CN1334659A - Bandwidth-variable channel estimation method for frequency-selective channel and its device - Google Patents

Abstract

The invention relates to a method of estimating channel with variable bandwidth in the condition of frequency selectivity channel. The method includes following steps. (1) With the received signal being conjugate multiplied by known pilot frequency code element, the estimated fading value of channel at K point of time can be found out. (2) The estimated fading value of channel is passed a filter suppressing noise. (3) A parameter of high-speed filter can be obtained by looking up a table. (4) With the estimated value of channel as well as received signal together being delayed, then it is sent to Viterbi decoder doing post treatment. The invented method complets channel estimation process accurately at different structure of pilot frequency with lesser operating quantity, also the requirement on communication quality of third generation movable communication system, especially in the situation that moving station is from low speed to high speed, is satisfied.

Description

Variable bandwidth channel method of estimation and device thereof under a kind of frequency-selective channel

The present invention is applied to receiver demodulation part in the code division multiple access system, The present invention be directed to the channel estimation problems of the received signal of the fast-changing fading channel of experience.The present invention is applicable to the consensus standard of DS-CDMA.

Along with professional volatile growth in communication field, also improving constantly for the requirement of aspects such as traffic capacity, flexibility and quality.The technology that is adopted from FDMA to TDMA, develops into present CDMA from analog to digital again.CDMA compares with the current TDMA system that generally adopts, have some special advantages, as frequency utilization efficient height, capacity is big, good confidentiality, the service quality height, cost is low etc., thereby in the third generation mobile system in future, direct sequence CDMA (DS-CDMA) system becomes the selection of the communication needs that can satisfy to increase rapidly just gradually.Different is with time-division or frequency division multiple access, and in DS-CDMA, all users use identical bandwidth simultaneously, and different users discerns by the PN sign indicating number sequence of unique distribution.

The channel of decline is a central issue that occurs in the digital mobile communication fast, in signals transmission, transmission channel meet with become when serious, the influence of frequency selective fading, unless correct estimation and equilibrium are carried out in such channel distortion, can not be recovered reliably otherwise transmit.And advanced, reliable channel estimating and detection technique will be alleviated the suffered restriction of other parts of communication system to a great extent, thereby can be made product to bigger capacity, littler size, and cost direction still less develops.Such as, the base station with outstanding receptivity can allow travelling carriage to use littler transmitting power to reduce the travelling carriage battery loss, perhaps improve the quality of link.

Consensus standard according to existing DS-CDMA, in WCDMA and CDMA2000 up channel, the data structure that adopts is that control channel DPCCH and Traffic Channel DPDCH transmit through mutually orthogonal disturbing code parallel, DPCCH channel wherein is the time-multiplexed pilot form, as shown in Figure 1, comprise frequency pilot sign and control information in the time slot, The data BPSK modulation.And in their down channel structure, then used pilot channel, just so-called continuous pilot form, as shown in Figure 2.

For transmitting of these forms, how correctly recover the raw information of data effectively at receiving terminal, be exactly the problem that demodulation need be considered.Comparatively effective correlation demodulation wherein is relevant with a reference signal by the signal that will receive, with the fading channel in the removal radio transmission for the bias effect that transmits on phase place and amplitude.This reference signal both had been the estimated value of the channel fading of signal experience.Obviously, link capacity is to reach the required snr of received signal of certain performance with receiver to be inversely proportional to, for the signal of experience decline, the signal to noise ratio more required than irrelevant detection owing to coherent detection is littler, thereby becomes the detection scheme that cdma system relatively is fit to.This just need be under quick multipath fading environments to the accurate estimation of the amplitude and the phase place of fading channel.To this, people have proposed a variety of methods, channel for the time-multiplexed pilot form, comprise and utilize the pilot frequency information interpolation to go out data segment channel fading value merely, and the self-adaptive forward that utilizes data segment and pilot to carry out is simultaneously predicted or adopt front and back to estimate channel or the like to the lattice filter of prediction simultaneously.These methods have also obtained reasonable effect respectively under certain environmental condition.But because the movement velocity excursion that the third generation mobile system need adapt to is very big, to per hour 500 kilometers, requirement simultaneously can transmit the data of high traffic from static.Existing channel estimation methods is difficult to satisfy simultaneously under the various conditions requirement for communication quality.

And in cdma system, owing to adopted fast power control (TPC), the statistical property of fading channel will no longer keep, and therefore the channel estimation filters based on the Weiner filter theory can not be applied to go in the reverse link.

