CN103095619B - Direct current elimination method and device - Google Patents

Abstract

Embodiments providing a kind of direct current elimination method and device, wherein the method includes: the reception data separate averaging method after analog digital conversion is carried out direct current elimination；The rear 128 chip data of described reception data Midamble code part are done original channel estimate, calculate the coefficient of each tap after channel is estimated；According to original estimation result, choose power and do remnant dc Processing for removing less than the tap setting thresholding.Direct current elimination method that the embodiment of the present invention is provided and device, digital information after analog digital conversion is first referred both to method and carries out direct current elimination, then according still further to remnant dc method of elimination carry out second time direct current eliminate, so enable to digital signal direct current eliminate more thorough.

Description

Direct current elimination method and device

Technical field

The present invention relates to communication technical field, particularly relate to a kind of direct current elimination method and device.

Background technology

At TD-SCDMA (TimeDivision-SynchronousCodeDivisionMultipleAccess, TD SDMA) in system, terminal, before carrying out descending joint detection algorithm, first had to the carrying out after radio frequency reception the digital signal that A/D (analog/digital) converts and carried out direct current Processing for removing.Different radio-frequency transmitter performance differences is very big, processes strategy for the direct current in the A/D conversion process of reception signal and is also not quite similar.

In TD-SCDMA system, subframe structure and regular time slots structure are as it is shown in figure 1, the persistent period of each subframe is 5ms, and each subframe one is divided into 7 regular time slots and 3 special time slots.Wherein, the regular time slots that can be used as downlink business in each subframe has 6 regular time slots such as TS0 and TS2～TS6.In each regular time slots, data symbol is positioned at Midamble (training sequence) code both sides, has protection interval (GP) of 16 chip (chip) at time slot afterbody.Descending joint-detection carries out data process in units of a time slot, for regular time slots, existing direct current elimination algorithm mainly has: segmentation disappears direct current method (abbreviation method of fractionation) and whole time slot average disappears direct current method (abbreviation averaging method), and the method (being called for short AGC method) of the DC component reading correspondence according to current AGC (receiver gain) value of invention in CN101304256, wherein

(1) method of fractionation: if being that the data of a time slot are divided into stem portion, every a part of IQ (inphase/orthogonal component) two paths of signals is averaged respectively, successively as the DC component of this segment data, IQ two paths of signals is individually subtracted DC component to eliminate direct current.Do so, likely because after data sectional, the average of every segment data differs, causes signal power skewness weighing apparatus in whole time slot, causes combined detection performance to decline.

(2) averaging method: being the most conventional simple direct current elimination method, IQ two paths of signals as the DC component of this time slot, is individually subtracted respective DC component to eliminate direct current by the average of whole time slot IQ two paths of signals by the method.As descending spreading factor SF=16, and the code channel number of configuration more when, receiving data is the superposition of multi-code track data, because sending data sample have certain randomness, use averaging method not it is likely that eliminate unclean remnant dc, the most likely can introduce more DC component.But in TD-SCDMA system, averaging method is the most excellent relative to method of fractionation performance.

(3) AGC method: need to first have to obtain receiver size of DC component under different AGC, the DC component that each reception signal reads correspondence according to AGC value eliminates.This way, needing after starting up of terminal takes long to sets up the corresponding relation of AGC and direct current, the direct current size that DC component is only limitted in sample, it is impossible to travel through various situation, and the result of joint-detection is very limited in different RF Components receptivities.

Above-mentioned DC detection method, it is impossible to the direct current realizing the digital signal after changing A/D accurately eliminates.

Summary of the invention

For solving the numeral after analog digital conversion cannot accurately be carried out direct current elimination problem by prior art, embodiments provide the method and apparatus that a kind of direct current eliminates.

A kind of direct current elimination method, including:

Reception data separate averaging method after analog digital conversion is carried out direct current elimination；

The rear 128 chip data of described reception data Midamble code part are done original channel estimate, calculate the coefficient of each tap after channel is estimated；

According to original estimation result, choose power and do remnant dc Processing for removing less than the tap setting thresholding.

