CN1901520A - Noise energy estimating method and device - Google Patents

Abstract

This invention discloses an estimation method of noise energy including: getting the being estimated signal containing real signals and target signals, eliminating the real signals from the signals to be estimated to get the intermediate noise and carrying out noise energy estimation to the target noise based on the got intermediate noise. This invention also discloses a device for estimating noise energy, which can increase the estimation accuracy by the signal-cancel method, in which, suitable transmission format can be selected to the HSDPA system to increase its throughput performance and the emission power can be adjusted based on the instant channel quality instruction, so it's more suitable for the instant communication environment to increase performance and efficiency of the communication system.

Description

A kind of noise energy estimating method and device

Technical field

The present invention relates to communication technical field, relate in particular to a kind of noise energy estimating method and device.

Background technology

In WCDMA (Wideband Code Division Multiple Access (WCDMA)) system, signal that transmitting terminal sends is through after the transmission of wireless channel, and can mix has various interference and noise, as thermal noise of other signal and receiver itself or the like.WCDMA is a spread spectrum system, distinguish unlike signal with different spreading codes, so, at receiving terminal, spreading code with actual signal carries out related operation (just usually said de-spreading operation) to total received signal, make the actual signal energy increase, and other signal become noise, is superimposed upon (as shown in Figure 1) on the actual signal with thermal noise.De-spreading operation makes that the signal to noise ratio after the despreading is the fixedly multiple (spreading factor doubly) of signal to noise ratio before the despreading, signal to noise ratio after the despreading can embody the quality (just signal is through the suffered interference of wireless channel and the size of noise) of wireless channel, so receiving terminal usually obtains the information of channel quality by the measurement of signal to noise ratio after the despreading.Receiving terminal obtains channel quality information a lot of important purposes, for example, at the evolution version HSDPA of WCDMA (High-Speed Downlink PacketAccess, high speed downlink packet inserts) in, use the Adaptive Modulation and Coding technology, this Technology Need is selected suitable transformat according to instantaneous channel quality indication, so that effectively utilize channel condition.So in the HSDPA system, accuracy of channel quality estimation requires than higher, just need despreading here after signal-to-noise ratio (SNR) estimation can have than higher accuracy.Signal-to-noise ratio (SNR) estimation generally is respectively signal energy and noise energy to be estimated, being divided by gets final product again, and the present invention estimates to be optimized to noise energy wherein, improves whole accuracy of channel quality estimation.

Signal can be expressed as after the despreading shown in Figure 1:

r _i＝S _i+n _i (1)

(i=1,2 ..., N, N are the symbolic numbers that is used for Noise Estimation).

Wherein, S _iBe actual signal, n _iIt is noise.Usually, carrying out signal-to-noise ratio (SNR) estimation is to utilize the despread signal of CPICH (Common Pilot Channel) channel to carry out, and the value of symbol of CPICH channel itself is invariable, so, the value of symbol S after the despreading _iBe exactly that channel factors multiply by a constant, S _iAnd S _jThe correlation that has is only by the situation of change decision of channel, if channel does not change in N symbol lengths, and then in this scope, S _iAnd S _jIdentical (also promptly relevant fully), otherwise correlation can decrease; n _iThe N of supposition obedience usually (0, σ _i ²) distribute its variances sigma _i ²Variation with channel changes, and when i ≠ j, n _iAnd n _jUncorrelated mutually; Usually suppose actual signal S _iWith noise n _iUncorrelated.

In the prior art, carry out noise energy and estimate, adopt following formula usually:

${\widehat{\mathrm{\σ}}}^2=\frac{1}{N-1}*\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{|r_i-(\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{r}_j/N)|}^2$

$=\frac{1}{N-1}*(\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\left|r_i\right|}^2-\frac{{\left|\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{r}_i\right|}^2}{N})---\left(2\right)$

The thought of this Noise Estimation algorithm is, in the constant substantially estimated length of actual signal, the variance of despread signal is exactly a noise energy.