The 3-G (Generation Three mobile communication system) that the present invention is directed to employing DS-CDMA technology is under multiple pilot configuration, utilize mobile station speed estimation technique or signal fadeout rate estimation technique, solve fast-changing fading channel environment lower channel estimation problem, the low pass filter bandwidth that the coefficient of adjusting in the channel estimation process to be adopted according to lookup table technology or corresponding empirical equation is variable and the method for coefficient, finish accurate channel estimation process under the different pilot configurations with less operand, thereby satisfy 3-G (Generation Three mobile communication system) for the travelling carriage requirement of communication quality under the mobile status from low speed to high speed, reach the purpose that can both obtain the channel estimating performance that meets the demands in the movement velocity excursion that is suitable in third generation mobile.

The technical scheme that realizes the object of the invention is:

Variable bandwidth channel method of estimation under a kind of frequency-selective channel comprises step:

A, received signal and known pilot frequency code element are carried out conjugate multiplication, obtain K channel fading estimated value constantly;

It is characterized in that, further comprising the steps of:

B, the channel fading estimated value that will obtain in a step are passed through a filter, with the noise in the further inhibition channel guess value;

After estimating through channel fading speed, c, the channel fading estimated value that will obtain simultaneously obtain the parameter of an adjustment filter again by look-up table in a step;

D, will pass through D with received signal through the channel estimation value that b, c step obtain at last _fSend into the Veterbi decoding part after the time-delay, do subsequent treatment.

Variable bandwidth channel method of estimation under the said frequencies selective channel, wherein, described filter is an average filter or the variable low pass filter of a coefficient.

Variable bandwidth channel method of estimation under the said frequencies selective channel, wherein, described average filter coefficient is constant to be 1, order parameter D _fBe the velocity estimation value that provides according to receiver channel rate of fading estimation unit, obtain according to corresponding relation calculating or look-up table again that the movement velocity of travelling carriage is fast more, corresponding D _fValue is more little.

Variable bandwidth channel method of estimation under the said frequencies selective channel, wherein, the order parameter D of the low pass filter that described coefficient is variable _fFixing, weight coefficient { h _i, i=0,1 ... 2D _fBe the velocity estimation value that provides according to receiver channel rate of fading estimation unit, and to calculate according to predetermined corresponding relation again or look-up table obtains, the movement velocity of travelling carriage is fast more, and the variable low pass filter two ends coefficient value of coefficient is more little.

The device that variable bandwidth channel under a kind of frequency-selective channel is estimated comprises one first multiplier, and the input of this multiplier connects a received signal and a pilot frequency code element signal; It is characterized in that, also comprise a filter, a channel fading speed estimation unit, a look-up table unit, a delay circuit, one second multiplier and a viterbi decoder; The output signal of described first multiplier is connected with the input of filter with channel fading speed estimation unit; The output signal of described channel fading speed estimation unit is connected with the input of look-up table unit; The output signal of described look-up table is connected with delay circuit with filter; Delay circuit is imported received signal simultaneously; The output signal of described filter and the output signal of delay circuit are decoded simultaneously through delivering to viterbi decoder behind second multiplier.

The device that variable bandwidth channel under above-mentioned a kind of frequency-selective channel is estimated, wherein, described filter is an average filter or the variable low pass filter of a coefficient.

Because the present invention has adopted above technical scheme, therefore under Limited resources, but down as certain processing channel number, total emission power etc., system can be according to accurate location of mobile station and velocity information, finish such as switching controls, for the travelling carriage allocated channel and reasonably (travelling carriage for microinching distributes a Microcell in the sub-district, the then distribution macrocell for rapid movement) task such as, can make more user obtain required service better, thereby reach the purpose of efficiently utilizing resource.

In addition, utilize comparatively accurate location of mobile station, velocity information, also can be mobile communcations system of future generation and be provided at more service outside voice, the data transmission.

Concrete feature of the present invention, performance are further specified by following embodiment and accompanying drawing thereof.

Fig. 1 is a prior art WCDMA uplink transport channel structure of time slot schematic diagram.

Fig. 2 is a prior art WCDMA descending transmission channel structure of time slot schematic diagram.

Fig. 3 is the channel estimation methods embodiment block diagram that adopts the linear interpolation decision-feedback in up.

Fig. 4 is the channel estimation methods embodiment block diagram that adopts LMS prediction decision-feedback in up.

Fig. 5 is a channel estimation method embodiment block diagram in descending.

Fig. 6 be under vehicle environment the error rate with D _fChange curve, Doppler frequency fd=(a) 100Hz.

Fig. 7 be under the vehicle environment error rate with D _fChange curve, Doppler frequency fd=(b) 900Hz.

See also accompanying drawing.

