CN1799234A - Vector equalizer and vector sequence estimator for block-coded modulation schemes - Google Patents

Abstract

A vector decision feedback equalizer (VDFE) device for the detection of codewords transmitted over a dispersive channel in systems employing a block code is disclosed. The VDFE device comprises a vector feedforward unit in which a received signal vector is processed to a feedforward-filtered vector; a vector feedback unit in which a delayed decision vector is processed to a feedback-filtered vector; a differential vector combiner which receives the feedforward-filtered vector and the feedback-filtered vector and delivers an estimated signal vector representing an estimate of a desired signal vector D<I#191 =B0#191 Xi#191 . The device further comprises a Euclidean distance minimizer unit in which a set of metrics Mi,q#191 corresponding to squared Euclidean distances Delta>2<i,q#191 is calculated between the estimated signal vector and a desired differential vector combiner output B0#191Cq#191 for each possible codeword Cq#191 being an element of the block code, and a decision index with a corresponding closest codeword Cqi#191 is selected such that the closest codeword Cqi#191 attains the minimum squared Euclidean distance in the set of metrics Mi,q#191. A codeword generator for selecting a delayed decision vector based on a delayed decision index that is derived from the decision index is further comprised in the device. Further is disclosed a reduced-state vector sequence estimator (RS-VSE) device for the detection of codewords transmitted over a dispersive channel. The RS-VSE device comprises a vector feedforward unit in which a received signal vector is processed to a feedforward-filtered vector; a plurality of state-metric processors each of which comprises a Euclidean distance minimizer unit outputting a set of metrics corresponding to smallest squared Euclidean distances and a vector feedback unit for computing a feedback-filtered vector based on a delayed decision vector that corresponds to a previous decided codeword Xi-1#191(k) .

Description

The vector equalization device and the vector sequence estimator that are used for block-coded modulation schemes

Technical field

The present invention relates in the wireless or cabled digital communication system that has employing block encoding modulation (BCM) on the noisy communication channel of linear dispersion.The present invention is applicable to wireless communication system, and including, but not limited to the wireless lan (wlan) that linear dispersion mainly causes because of multipath transmisstion, the present invention also is applicable to wired communication system.The present invention relates to based on the receiver at the DFF (DFE) or the approximate maximum-likelihood sequence estimator (MLSE) of BCM scheme, the BCM scheme for example is the complementary code keying (CCK) that adopts in WLAN according to IEEE802.11 standard family.More specifically say, the present invention relates to vector decision-feedback equalizer (VDFE) equipment and reduce state vector sequential estimator (RS-VSE) equipment, be used for detecting the code word that on dispersive channel, sends in the system that adopts block code.

Background technology

In digital communication system, there is the channel of linear dispersion, the multipath channel in the wireless lan (wlan) for example, the individual data code element that causes sending produces a receiver response of crossing over a plurality of data symbols cycles.When adopting block encoding modulation (BCM), the code word of transmission often is called code element, and especially for the wireless system based on direct sequence spread spectrum (DSSS), the code word element of transmission is called chip.Adopting block length is that effect of linear dispersion can be divided into two components conceptive, interference (ICI) between intersymbol interference just (ISI) and sheet in the system of BCM of N sample.In the interval of the given N sample that is associated with current code word, ISI is the received signal component that adjacent code word produces, and ICI is the reception sample component that the adjacent chip because of current code word produces.

DFF (DFE) is a kind of well-known equipment of eliminating ISI when code element comprises single real number value or complex values sample.In recurrence each time, this classification (scalar) DFE is according to receiving sample and the single transmission sample of the estimation of judgement (just having sent the estimation of sample) in the past, by from the signaling point set that is called signal constellation (in digital modulation), selecting the signaling point of the most approaching estimated single transmission sample, come the judgement that makes new advances, and will newly adjudicate and feed back to feedback fraction immediately.

But in the system of coded modulation (BCM) that adopt to determine, code element (also becoming code word) comprises the piece of N real number value or complex values sample (being also referred to as chip), and wherein N is a fixed integer, for example 4 or 8.Therefore, bit decision can carry out by every N sample, and it has been got rid of and adopts above-mentioned sample one by one to carry out the situation of the classification DFE recurrence of decision-feedback.

