CN1913380A - Method of channel after processing in time division duplex system - Google Patents

Abstract

This invention relates to a channel post processing method in a TD duplex system including: processing data at the emitting end, expanding, modulating, scrambling and rising cosine filtering the data then to be emitted via an empty, the receiving end separates training sequence receiving signals from received data to finish the channel estimation and post process of users, the received training sequences can be expressed as e=Gh+n, in which, matrix G is a circulating matrix moving to the right composed of a basic Midamble code, h is the channel impulse response of users and n is the Gaussian white noise, which utilizes time delay information to enhance channel post process and utilizes the correlated character of channel between time slots to smooth time slot so as to overcome the influence of noises to channel estimation.

Description

The method of channel post-processing in a kind of tdd systems

Technical field

The present invention relates to a kind of code division multiple access system of communication field, particularly the channel post-processing method of TD-SCDMA mobile communication system is in particular a kind of based on utilizing channel relevancy to improve the method for channel post-processing in the TDD system.

The inventive method can also be used for moving communicating field and be applied to all TDD-CDMA systems and SCDMA system.

Background technology

In the prior art TD-CDMA particularly the TD-SCDMA system be based on the 3-G (Generation Three mobile communication system) of time division duplex (TDD), wherein relate to a lot of key technologies, as Fast Channel estimation technique, associated detection technique, intelligent antenna technology etc.In the TD-SCDMA system, use associated detection technique, can effectively suppress MAI and ISI, associated detection technique is to be based upon on the basis of channel estimating, the effect of channel estimating has very significant effects to combined detection performance, and good channel estimation technique becomes the necessary condition of decision joint-detection.

In the TD-CDMA system, the Fast Channel estimation technique has important effect, it is not only the precondition that some key technologies realize, but also is the basis that various Measurement Algorithm are achieved, and the whether accurate of it will directly have influence on the performance of this series of algorithms.The multipath transmisstion of signal has proposed stern challenge to channel estimating under the wireless mobile environment, how to obtain channel estimating accurately and become the key in the key technology of TD-SCDMA under abominable dissemination channel condition.Therefore, reducing channel estimation error, promptly weaken or eliminate noise in the channel response, is the problem that presses for solution.

Existing TD-CDMA system, particularly the channel estimation technique of TD-SCDMA generally all is a utilization Steiner method of estimation, see also B.Steiner and P.W.Baier.Low cost channelestimation in the uplink receiver of CDMA mobile radio systems.FRE-QUENZ, vol.47,1993,292-298, the Steiner estimator is a kind of low-cost channel estimation methods, it passes through the reasonably training sequence of designing user, the linear convolution of complexity is converted into simple circular convolution, utilize the cyclic shift characteristic of matrix, can realize by FFT and IFFT computing.

Yet the performance of Steiner estimator is subjected to the influence of receiving terminal additive noise, and its channel response that obtains responds with ideal communication channel and compares, and comprises noise component(s), and this is channel estimation errors just; Can enlarge noise power simultaneously, cause the relative input of output signal to noise ratio that a certain amount of loss is arranged.At this problem, in the prior art open source literature " Kang Shaoli, Qiu Zhengding; Li Shihe, the improvement of low-cost channel estimation methods in the TD-SCDMA system, communication journal; Vol.23 No.10; 2002,108-113 ", a kind of noise-reduction method of crossing Threshold detection has been proposed, this method has very little amount of calculation, and can make system obtain certain performance improvement, but need dynamic noise power and the noise gate of estimating in the current time slots, increase certain overhead.

It needs to be noted simultaneously: in the prior art, even accurately estimate noise power, but owing to noise gate is fixed, when thresholding is too high, though can suppress the noise tap better, also suppressed available small-signal tap simultaneously, cause combined detection performance to descend; When thresholding is too low,, equally also can cause combined detection performance to descend though stick signal tap better simultaneously also causes many noise taps also to be retained.Though the thought of adaptive threshold can head it off, does not have realizability at present.Show by simulation study that simultaneously " desirable time delay " (time delay as can be known)+reprocessing is a kind of a kind of method of approaching " ideal communication channel " (noiseless) most.