Variable bandwidth channel method of estimation under a kind of frequency-selective channel of the present invention is carried out conjugate multiplication with received signal and known pilot frequency code element earlier, obtains K channel fading estimated value constantly; And then try to achieve variable bandwidth channel and estimate, may further comprise the steps:

The channel fading estimated value that will obtain earlier is by a filter, with the noise in the further inhibition channel guess value; After estimating through channel fading speed, the channel fading estimated value that will formerly obtain simultaneously obtains the parameter of an adjustment filter again by look-up table; To pass through D with received signal through the channel estimation value that above step obtains at last _fSend into the Veterbi decoding part after the time-delay, do subsequent treatment.

Described filter is an average filter or the variable low pass filter of a coefficient.

If the employing average filter, its coefficient is constant to be 1, order parameter D _fBe the velocity estimation value that provides according to receiver channel rate of fading estimation unit, obtain according to corresponding relation calculating or look-up table again that the movement velocity of travelling carriage is fast more, corresponding D _fValue is more little.

If adopt the variable low pass filter of coefficient, then its order parameter D _fFixing, weight coefficient { h _i, i=0,1 ... 2D _fBe the velocity estimation value that provides according to receiver channel rate of fading estimation unit, and to calculate according to predetermined corresponding relation again or look-up table obtains, the movement velocity of travelling carriage is fast more, and the variable low pass filter two ends coefficient value of coefficient is more little.

The described channel fading value of this method can obtain in several ways, can adopt Gauss interpolation method or adaptive prediction filter method to obtain for time-multiplexed pilot form (up channel structure); Can directly obtain by the pilot frequency code element generator for connecting pilot frequency format (down channel structure).

Suppose a single user system, transmit through BPSK or QPSK modulation.In channel time slot, adopt the time-multiplexed pilot form.Each slot length T _Slot=(N _p+ N _d) T, the T here is single symbol time, and Nd is a data symbols length, and Np is a pilot symbol length.For time-multiplexed pilot, N _d＞0, for the situation of continuous pilot, pilot channel does not have data symbols, N _d=0.Structure of time slot is seen Fig. 1.

Suppose multipath channel be by the L road can decompose propagate the footpath (l=0,1, ^*, L-1) to form, the signal that receives can be expressed as: $r\left(\mathrm{kT}\right)=\underset{i=0}{\overset{L=1}{\mathrm{\Σ}}}{c}_l\left(\mathrm{kT}\right)s(\mathrm{kT}-{\mathrm{\τ}}_l)+n_l\left(\mathrm{kT}\right)------\left(1\right)$

Wherein, n (t) is a background noise, can be considered additivity Gauss white noise, and its one-sided power spectrum density is _N0c _l(t) and τ _lThen be the 1st footpath complex channel gain and time-delay; S (t) is corresponding transmitting baseband signal.

Below by several embodiment channel estimation methods of the present invention is described in more detail.

Embodiment one, linear interpolation feedback channel method of estimation:

Adopt the Gauss interpolation method to obtain the channel estimation methods of pilot frequency code element, can utilize time-multiplexed pilot section emission data known, obtain the fading channel information of pilot earlier, estimate the channel fading value of data segment again by the method for interpolation.The method that this Gauss interpolation method obtains pilot frequency code element is: at first utilize known pilot frequency code element, obtain corresponding channel fading value, be used for interpolation after the denoising that adds up to obtain the initial estimate of the channel information of data symbols section correspondence in each time slot; The modulation value of writing to of each footpath data symbols is tried to achieve according to resulting channel initial estimate in each footpath, carries out RAKE again and merges, and the result carries out an interim judgement to its output, recovers the estimation for transmit data symbols; Remove the modulation that transmits that above-mentioned steps obtains in each footpath received signal, obtain the comparatively accurate valuation of channel constantly, merge, constitute the data sequence of similar continuous pilot with the pilot number of writing to modulation value after delaying time for n.

Try to achieve channel estimation value by method of the present invention again, step is: a filter is passed through in the comparatively accurate valuation to channel that will obtain in above step, with the noise in the further inhibition channel guess value; Comparatively accurate valuation obtains a parameter of adjusting filter according to predetermined corresponding relation calculating or look-up table to channel simultaneously; At last will be through the channel guess value and process D of above step _fThe received signal of time-delay is sent into the Veterbi decoding part of back together, does subsequent treatment.

The top filter of mentioning can be a simple average filter, also can be a low pass filter that coefficient is variable.

If the employing average filter, its coefficient is constant to be 1, order parameter D _fBe the velocity estimation value that provides according to the receiver velocity estimation apparatus, obtain according to predetermined corresponding relation calculating or look-up table again that the movement velocity of travelling carriage is fast more, corresponding D _fValue is more little.