Also there is known receiver technology, for example the technology of in United States Patent (USP) 6233273 B1 " RAKE Receiver with Embedded DecisionFeedback Equalizer " and U.S. Patent application 2001/0036223 A1, describing based on classification DFE at BCM.

Fig. 1 has provided the improved classification DFE according to United States Patent (USP) 6233273 B1, and it is different with classification DFE aspect two.At first because linear dispersion conceptive can the branch origin cause of formation code word and the ISI that produces and the ICI that produces because of the past chip of current code word in the past, so only from the output 111 of feedforward part 110, deduct the estimation 161 of ISI.Secondly, but still be subjected to the ICI interference signals, be admitted to correlator bank (bank) 130 at the ISI that do not have of the output of differential combiner 120.Utilize maximum-value selector 140, by selecting the codewords indexes corresponding to maximal correlation, every N sample obtains a decision index 141.Decision index 141 is admitted to data decoder (not shown in figure 1) and codeword generator 150, and the latter generates the BCM code word 151 corresponding to decision index 141, is used for estimating ISI at feedback fraction 160.

Those skilled in the art will appreciate that, to the BCM scheme, the CCK that for example has unitized Reed-Muller (GRM) code structure, correlator bank 130 can be passed through fast Walsh (Walsh) conversion (FWT) and effectively realize, the GRM code structure is on IEEE Transactions on Information Theory the 7th phase the 45th volume November in 1999 J.A.Davis and J.Jedwab " Peak-to-Mean power control in OFDM; Golay complementary sequences; and Reed-Muller codes " for example, and upward " Generalized Reed-Muller codes and power control inOFDM modulation " the middle description of K.G.Paterson of IEEETransactions on Information Theory the 1st phase the 46th of January calendar year 2001 volume.

As United States Patent (USP) 6233273 B1 and the above, the performance of the improved classification DFE shown in Fig. 1 descends because of residual ICI distortion.

Fig. 2 shows " RAKE Receiver withEmbedded Scalar DFEs (the having the RAKE receiver that embeds classification DFE) " according to United States Patent (USP) 6233273 B1, and it is configured to eliminate ISI and ICI.Feedforward part 210 is calculated output signal according to the sample 200 that receives according to following formula:

$Z_{\mathrm{Ni}+n}=\underset{m=-L_F+1}{\overset{0}{\mathrm{\&Sigma;}}}{f}_m{y}_{\mathrm{Ni}+n-m-\mathrm{\&delta;},}---[\mathrm{Equ}.1]$

Wherein δ is suitable delay parameter.The index that sends for Q hypothesis is each of current code word of q=0...Q-1, and an independent receiver branch is provided, and it comprises feedback fraction 280.q, differential combiner 220.q and codeword correlator 230.q.Feedback fraction 280.q calculates output signal 281.q according to following formula:

$V_{\mathrm{Ni}+n,q}=\underset{m=1}{\overset{n}{\mathrm{\&Sigma;}}}{b}_m{c}_{n-m,q}+\underset{m=n+1}{\overset{M}{\mathrm{\&Sigma;}}}{b}_m{\hat{x}}_{\mathrm{Ni}+n-m},---[\mathrm{Equ}.2]$

Its discipline Be to estimate according to the ICI that the current code word that hypothesis sends calculates,

Be to estimate according to the ISI that the code word with leading decision calculates.Differential combiner 220.q is according to following formula signal calculated 221.q:

${\hat{d}}_{\mathrm{Ni}+n,q}\underset{\&OverBar;}{\mathrm{\&Delta;}}{z}_{\mathrm{Ni}+n}-v_{\mathrm{Ni}+n,q}=z_{\mathrm{Ni}+n}-\underset{m=1}{\overset{n}{\mathrm{\&Sigma;}}}{b}_m{c}_{n-m,q}-\underset{m=n+1}{\overset{M}{\mathrm{\&Sigma;}}}{b}_m{\hat{x}}_{\mathrm{Ni}+n-m},---[\mathrm{Equ}.3]$

Wherein having provided in current transmission code word is C _qSupposition issue feed signals element X _Ni+nEstimation.Signal 221.q is sent to codeword correlator 230.q.At last, maximum-value selector 240 is according to the definite decision index 241 corresponding to the code word with maximal correlation of relevant 231.0...231.Q-1, and decision index 241 is delivered to data decoder (not shown among Fig. 2) and feedback path by retardation element 250.