Summary of the invention

The method that the purpose of this invention is to provide channel post-processing in a kind of tdd systems, at above-mentioned prior art problem, overcome the noise problem in the channel impulse response of channel estimating, provided the correlation of utilizing channel energy, strengthen the channel post-processing method with its method of approaching " desirable time delay " and reprocessing, reduce channel estimation error; The present invention is a kind of method of utilizing time delay information to strengthen channel post-processing, this method is utilized the relevant characteristic of channel power between time slot, it is level and smooth to carry out time slot, has overcome the influence of noise to channel estimating, has improved the estimated accuracy of each user's channel impulse response.

Technical scheme of the present invention comprises:

The method of channel post-processing in a kind of tdd systems, it comprises:

A, data are done transmitting terminal handle, spread spectrum, scrambling, modulation and raised cosine filtering are launched by eating dishes without rice or wine;

B, receiving terminal are isolated the training sequence received signal from receive data, finish channel estimating and channel post-processing to each user; The described training sequence that receives can be expressed as

e＝Gh+n (1)

Wherein, the G matrix is the ring shift right matrix that is made of a basic Midamble sign indicating number, and h is user's a channel impulse response, and n is a white Gaussian noise.

Described method wherein, also comprises the following steps:

A1: the training sequence that will receive data proposes to do channel estimating;

The maximal possibility estimation of channel impulse response $\hat{h}=G^{-1}e;$

B1: carry out channel post-processing;

C1: obtain each user's channel impulse response, from

In be that length is taken out K value and constituted each subscriber channel and estimate that promptly each user's channel response window length is W with W;

D1: the joint-detection data demodulates of carrying out the multi-user.

Described method, wherein, described step B1 also comprises, the power iteration mean value of estimating each tap with multi-slot channel is done channel post-processing with noise gate to the power average value of this time slot as the value of carrying out channel post-processing, the steps include:

B11, calculating $\stackrel{\‾}{{\left|\right|h_{n\_\mathrm{slot}}\left|\right|}^2}=(1-p)\·\stackrel{\‾}{{\left|\right|h_{n-1\_\mathrm{slot}}\left|\right|}^2}+p\·{\left|\right|{\hat{h}}_{n\_\mathrm{slot}}\left|\right|}^2,$

The channel impulse response estimation value of representing this time slot,

The power p that is each tap of channel impulse response of n-1 time slot is a forgetting factor, the scope 0～1 of getting;

B12, a noise gate T are to what estimate

Carry out channel post-processing, zero greater than the reservation of decision threshold less than being changed to of decision threshold, the position of tracer signal tap, the i.e. position of nonzero element;

B13, general

Corresponding tap keeps, and channel post-processing is finished in other tap zero setting, obtains the channel impulse response through reprocessing of this time slot

B14, general

Pass to next time slot.

Described method, wherein, described step D1 also comprises:

D11, with the user separately the spread spectrum scrambler and through the channel impulse response of channel post-processing

Generator matrix A;

D12, to the user data demodulation, $\hat{d}={(A^H\·A)}^{-1}{A}^{H}e.$

Described method, wherein, the processing of described channel estimating also comprises: use a basic Midamble sign indicating number, be total to NK voice user, K is the number of users of a scrambler of use, the scrambler number that N uses simultaneously for this sub-district, the long W of each user's channel estimation window is reduced to

The maximal possibility estimation of channel impulse response h is

Promptly

$\hat{h}={\left(G^{*T}G\right)}^{-1}{G}^{*T}e---\left(2\right)$

Because the G matrix is the ring shift right matrix, has

$\hat{h}=G^{-1}e---\left(3\right)$

；

According to:

$\hat{h}=\mathrm{ifft}(\mathrm{fft}\left(e\right)/\mathrm{fft}\left(m\right))---\left(4\right)$

To channel impulse response

The power of each tap judge, if more than or equal to the threshold value that is provided with, then keep the position of this tap; Otherwise, remove the response of this tap, i.e. the channel post-processing process.