If adopt the variable low pass filter of coefficient, its order parameter D _fFixing, weight coefficient { h _i, i=0,1 ... 2D _fBe the velocity estimation value that provides according to the receiver velocity estimation apparatus, and to calculate according to predetermined corresponding relation again or look-up table obtains, the movement velocity of travelling carriage is fast more, and the variable low pass filter two ends coefficient value of coefficient is more little.

Received signal is broken down into the signal of L road different delay through after the relevant Rake receiver, and m data of n time slot of matched filter output can be expressed as: $\hat{d}(m,n)=\underset{l=0}{\overset{L-1}{\mathrm{\Σ}}}{r}_l(m,n){\hat{c}}_l^{*}(m,n)=N_p,N_{p+1},\·\·\·N_p+N_d-1------\left(2\right)$

Because pilot signal is in advance known, therefore, the 1st footpath is at the gain c at pilot frequency code element place _l(m, n), m=0,1 ... N _p-also can estimate, right _NpIndividual pilot frequency code element is done the valuation that demodulation can obtain a pilot channel fading after average

Interpolation class algorithm is according to known

It is pairing that interpolation goes out data symbols ${\hat{c}}_l(m,n),m=N_p,N_P+1,\·\·\·N_p+N_d-,$ Thereby try to achieve d (m, n).

Usually the interpolation channel estimation method that adopts has the Gauss interpolation method of introducing below.

The Gauss interpolation filter belongs to the category of the variable low pass filter of coefficient, and it also is different that the Gaussian filter coefficient of different rank is chosen, and three kinds of zeroth orders, single order, second order are generally arranged, and the interpolation formula of their correspondences is respectively: ${\hat{c}}_l(m,n)=Q_{-1}\left(\frac{m^{\′}}{N}\right){\hat{c}}_l(n-1)+Q_0\left(\frac{m^{\′}}{N}\right){\hat{c}}_l\left(n\right)+Q_1\left(\frac{m^{\′}}{N}\right){\hat{c}}_l(n+1)----\left(3\right)$ Wherein, N=N _d+ 1, m '=m-N _p+ 1=1,2 ... N _d

Weight coefficient separately is respectively:

Gauss's zeroth order: $Q_{-1}\left(\frac{m}{N}\right)=0$ $Q_0\left(\frac{m}{N}\right)=1$ $Q_1\left(\frac{m}{N}\right)=0----\left(4\right)$

Gauss's single order: $Q_{-1}\left(\frac{m}{N}\right)=0$ $Q_0\left(\frac{m}{N}\right)=1-\left(\frac{m}{N}\right)$ $Q_1\left(\frac{m}{N}\right)=\frac{m}{N}----\left(5\right)$ Gauss's second order: $Q_{-1}\left(\frac{m}{N}\right)=\frac{1}{2}\{{\left(\frac{m}{N}\right)}^2-\frac{m}{N}\}$ $Q_0\left(\frac{m}{N}\right)=1-{\left(\frac{m}{N}\right)}^2$ $Q_1\left(\frac{m}{N}\right)=\frac{1}{2}\{{\left(\frac{m}{N}\right)}^2+\frac{m}{N}\}----\left(6\right)$

The advantage of interpolation algorithm is that algorithm is simple, and computing is little, but time-delay is bigger, is a time slot as the time-delay of Gauss one, second order interpolation.Relative, adopt self adaptation to estimate that channel does not then have latency issue.

Fig. 3 is the channel estimation methods embodiment block diagram that adopts the linear interpolation decision-feedback in up.

The channel estimation methods embodiment block diagram of this linear interpolation decision-feedback comprises: a splitter, a multiplier that is connected with the output signal of splitter and an interpolating portion subdivision, one adder that is connected with the output signal of multiplier, and an interim judgement unit that is connected with the output signal of this adder, the multiplier that a mixer that is connected with the output signal of interim judgement unit and is connected with received signal with mixer; Also comprise:

One second delay unit that is connected with received signal; One low pass/the average filter that is connected with the output signal of multiplier, a Viterbi decoding unit that is connected with the output signal of the output signal of this low pass/average filter and second delay unit; The one channel fading speed estimation unit that is connected with the output signal of multiplier, a look-up table unit that is connected with the output signal of this channel fading speed estimation unit; The output signal of this look-up table unit is delivered to the variable low pass filter of coefficient and second delay unit respectively, and the output signal of this look-up table unit feeds back to the interpolating portion subdivision simultaneously.