As described in United States Patent (USP) 6233273 B1,, can remove codeword correlator 230.0...230.Q-1 by differential combiner 220.0...220.Q-1 for raising the efficiency.This caused producing on the function shown in Figure 3 quite but on calculating more efficiently receiver, this receiver is characterised in that to have correlator bank 320, one group of codeword correlator just at identical input signal 311 enterprising line operates, and has second codeword correlator set, it is by calculating feedback filter and the opposite in time convolution of gripping code word altogether in advance, and described convolution is stored in the look-up table that feedback fraction 380.0...380.Q-1 can visit, and in feedback fraction 370.0...370.Q-1, make up with feedback filter coefficient.As mentioned above, correlator bank 320 can be passed through FWT, effectively realizes at the BCM scheme with GRM code structure.Although passed through above simplification, the receiver among Fig. 3 still complexity is very high, because need Q Parallel Feedback part and differential combiner, and Q in fact can be up to 16,64 or 256.

For those skilled in the art, obviously ICI can be correctly synthetic in the feedback fraction of the Fig. 2 that transmits code word corresponding to reality or Fig. 3.The feedback of the current code word of supposing on the other hand, will cause remaining the additional distortion in the feedback fraction.Therefore, by the individual follow-up output of the N of the differential combiner 220.0...220.Q-1 among Fig. 2 The vector of forming

${\hat{d}}_{i,q}={[{\hat{d}}_{\mathrm{Ni}+0,q},{\hat{d}}_{\mathrm{Ni}+1,q},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{d}}_{\mathrm{Ni}+N-1,q}]}^T$

The energy that is different in essence is often arranged.In these cases, well-known, decision device should make square minimum of Euclidean distance on index q=0...Q-1:

$E_{i,q}\underset{\&OverBar;}{\mathrm{\&Delta;}}{\left|\right|{\hat{d}}_{i,q}-c_q\left|\right|}^2---[\mathrm{Equ}.4]$

C wherein _q=[C _{0, q}, C _{1, q}..., C _{N-1, q}] ^TQ code word of expression BCM sign indicating number, rather than make (real part) relevant maximum,

$r_{i,q}\underset{\&OverBar;}{\mathrm{\&Delta;}}{c}_q^{*}{\hat{d}}_{i,q}=\underset{n=0}{\overset{N-1}{\mathrm{\&Sigma;}}}{c}_{n,q}^{*}{\hat{d}}_{i,q},---[\mathrm{Equ}.5]$

This is because E _{I, q}Can expand to:

$E_{i,q}={\left|\right|{\hat{d}}_{i,q}\left|\right|}^2-2\mathrm{Re}\left\{r_{i,q}\right\}+{\left|\right|c_q\left|\right|}^2---[\mathrm{Equ}.6]$

Energy wherein Can be different in essence with above-mentioned.Therefore, make suboptimum correlated measure maximum, rather than make Euclidean distance minimum, degrade performance in fact.

Should be noted that adopting the BCM scheme of the QPSK (quaternary PSK) under uniform amplitude modulation, for example CCK situation, send the energy ‖ c of code word _q‖ ²Be identical.

Trial is with energy A kind of new method of taking into account of variation be that q=0...Q-1 is calculated relevant r _{I, q}And energy Make up these so that square minimum of the Euclidean distance shown in [Equ.6].Unfortunately, also there is not which kind of algorithm can be to the quick calculating energy of q=0...Q-1

This energy calculates than FWT more complicated fact on calculating and makes this method can't obtain practical application.

From the above, have a kind of like this demand in this area, promptly need a kind of being fit to receive improving equipment of BCM by dispersive channel, it does not rely on the maximization to the suboptimum correlated measure.This receiver device should provide a kind of fast and chip-area install efficiently, be used to calculate Euclidean distance square.

In addition, need a kind of improved receiver device, offset the error propagation effect that occurs in the receiver based on DFE.