Can prove, (2) the noise energy algorithm for estimating of formula, in the constant substantially estimated length of actual signal, be only not have partially and estimate, if actual signal changes in estimated length, the nothing that then is not noise energy is estimated partially, the noise energy that estimates is bigger than normal, the simple proof process that this problem is described below is (for for simplicity, following derivation all is to carry out in real number field, actual signal and noise are plural number, but with real number representation with to derive with complex representation be equivalent, do not influence the result of final derivation and the conclusion that draws):

In (1), because actual signal S _iWith noise n _iEnergy all relevant with channel, in N symbol, may change, so, suppose that the actual signal average power in N the symbol is $\stackrel{\‾}{S^2}=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{S}_i^2/N,$ Noise average power is $\stackrel{\‾}{{\mathrm{\σ}}^2}=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2/N.$ When noise energy is estimated, estimation be exactly

$\stackrel{\‾}{{\mathrm{\σ}}^2}=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2/N.$

Owing to derive, then (2) formula can be written as in real number field:

${\widehat{\mathrm{\σ}}}^2=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{[r_i-(\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{r}_j/N)]}^2$

$=\frac{1}{N-1}(\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{r}_i^2-\frac{{\left(\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{r}_i\right)}^2}{N})---\left(3\right)$

(this average approaches the average of this estimated value of calculating more now $\stackrel{\‾}{{\mathrm{\σ}}^2}=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2/N,$ Then Noise Estimation is accurate more, if just equal, then is not have partially to estimate):

$E\left({\widehat{\mathrm{\σ}}}^2\right)=\frac{1}{N-1}[E\left(\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{r}_i^2\right)-\frac{E[{\left(\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{r}_i\right)}^2}{N}]---\left(4\right)$

Because:

$E\left(\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{r}_i^2\right)=E\left[\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{(S_i+n_i)}^2\right]=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}E\left[{(S_i+n_i)}^2\right]=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}E\left(S_i^2\right)+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2---\left(5\right)$

$E\left[{\left(\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{r}_i\right)}^2\right]=E\left[{(S_1+n_1+S_2+n_2+......+S_N+n_N)}^2\right]$

$=E[S_1\·\left(\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{S}_k\right)+S_2\·\left(\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{S}_k\right)+......+S_N\·\left(\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{S}_k\right)]+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2,$

$=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}E[S_i\·\left(\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{S}_k\right)]+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2$ (6)

So:

$E\left({\widehat{\mathrm{\σ}}}^2\right)=\frac{1}{N-1}[\underset{i=1}{\overset{N}{\mathrm{\Σ}}}E\left(S_i^2\right)+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2-\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}E[S_i\·\left(\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{S}_k\right)]}{N}-\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2}{N}$

$=\frac{1}{N-1}\left\{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\frac{[E\left(S_i^2\right)-E(S_i\·S_1)]+[E\left(S_i^2\right)-E(S_i\·S_2)+\·\·\·+[E\left(S_i^2\right)-E(S_i\·S_N)]}{N}\right\}+\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2}{N}$

(7)

If 1. channel constant in N symbol lengths (under the low speed situation), then:

S ₁＝S ₂＝...＝S _N＝S (8)

${\mathrm{\σ}}_1^2={\mathrm{\σ}}_2^2=...={\mathrm{\σ}}_N^2={\mathrm{\σ}}^2---\left(9\right)$

So:

$E\left({\widehat{\mathrm{\σ}}}^2\right)-\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2}{N}=\frac{1}{N-1}\{{\mathrm{NS}}^2+{\mathrm{N\σ}}^2-{\mathrm{NS}}^2-{\mathrm{\σ}}^2\}-{\mathrm{\σ}}^2=0,---\left(10\right)$

Also promptly, at this moment, such Noise Estimation is not have partially to estimate.

If 2. channel changes (under the high-speed case) in N symbol lengths, at this moment,

$E\left({\widehat{\mathrm{\σ}}}^2\right)-\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2}{N}$

$=\frac{1}{N-1}\left\{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\frac{[E\left(S_i^2\right)-E(S_i\·S_1)]+[E\left(S_i^2\right)-E(S_i\·S_2)]+...+[E\left(S_i^2\right)-E(S_i\·S_N)]}{N}\right\},$

(11)

Because channel changes in N symbol lengths, so, E (S _i ²)＞E (S _iS _j) (i ≠ j), so:

$E\left({\widehat{\mathrm{\σ}}}^2\right)-\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2}{N}>0---\left(12\right)$

This shows that if channel changes (under the situation of travelling carriage high-speed motion), then this traditional noise estimation method of (2) formula makes that the noise energy that estimates is higher, causes the signal-to-noise ratio (SNR) estimation value on the low side in estimated length.This point shows particularly evidently in the HSDPA system, because under the high-speed case, the signal-to-noise ratio (SNR) estimation value is on the low side, then the transformat of Xuan Zeing is on the low side, causes the throughput of this moment on the low side, does not utilize the advantage of Adaptive Modulation and Coding mode among the HSDPA well.