At k constantly, receiver receives the DPCCH signal y in the 1st footpath _Kj, (K the code element in received signal the 1st footpath) is divided into pilot through splitter 211 and data segment receives data two parts.Pilot logarithmic data (K code element of pilot data of the 1st footpath splitter output) is admitted to Gauss's single order interpolating module 210, through 3,4,6 formulas, provides data segment the 1st footpath k the channel estimating value of k data constantly

Receive data with data segment after the conjugation and multiply each other, obtain k the data sign estimation value in the 1st footpath by multiplication unit

It and other footpaths are estimated value accordingly

Finish RAKE by adder unit 202 together and merge, obtain the merging value

It is through interim judgement unit 203, and output transmitted data symbols BPSK modulates estimated value x _k, the transmit symbol of this result and known pilot merges through mixer 212, reverts to original structure of time slot, after the conjugation again with the k received signal y in the 1st footpath constantly _{K, l}Multiply each other through multiplication unit 204, obtain channel estimated value more accurately constantly for k This estimated value is admitted to the variable low pass filter of coefficient 206, through (10) formula, obtains the channel guess value c through time-delay and inhibition Gaussian noise _{K-Df, l}, again after the conjugation with through D _fThe 1st footpath received signal of individual symbol delay unit 205 is y _{K-Df, l}Multiply each other by multiplier unit 213, its output is as the input of Viterbi decoder 207.

Simultaneously,

Also can be used as the input of channel fading speed estimation unit 208, by adding up in the unit interval The frequency that the anglec of rotation changes, perhaps other speed estimation algorithms can obtain an estimated value ＇ f roughly about current channel fading speed _e, in look-up table 209, stored channel fading rate and the corresponding suitable variable low pass filter length of coefficient in advance, according to input ＇ f _e, obtain the filter length that ought should adopt for the previous period, come the variable low pass filter of control coefrficient 206 by it, and delay unit 205 etc.

Embodiment two, and this is an adaptive prediction LMS judgment feedback channel estimation method:

Adopt the time-multiplexed pilot form in channel time slot, it adopts adaptive prediction filter LMS to dope each footpath at next one channel fading coefficient constantly, by decision-feedback adaptive prediction filter LMS is corrected again.The method of trying to achieve of its pilot frequency code element is: at first at the pilot that receives data slot, utilize known pilot frequency code element, obtain corresponding channel fading value, the coefficient of adaptive prediction filter LMS is trained; The data symbols section, the channel estimating value that each footpath provides at a last n-1 constantly according to corresponding adaptive prediction filter LMS for the current n moment, try to achieve the modulation value of writing to of each footpath data symbols, carrying out RAKE again merges, the result carries out an interim judgement to its output, recovers the estimation for transmit data symbols; Modulation that transmits that step obtains on each footpath received signal is removed, obtain the comparatively accurate valuation of channel constantly for n, adjust the coefficient of each diameter adaptive predictive filter LMS with it as calibrating signal, try to achieve next n+1 channel fading value constantly again, so iteration is gone down;

And then continue to obtain channel estimation value: a filter is passed through in the comparatively accurate valuation to channel that will obtain, with the noise in the further inhibition channel guess value; Simultaneously the comparatively accurate valuation of channel is calculated according to predetermined corresponding relation or look-up table obtains the parameter of an adjustment average filter; At last will be through the channel guess value and process D of above-mentioned steps _fThe received signal of time-delay is sent into the Veterbi decoding part of back together, does subsequent treatment.

The top filter of mentioning can be an average filter, also can be a low pass filter that coefficient is variable.

If the employing average filter, its coefficient is constant to be 1, order parameter D _fBe the velocity estimation value that provides according to the receiver velocity estimation apparatus, obtain according to predetermined corresponding relation calculating or look-up table again that the movement velocity of travelling carriage is fast more, corresponding D _fValue is more little.

If adopt the variable low pass filter of coefficient, its order parameter D _fFixing, weight coefficient { h _i, i=0,1 ... 2D _fBe the velocity estimation value that provides according to the receiver velocity estimation apparatus, and to calculate according to predetermined corresponding relation again or look-up table obtains, the movement velocity of travelling carriage is fast more, and the variable low pass filter two ends coefficient value of coefficient is more little.

Adaptive prediction decision feedback algorithms, its basic ideas are to utilize the LMS filter to dope the channel fading value: $c_k\&ap;y_k/x_k\underset{\_}{\mathrm{\&Delta;}}{\tilde{c}}_k,$ ${\hat{c}}_k=\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{b}_i^{*}{\tilde{c}}_{k-i}\underset{-}{\mathrm{\&Delta;}}\stackrel{-}{b}{\left(k\right)}^H\stackrel{-}{\stackrel{~}{c}}\left(k\right),----\left(7\right)$ $\stackrel{-}{\stackrel{~}{c}}\left(k\right)={({\tilde{c}}_{k-1},{\tilde{c}}_{k-2},\&hellip;,{\tilde{c}}_{K-N})}^T,\stackrel{-}{b}\left(k\right)={({\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="A0011947100181.png"\; state="\{\{state\}\}">b</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="A0011947100181.png,A0011947100182.png"\; state="\{\{state\}\}">b</figure-callout>}}_2,\&hellip;b_N)}^T$ Wherein, y _kBe k received signal, be equivalent to the y (kT) in (1) formula, c _kBe actual channel fading value,

For channel fading estimated value, by received signal y through overcorrection _kWith through really up to the markly declare, the detected symbol x after the BPSK modulation once more _kCompare and obtain.