Described method, wherein, described method also comprises:

According to channel impulse response

Estimation, utilize correlation of each footpath energy, the power of multi-slot channel impulse response carries out channel post-processing, differentiating h is user's signal footpath, still the false footpath that is caused by noise jamming.

The method of channel post-processing in a kind of tdd systems provided by the present invention, utilize time delay information to strengthen channel post-processing, utilized the relevant characteristic of channel between time slot, it is level and smooth to carry out time slot, overcome the influence of noise, improved the estimated accuracy of each user's channel impulse response channel estimating; And be a kind of simple, easy realization, the higher TD-SCDMA system channel estimation method of efficient, have better generality simultaneously, and finally improve the receptivity of uplink receiver.Adopt the method for utilizing time delay information to strengthen channel post-processing of the present invention, compared with prior art, improved the estimated accuracy of each user's channel impulse response, effectively reduce the error rate, improved systematic function, increased power system capacity.

Description of drawings

Fig. 1 is the schematic diagram of the pairing system signal receiving demodulation of the inventive method;

Fig. 2 is that the time delay information of utilizing of the inventive method strengthens the channel post-processing simple block diagram.

Embodiment

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

The method of channel post-processing in the tdd systems of the present invention is done transmitting terminal to data and is handled, and (being equivalent to wireless channel) launched in spread spectrum, scrambling (grouping scrambling), modulation and raised cosine filtering by eating dishes without rice or wine.From receive data, isolate the training sequence received signal then, finish channel estimating and channel post-processing each user.The training sequence that receives can be expressed as

e＝Gh+n (1)

Wherein, the G matrix is the ring shift right matrix that is made of a basic Midamble sign indicating number, and h is user's a channel impulse response, and n is a white Gaussian noise.

Suppose to adopt simultaneously in the sub-district N scrambler, then number of users increases to N doubly.For this reason, the present invention adopts and carries out channel estimating in the following method: use a basic Midamble sign indicating number, be total to NK voice user, K is for using the number of users of a scrambler, and the long W of each user's channel estimation window is reduced to

Finish the maximal possibility estimation of channel impulse response h with formula (2) Promptly

$\hat{h}={\left(G^{*T}G\right)}^{-1}{G}^{*T}e---\left(2\right)$

Because the G matrix is the ring shift right matrix, has

$\hat{h}=G^{-1}e---\left(3\right)$

Therefore, utilize the cyclic shift characteristic of matrix, above-mentioned channel estimating can adopt the FFT/IFFT fast method to realize, thereby has improved arithmetic speed significantly.

As shown in the formula (4):

$\hat{h}=\mathrm{ifft}(\mathrm{fft}\left(e\right)/\mathrm{fft}\left(m\right))---\left(4\right)$

To channel impulse response

The power of each tap judge, if more than or equal to the threshold value that is provided with, then keep the position of this tap; Otherwise, removing the response of this tap, the channel impulse response that can claim this method to obtain is a post process channel response.

The sign indicating number design of TD-SCDMA system, frame structure, spreading rate etc. provide feasibility for interference mitigation technology.Because the channel variation between a last time slot and the next time slot is very unconspicuous, particularly each the footpath energy information, the channel estimating information (mainly being energy information) of front and back time slot can be utilized, so we utilize the time delay correlation to improve channel estimating.

Suppose Be the estimation of channel impulse response, the present invention further utilizes the correlation of each footpath energy, and the power of the channel impulse response that multi-slot (energy) is level and smooth carries out channel post-processing, and differentiating h is user's signal footpath, still the false footpath that is caused by noise jamming.The power iteration mean value of estimating each tap with multi-slot channel is done channel post-processing with noise gate to the power average value of this time slot as the value of carrying out channel post-processing.

The detailed process of this method method is as follows:

$\stackrel{\‾}{{\left|\right|h_{n\_\mathrm{slot}}\left|\right|}^2}=(1-p)\·\stackrel{\‾}{{\left|\right|h_{n-1\_\mathrm{slot}}\left|\right|}^2}+p\·{\left|\right|{\hat{h}}_{n\_\mathrm{slot}}\left|\right|}^2---\left(5\right)$

In the formula,

The channel impulse response estimation value of representing this time slot,

Be the power of each tap of channel impulse response of n-1 time slot.P is a forgetting factor, span 0～1.