A kind of direct current cancellation element, including:

Whole time slot average direct current cancellation element, for carrying out direct current elimination to the reception data separate averaging method after analog digital conversion；

Channel estimating apparatus, for described reception data are carried out original channel algorithm for estimating, calculates the coefficient of each tap after channel is estimated；

Remnant dc estimation unit, for realizing the best estimate to remnant dc；

Remnant dc cancellation element, eliminates the residual DC estimated in described remnant dc estimation unit for realization from described reception data.

Direct current elimination method that the embodiment of the present invention is provided and device, digital information after analog digital conversion is first carried out averaging method and carries out direct current elimination, then according still further to remnant dc method of elimination carry out second time direct current eliminate, so enable to digital signal direct current eliminate more thorough.

Accompanying drawing explanation

Fig. 1 is TD-SCDMA system regular time slots structural representation；

Fig. 2 is Midamble code structural representation；

Fig. 3 is the reception data simulation figure containing remnant dc component；

Fig. 4 is averaging method channel impulse response；

Fig. 5 is that averaging method demodulates planisphere；

Fig. 6 is channel impulse response analogous diagram in the embodiment of the present invention；

Fig. 7 is demodulation planisphere in the embodiment of the present invention；

Fig. 8 is that reception data SNR containing remnant dc component measure distribution schematic diagram

Fig. 9 is the analogous diagram of embodiment of the present invention block error rate BLER；

Figure 10 is throughput schematic diagram in the embodiment of the present invention；

Figure 11 is embodiment of the present invention direct current cancellation element schematic diagram.

Detailed description of the invention

Below in conjunction with the accompanying drawings, one direct current elimination method of the present invention and device are described in detail.

Embodiments provide a kind of direct current elimination method and device, including:

S1, the reception data separate averaging method after analog digital conversion is carried out direct current elimination；

S2, the rear 128chip data receiving data Midamble code part are done original channel estimate, calculate the coefficient of each tap after channel is estimated；

S3, according to original estimation result, choose power and do remnant dc Processing for removing less than the tap setting thresholding.

In embodiments of the present invention, the thresholding of this setting can do different process according to concrete application scenarios, if several reference mobile channels set in third generation partner program 3GPP agreement are (such as PA3, PB3, VA30, VA120) under the conditions of, according to the minimum BLER value required for different business performance, emulation obtains a high threshold met required for the BLER that business is allowed, the thresholding eliminated as direct current, remnant dc Processing for removing is all made less than the tap of this setting thresholding, based on experience value, the span of this thresholding can be [1/64, 1/512] between.

In one embodiment of the invention, described reception data separate averaging method after analog digital conversion is carried out direct current elimination, including:

In described reception data in each time slot, using the average of whole time slot IQ two paths of signals as the DC component of this time slot, IQ two paths of signals is individually subtracted respective DC component to eliminate direct current.

In another embodiment of the presently claimed invention, described according to original estimation result, choose power and do remnant dc Processing for removing less than the tap setting thresholding, including:

Calculate the minima of the power sum of all taps less than described thresholding；

The minima of the power sum according to described all taps calculates the size of described remnant dc；

Size according to described remnant dc carries out direct current elimination.、

In embodiments of the present invention, by deriving remnant dc component and the relation of original channel estimation, and then calculate the way meeting this part tap power sum minimum, estimate the size of remnant dc, again this Partial DC is eliminated from system, thus the direct current in the digital signal after analog digital conversion is completely eliminated, promote systematic function.

Embodiments provide a kind of direct current elimination method and device, including:

(1) reception data acquisition averaging method is done direct current to eliminate, to eliminate the data before big DC component assumes joint-detectioneFor through QPSK QPSK or the data of quadrature amplitude modulation QAM modulation, i.e. data after A/D changes, its symbol sebolic addressingeN () is with being expressed as

e(n)=e_I(n)+je_Q(n), n=0,1, LN-1, N=864 (1)

Wherein e_I(n) and e_QN () is respectively in-phase component sequence (I road) and the quadrature component sequence (Q road) of data.

