signal constellation point of modulation signal constellation point, all can not guarantee the corresponding relation between correctly definite restituted signal constellation point and the modulation signal constellation point, thereby can cause the error of calculation of noise power, further cause the evaluated error between SNR estimated value and the actual value bigger.

Summary of the invention

In view of this, main purpose of the present invention is to provide the method for a kind of SNR of estimation, reduces the estimated value of SNR and the error between the actual value.

Another main purpose of the present invention is to provide the device of a kind of SNR of estimation, reduces the estimated value of SNR and the error between the actual value.

In order to reach above-mentioned first purpose, the invention provides a kind of method of estimated snr, this method comprises:

A. in the restituted signal constellation point, select the signal component absolute value greater than the peaked restituted signal constellation point of modulation signal edge constellation point signal component absolute value, utilize selected restituted signal constellation point calculating noise power;

B. calculate transmit signal power;

C. calculate the ratio of transmit signal power and noise power, obtain the estimated value of signal to noise ratio;

Wherein, describedly utilize selected restituted signal constellation point calculating noise power to comprise,

When the power calculation of two dimension is converted into the power calculation of one dimension, according to following formula calculating noise power:

${\mathrm{\σ}}^2=\frac{1}{K}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{|z_k-c_m|}^2\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}^{\′}}\exp (-\frac{{|z_k-c_m|}^2}{2{\mathrm{\σ}}^{\′2}})\×2,$

Wherein, the in-phase component of selected restituted signal constellation point and quadrature component are total up to K, constitute sequence z _k, k=1 ... .K, M are modulation signal edge constellation point number, c _mThe analog value that is the coordinate transformation of m modulation signal edge constellation point in two-dimentional planisphere in the one dimension planisphere,

The expression probability factor, σ ²And σ ' ²Be noise power, σ ' wherein ²Be the initial value of noise power or the noise power-value that obtains of iteration each time;

Calculating noise power σ at every turn ²After, judge each resulting noise power σ ²With the initial value of noise power in the described probability factor or the noise power-value σ ' that iteration obtains ²Difference whether less than the difference threshold that is provided with, if, execution in step b then; Otherwise, with this resulting noise power as the variances sigma of probability factor ' ², recomputate noise power.

Described signal component is in-phase component and quadrature component;

Step a is described in the restituted signal constellation point to be: select the in-phase component absolute value greater than the peaked restituted signal constellation point of modulation signal edge constellation point in-phase component absolute value in the restituted signal constellation point, select the quadrature component absolute value greater than the peaked restituted signal constellation point of modulation signal edge constellation point quadrature component absolute value.

The described in-phase component absolute value of selecting in selected restituted signal constellation point is greater than the peaked restituted signal constellation point of modulation signal edge constellation point in-phase component absolute value, select the quadrature component absolute value to comprise greater than the step of the peaked restituted signal constellation point of modulation signal edge constellation point quadrature component absolute value: in-phase component and quadrature component to selected restituted signal constellation point take absolute value respectively, obtain positive restituted signal constellation point, whether the in-phase component of judging resulting positive restituted signal constellation point is greater than modulation signal edge constellation point in-phase component absolute value maximum, if then select described in-phase component; Whether the quadrature component of judging resulting positive restituted signal constellation point is greater than modulation signal edge constellation point quadrature component absolute value maximum, if then select described orthogonal variable.

The described calculating transmit signal power of step b is:

Utilize all restituted signal constellation point to calculate received signal power, deduct the resulting noise power of step a then and obtain transmit signal power;

Perhaps, when the modulation signal constellation point power normalization that transmits is 1, then transmit signal power is 1.

Described restituted signal constellation point is: the restituted signal constellation point that the modulation signal constellation point obtains by the additive white Gaussian noise awgn channel;

Perhaps, described restituted signal constellation point is: the constellation point that the modulation signal constellation point obtains by Rayleigh Rayleigh channel is carried out the restituted signal constellation point that equilibrium obtains.

Described modulation signal constellation point is: utilize to have the modulation system of square symmetry planisphere, carry out the constellation point that the planisphere mapping obtains.

In order to reach above-mentioned second purpose, the invention provides a kind of device of estimated snr, this device comprises: constellation diagram map unit, constellation point selected cell, noise power calculation unit, transmit signal power computing unit and snr computation unit;

Wherein, constellation diagram map unit is used for carrying out to received signal constellation diagram mapping, obtains the restituted signal constellation point;

The constellation point selected cell is used for separating the resulting restituted signal constellation point of map unit from constellation point and selects the signal component absolute value greater than the peaked restituted signal constellation point of modulation signal edge constellation point signal component absolute value;

The noise power calculation unit is used to the restituted signal constellation point of utilizing the constellation point selected cell selected, when the power calculation of two dimension is converted into the power calculation of one dimension, according to

Formula calculating noise power; Wherein, the in-phase component of selected restituted signal constellation point and quadrature component are total up to K, constitute sequence z _k, k=1 ... .K, M are modulation signal edge constellation point number, c _mThe analog value that is the coordinate transformation of m modulation signal edge constellation point in two-dimentional planisphere in the one dimension planisphere,

The expression probability factor, σ ²And σ ' ²Be noise power, and, σ ' ²Be the initial value of noise power or the noise power-value that obtains of iteration each time; Calculating noise power σ at every turn ²After, judge each resulting noise power σ ²With the initial value of noise power in the described probability factor or the noise power-value σ ' that iteration obtains ²Difference whether less than the difference threshold that is provided with, if then obtain noise power; Otherwise, with this resulting noise power as the variances sigma of probability factor ' ², recomputate noise power;

The transmit signal power computing unit is used to calculate transmit signal power;

The snr computation unit is used for according to resulting noise power in noise power calculation unit and the resulting transmit signal power of transmit signal power computing unit, calculates the estimated value that transmit signal power and the ratio of noise power obtain signal to noise ratio.