Summary of the invention

Generally speaking, a kind of vector decision-feedback equalizer (VDFE) that the length that sends on the dispersive channel is the BCM code word of N that is fit to be received in is disclosed.Different with traditional receiver based on classification DFE that is used for the single output of single input dispersive channel, VDFE is based on the vector channel that dispersive channel is interpreted as with BCM scheme coupling, and avoided the decision-feedback of sample one by one intrinsic in traditional DFE structure.VDFE can be operated on the vector that length is N, and adopts N * N matrix value filter coefficient.Depend on the BCM scheme, VDFE equipment can be operated on real number value or the complex value vector.VDFE comprises vector feedforward filter unit (perhaps being exactly vector feedforward unit simply) and vector feedback filter unit (perhaps being exactly vector feedback unit simply), the two all is connected to differential vector combiner, and the output vector of differential vector combiner is represented the equalizer target vector estimation.This output vector is multiplied by rectangular matrix alternatively, sends to quick euclidean distance minimizer (minimizer) then, is used to select the nearest code word at Euclidean distance.More particularly, fast euclidean distance minimizer receives on the Euclidean distance collection between the vector at the reception vector of estimating with from transmitting hypothesis that code word set draws, carries out minimizing of Euclidean distance.By vector decision feedback, VDFE has eliminated the ISI that causes because of last code word.

According to a first aspect of the present invention, provide a kind of vector decision-feedback equalizer (VDFE) equipment that in the system that adopts block code, is used to detect the code word that on dispersive channel, sends.This equipment comprises: vector feedforward unit, and wherein the signal phasor of Jie Shouing is processed into the feedforward filtered vector; Vector feedback unit, wherein delay judgment vector is processed into the feedback filtering vector; Differential vector combiner, it receives feedforward filtered vector and feedback filtering vector, and transmits expression desired signal vector D _i=B ₀X _iThe estimated signal vector of estimation.This equipment also comprises euclidean distance minimizer unit, wherein is each the possible code word C as the block code element _q, corresponding to estimated signal vector and required differential vector combiner output B ₀C _qBetween square Δ of Euclidean distance _{I, q} ², calculate a set of metrics M _{I, q}, and select to have corresponding nearest code word

Decision index, make nearest code word

Be implemented in this set of metrics M _{I, q}In minimum euclid distance square.This equipment also comprises codeword generator, and it selects delay judgment vector based on the time-delay decision index that obtains according to decision index.

The Euclidean distance that VDFE equipment is adjudicated by compute codeword, and just relevant, reduce the probability of error, and provide a kind of by utilizing fast Walsh transform (FWT) to calculate the fast method of Euclidean distance collection.

Desired signal vector D _i=B ₀X ₁Can comprise a filter factor matrix B ₀, they obtain according to the impulse response that dispersive channel records.This allows to optimize filter factor B according to least mean-square error (MMSE) principle ₀Thereby, reduce the distortion of estimated signal vector with respect to the desired signal vector.

Euclidean distance minimizer unit can comprise linear transform unit, and wherein estimated signal vector is multiplied by the filter factor matrix B ₀The complex conjugate transposed matrix, obtain the correlator bank input vector.Like this, can calculate relevant with the desired signal vector by two steps: at first, with the filter factor matrix B ₀Relevant, secondly relevant with each code word.

By in vector feedforward unit and vector feedback unit, carrying out and the filter factor matrix B ₀Relevant, can reduce complexity, make the correlator bank input vector corresponding to the output of differential vector combiner, and vector feedforward unit and vector feedback unit are carried out and the filter factor matrix B ₀The multiplication of complex conjugate transposed matrix.

Euclidean distance minimizer unit can be passed through from calculating good energy term ε in advance _qIn deduct continuous item ρ _{I, q}The multiple of real part, calculate set of metrics M _{I, q}This is comparatively favourable in slow variation or time-independent channel, because can calculate energy term ε in advance in channel estimation phase _q, and use it for the follow-up data reception stage.

Calculating each continuous item ρ by correlator bank _{I, q}The time, can walk abreast that it is relevant to draw code word, aforementioned correlator bank is with the correlator bank input vector with from each code word C of block code _qRelevant.

Correlator bank can realize with the Fast transforms unit form of carrying out fast Walsh transform (FWT), thereby caused the quick realization of euclidean distance minimizer unit.This will advantageously utilize the GRM code structure, and more effectively compute codeword is relevant.