Summary of the invention

The purpose of this invention is to provide a kind of noise energy estimating method and device, can determine noise energy more exactly, for the HSDPA system, can select transformat, thereby improve the throughput of system according to signal-to-noise ratio (SNR) estimation value more accurately.

The invention discloses a kind of noise energy estimating method, comprising:

Acquisition signal to be estimated, described signal to be estimated comprises actual signal and target noise;

Wait to estimate from each and eliminate actual signal signal, with noise in the middle of obtaining;

According to the middle noise that obtains target noise being carried out noise energy estimates.

Described actual signal has the characteristic that can be eliminated.

Described waiting to estimate the signal from each eliminated actual signal and specifically comprised with the step of noise in the middle of obtaining:

In certain estimated length, choose the symbol that actual signal can be eliminated the signal from waiting to estimate, eliminate the actual signal in the symbol choose, the institute in estimated length remains to be estimated actual signal in the signal and eliminates and finish, to obtain the centre noise.

Described middle noise according to acquisition carries out noise energy to target noise and estimates specifically to comprise:

The middle noise that obtains is carried out noise energy estimate,, obtain the target noise energy according to middle noise energy again with noise energy in the middle of obtaining.

Described actual signal has the characteristic that can be eliminated and comprises: the actual signal in the adjacent-symbol equates.

Described waiting to estimate the signal from each eliminated actual signal and specifically comprised with the step of noise in the middle of obtaining:

In certain estimated length, choose two adjacent symbols the signal from waiting to estimate, eliminate the actual signal in two symbols; Choose two adjacent symbols successively and carry out above-mentioned processing, till in estimated length, remaining to be estimated actual signal in the signal and being eliminated, with noise in the middle of obtaining.

Described middle noise according to acquisition carries out noise energy to target noise and estimates specifically to comprise:

The middle noise that obtains is carried out noise energy estimate,, middle noise energy is changed to obtain the target noise energy with noise energy in the middle of obtaining.

The invention also discloses a kind of noise energy estimation unit, comprising:

Obtain signal element to be estimated, be used to obtain signal to be estimated;

Eliminate the actual signal unit, be used for waiting to estimate the signal to be estimated that signal element obtains and eliminating actual signal, with noise in the middle of obtaining from described obtaining;

The noise energy estimation unit is used for according to the middle noise that obtains target noise being carried out noise energy and estimates.

Described elimination actual signal unit specifically comprises:

Selected cell is used for from what obtain that signal element to be estimated obtains waiting to estimate the symbol that signal selects actual signal to be eliminated;

Eliminate the unit, be used for eliminating actual signal from the symbol that selected cell is selected.

According to the present invention, by adopting " signal cancellation " noise energy estimating method, can improve the estimated accuracy of noise, for the HSDPA system, can select more suitable transformat, thereby improve the HSDPA throughput performance under the high speed situation; For rapid power control, can adjust transmitting power according to instantaneous channel quality indication, thereby make the transmitting power of adjustment be fit to the instantaneous communication environment more, improve communication system performance and efficient.

Description of drawings

Fig. 1 shows noise and is superimposed upon despread signal schematic diagram on the actual signal;

Fig. 2 shows the schematic diagram of noise energy estimation unit of the present invention.

Embodiment

Understand and realization the present invention the existing embodiments of the invention of describing in conjunction with the accompanying drawings for the ease of persons skilled in the art.

Basic thought of the present invention is, by adopting the noise energy estimating method of " signal cancellation ", can solve the noise energy estimated value problem bigger than normal that the variation because of actual signal causes preferably.Therefore, in the WCDMA system, can improve the channel estimating quality under the high speed situation, especially can obviously improve the throughput performance of HSDPA system under the high speed situation.

By the analysis of background technology as can be known, the higher reason of noise energy estimated value is to have comprised the actual signal composition in the noise energy that estimates under the high-speed case, can expect thus, if the actual signal composition in the despread signal can be removed, only remaining noise then can be eliminated such deviation.The thinking of Here it is " signal cancellation " noise estimation method.The specific implementation of this noise estimation method is: for the despread values of N CPICH symbol, whenever adjacent 2 are subtracted each other, because channel can think constant at 2 symbols in the short like this time, the actual signal composition of adjacent 2 symbols can think equal, such one subtracts, just remaining noise section only.In estimated length, can obtain the individual such subtraction value of N/2 (estimated length N is taken as even number), this N/2 value is directly asked the mould mean value of square, can obtain the noise energy estimated value.But, this noise energy estimated value is 2 times of desired noise energy, so, also will be divided by 2.