Use the mean value in each footpath of time slot

With current time slots Calculate this time slot

Then with the power average value of noise gate to this time slot Doing channel post-processing, will in the position of tracer signal tap again

Corresponding tap keeps, and channel post-processing is finished in other tap zero setting, obtains the channel impulse response through reprocessing of this time slot

Simultaneously will Pass to next time slot.

After user data passes through spectrum-spreading and scrambling, wireless channel, arrive receiver, received signal e can be expressed as:

e＝Ad+n (6)

D is the column vector that all users' symbol is formed, the user separately the spread spectrum scrambler and the channel impulse response of reprocessing Generation system transmission matrix A,

$A=\left[\begin{array}{cccc}{A}^{k_1}& {\mathrm{<figure-callout\; id="2"\; label="A\; k"\; filenames="A20051009008300151.png"\; state="\{\{state\}\}">A</figure-callout>}}^{k_{}}& \&CenterDot;\&CenterDot;\&CenterDot;& A^{\mathrm{ka}}\end{array}\right],$ Ka is the antenna number

(7)

To the user data demodulation,

$\hat{d}={(A^H\&CenterDot;A)}^{-1}{A}^{H}e---\left(8\right)$

The concrete steps of the inventive method comprise:

The first step: the training sequence that will receive data proposes to do channel estimating, and the maximal possibility estimation of channel impulse response is $\hat{h}=G^{-1}e.$

Second step: carry out channel post-processing;

This step can comprise following substep again:

1 calculates $\stackrel{\&OverBar;}{{\left|\right|h_{n\_\mathrm{slot}}\left|\right|}^2}=(1-p)\&CenterDot;\stackrel{\&OverBar;}{{\left|\right|h_{n-1\_\mathrm{slot}}\left|\right|}^2}+p\&CenterDot;{\left|\right|{\hat{h}}_{n\_\mathrm{slot}}\left|\right|}^2,$ P is a forgetting factor, the scope 0～1 of getting;

2 one noise gate T are to what estimate Carry out channel post-processing, zero greater than the reservation of decision threshold less than being changed to of decision threshold, the position of tracer signal tap, the i.e. position of nonzero element;

3 will again Corresponding tap keeps, and channel post-processing is finished in other tap zero setting, obtains the channel impulse response through reprocessing of this time slot

4 will Pass to next time slot.

The 3rd step: the channel impulse response that obtains each user;

From In be that length is taken out K value and just constituted each subscriber channel and estimate that promptly each user's channel response window is long is W with W.

The 4th step: carry out multi-user's joint-detection data demodulates, that is:

The 1 usefulness user spread spectrum scrambler separately and the channel impulse response of process channel post-processing Generator matrix A;

2 pairs of user data demodulation, $\hat{d}={(A^H\&CenterDot;A)}^{-1}{A}^{H}e.$

Be described in further detail below in conjunction with the enforcement of accompanying drawing technical scheme:

At first, the inventive method must be done transmitting terminal to data and handle, and spread spectrum, scrambling, modulation and raised cosine filtering are launched by antenna.

The training sequence that receives data is proposed to do channel estimating, and data portion is carried out multi-user's joint-detection demodulation, finishes data separating.

Then, from receive data, isolate the training sequence received signal, finish channel estimating and channel post-processing each user.The training sequence that receives can be expressed as

e＝Gh+n (1)

Wherein, the G matrix is the ring shift right matrix that is made of a basic Midamble sign indicating number, and h is user's a channel impulse response, and n is a white Gaussian noise.

Finish the maximal possibility estimation of channel impulse response h with formula (1) Promptly

$\hat{h}={\left(G^{*T}G\right)}^{-1}{G}^{*T}e---\left(2\right)$

Because the G matrix is the ring shift right matrix, has

$\hat{h}=G^{-1}e---\left(3\right)$

Therefore above-mentioned channel estimating can adopt the FFT/IFFT fast method to realize, thereby has improved arithmetic speed significantly, as shown in the formula:

$\hat{h}=\mathrm{ifft}(\mathrm{fft}\left(e\right)/\mathrm{fft}\left(m\right))---\left(4\right)$

The product of G matrix and H matrix is realized with first row of G matrix and the convolution of H matrix in the formula, after doing the first row FFT of G matrix and H matrix and being FFT and multiplying each other, asks IFFT again, and the convolution of the complexity of time domain is converted into efficiently frequency domain FFT/IFFT realization fast.