Described transmit signal power computing unit calculates received signal power for utilizing the resulting restituted signal constellation point of constellation diagram map unit, from the noise power calculation unit, obtain noise signal power, and the difference of calculating received signal power and noise power obtains the unit of transmit signal power; Perhaps for when the modulation signal constellation point power normalization that transmits is 1, directly draw transmit signal power and be 1 unit.

According to method provided by the invention, only utilized in the restituted signal constellation point signal component absolute value greater than the peaked part restituted signal of modulation signal edge constellation point signal component absolute value constellation point when estimating SNR, avoid the error of the noise power that adjacent constellation point brings, thereby improved the accuracy of estimating SNR.And, the present invention also be multiply by probability factor to the noise power of each signal component, the big noise power of changes in amplitude be multiply by less probability factor, the noise power error of avoiding the restituted signal constellation point changes in amplitude brought owing to channel effect to cause greatly, thus the accuracy of estimating SNR further improved.Also have, the present invention tends towards stability noise power by the method for continuous iteration gradually, makes noise power near actual noise power, from and further improved the accuracy of estimating SNR.

Description of drawings

Figure 1 shows that the structured flowchart of digital communication system in the prior art;

Figure 2 shows that the flow chart of estimating SNR in the prior art;

Figure 3 shows that qpsk modulation signal planisphere in the prior art;

Figure 4 shows that the QPSK restituted signal planisphere through obtaining behind the awgn channel in the prior art;

Figure 5 shows that the QPSK restituted signal planisphere through obtaining by channel estimating and equilibrium behind the Rayleigh channel in the prior art;

Figure 6 shows that 16QAM modulated signal constellation in the prior art;

Figure 7 shows that the 16QAM restituted signal planisphere through obtaining behind the awgn channel in the prior art;

Figure 8 shows that the flow chart of estimating SNR among the present invention;

Figure 9 shows that the schematic diagram that utilizes constellation point when calculating noise power in the embodiment of the invention one;

Figure 10 shows that the flow chart of estimating SNR in the embodiment of the invention one;

Figure 11 shows that the schematic diagram that utilizes constellation point when calculating noise power in the embodiment of the invention two;

Figure 12 shows that the schematic diagram that utilizes constellation point when calculating noise power in the embodiment of the invention two;

Figure 13 shows that the flow chart of estimating SNR in the embodiment of the invention two;

Figure 14 shows that the flow chart of estimating SNR in the embodiment of the invention three;

Figure 15 shows that the flow chart of estimating SNR in the embodiment of the invention four;

Figure 16 shows that the structured flowchart of realizing signal-to-noise ratio (SNR) estimation among the present invention.

Embodiment

For making the purpose, technical solutions and advantages of the present invention clearer, lift specific embodiment below, the present invention is further detailed explanation.

SNR method of estimation of the prior art exists the reason of big evaluated error to be, adopt all restituted signal constellation point during calculating noise power, and some restituted signal constellation point can't be determined corresponding modulation signal constellation point exactly, thereby causes evaluated error bigger.The present invention removes these to the bigger restituted signal constellation point of evaluated error influence in estimating the SNR process, only utilize part restituted signal constellation point calculating noise power in the restituted signal constellation point, thereby reduce evaluated error, improves accuracy of estimation.

The SNR method of estimation that the present invention proposes comprises following step as shown in Figure 8:

Step 801: receiving terminal is selected peaked part restituted signal constellation point in the absolute value of signal component absolute value greater than all modulation signal edge constellation point signal components in the restituted signal constellation point, and utilizes selected part restituted signal constellation point calculating noise power σ ²

Wherein, described modulation signal edge constellation point is an outermost constellation point in planisphere distributes, all constellation point all are modulation signal edge constellation point in the qpsk modulation signal planisphere as shown in Figure 3, and the absolute value of the in-phase component of normalization constellation point or quadrature component is in the 16QAM modulated signal constellation as shown in Figure 6

Constellation point be modulation signal edge constellation point.

Step 802: calculate transmit signal power P _sHere, the method for calculating transmit signal power is consistent with step 202 described in the prior art.

Wherein, if calculate transmit signal power by the method for received signal power noise abatement acoustical power, then owing to the noise power error that obtains by step 801 is less, therefore, resulting transmit signal power error is also less.

Step 803: calculate the estimated value of SNR, i.e. SNR=P _s/ σ ²Here, the method for calculating SNR is consistent with step 203 described in the prior art.

Below in conjunction with specific embodiment, describe signal-noise ratio estimation method in detail.Being normalized to 1 with transmit signal power in following examples is example, provides the planisphere of modulation signal constellation point and restituted signal constellation point.

Embodiment one: the SNR under the awgn channel QPSK modulation case estimates

In the present embodiment, when estimating SNR, the absolute value of not considering in-phase component in the QPSK restituted signal constellation point is less than the absolute value of qpsk modulation signal edge constellation point in-phase component absolute value and the quadrature component constellation point less than qpsk modulation signal edge constellation point quadrature component absolute value, and the absolute value of only considering in-phase component in the QPSK restituted signal constellation point is greater than the absolute value of qpsk modulation signal edge constellation point in-phase component absolute value or the quadrature component part QPSK restituted signal constellation point greater than qpsk modulation signal edge constellation point quadrature component absolute value.