Next, above-described this process is carried out the derivation of equation:

Suppose that estimated length N is an even number, makes M=N/2.

With despread signal r _i=S _i+ n _i(i=1,2 ..., N, N are the symbolic numbers that is used for Noise Estimation) every adjacent two subtract each other:

Owing to can think that channel is constant in two symbols, so, can think S ₁=S ₂So, obtain R ₁=r ₂-r ₁=S ₂+ n ₂-S ₁-n ₁=n ₂-n ₁, R ₁Average be 0, variance equals σ ₁ ²+ σ ₂ ²

Again because can think n ₁Energy σ ₁ ²Equal n ₂Energy σ ₂ ²(note δ ₁ ²=2 σ ₁ ²=2 σ ₂ ²), so, R ₁Be exactly one obey N (0, δ ₁ ²) stochastic variable that distributes.

In like manner:

R ₂=r ₄-r ₃=n ₄-n ₃Be obey N (0, δ ₂ ²) stochastic variable (δ that distributes ₂ ²=2 σ ₃ ²=2 σ ₄ ²) ...

R _M=r _N-r _N-1=n _N-n _N-1Be obey N (0, δ M ²) stochastic variable (δ that distributes _M ²=2 σ _N-1 ²=2 σ _N ²).

Top R ₁, R ₂..., R _MBetween be mutually independently.

To the R that obtains like this ₁, R ₂..., R _MThis M value is asked the mould mean value of square, obtains an estimated value:

${\widehat{\mathrm{\δ}}}^2=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{\left|R_i\right|}^2}{M}---\left(13\right)$

This estimated value is averaged:

$E\left({\widehat{\mathrm{\δ}}}^2\right)=\frac{E\left(\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{\left|R_i\right|}^2\right)}{M}=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{\mathrm{\δ}}_i^2}{M}---\left(14\right)$

Since front supposition M=N/2, σ _i ²=σ _I+1 ²(i=1,3,5 ..., N-1), deriving easily obtains

$2\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2/N=2\·\underset{i=\mathrm{1,3,5},...,N-1}{\mathrm{\Σ}}2{\mathrm{\σ}}_i^2/\left(2M\right)=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{\mathrm{\δ}}_i^2}{M}---\left(15\right)$

That is to say that the average of the estimated value that this method obtains is 2 times of our noise energy that need estimate.So, with estimated value

Divided by 2, what obtain is exactly noise energy again $\stackrel{\‾}{{\mathrm{\σ}}^2}=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^2/N$ Nothing estimate partially.

When estimated length N is odd number, can give up last symbol, N-1 symbol carries out Noise Estimation as stated above before only getting.

In sum, the central idea of " signal cancellation " noise energy estimating method be by eliminate actual signal composition in the despread signal (as, subtracting each other of CPICH despread signal adjacent two symbols value, actual signal composition is wherein removed), only stay noise, so just can solve the noise energy estimated value problem bigger than normal that the actual signal composition changes and causes in estimated length.

Except the WCDMA system, for other the signal to be estimated that is rendered as " actual signal+noise ", also can adopt thought of the present invention to improve the accuracy of signal to noise ratio valuation, just improve accuracy of channel quality estimation, key wherein is exactly to carry out noise energy again after utilizing the known correlation of actual signal that actual signal is eliminated to estimate.The correlation that described actual signal is known is meant that can carry out certain processing by known correlation eliminates actual signal.For example, actual signal in the foregoing description in the adjacent-symbol is known to be equated, so, adjacent-symbol can be subtracted each other, eliminate actual signal, simultaneously, the derivation of equation by the front as can be known, middle the noise energy of this moment and the relation of waiting to estimate the target noise energy in the signal originally are 2 times relations; If the actual signal in the adjacent-symbol is other relation, such as opposite, 2 times or the like, accordingly, can subtract each other afterwards by the adjacent-symbol addition, divided by 2 and eliminate actual signal, and obtain middle noise energy and the relation of waiting to estimate the target noise energy in the signal originally by the similar derivation of equation in front.