Be that the process coefficient is b (k)=(b ₁, b ₂... b _N) ^TThe predicted value for k channel fading constantly of LMS filter.This filter coefficient is tried to achieve by following LMS recursive algorithm: $\stackrel{-}{b}(k+1)=\stackrel{-}{b}\left(k\right)+\mathrm{\&mu;}({\tilde{c}}_k-{\hat{c}}_k)*\stackrel{-}{\stackrel{~}{c}}\left(k\right)----\left(8\right)$

μ is the step-length of LMS filter in the formula.

Try to achieve

After, can the data estimator symbol ${\hat{x}}_k=\frac{y_k}{{\hat{c}}_k},$ Again it is modulated to corresponding x by minimum range decision rule _kGet on: $\mathrm{xk}=\underset{x_k\&Element;D}{\min }|x_k-x_k|,D=\{+1,-1\}\left(9\right)$

At the pilot of each time slot, frequency pilot sign x _kBe in advance known, the channel value c that obtains by it _kComparatively accurate, can be used to train the coefficient b (k) of current LMS filter.

Fig. 4 is the channel estimation methods embodiment block diagram that adopts LMS prediction decision-feedback in up.

In up, adopt the channel estimation methods embodiment block diagram of LMS prediction decision-feedback to comprise one first multiplier, an adder, an interim judgement unit, one second multiplier, one first delay unit and an adaptive prediction filter LMS; First multiplier input receives data, and its output valve is delivered to an adder and merged, and obtains a merging value and delivers to the interim judgement unit and export a data segment estimated value, and this output signal and received signal are sent into second multiplier together;

Also comprise:

One second delay unit that is connected with input signal; One low pass/the average filter that is connected with the output signal of second multiplier, a Viterbi decoding unit that is connected with the output signal of the output signal of this low pass/average filter and second delay unit; The one channel fading speed estimation unit that is connected with the output signal of second multiplier, a look-up table unit that is connected with the output signal of this channel fading speed estimation unit; The output signal of this look-up table unit is delivered to the variable low pass filter of coefficient and second delay unit respectively; Deliver to adaptive prediction filter LMS from the output signal of channel fading speed estimation unit after through a delayer simultaneously, deliver to the input of first multiplier from the output signal of adaptive prediction filter LMS again.

At k constantly, receiver receives the DPCCH signal in the 1st footpath _{Yk, l}, through a multiplication unit 101 and the 1st footpath channel estimating value of the k-1 moment for the k moment Conjugate multiplication obtains the 1st footpath data symbol estimated value

It and other footpaths are estimated value accordingly Finish RAKE by adder unit 102 together and merge, obtain the merging value Its is through interim judgement unit 103, output transmitted data symbols BPSK modulation estimated value xk, this as a result after the conjugation again with the k received signal y in the 1st footpath constantly _{K, l}Multiply each other through multiplication unit 104, obtain channel estimated value more accurately constantly for k It is through being admitted to LMS adaptive prediction filter 111, with preceding N-1 channel estimation value constantly after the delay unit 110

As the input of N rank filter, through the computing of (7), (8) formula, output is to next one channel estimating constantly together for l

Calculate constantly for the next one.

Step in 112 is applicable to the data segment of time slot, at pilot, only needs directly with known pilot frequency code element x _k Input 104 gets final product.

Also be admitted to the variable low pass filter of coefficient 106,, obtain channel guess value c through time-delay and inhibition Gaussian noise through (10) formula _{K-Df, l}, again after the conjugation with through D _fThe 1st footpath received signal of individual symbol delay unit 105 is y _{K-Df, l}Together as the input of Viterbi decoder 107.

Simultaneously, Also can be used as the input of channel fading speed estimation unit 108, by adding up in the unit interval

The frequency that the anglec of rotation changes, perhaps other speed estimation algorithms can obtain an estimated value ＇ f roughly about current channel fading speed _e, in look-up table 109, stored channel fading rate and the corresponding suitable variable low pass filter length of coefficient in advance, according to input ＇ f _e, obtain the length that ought should adopt for the previous period, come the variable low pass filter of control coefrficient 106 by it, and delay unit 105 etc.