If Data after then averaging method direct current eliminatese' symbol sebolic addressinge' (n) is represented by

${\underset{\‾}{e}}^{\′}\left(n\right)=(e_I\left(n\right)-{\stackrel{\‾}{e}}_I)+j(e_Q\left(n\right)-{\stackrel{\‾}{e}}_Q)={e^{\′}}_I\left(n\right)+j{e^{\′}}_Q\left(n\right)---\left(2\right)$

(2) use the rear 128chip of Midamble code to do original channel to estimate；

Because by such environmental effects such as multipaths, the front 16chip of the Midamble code in above-mentioned digital signal after A/D changes is easily subject to the impact of Datasymbols0 data division in Fig. 1, so only taking rear 128chip to carry out channel estimation, if the Midamble code data of the 128chip after step (1) aree′_mIf 128 Fourier's change (DFT) and anti-Fourier change (IDFT) can represent by change square formation DFTX and IDFTX respectively, then original estimation result can be expressed ash _raw

h _raw=IDFTXG ^-1·DFTX·e′_m(3)

WhereinGBeing diagonal matrix, its diagonal entry vector is:

g=DFTXm _P(4)

Whereinm _P=(m ₁,m ₂...,m _P) it is the local BasicMidamble code after modulating, wherein P=128, if remnant dc vector isDC=(DC,DC, KDC)_P, whereinDCFor remnant dc amount, and Midamble code is identical with data division remnant dc size, it is assumed that eliminate remnant dcDCAfter channel estimation results beh′_raw, then have

h′_raw=IDFTXG ^-1·DFTX·(e′_m-DC)

=IDFTXG ^-1·DFTX·e′_m-IDFTX·G ^-1·DFTX·DC(5)

=h _raw-DC·IDFTX·G ^-1·P·δ

Wherein $\mathrm{\δ}={\left[\begin{array}{c}1\\ 0\\ M\\ 0\\ 0\end{array}\right]}_{P\×1},$ If e_mid(n)=m _P=(m ₁,m ₂, K, m_P),

Wherein e_mid(0)=m ₁, e_mid(1)=m ₂, L, emid(P-1)=m _P

By $\underset{\‾}{g}=\mathrm{DFTX}\·{\underset{\‾}{m}}_P\⇔g\left(k\right)=\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)W_P^{\mathrm{kn}},$ Wherein $W_P^{\mathrm{nk}}=\exp (-j\frac{2\mathrm{\π}}{P}\mathrm{nk})$

$\underset{\‾}{G}=\mathrm{diag}(\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right),\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)W_P^n,L\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)W_p^{n(k-1)},\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)W_P^{\mathrm{nk}})$

${\underset{\‾}{G}}^{-1}\·\mathrm{\δ}=\mathrm{diag}(\frac{1}{\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)},\frac{1}{\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)W_P^n},L,\frac{1}{\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)W_P^{n(k-1)}},\frac{1}{\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)W_P^{\mathrm{nk}}})\·{\left[\begin{array}{c}1\\ 0\\ M\\ 0\\ 0\end{array}\right]}_{P\×1}$

$=\frac{1}{\underset{n=0}{\overset{P-1}{\mathrm{\Σ}}}{e}_{\mathrm{mid}}\left(n\right)}\·\mathrm{\δ}={\underset{\‾}{E}}_{\mathrm{mid}}\·\mathrm{\δ}---\left(6\right)$

Wherein diag (X) represent by vectorxThe diagonal matrix of composition,E _midInverse for local Midamble code P point plural number sum.Can release time thus

h′_raw=h _raw-DC·E _midIDFTX (P δ)=h _raw-E _mid·DC(7)

IfH′_raw(n)=h′_raw,H _raw(n)=h _raw, n=0,1, L, P-1, then, after eliminating remnant dc, the channel estimation sequence result relation of channel estimation results and the remnant dc that do not disappears is as follows

H′_raw(n)=H _raw(n)-DC·E _mid(8)

(3) choose the intercepting thresholding of original estimation result, the tap less than threshold value is carried out computing, it is therefore an objective to make the power of these taps power after remnant dc eliminates and minimum.