Be the estimation of channel impulse response, the present invention further utilizes the time delay correlation, and the channel impulse response power that multi-slot is level and smooth carries out channel post-processing, and differentiating h is user's signal footpath, still the false footpath that is caused by noise jamming.The power iteration mean value of estimating each tap with multi-slot channel is done channel post-processing with noise gate to the power average value of this time slot as the value of carrying out channel post-processing.

$\stackrel{\&OverBar;}{{\left|\right|h_{n\_\mathrm{slot}}\left|\right|}^2}=(1-p)\&CenterDot;\stackrel{\&OverBar;}{{\left|\right|h_{n-1\_\mathrm{slot}}\left|\right|}^2}+p\&CenterDot;{\left|\right|{\hat{h}}_{n\_\mathrm{slot}}\left|\right|}^2---\left(5\right)$

In the formula,

The channel impulse response estimation value of representing this time slot, Be the power of each tap of channel impulse response of n-1 time slot.P is a forgetting factor, span 0～1.

Use the mean value of time slot With current time slots Calculate this time slot

Then with the power average value of noise gate to this time slot Do channel post-processing, to channel impulse response The power of each tap judge, if more than or equal to the threshold value that is provided with, then keep the position of this tap; Otherwise, remove the response of this tap.Again will in the position of tracer signal tap

Corresponding tap keeps, and channel post-processing is finished in other tap zero setting, obtains the channel impulse response through reprocessing of this time slot Simultaneously will

Pass to next time slot.

So, user data arrives receiver by spectrum-spreading and scrambling, air traffic channel, and received signal e can be expressed as:

$\left[\begin{array}{c}{e}^1\\ {\mathrm{<figure-callout\; id="2"\; label="e"\; filenames="A20051009008300151.png"\; state="\{\{state\}\}">e</figure-callout>}}^2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ e^{k_a}\end{array}\right]=\left[\begin{array}{c}{A}^1\\ {\mathrm{<figure-callout\; id="2"\; label="A"\; filenames="A20051009008300151.png"\; state="\{\{state\}\}">A</figure-callout>}}^2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ A^{k_a}\end{array}\right]d+\left[\begin{array}{c}{n}^1\\ {\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A20051009008300151.png"\; state="\{\{state\}\}">n</figure-callout>}}^2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ n^{k_a}\end{array}\right]---\left(14\right)$

D is the column vector that all users' symbol is formed, the user separately the spread spectrum scrambler and the channel impulse response of reprocessing

Generation system transmission matrix A.

To the user data demodulation, can use whitened matched filter algorithm (WMF) with matrix A, ZF linear block balance device algorithm (ZF), lms algorithm multiple joint detection algorithms such as (MMSE).Here be that example is carried out data estimation with the lms algorithm, have

$\hat{d}={(A^H{R_n}^{-1}A+{R_d}^{-1})}^{-1}{A}^H{R_n}^{-1}e---\left(15\right)$

Said method technical scheme of the present invention is applicable to that all are applied to all cdma systems, particularly TDD-CDMA system and SCDMA system of institute.Any have a signal processing, and the engineer of knowledge background such as communication can design corresponding channel estimating and joint-detection device according to the present invention, and it all should be included in thought of the present invention and the scope.

The inventive method has overcome the noise problem in the channel impulse response of channel estimating, provided the correlation of utilizing channel energy, strengthen a kind of processing method of channel post-processing, approach " desirable time delay "+post-processing approach with it and reduced channel estimation error; The advantage that has kept institute's extracting method in the above-mentioned prior art document, solved simultaneously the defective that exists in the said method again, be a kind of simple, easy realization, the higher TD-SCDMA system channel estimation method of efficient, have better generality simultaneously, and finally improve the receptivity of uplink receiver.