4 qpsk modulation signal constellation point have symmetry as shown in Figure 3, therefore, in the present embodiment, the in-phase component and the quadrature component of as shown in Figure 4 the QPSK restituted signal constellation point that obtains through awgn channel are all taken absolute value, and according to the symmetry of constellation point, QPSK restituted signal constellation point all is mapped in the first quartile, obtains planisphere as shown in Figure 9.Be presented in the specific implementation flow process of estimating SNR under the QPSK modulation case under the awgn channel in detail below in conjunction with Fig. 9, its flow process comprises following step as shown in figure 10:

Step 1001: N the QPSK restituted signal constellation point y of receiving terminal to receiving _nIn-phase component and quadrature component all take absolute value, obtain the positive QPSK restituted signal constellation point y ' of N _n, y ' _n=abs[Re (y _n)]+i*abs[Im (y _n)], wherein, n=1...N takes absolute value the planisphere that obtains as shown in Figure 9.

Step 1002: with resulting positive QPSK restituted signal constellation point y ' in the step 1001 _nIn-phase component compare with the in-phase component of qpsk modulation signal edge constellation point and the absolute value of quadrature component respectively with quadrature component, in the in-phase component of positive QPSK restituted signal constellation point, select in-phase component greater than the in-phase component absolute value of qpsk modulation signal edge constellation point, in the quadrature component of positive QPSK restituted signal constellation point, select quadrature component greater than the quadrature component absolute value of qpsk modulation signal edge constellation point, selected in-phase component and quadrature component are total up to K, constitute sequence z _k, k=1...K.

Wherein, select the step of in-phase component and quadrature component to be: to judge that whether resulting positive QPSK restituted signal constellation point in-phase component is greater than qpsk modulation signal edge constellation point in-phase component absolute value, if, then select described positive QPSK restituted signal constellation point in-phase component, otherwise abandon described positive QPSK restituted signal constellation point in-phase component; Simultaneously, judge that whether resulting positive QPSK restituted signal constellation point quadrature component is greater than qpsk modulation signal edge constellation point quadrature component absolute value, if then select described positive QPSK restituted signal constellation point quadrature component, otherwise abandon described positive QPSK restituted signal constellation point quadrature component.

Because transmitted power is normalized to 1 in the present embodiment, the absolute value of qpsk modulation signal edge constellation point in-phase component and quadrature component all is

In-phase component and the quadrature component selecting to satisfy above-mentioned condition from positive QPSK restituted signal constellation point are: select in-phase component greater than

In-phase component, select quadrature component greater than

Quadrature component.Do not carry out the situation of power normalization for transmitted power, the absolute value of qpsk modulation signal edge constellation point in-phase component and quadrature component all is 1, needs to select in-phase component greater than 1 in-phase component, selects quadrature component greater than 1 quadrature component.

As shown in Figure 9, the selected in-phase component of this step is the in-phase component of solid line the right constellation point, and selected quadrature component is the quadrature component of dotted line top constellation point.

Step 1003: utilize resulting sequence z _kCalculating noise power σ ², noise power σ ²Be z _kIn the twice of the average of squared difference between each element and qpsk modulation signal edge constellation point in-phase component or the quadrature component absolute value c, i.e. σ ²=E[(z _k-c) ²] * 2, k=1...K perhaps can be expressed as

${\mathrm{\σ}}^2=\frac{1}{K}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{(z_k-c)}^2\×2.$

Because transmitted power is normalized to 1 in the present embodiment, therefore, the absolute value c of described qpsk modulation signal edge constellation point in-phase component or quadrature component is

Do not carry out the situation of power normalization for transmitted power, the absolute value c of described qpsk modulation signal edge constellation point in-phase component or quadrature component is 1.

Because in this step during calculating noise power, the power calculation of two dimension is converted into the power calculation of one dimension, therefore, need be to z _kAnd the average of squared difference multiply by 2 between qpsk modulation signal edge constellation point in-phase component or the quadrature component absolute value c.

Step 1004: obtain transmit signal power P _s

Because transmitted power is normalized to 1 in the present embodiment, therefore, transmit signal power P _s=1.Do not carry out the situation of power normalization for transmitted power, utilize all restituted signal constellation point to calculate received signal powers after, calculate described received signal power and noise power σ ²Difference obtain transmit signal power P _s

Step 1005: calculate the estimated value of SNR, the estimated value of SNR is transmit signal power P _sWith the ratio of noise power, i.e. SNR=P _s/ σ ²