Embodiment three: adopt the continuous pilot form in channel time slot, at first utilize known pilot frequency code element, obtain corresponding channel fading value, obtain the initial estimation of channel information in each time slot; The modulation value of writing to of each footpath data symbols is tried to achieve according to the channel initial estimate that obtains in each footpath, recovers the estimation for transmit data symbols.And then obtain channel estimation value, the steps include: a filter is passed through in the above comparatively accurate valuation to channel that obtains, with the noise in the further inhibition channel guess value; Simultaneously the comparatively accurate valuation of channel is calculated according to predetermined corresponding relation or look-up table obtains the parameter of an adjustment average filter; At last will be through the channel guess value and process D of above-mentioned steps _fThe received signal of time-delay is sent into the Veterbi decoding part of back together, does subsequent treatment.

The coefficient of the average filter in the above-mentioned step is constant to be 1, order parameter D _fBe the velocity estimation value that provides according to the receiver velocity estimation apparatus, obtain according to predetermined corresponding relation calculating or look-up table again that the movement velocity of travelling carriage is fast more, corresponding D _fValue is more little.

For the situation that adopts continuous pilot, preliminary channel estimating only need simply be removed the modulation that transmits and get final product: ${\tilde{c}}_k=\mathrm{ykexp}[-\mathrm{jK\&pi;}]----\left(10\right)$

(10) value of K, is 0 and gets 0 if be 1 then get 1 by the decision of the pattern of corresponding k pilot tone in the formula.

No matter be continuous pilot, still the time-multiplexed pilot situation that adopts interpolation class algorithm or adaptive prediction algorithm to finish, before the reception data delivery with channel estimation results and process time-delay arrives Viterbi decoder and corresponding DPDCH channel interpolating portion branch, in order to suppress the Gaussian noise in the channel, not direct input

But be 2D with it by a length earlier _fThe variable low pass filter of+1 coefficient increases the accuracy of channel estimating to improve signal to noise ratio: ${\stackrel{\_}{c}}_{k-\mathrm{Df}}=\underset{i=0}{\overset{2D_f}{\mathrm{\&Sigma;}}}{h}_j{\tilde{c}}_{k-1}----\left(11\right)$

For the variable low pass filter length parameter D of coefficient _fChoose, should be subjected to the restriction of signal correction length.For specific channel fading rate f _m, the exponent number 2D of the low pass filter that coefficient is variable _f+ 1 with the long-pending (2D of symbol period T _f+ 1) T, the cycle that yet promptly participates in relevant signal correspondence should not surpass maximum Nyquist frequency 2f _mInverse.

Fig. 5 is a kind of embodiment block diagram of channel estimation method in descending.

One local pilot frequency code element generator, a multiplier that is connected with the output signal of local pilot frequency code element generator;

Also comprise: second delay unit that is connected with input signal; One low pass/the average filter that is connected with the output signal of multiplier, a Viterbi decoding unit that is connected with the output signal of the output signal of this low pass/average filter and second delay unit; The one channel fading speed estimation unit that is connected with the output signal of multiplier, a look-up table unit that is connected with the output signal of this channel fading speed estimation unit; The output signal of this look-up table unit is delivered to the variable low pass filter of coefficient and second delay unit respectively.

At k constantly, receiver receives the DPCCH signal y in the 1st footpath _Kj, through a multiplication unit 101 and the k moment transmitting terminal pilot frequency code element x that sends by local pilot frequency code element generator 107 _kConjugate multiplication obtains the 1st footpath channel fading estimated value

Be admitted to the variable low pass filter of coefficient 302,, obtain channel guess value c through time-delay and inhibition Gaussian noise through (11) formula _{K-Df, l}, again after the conjugation with through D _fThe 1st footpath received signal of individual symbol delay unit 305 is y _{K-Df, j}Together as the input of Viterbi decoder 306.Simultaneously, Also can be used as the input of channel fading speed estimation unit 303, by adding up in the unit interval

The frequency that the anglec of rotation changes, perhaps other speed estimation algorithms can obtain an estimated value ＇ f roughly about current channel fading speed _e, in look-up table 304, stored channel fading rate and the corresponding suitable variable low pass filter length of coefficient in advance, according to input ＇ f _e, obtain the length that ought should adopt for the previous period, control the low pass filter that coefficient is variable by it, and delay unit 305 etc.