Vector feedforward unit can adopt fractional tap spacing (fractional tap spacing), and except that minimum Nyquist (Nyquist) bandwidth that requires, it also allows to utilize extra bandwidth to improve systematic function.

Vector feedforward unit, vector feedback unit and differential vector combiner can be exported the decimal system version of feedforward filtered vector, feedback filtering vector and estimated signal vector respectively, are used for eliminating the redundancy of code word.So, can realize lower data speed, for example the complexity for the feed forward element of the 5.5Mb/s of IEEE802.11b reduces.

According to a second aspect of the present invention, a kind of minimizing state vector sequential estimator (RS-VSE) equipment is provided, be used for detecting the code word that on dispersive channel, sends in the system that adopts block code.RS-VSE equipment can reduce the error propagation of VDFE equipment, keeps the efficient calculation of Euclidean distance.RS-VSE equipment comprises: vector feedforward unit, and wherein the signal phasor of Jie Shouing is processed into the feedforward filtered vector; A plurality of status quantitavie processors, each all comprises an euclidean distance minimizer unit, its output is corresponding to the set of metrics of minimum euclid distance square, and vector feedback unit, is used for according to the code word corresponding to last judgement

Delay judgment vector, calculate the feedback filtering vector; And Viterbi residue (survivor) measures selector, and it receives set of metrics from a plurality of status quantitavie processors, and therefrom selects final minimum set of metrics.

Reduce state vector sequential estimator (RS-VSE) equipment and be fit to be received in the BCM code word that the length that sends on the dispersive channel is N.Be similar to VDFE, RS-VSE is based on dispersive channel being construed to the vector channel that is complementary with the BCM scheme.RS-VSE can be operated on the vector that length is N, adopts N * N matrix value filter factor.Depend on the BCM scheme, RS-VSE equipment can be operated on real number value or the complex value vector.RS-VSE equipment comprises: vector feedforward part or vector whitened matched filter and one reduce the state viterbi algorithm, and this algorithm is used to select the approximate maximal possibility estimation of the codeword sequence that sends.RS-VSE equipment provides the state of lesser amt, and they are corresponding to most probable code word set in the past, and by the quick branch metric calculating relevant with state, the ISI that will be caused by previous code word takes into account.Like this, RS-VSE equipment provides the receiver performance of approaching optimum, with the suitable increase of complexity, has offset well-known error propagation effect.The branch of RS-VSE Viterbi lattice (trellis) is corresponding to the transmission of whole N chip codeword.Some branches can send from the single status of these lattice, and each branch is corresponding to the transmission of N chip codeword.In order to raise the efficiency, it is smaller value (for example 2 or 4) that people wish to limit from the quantity of the branch that each state sends, with (for example 2 or 4) code word corresponding to minimum euclid distance.Memory at vector channel is 1 o'clock, and RS-VSE equipment is particularly useful.

After the estimation of given dispersive channel and filtering parameter, VDFE and RS-VSE can both optimize equalizer target vector.Specifically, equalizer target vector does not need to be equal to the code word of transmission.By making equalizer target vector, can obtain extra performance gain corresponding to the optimization linear transformation that sends code word.

The matrix value filter factor of VDFE and RS-VSE can utilize the least mean-square error (MMSE) of vector quantization or compel 0 (zero-forcing, ZF) estimation technique calculates and/or adjusts, and it can carry out based on the channel estimating that obtains during preamble or the beginning sequence or by adaptive technique.

For having the BCM scheme of GRM code structure, VDFE and RS-VSE allow to utilize fast Walsh transform (FWT) to calculate the fast method that makes the required Euclidean distance collection of distance minimization.

Compare with the traditional receiver that is equipped with embedded classification DFE, adopt the performance of the receiver of VDFE equipment to be improved, traditional receiver maximization suboptimum correlated measure, and the energy changing of the code word that can't receive by multipath channel is taken into account.Simultaneously, the VDFE equipment that provides does not need one group of Q parallel feedback fraction, thereby has reduced complexity.

VDFE and/or RS-VSE equipment can realize by special use or programmable hardware device, for example ASIC or FPGA or as computer program, and they can provide with the form of the computer program stored on the computer usable medium.

Description of drawings

Below by example,, describe the preferred embodiments of the present invention in detail in conjunction with appended schematic diagram.