Embodiment two: the SNR under the awgn channel 16QAM modulation case estimates

In the present embodiment, when estimating SNR, the absolute value of not considering in-phase component in the 16QAM restituted signal constellation point is less than 16QAM modulation signal edge constellation point in-phase component absolute value maximum, and the absolute value of quadrature component is less than the peaked constellation point of 16QAM modulation signal edge constellation point quadrature component absolute value, only consider in-phase component absolute value in the 16QAM restituted signal constellation point greater than the absolute value of 16QAM modulation signal edge constellation point in-phase component absolute value maximum or quadrature component greater than the peaked part 16QAM restituted signal of 16QAM modulation signal edge constellation point quadrature component absolute value constellation point.Wherein, described 16QAM modulation signal edge constellation point is the constellation point on the solid line boxes position in the 16QAM restituted signal planisphere shown in Figure 11.Because 16QAM modulation signal constellation point has a plurality of edges constellation point, so can there be a plurality of values in the signal component of edge constellation point, thereby can there be a plurality of values in the absolute value of in-phase component and quadrature component.Among the present invention, selected maximum is called the absolute value maximum of signal component from the absolute value of the signal component of a plurality of edges constellation point, promptly selected maximum is called the absolute value maximum of in-phase component from the in-phase component absolute value of a plurality of edges constellation point, selected maximum is called the absolute value maximum of quadrature component from the quadrature component absolute value of a plurality of edges constellation point, and described absolute value maximum is used for the benchmark of the restituted signal constellation point of calculating noise power as selection.Do not consider solid line boxes shown in Figure 11 when estimating SNR in the present embodiment with interior 16QAM restituted signal constellation point, and consider solid line boxes 16QAM restituted signal constellation point in addition.

16 16QAM modulation signal constellation point have symmetry as shown in Figure 6, therefore, in the present embodiment, the in-phase component and the quadrature component of as shown in Figure 7 the 16QAM restituted signal constellation point that obtains through awgn channel are all taken absolute value, and according to the symmetry of constellation point, 16QAM restituted signal constellation point all is mapped in the first quartile, obtains planisphere as shown in figure 12.Be presented in the specific implementation flow process of estimating SNR under the 16QAM modulation case under the awgn channel in detail below in conjunction with Figure 12, its flow process comprises following step as shown in figure 13:

Step 1301: N the 16QAM restituted signal constellation point y of receiving terminal to receiving _nIn-phase component and quadrature component all take absolute value, obtain the positive 16QAM restituted signal constellation point y ' of N _n, y ' _n=abs[Re (y _n)]+i*abs[Im (y _n)], wherein, n=1...N takes absolute value the planisphere that obtains as shown in figure 12.

Step 1302: with resulting positive 16QAM restituted signal constellation point y ' in the step 1301 _nIn-phase component compare with the in-phase component of 16QAM modulation signal edge constellation point and the absolute value of quadrature component respectively with quadrature component, in the in-phase component of positive 16QAM restituted signal constellation point, select the peaked in-phase component of in-phase component absolute value greater than 16QAM modulation signal edge constellation point, in the quadrature component of positive 16QAM restituted signal constellation point, select the peaked quadrature component of quadrature component absolute value greater than 16QAM modulation signal edge constellation point, selected in-phase component and quadrature component are total up to K, constitute sequence z _k, k=1...K.

Wherein, select the step of in-phase component and quadrature component to be: to judge that whether resulting positive 16QAM restituted signal constellation point in-phase component is greater than 16QAM modulation signal edge constellation point in-phase component absolute value maximum, if, then select described positive 16QAM restituted signal constellation point in-phase component, otherwise abandon described positive 16QAM restituted signal constellation point in-phase component; Simultaneously, judge that whether resulting positive 16QAM restituted signal constellation point quadrature component is greater than 16QAM modulation signal edge constellation point quadrature component absolute value maximum, if, then select described positive 16QAM restituted signal constellation point quadrature component, otherwise abandon described positive 16QAM restituted signal constellation point quadrature component.

Because transmitted power is normalized to 1 in the present embodiment, the absolute value maximum of 16QAM modulation signal edge constellation point in-phase component and quadrature component all is In-phase component and the quadrature component selecting to satisfy above-mentioned condition from positive 16QAM restituted signal constellation point are: select in-phase component greater than

In-phase component, select quadrature component greater than

Quadrature component.Do not carry out the situation of power normalization for transmitted power, the absolute value maximum of 16QAM modulation signal edge constellation point in-phase component and quadrature component all is 3, needs to select in-phase component greater than 3 in-phase component, selects quadrature component greater than 3 quadrature component.

As Figure 12, the selected in-phase component of this step is the in-phase component of solid line the right constellation point, and selected quadrature component is the quadrature component of dotted line top constellation point.

Step 1303: utilize resulting sequence z _kCalculating noise power σ ², noise power σ ²Be z _kIn the twice of the average of squared difference between each element and 16QAM modulation signal edge constellation point in-phase component or the quadrature component absolute value maximum c, i.e. σ ²=E[(z _k-c) ²] * 2, k=1...K perhaps can be expressed as

In the present embodiment, because transmitted power is normalized to 1 in the present embodiment, therefore, the in-phase component of described 16QAM modulation signal edge constellation point or the absolute value maximum c of quadrature component are

Do not carry out the situation of power normalization for transmitted power, the absolute value maximum c of described 16QAM modulation signal edge constellation point in-phase component or quadrature component is 3.

Because in this step during calculating noise power, the power calculation of two dimension is converted into the power calculation of one dimension, therefore, need be to z _kAnd the average of squared difference multiply by 2 between 16QAM modulation signal edge constellation point in-phase component or the quadrature component absolute value c.