Fig. 6,7 is for all being 1 when the variable low pass filter weight coefficient of coefficient, and the error rate is with D when also promptly carrying out simple average _fThe curve that changes.(a) be channel Doppler frequency f _d=100Hz obtains, and (b) is channel Doppler frequency f _d=900Hz obtains.In (a), error rate order from high to low is followed successively by: D _f=2.4.6.8, obviously the increase along with filter order descends.And in (b), this descending order becomes: D _f=8.6.2.4.This is because under the slower situation of channel variation, the channel value in a very long time all changes not quite, good relationship, the smooth length of getting within the specific limits, to add up, suppress anti noise obvious more for long correlation more; And when channel variation is very fast, within a short period of time, do not possessed correlation between the channel fading value, if the exponent number of filter is still longer in this case, to no longer include the relevant advantage that adds up to adding up of channel, on the contrary may be because the result who makes output that cancels out each other between the uncorrelated signal more departs from actual value, still, too short as correlation length, as D _f=2, can not guarantee enough denoising effects again, so D _f=4 is more suitable values, in fact, and in the filter order length of this value, can be in the hope of, channel fading value still relevant.

From the result who obtains, when channel fading speed was low, the length of filter can obtain longer, otherwise, when channel fading very fast, during as the maximum speed 900Hz left and right sides that requires at 3G, the long rapid decline that will cause receiver performance of filter length.As seen, D _fChoose final channel estimating performance had considerable influence.In order to address this problem, the result that the present invention proposes the velocity estimation apparatus according to front end provides determines the thought of the low pass filter length that coefficient is variable.It mainly is before filtering and noise reduction is carried out in the channel fading valuation, roughly estimates earlier the Doppler frequency ＇ f that current channel fading changes _e, adopt the method for tabling look-up to determine filter length according to this result, thereby reach the purpose of adjusting in real time according to motion velocity of mobile station.

The present invention is directed to the 3-G (Generation Three mobile communication system) that adopts the DS-CDMA technology, under multiple pilot configuration, utilize mobile station speed estimation technique or signal fadeout rate estimation technique, the low pass filter bandwidth that the coefficient of adjusting in the channel estimation process to be adopted according to lookup table technology or corresponding empirical equation is variable and the method for coefficient, finish accurate channel estimation process under the different pilot configurations with less operand, thereby satisfy 3-G (Generation Three mobile communication system) for the travelling carriage requirement of communication quality under the mobile status from low speed to high speed.

Claims (6)

1, the variable bandwidth channel method of estimation under a kind of frequency-selective channel comprises step:

A, received signal and known pilot frequency code element are carried out conjugate multiplication, obtain K channel fading estimated value constantly;

It is characterized in that, further comprising the steps of:

B, the channel fading estimated value that will obtain in a step are passed through a filter, with the noise in the further inhibition channel guess value;

After estimating through channel fading speed, c, the channel fading estimated value that will obtain simultaneously obtain the parameter of an adjustment filter again by look-up table in a step;

D, will pass through D with received signal through the channel estimation value that b, c step obtain at last _fSend into the Veterbi decoding part after the time-delay, do subsequent treatment.

2, the variable bandwidth channel method of estimation under the frequency-selective channel according to claim 1 is characterized in that, described filter is an average filter or the variable low pass filter of a coefficient.

3, the variable bandwidth channel method of estimation under the frequency-selective channel according to claim 2 is characterized in that, described average filter coefficient is constant to be 1, order parameter D _fBe the velocity estimation value that provides according to receiver channel rate of fading estimation unit, obtain according to corresponding relation calculating or look-up table again that the movement velocity of travelling carriage is fast more, corresponding D _fValue is more little.

4, the variable bandwidth channel method of estimation under the frequency-selective channel according to claim 2 is characterized in that, the order parameter D of the low pass filter that described coefficient is variable _fFixing, weight coefficient (h _i, i=0,1 ... 2D _f) be the velocity estimation value that provides according to receiver channel rate of fading estimation unit, to calculate according to predetermined corresponding relation again or look-up table obtains, the movement velocity of travelling carriage is fast more, and the variable low pass filter two ends coefficient value of coefficient is more little.

5, the device of the estimation of the variable bandwidth channel under a kind of frequency-selective channel comprises one first multiplier, and the input of this multiplier connects a received signal and a pilot frequency code element signal; It is characterized in that, also comprise a filter, a channel fading speed estimation unit, a look-up table unit, a delay circuit, one second multiplier and a viterbi decoder; The output signal of described first multiplier is connected with the input of filter with channel fading speed estimation unit; The output signal of described channel fading speed estimation unit is connected with the input of look-up table unit; The output signal of described look-up table is connected with delay circuit with filter; Delay circuit is imported received signal simultaneously; The output signal of described filter and the output signal of delay circuit are decoded simultaneously through delivering to viterbi decoder behind second multiplier.

6, the device of the estimation of the variable bandwidth channel under a kind of frequency-selective channel according to claim 5 is characterized in that described filter is an average filter or the variable low pass filter of a coefficient.

Images