First obtain original channel to estimateH _rawThe maximum tap power H of (n)_{raw_max}, set η_ratioThe threshold coefficient intercepted for tap, η_ratioBetween 0 to 1, typically can be given by emulation.If the tap exceeding thresholding has w, the sequence of tap sequence number is represented by n_r=[n₁, n₂, Ln_w], n₁, n₂, Ln_w∈ 0,1, KP-1 then has |H _raw(n_r)|²＞ η_ratio·H_{raw_max}.Without loss of generality, Wo Menshe

$\left\{\begin{array}{c}{\underset{\‾}{H}}_{\mathrm{raw}}\left(n\right)=H_{\mathrm{raw}\_I}\left(n\right)+j{H}_{\mathrm{raw}\_Q}\left(n\right)\\ {\underset{\‾}{E}}_{\mathrm{mid}}=E_{\mathrm{mid}\_I}+j{E}_{\mathrm{mid}\_Q}\\ \underset{\‾}{\mathrm{DC}}={\mathrm{DC}}_I+{\mathrm{jDC}}_Q\end{array}\right.---\left(9\right)$

Wherein H_{raw_I}(n) and H_{raw_Q}N () is respectively IQ (inphase/orthogonal component) two paths of data that original channel is estimated, E_{mid_I}And E_{mid_Q}For the IQ two paths of data reciprocal of local Midamble code P point plural number sum, DC_IAnd DC_QIt is respectively the remnant dc component of IQ two-way, and separate.Then intercept the power of the tap of below thresholding and be $\mathrm{Power}=\underset{n=0,n\≠n_r}{\overset{P-1}{\mathrm{\Σ}}}{\left|{{\underset{\‾}{H}}^{\′}}_{\mathrm{raw}}\left(n\right)\right|}^2,$ That is:

$\mathrm{Power}=\underset{n=0,n\≠n_r}{\overset{P-1}{\mathrm{\Σ}}}{\left((H_{\mathrm{raw}\_I}\left(n\right)-{\mathrm{DC}}_I{E}_{\mathrm{mid}\_I}+{\mathrm{DC}}_Q{E}_{\mathrm{mid}\_Q})\right)}^2+(H_{\mathrm{raw}\_Q}\left(n\right)-{\mathrm{DC}}_I{E}_{\mathrm{mid}\_Q}-{\mathrm{DC}}_Q{E}_{\mathrm{mid}\_I}{\left)\right)}^2)$

$={\mathrm{coef}}_1{\mathrm{DC}}_I^2+{\mathrm{cofe}}_2{\mathrm{DC}}_I+{\mathrm{coef}}_3{\mathrm{DC}}_Q^2+{\mathrm{coef}}_4{\mathrm{DC}}_Q+{\mathrm{coef}}_5---\left(10\right)$

Wherein $\left\{\begin{array}{c}{\mathrm{coef}}_1={\mathrm{coef}}_3=(N-w)(E_{\mathrm{mid}\_I}^2+E_{\mathrm{mid}\_Q}^2)\\ {\mathrm{cofe}}_2=-2\underset{n=0,n\≠n_r}{\overset{P-1}{\mathrm{\Σ}}}(H_{\mathrm{raw}\_I}\left(n\right)E_{\mathrm{mid}\_I}+H_{\mathrm{raw}\_Q}\left(n\right)E_{\mathrm{mid}\_Q})\\ {\mathrm{cofe}}_4=-2\underset{n=0,n\≠n_r}{\overset{P-1}{\mathrm{\Σ}}}(H_{\mathrm{raw}\_Q}\left(n\right)E_{\mathrm{mid}\_I}-H_{\mathrm{raw}\_I}\left(n\right)E_{\mathrm{mid}\_Q})\\ {\mathrm{coef}}_5=\underset{n=0,n\≠n_r}{\overset{P-1}{\mathrm{\Σ}}}(H_{\mathrm{raw}\_I}^2\left(n\right)+H_{\mathrm{raw}\_Q}^2\left(n\right))\end{array}\right.$

Because coef₁=coef₃＞ 0, and DC_IAnd DC_QIndependent mutually, therefore the condition that (10) formula obtains minima isAndI.e. (10) formula is respectively for variables D C_IAnd DC_QFirst-order partial derivative be 0, thus can try to achieve

$\left\{\begin{array}{c}{\mathrm{DC}}_I=\frac{\underset{n=0,n\≠n_r}{\overset{N-1}{\mathrm{\Σ}}}(H_{\mathrm{raw}\_I}\left(n\right)E_{\mathrm{mid}\_I}+H_{\mathrm{raw}\_Q}\left(n\right)E_{\mathrm{mid}\_Q})}{(N-w)(E_{\mathrm{mid}\_I}^2+E_{\mathrm{mid}\_Q}^2)}\\ {\mathrm{DC}}_Q=\frac{\underset{n=0,n\≠r}{\overset{N-1}{\mathrm{\Σ}}}(H_{\mathrm{raw}\_Q}\left(n\right)E_{\mathrm{mid}\_I}-H_{\mathrm{raw}\_I}\left(n\right)E_{\mathrm{mid}\_Q})}{(N-w)(E_{\mathrm{mid}\_I}^2+E_{\mathrm{mid}\_Q}^2)}\end{array}\right.---\left(11\right)$