Step 1304: obtain transmit signal power P _s

Because transmitted power is normalized to 1 in the present embodiment, therefore, transmit signal power P _s=1.Do not carry out the situation of power normalization for transmitted power, utilize all restituted signal constellation point to calculate received signal powers after, calculate described received signal power and noise power σ ²Difference obtain transmit signal power P _s

Step 1305: calculate the estimated value of SNR, the estimated value of SNR is transmit signal power P _sWith the ratio of noise power, i.e. SNR=P _s/ σ ²

Above-described embodiment one, two is respectively the embodiment that estimates SNR under awgn channel, is given in the embodiment that estimates SNR under the Rayleigh channel below.In the Rayleigh channel, there are influences such as multidiameter, phase deviation and decline, receiving terminal need utilize received signal to carry out channel estimating and its signal is carried out the restituted signal constellation point that equilibrium obtains removing channel effect, but because channel estimating and balanced inaccuracy, the amplitude that part restituted signal constellation point departs from the modulation signal constellation point is very big, and these restituted signal constellation point estimate to produce bigger error to SNR.At this problem, the present invention proposes when calculating noise power, the noise power of each restituted signal constellation point be multiply by the probability factor of its present position, noise power to resulting each restituted signal constellation point is averaged the average power that obtains received signal again, thereby reduces the evaluated error that the big restituted signal constellation point of changes in amplitude is brought.

From Probability as can be known, each restituted signal constellation point departs from the common Gaussian distributed of probability of modulation signal constellation point.Therefore in embodiments of the present invention during calculating noise power, the noise power of each restituted signal constellation point is multiplied by the noise power that gaussian probability that this restituted signal constellation point departs from each modulation signal constellation point obtains this restituted signal constellation point, the noise power of all restituted signal constellation point that obtain according to this method is averaged the average noise power that obtains received signal then, shown in the following formula of resulting noise power:

${\mathrm{\σ}}^2=\frac{1}{N}\underset{n=1}{\overset{N}{\mathrm{\Σ}}}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{|y_n-c_m|}^2\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}^{\′}}\exp (-\frac{{|y_n-c_m|}^2}{2{\mathrm{\σ}}^{\′2}})---\left(I\right)$

Among the formula I, N is the restituted signal constellation point number of received signal, y _nBe n restituted signal constellation point, M is a modulation signal constellation point number, c _mBe the coordinate of m modulation signal constellation point in planisphere, | y _n-c _m| ²The noise power of representing n relative m the modulation signal constellation point of restituted signal constellation point,

Represent that n restituted signal constellation point departs from the probability factor of m modulation signal constellation point, this probability factor is with σ ' ²For variance, with restituted signal constellation point and modulation signal constellation point distance is the probability factor of the normal distribution of stochastic variable, σ ²And σ ' ²It is noise power.Formula I is the iterative formula of calculating noise power, utilizes σ ' ²Iteration obtains σ ², σ ' wherein ²Be the initial value of noise power or the noise power-value that obtains of iteration each time, the noise power σ that obtains tending towards stability by iteration ²

Below in conjunction with specific embodiment, be presented in detail and carry out the SNR estimation approach under the Rayleigh channel.

Embodiment three: the SNR under the Rayleigh channel QPSK modulation case estimates

After Fig. 5 receives the qpsk modulation signal of process Rayleigh channel for receiving terminal, carry out channel estimating and the balanced constellation point that obtains to received signal.As can be seen from Figure 5, because balanced inaccuracy, the changes in amplitude that a part receives constellation point is very big, and the constellation point that these changes in amplitude are big is bigger to the estimation influence of SNR, therefore, utilize the above-mentioned method minimizing SNR evaluated error that noise power be multiply by probability factor in the present embodiment.

In the present embodiment, do not consider in-phase component absolute value in the QPSK restituted signal constellation point less than the absolute value of qpsk modulation signal edge constellation point in-phase component absolute value and quadrature component constellation point yet, only consider that in-phase component absolute value in the QPSK restituted signal constellation point is greater than the absolute value of qpsk modulation signal edge constellation point in-phase component absolute value or the quadrature component part QPSK restituted signal constellation point greater than qpsk modulation signal edge constellation point quadrature component absolute value less than qpsk modulation signal edge constellation point quadrature component absolute value.

SNR under QPSK modulation case under the Rayleigh channel estimates flow process as shown in figure 14, comprises following step:

Step 1401,1402: respectively with the step 1001 described in the embodiment one, 1002 description with handle consistent.

Step 1403: utilize resulting sequence z _k,, the steps include: by alternative manner calculating noise power

Step 1403-1: the initial value σ ' that is provided for the noise power of iteration ²Wherein, described initial value σ ' ²Can be set to arbitrary value, more appropriate value for example rule of thumb is set, the initial value of setting appropriately can reduce number of iterations.

Step 1403-2: utilize sequence z _kIn each element, according to formula I calculating noise power σ ²

Because by step 1401 and 1402, the restituted signal constellation point of received signal is transformed into one dimension, the modulation signal edge constellation point among the iterative formula I of calculating noise power also needs to be converted into accordingly one dimension.The qpsk modulation signal constellation point has only+1 or-1 two constellation point in the one dimension, therefore, and constellation point number M=2 among the formula I, corresponding c _mValue is respectively when power normalization

For the situation that does not have power normalization, c _mValue is respectively c ₁=1, c ₂=-1.