As shown in Figure 2, the invention also discloses a kind of noise energy estimation unit, comprising: obtain signal element to be estimated, be used to obtain signal to be estimated; Eliminate the actual signal unit, be used for waiting to estimate the signal to be estimated that signal element obtains and eliminating actual signal from described obtaining, in the signal after the processing only remaining in the middle of noise, and can draw noise energy and original waiting in the middle of this by theoretical derivation and estimate relation between the target noise energy in the signal; The noise energy estimation unit, be used for middle the noise of eliminating the output of actual signal unit is carried out the noise energy estimation, and according to middle noise energy and waited to estimate relation between the target noise energy in the signal originally and waited to estimate noise energy estimated value in the signal originally.

Described elimination actual signal unit specifically comprises: selected cell, be used for from obtain wait to estimate signal element obtain wait to estimate the symbol that signal selects actual signal to be eliminated; Eliminate the unit, be used for eliminating actual signal from the symbol that selected cell is selected.

According to the present invention, by adopting " signal cancellation " noise energy estimating method, can improve the estimated accuracy of noise, so that select more suitable transformat, thereby improve the HSDPA throughput performance under the high speed situation.Below form listed under the single antenna pattern, during typical IA High Speed Channel (VA120), the HSDPA throughput performance of the prior art in the background technology and " signal cancellation " of the present invention noise estimation method relatively:

In the single antenna link, the throughput performance of the following two kinds of Noise Estimation algorithms of VA120 channel relatively

	Average SNR (signal interference ratio)	Average throughput (kbps)
			Prior art	9.210	1078.794
" signal cancellation " method	11.646	1306.952

As seen from the above table, under the IA High Speed Channel, " signal cancellation " noise estimation method can solve the noise energy estimated value problem bigger than normal that the actual signal composition changes and causes in estimated length, and the SNR of estimation obviously increases, and throughput has improved about 21%.

Though described the present invention by embodiment, those of ordinary skills know, without departing from the spirit and substance in the present invention, just can make the present invention that many distortion and variation are arranged, and scope of the present invention is limited to the appended claims.

Claims (9)

1, a kind of noise energy estimating method is characterized in that, comprising:

Acquisition signal to be estimated, described signal to be estimated comprises actual signal and target noise;

Wait to estimate from each and eliminate actual signal signal, with noise in the middle of obtaining;

According to the middle noise that obtains target noise being carried out noise energy estimates.

2, method according to claim 1 is characterized in that, described actual signal has the characteristic that can be eliminated.

3, method according to claim 2 is characterized in that, described waiting to estimate the signal from each eliminated actual signal and specifically comprised with the step of noise in the middle of obtaining:

In certain estimated length, choose the symbol that actual signal can be eliminated the signal from waiting to estimate, eliminate the actual signal in the symbol choose, the institute in estimated length remains to be estimated actual signal in the signal and eliminates and finish, to obtain the centre noise.

4, method according to claim 3 is characterized in that, described middle noise according to acquisition carries out noise energy to target noise and estimates specifically to comprise:

The middle noise that obtains is carried out noise energy estimate,, obtain the target noise energy according to middle noise energy again with noise energy in the middle of obtaining.

5, method according to claim 2 is characterized in that, described actual signal has the characteristic that can be eliminated and comprises: the actual signal in the adjacent-symbol equates.

6, method according to claim 5 is characterized in that, described waiting to estimate the signal from each eliminated actual signal and specifically comprised with the step of noise in the middle of obtaining:

In certain estimated length, choose two adjacent symbols the signal from waiting to estimate, eliminate the actual signal in two symbols; Choose two adjacent symbols successively and carry out above-mentioned processing, till in estimated length, remaining to be estimated actual signal in the signal and being eliminated, with noise in the middle of obtaining.

7, method according to claim 6 is characterized in that, described middle noise according to acquisition carries out noise energy to target noise and estimates specifically to comprise:

The middle noise that obtains is carried out noise energy estimate,, middle noise energy is changed to obtain the target noise energy with noise energy in the middle of obtaining.

8, a kind of noise energy estimation unit is characterized in that, comprising:

Obtain signal element to be estimated, be used to obtain signal to be estimated;

Eliminate the actual signal unit, be used for waiting to estimate the signal to be estimated that signal element obtains and eliminating actual signal, with noise in the middle of obtaining from described obtaining;

The noise energy estimation unit is used for according to the middle noise that obtains target noise being carried out noise energy and estimates.

9, device according to claim 8 is characterized in that, described elimination actual signal unit specifically comprises:

Selected cell is used for from what obtain that signal element to be estimated obtains waiting to estimate the symbol that signal selects actual signal to be eliminated;

Eliminate the unit, be used for eliminating actual signal from the symbol that selected cell is selected.

Images