When this i.e. uses the present invention, the IQ two paths of signals remnant dc estimated disappears component.Data result after the direct current elimination that whole time slot is final ise"=e′-DC _N, whereinDC _N=(DC,DC, KDC)_N, (9) formula and (11) formula jointly describe.

Experiments prove that, the program has effectively eliminated the indelible DC component of averaging method, and for being configured to the data of the descending SF=16 time slot format of 16 code channels altogether, SNR (signal to noise ratio) measurement result in joint-detection has and substantially carries.For high-speed downlink packet access HSDPA business based on CQI (channel quality instruction) mechanism, network is all indicated by CQI for the resource distribution of terminal and feeds back, under equal physics resource distribution, SNR measured value is the biggest, CQI feedback value is the biggest, network is the fullest for the scheduling of resource of terminal, and under constant decoding stability, the performance of throughput is the highest.

Meanwhile, the signal constellation (in digital modulation) figure that the lifting of SNR more can make joint-detection go out more is concentrated, and for the most getable lifting of decoding reliability after joint-detection, can improve system throughput further.

Test simulation result:

The analogous diagram of the reception data containing remnant dc component is not as it is shown on figure 3, it can be seen that the symbol of this reception data trailer is found out, also remnant dc eliminates totally.

Only with the joint detection results receiving signal after average direct current such as Fig. 4 figure, shown in 5, wherein Fig. 4 is the averaging method shock response to channel, Fig. 5 is that averaging method demodulates planisphere, wherein see from channel estimation results, tap power outside the main footpath that direct current causes is relatively big, and demodulation planisphere compares and dissipates.

After using the direct current elimination method that the embodiment of the present invention provides, to the joint detection results of same group of sample data as shown in Figure 6, Figure 7, in terms of channel estimation results, the power of the little tap that direct current causes substantially reduces, demodulation planisphere is concentrated very much, in terms of SNR measurement result, (see table 1), improve 11dB.

Table 1

	SNR(dB)
		Averaging method	17.1dB
The present invention	28.18dB

SNR measurement result (seeing table 2) based on the more serious sample data of DC influence, sees accompanying drawing 8, and figure below is for being verified by 1000 groups of sample datas that DC influence is more serious in actual test environment.

Table 2

	Average SNR (dB)
		Averaging method	21.2dB
The present invention	28.5dB

In terms of the result, having clear improvement the measurement result of same group of sample SNR, the distribution of SNR is more concentrated, and average SNR has the performance gain of about 7dB.

HSDPAFRC simulation, wherein simulated conditions is as shown in table 3:

Table 3

Being not difficult to find out, simulated environment is the highest configuration of HSDPA under 1.28M bandwidth, reaches as high as the Theoretical Rate of 2.8Mbps under this kind of configuration, but the encoding rate under this kind of configuration is up to 99%, coding gain is the least, wants to reach reasonable performance, and the requirement to receiving data is at a relatively high.In this case, the present invention can bring the gain on BLER (error block ratio) and throughput equally, and BLER can bring the gain of 1.5dB, and throughput is the highest improves more than 10 times.The visible present invention, for the penalty caused because of direct current reason, has reasonable inhibitory action, can reduce the performance difference brought because of the difference of different radio frequency receiver simultaneously.

The embodiment of the present invention additionally provides a kind of direct current cancellation element, including:

Whole time slot average direct current cancellation element 1101, for carrying out direct current elimination to the reception data separate averaging method after analog digital conversion；

Channel estimating apparatus 1102, for described reception data are carried out original channel algorithm for estimating, calculates the coefficient of each tap after channel is estimated；

Remnant dc estimation unit 1103, for realizing the best estimate to remnant dc；

Remnant dc cancellation element 1104, eliminates the residual DC estimated in described remnant dc estimation unit 1103 for realization from described reception data.