By above analysis, the iterative formula of calculating noise power in the present embodiment is as follows:

${\mathrm{\σ}}^2=\frac{1}{K}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\underset{m=1}{\overset{2}{\mathrm{\Σ}}}{|z_k-c_m|}^2\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}^{\′}}\exp (-\frac{{|z_k-c_m|}^2}{2{\mathrm{\σ}}^{{\′}^2}})\×2---\left(\mathrm{II}\right)$

Owing to only utilized in-phase component or quadrature component absolute value QPSK restituted signal constellation point in the present embodiment greater than the sideband position of qpsk modulation signal edge constellation point in-phase component or quadrature component absolute value, and during calculating noise power, square distance between each in-phase component or quadrature component and qpsk modulation signal edge constellation point in-phase component or the quadrature component absolute value be multiply by probability factor, therefore, qpsk modulation signal edge constellation point for remote location is little to the noise power influence, can consider the qpsk modulation signal edge constellation point of remote location in the present embodiment, thereby the formula II of calculating noise power can further be reduced to formula as follows:

${\mathrm{\σ}}^2=\frac{1}{K}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{\left|(z_k-c)\right|}^2\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}^{\′}}\exp (-\frac{{|z_k-c|}^2}{2{\mathrm{\σ}}^{{\′}^2}})\×2---\left(\mathrm{III}\right)$

Among the formula III, during power normalization,

When not having power normalization, c=1.

According to above analysis, utilize formula III calculating noise power in this step.

Step 1403-3: judge noise power σ ²Whether tend towards stability, promptly compare the noise power σ that iteration obtains ²The noise power σ ' that obtains with its initial value or a preceding iteration ²Difference whether less than a difference threshold that is provided with, if then think noise power σ ²Tend towards stability execution in step 1404; Otherwise, think noise power σ ²Do not tend towards stability execution in step 1403-4.Wherein, difference threshold can for example, can be set to 0.001 according to concrete applied environment setting.

Step 1403-4: with the resulting noise power σ of step 1403-2 ²Be set to the initial value σ ' of the noise power of next iteration ², return execution in step 1403-2 then.

Step 1404,1405: respectively with the step 1004 described in the embodiment one, 1005 description with handle consistent.

Embodiment four: the SNR under the Rayleigh channel 16QAM modulation case estimates

In the present embodiment, the 16QAM modulation signal of process Rayleigh channel is because balanced inaccuracy, the changes in amplitude that receives constellation point also can be very big, therefore, also reduces the big constellation point of changes in amplitude is estimated generation to SNR influence according to the method for noise power being taken advantage of probability factor.SNR under 16QAM modulation case under the Rayleigh channel estimates flow process as shown in figure 15, comprises following step:

Step 1501,1502: respectively with the step 1301 described in the embodiment two, 1302 description with handle consistent.

Step 1503: utilize resulting sequence z _k,, the steps include: by alternative manner calculating noise power

Step 1503-1: the initial value σ ' that is provided for the noise power of iteration ²Wherein, described initial value σ ' ²Can be set to arbitrary value, more appropriate value for example rule of thumb is set, the initial value of setting appropriately can reduce number of iterations.

Step 1503-2: utilize sequence z _kIn each element, according to formula I calculating noise power σ ²

Because by step 1501 and 1502, the restituted signal constellation point of received signal is transformed into one dimension, the modulation signal constellation point among the iterative formula I of calculating noise power also needs to be converted into accordingly one dimension.In the one dimension 16QAM modulation signal constellation point have only-3 or-1 or+1 or+3 four constellation point, therefore, constellation point number M=4 among the formula I.Corresponding c _mValue is respectively when power normalization

For the situation that does not have power normalization, c _mValue is respectively c ₁=-3, c ₂=-1, c ₃=1, c ₄=3.

By above analysis, the iterative formula of calculating noise power in the present embodiment is as follows:

${\mathrm{\σ}}^2=\frac{1}{K}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\underset{m=1}{\overset{4}{\mathrm{\Σ}}}{|z_k-c_m|}^2\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}^{\′}}\exp (-\frac{{|z_k-c_m|}^2}{2{\mathrm{\σ}}^{{\′}^2}})\×2---\left(\mathrm{IV}\right)$

Owing to only utilized in-phase component or quadrature component absolute value 16QAM restituted signal constellation point in the present embodiment greater than the sideband position of 16QAM modulation signal edge constellation point in-phase component or quadrature component absolute value, and during calculating noise power, square distance between each in-phase component or quadrature component and 16QAM modulation signal edge constellation point in-phase component or the quadrature component absolute value be multiply by probability factor, therefore, 16QAM modulation signal edge constellation point for remote location is little to the noise power influence, can consider the 16QAM modulation signal edge constellation point of remote location in the present embodiment, thereby the formula IV of calculating noise power can further be reduced to formula as follows:

${\mathrm{\σ}}^2=\frac{1}{K}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{|z_k-c|}^2\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}^{\′}}\exp (-\frac{{|z_k-c|}^2}{2{\mathrm{\σ}}^{{\′}^2}})\×2---\left(V\right)$

Among the formula V, when power normalization,

When not having power normalization, c=3.

According to above analysis, utilize formula V calculating noise power in this step.

Step 1503-3: judge noise power σ ²Whether tend towards stability, promptly compare the noise power σ that iteration obtains ²The noise power σ ' that obtains with its initial value or a preceding iteration ²Difference whether less than a difference threshold that is provided with, if then think noise power σ ²Tend towards stability execution in step 1504; Otherwise, think noise power σ ²Do not tend towards stability execution in step 1503-4.Wherein, difference threshold can for example, can be set to 0.001 according to concrete applied environment setting.

Step 1503-4: with the resulting noise power σ of step 1503-2 ²Be set to the initial value σ ' of the noise power of next iteration ², return execution in step 1503-2 then.

Step 1504,1505: respectively with embodiment two in step 1304,1305 description and handle consistent.

In embodiment one and two described SNR estimations under awgn channel, also can use the method for utilizing probability factor and alternative manner calculating noise power in embodiment three and the four described SNR estimations under the Ray1eigh channel.