Should be understood that the above-mentioned description at specific embodiment of the present invention is comparatively concrete, can not therefore think the restriction to scope of patent protection of the present invention, scope of patent protection of the present invention should be as the criterion with claims.

Claims (6)

1, the method for channel post-processing in a kind of tdd systems, it comprises:

A, data are done transmitting terminal handle, spread spectrum, scrambling, modulation and raised cosine filtering are launched by eating dishes without rice or wine;

B, receiving terminal are isolated the training sequence received signal from receive data, finish channel estimating and channel post-processing to each user; The described training sequence that receives can be expressed as

e＝Gh+n (1)

Wherein, the G matrix is the ring shift right matrix that is made of a basic Midamble sign indicating number, and h is user's a channel impulse response, and n is a white Gaussian noise.

2, method according to claim 1 is characterized in that, described method also comprises the following steps:

A1: the training sequence that will receive data proposes to do channel estimating;

The maximal possibility estimation of channel impulse response $\hat{h}=G^{-1}e;$

B1: carry out channel post-processing;

C1: obtain each user's channel impulse response, from

In be that length is taken out K value and constituted each subscriber channel and estimate that promptly each user's channel response window length is W with W;

D1: the joint-detection data demodulates of carrying out the multi-user.

3, method according to claim 2, it is characterized in that described step B1 also comprises, estimates that with multi-slot channel the power iteration mean value of each tap is as the value of carrying out channel post-processing, with noise gate the power average value of this time slot is done channel post-processing, the steps include:

B11, calculating $\stackrel{\&OverBar;}{{\left|\right|h_{n\_\mathrm{slot}}\left|\right|}^2}=(1-p)\&CenterDot;\stackrel{\&OverBar;}{{\left|\right|h_{n-1\_\mathrm{slot}}\left|\right|}^2}+p\&CenterDot;{\left|\right|{\hat{h}}_{n\_\mathrm{slot}}\left|\right|}^2,$ The channel impulse response estimation value of representing this time slot, The power p that is each tap of channel impulse response of n-1 time slot is a forgetting factor, the scope 0-1 that gets;

B12, a noise gate T are to what estimate

Carry out channel post-processing, zero greater than the reservation of decision threshold less than being changed to of decision threshold, the position of tracer signal tap, the i.e. position of nonzero element:

B13, general

Corresponding tap keeps, and channel post-processing is finished in other tap zero setting, obtains the channel impulse response through reprocessing of this time slot

B14, general

Pass to next time slot.

4, method according to claim 2 is characterized in that, described step D1 also comprises:

D11, with the user separately the spread spectrum scrambler and through the channel impulse response of channel post-processing Generator matrix A;

D12, to the user data demodulation, $\hat{d}={(A^H\&CenterDot;A)}^{-1}{A}^{H}e.$

5, method according to claim 1, it is characterized in that, the processing of described channel estimating also comprises: use a basic Midamble sign indicating number, be total to NK voice user, K is for using the number of users of a scrambler, the scrambler number that N uses simultaneously for this sub-district, the long W of each user's channel estimation window is reduced to

The maximal possibility estimation of channel impulse response h is

Promptly

$\hat{h}={\left(G^{*T}G\right)}^{-1}{G}^{*T}e---\left(2\right)$

Because the G matrix is the ring shift right matrix, has

$\hat{h}=G^{-1}e---\left(3\right)$

；

According to:

$\hat{h}=\mathrm{ifft}(\mathrm{fft}\left(e\right)/\mathrm{fft}\left(m\right))---\left(4\right)$

To channel impulse response

The power of each tap judge, if more than or equal to the threshold value that is provided with, then keep the position of this tap; Otherwise, remove the response of this tap, i.e. the channel post-processing process.

6, method according to claim 5 is characterized in that, described method also comprises:

According to channel impulse response

Estimation, utilize correlation of each footpath energy, the power of multi-slot channel impulse response carries out channel post-processing, differentiating h is user's signal footpath, still the false footpath that is caused by noise jamming.

Images