In another embodiment of the presently claimed invention, described whole time slot average direct current cancellation element 1101, specifically in each time slot in described reception data, using the average of whole time slot IQ two paths of signals as the DC component of this time slot, IQ two paths of signals is individually subtracted respective DC component to eliminate direct current.

In another embodiment of the presently claimed invention, described remnant dc estimation unit 1103, specifically for calculating the minima of the power sum of all taps less than described thresholding；The minima of the power sum according to described all taps calculates the size of described remnant dc.

The direct current cancellation element that the embodiment of the present invention is provided, is part that is completely corresponding with the direct current elimination method in previous embodiment, that omit in the present embodiment, is referred to despising of embodiment of the method part.It is above the embodiment of the present invention one preferably embodiment, anyone is on the premise of skilled, the spirit without departing substantially from the present invention and without departing from the technical scope that the present invention relates on the premise of, can to the present invention describe details make various supplement and amendment.Protection scope of the present invention is not limited to the scope cited by embodiment, and protection scope of the present invention is as the criterion with claim.

Claims (5)

1. a direct current elimination method, for TD-SCDMA standard, it is characterised in that including:

Reception data separate averaging method after analog digital conversion is carried out direct current elimination；

The rear 128 chip data of described reception data training sequence Midamble code part are done original channel estimate, calculate the coefficient of each tap after channel is estimated；

According to original estimation result, choose power and do remnant dc Processing for removing less than the tap setting thresholding；

Wherein, described according to original estimation result, choose power and do remnant dc Processing for removing less than the tap setting thresholding, including:

Calculate the minima of the power sum of all taps less than described thresholding；

The minima of the power sum according to described all taps calculates the size of described remnant dc；

Size according to described remnant dc carries out direct current elimination.

Direct current elimination method the most according to claim 1, it is characterised in that described reception data separate averaging method after analog digital conversion is carried out direct current elimination, including:

In described reception data in each time slot, using the average of whole time slot inphase/orthogonal component IQ two paths of signals as the DC component of this time slot, IQ two paths of signals is individually subtracted respective DC component to eliminate direct current.

Direct current elimination method the most according to claim 1, it is characterised in that the minima of the described power sum according to described all taps calculates the size of described remnant dc, including:

$\left\{\begin{array}{c}{\mathrm{DC}}_I=\frac{\underset{n=0,n\≠n_r}{\overset{N-1}{\Σ}}\left(H_{raw\_I}\left(n\right)E_{mid\_I}+H_{raw\_Q}\left(n\right)E_{mid\_Q}\right)}{\left(N-w\right)\left(E_{mid\_I}^2+E_{mid\_Q}^2\right)}\\ {\mathrm{DC}}_Q=\frac{\underset{n=0,n\≠n_r}{\overset{N-1}{\Σ}}\left(H_{raw\_Q}\left(n\right)E_{mid\_I}-H_{raw\_I}\left(n\right)E_{mid\_Q}\right)}{\left(N-w\right)\left(E_{mid\_I}^2+E_{mid\_Q}^2\right)}\end{array}\right.$

Wherein H_{raw_I}(n) and H_{raw_Q}N () is respectively the IQ two paths of data that original channel is estimated, E_{mid_I}And E_{mid_Q}For the IQ two paths of data reciprocal of local Midamble code P point plural number sum, DC_IAnd DC_QIt is respectively the remnant dc component of IQ two-way.

4. a direct current cancellation element, for TD-SCDMA standard, it is characterised in that including:

Whole time slot average direct current cancellation element, for carrying out direct current elimination to the reception data separate averaging method after analog digital conversion；

Channel estimating apparatus, for described reception data are carried out original channel algorithm for estimating, calculates the coefficient of each tap after channel is estimated；

Remnant dc estimation unit, for realizing the best estimate to remnant dc；

Remnant dc cancellation element, eliminates the residual DC estimated in described remnant dc estimation unit for realization from described reception data；

Wherein, described remnant dc estimation unit, specifically for calculating the minima of the power sum less than all taps setting thresholding；The minima of the power sum according to described all taps calculates the size of described remnant dc.

Device the most according to claim 4, it is characterised in that

Described whole time slot average direct current cancellation element, specifically in each time slot in described reception data, using the average of whole time slot inphase/orthogonal component IQ two paths of signals as the DC component of this time slot, IQ two paths of signals is individually subtracted respective DC component to eliminate direct current.