The above embodiment is applicable to that the SNR with symmetric modulated signal constellation estimates, because the modulation signal constellation point of QPSK, 16QAM has symmetry, the restituted signal constellation point can be mapped to first quartile, simplify the step of calculating noise power, equally, also the restituted signal constellation point can be mapped to other quadrant.

Above embodiment is described, and the restituted signal constellation point is mapped to the just convenience in order to calculate of a quadrant, and in the present invention, also can be mapped to a certain quadrant to the restituted signal constellation point also can calculating noise power, is described below described:

At first, the in-phase component and the quadrature component of certain condition satisfied in selection, the method of selecting is: the quadrant of determining restituted signal constellation point place, from the restituted signal constellation point of first quartile, select in-phase component greater than the peaked in-phase component of first quartile modulation signal edge constellation point in-phase component, select quadrature component greater than the peaked quadrature component of first quartile modulation signal edge constellation point quadrature component; From the restituted signal constellation point of second quadrant, select the in-phase component of in-phase component, select quadrature component greater than the peaked quadrature component of the second quadrant modulation signal edge constellation point quadrature component less than the second quadrant modulation signal edge constellation point in-phase component minimum value; From the restituted signal constellation point of third quadrant, select the in-phase component of in-phase component, select the quadrature component of quadrature component less than third quadrant modulation signal edge constellation point quadrature component minimum value less than third quadrant modulation signal edge constellation point in-phase component minimum value; From four-quadrant restituted signal constellation point, select in-phase component greater than the peaked in-phase component of four-quadrant modulation signal edge constellation point in-phase component, select the quadrature component of quadrature component less than four-quadrant modulation signal edge constellation point quadrature component minimum value.

Then, utilize selected in-phase component and quadrature component calculating noise power, its method is: when calculating the noise power of selected in-phase component and quadrature component in the first quartile, the modulation signal edge peaked squared difference of constellation point in-phase component is the noise power of this in-phase component in in-phase component and the first quartile, and the modulation signal edge peaked squared difference of constellation point quadrature component is the noise power of this quadrature component in quadrature component and the first quartile; When calculating the noise power of selected in-phase component and quadrature component in second quadrant, the squared difference of modulation signal edge constellation point in-phase component minimum value is the noise power of this in-phase component in the in-phase component and second quadrant, and the modulation signal edge peaked squared difference of constellation point quadrature component is the noise power of this quadrature component in the quadrature component and second quadrant; When calculating the noise power of selected in-phase component and quadrature component in the third quadrant, the squared difference of modulation signal edge constellation point in-phase component minimum value is the noise power of this in-phase component in in-phase component and the third quadrant, and the squared difference of modulation signal edge constellation point quadrature component minimum value is the noise power of this quadrature component in quadrature component and the third quadrant; When calculating the noise power of selected in-phase component and quadrature component in the four-quadrant, the modulation signal edge peaked squared difference of constellation point in-phase component is the noise power of this in-phase component in in-phase component and the four-quadrant, and the squared difference of modulation signal edge constellation point quadrature component minimum value is the noise power of this quadrature component in quadrature component and the four-quadrant.The twice of the mean value of the noise power sum of in-phase component and quadrature component obtains noise power in the calculating All Quardrants.

The above method of directly utilizing resulting restituted signal constellation point to calculate SNR still belongs within the protection range of the present invention.

The method of estimation SNR of the present invention is not only applicable to awgn channel, Rayleigh channel, also is applicable to the channel with other characteristic; Simultaneously, be not only applicable to QPSK and 16QAM modulation, also be applicable to other modulation system, for example 64QAM modulation etc. with square symmetry planisphere.

The method of above-described estimation SNR can realize by following structure, Figure 16 shows that the structured flowchart of realizing signal-to-noise ratio (SNR) estimation, comprise in this structure: constellation diagram map unit, constellation point selected cell, noise power calculation unit, transmit signal power computing unit and snr computation unit.

Wherein, constellation diagram map unit is carried out constellation diagram mapping to received signal, obtains the restituted signal constellation point.The constellation point selected cell is separated the restituted signal constellation point that map unit obtains from constellation point, select peaked part restituted signal constellation point in the absolute value of signal component absolute value greater than all modulation signal edge constellation point signal components, pass to the noise power calculation unit then.The selected restituted signal constellation point of noise power calculation unit by using constellation point selected cell calculating noise power.And the transmit signal power computing unit calculates transmit signal power, at first utilize resulting all restituted signal constellation point calculating received signal powers in the constellation diagram map unit, obtain noise signal power then from the noise power calculation unit, the difference of calculating received signal power and noise power again obtains transmit signal power.When the modulation signal constellation point power normalization that transmits was 1, it was 1 that the transmit signal power computing unit can directly draw transmit signal power.The transmit signal power that noise power that the snr computation unit obtains according to the noise power calculation unit and transmit signal power computing unit obtain calculates the estimated value that transmit signal power and the ratio of noise power obtain signal to noise ratio.

The above only is preferred embodiment of the present invention, and is in order to restriction the present invention, within the spirit and principles in the present invention not all, any modification of being done, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (8)

1. the method for an estimated snr comprises:

A. in the restituted signal constellation point, select the signal component absolute value greater than the peaked restituted signal constellation point of modulation signal edge constellation point signal component absolute value, utilize selected restituted signal constellation point calculating noise power;

B. calculate transmit signal power;

C. calculate the ratio of transmit signal power and noise power, obtain the estimated value of signal to noise ratio;

It is characterized in that, describedly utilize selected restituted signal constellation point calculating noise power to comprise,

When the power calculation of two dimension is converted into the power calculation of one dimension, according to following formula calculating noise power:

Wherein, the in-phase component of selected restituted signal constellation point and quadrature component are total up to K, constitute sequence z _k, k=1 ... .K, M are modulation signal edge constellation point number, c _mThe analog value that is the coordinate transformation of m modulation signal edge constellation point in two-dimentional planisphere in the one dimension planisphere,

The expression probability factor, σ ²And σ ' ²Be noise power, and, σ ' ²Be the initial value of noise power or the noise power-value that obtains of iteration each time;

Calculating noise power σ at every turn ²After, judge each resulting noise power σ ²With the initial value of noise power in the described probability factor or the noise power-value σ ' that iteration obtains ²Difference whether less than the difference threshold that is provided with, if, execution in step b then; Otherwise, with this resulting noise power as the variances sigma of probability factor ' ², recomputate noise power.

2. method according to claim 1 is characterized in that,

Described signal component is in-phase component and quadrature component;

The described signal component absolute value of selecting in the restituted signal constellation point greater than the modulation signal edge peaked restituted signal constellation point of constellation point signal component absolute value is: select the in-phase component absolute value greater than the peaked restituted signal constellation point of modulation signal edge constellation point in-phase component absolute value in the restituted signal constellation point, select the quadrature component absolute value greater than the peaked restituted signal constellation point of modulation signal edge constellation point quadrature component absolute value.

3. method according to claim 2 is characterized in that,

The described in-phase component absolute value of selecting in the restituted signal constellation point is greater than the peaked restituted signal constellation point of modulation signal edge constellation point in-phase component absolute value, select the quadrature component absolute value to comprise greater than the step of the peaked restituted signal constellation point of modulation signal edge constellation point quadrature component absolute value: in-phase component and quadrature component to selected restituted signal constellation point take absolute value respectively, obtain positive restituted signal constellation point, whether the in-phase component of judging resulting positive restituted signal constellation point is greater than modulation signal edge constellation point in-phase component absolute value maximum, if then select described in-phase component; Whether the quadrature component of judging resulting positive restituted signal constellation point is greater than modulation signal edge constellation point quadrature component absolute value maximum, if then select described orthogonal variable.

4. according to each described method of claim 1 to 3, it is characterized in that the described calculating transmit signal power of step b is:

Utilize all restituted signal constellation point to calculate received signal power, deduct the resulting noise power of step a then and obtain transmit signal power;

Perhaps, when the modulation signal constellation point power normalization that transmits is 1, then transmit signal power is 1.

5. according to each described method of claim 1 to 3, it is characterized in that,

Described restituted signal constellation point is: the restituted signal constellation point that the modulation signal constellation point obtains by the additive white Gaussian noise awgn channel;

Perhaps, described restituted signal constellation point is: the modulation signal constellation point is carried out the constellation point that equilibrium obtains by the constellation point that Rayleigh Rayleigh channel obtains.

6. method according to claim 5 is characterized in that,

Described modulation signal constellation point is: utilize to have the modulation system of square symmetry planisphere, carry out the constellation point that the planisphere mapping obtains.

7. the device of an estimated snr is characterized in that, this device comprises: constellation diagram map unit, constellation point selected cell, noise power calculation unit, transmit signal power computing unit and snr computation unit;

Wherein, constellation diagram map unit is used for carrying out to received signal constellation diagram mapping, obtains the restituted signal constellation point;

The constellation point selected cell is used for separating the resulting restituted signal constellation point of map unit from constellation point and selects the signal component absolute value greater than the peaked restituted signal constellation point of modulation signal edge constellation point signal component absolute value;

The noise power calculation unit is used to utilize the selected restituted signal constellation point of constellation point selected cell, when the power calculation of two dimension is converted into the power calculation of one dimension, according to

Formula calculating noise power; Wherein, the in-phase component of selected restituted signal constellation point and quadrature component are total up to K, constitute sequence z _k, k=1 ... .K, M are modulation signal edge constellation point number, c _mThe analog value that is the coordinate transformation of m modulation signal edge constellation point in two-dimentional planisphere in the one dimension planisphere,

The expression probability factor, σ ²And σ ' ²Be noise power, and σ ' ²Be the initial value of noise power or the noise power-value that obtains of iteration each time; Calculating noise power σ at every turn ²After, judge each resulting noise power σ ²With the initial value of noise power in the described probability factor or the noise power-value σ ' that iteration obtains ²Difference whether less than the difference threshold that is provided with, if then obtain noise power; Otherwise, with this resulting noise power as the variances sigma of probability factor ' ², recomputate noise power;

The transmit signal power computing unit is used to calculate transmit signal power;

The snr computation unit is used for according to resulting noise power in noise power calculation unit and the resulting transmit signal power of transmit signal power computing unit, calculates the estimated value that transmit signal power and the ratio of noise power obtain signal to noise ratio.

8. device according to claim 7, it is characterized in that, described transmit signal power computing unit calculates received signal power for utilizing the resulting restituted signal constellation point of constellation diagram map unit, from the noise power calculation unit, obtain noise signal power, and the difference of calculating received signal power and noise power obtains the unit of transmit signal power; Perhaps for when the modulation signal constellation point power normalization that transmits is 1, directly draw transmit signal power and be 1 unit.

Images