Fig. 1 shows the DFE that is used for the BCM scheme according to prior art, it is characterized in that eliminating the feedback fraction of the ISI that the code word because of the past the occurs caused ICI of past chip of current code word (rather than because of), its feature also is, combination correlator bank and the maximum-value selector that is used to select the code word with maximal correlation.

Fig. 2 shows the receiver according to a plurality of parallel embedding feedback fractions of having of prior art, it is characterized in that eliminating the device of ISI and ICI, and the maximum-value selector that is to make up correlator and is used to select the code word with maximal correlation.

Fig. 3 shows the receiver according to the optimization of a plurality of parallel embedding feedback fractions of having of prior art, it is characterized in that eliminating the device of ISI and ICI, and the maximum-value selector that is to make up correlator bank and is used to select the code word with maximal correlation.

Fig. 4 shows the block diagram that has the vector decision-feedback equalizer (VDFE) of quick euclidean distance calculator according to of the present invention.

Fig. 5 shows the block diagram of the minimizing state vector sequential estimator (RS-VSE) with quick euclidean measuring calculator, and the state of four minimizings is corresponding to four best previous code words in the exemplary realization that provides.

Provide these figure just for convenience of explanation, might not represent concrete instance of the present invention.

Embodiment

Fig. 4 shows the schematic block diagram according to vector decision-feedback equalizer of the present invention (VDFE) 400.VDFE equipment 400 abbreviates VDFE afterwards as, comprising: vector feedforward unit 410; Differential vector combiner 420; Optional matched filter 430; Euclidean distance minimizer unit 440 is also referred to as quick euclidean distance minimizer 440, and it is suitable for finding out nearest code word according to given block code; Feedback path, it comprises delay cell 450, codeword generator 460, vector feedback unit 470.In the detailed description of back, introducing length is the sample vector 401,411,421,461 and 471 of N, and is defined as follows respectively:

$Y_i\underset{\&OverBar;}{\mathrm{\&Delta;}}{[y_{\mathrm{Ni}+N-1},\&CenterDot;\&CenterDot;\&CenterDot;,y_{\mathrm{Ni}+1},y_{\mathrm{Ni}+0}]}^T$

$Z_i\underset{\&OverBar;}{\mathrm{\&Delta;}}{[z_{\mathrm{Ni}+N-1},\&CenterDot;\&CenterDot;\&CenterDot;,z_{\mathrm{Ni}+1},z_{\mathrm{Ni}+0}]}^T$

${\hat{D}}_i\underset{\&OverBar;}{\mathrm{\&Delta;}}{[{\hat{d}}_{\mathrm{Ni}+N-1},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{d}}_{\mathrm{Ni}+1},{\hat{d}}_{\mathrm{Ni}+0}]}^T$

${\hat{X}}_i\underset{\&OverBar;}{\mathrm{\&Delta;}}{[{\hat{x}}_{\mathrm{Ni}+N-1},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{x}}_{\mathrm{Ni}+1},{\hat{x}}_{\mathrm{Ni}+0}]}^T$

With

$V_i\underset{\&OverBar;}{\mathrm{\&Delta;}}{[V_{\mathrm{Ni}+N-1},\&CenterDot;\&CenterDot;\&CenterDot;,V_{\mathrm{Ni}+1},V_{\mathrm{Ni}+0}]}^T$

Sample vector 401,411,421,461 and 471 also is called received signal vector 401, feedforward filtered vector 411, estimated signal vector 421, delay judgment vector 461 and feedback filtering vector 471.Though the length of received signal vector 401 is N hereinafter, the present invention has been contained the longer input vector of the utilization of generally adopting in balanced field and has been carried out the enhancing of fractional tap spacing.Vector feedforward unit 410 adopts N * N matrix value feedforward filter coefficients F _m, m=-L _F... 0, and vector feedback unit 470 adopts N * N matrix value feedback filtering coefficient B _m, m=1...M _B Vector feedforward unit 410 is calculated feedforward filtered vector 411 according to following formula according to the vector 401 that receives:

$Z_i=\underset{m=-L_F+1}{\overset{0}{\mathrm{\&Sigma;}}}{F}_m{Y}_{i-m-{\mathrm{\&delta;}}_F,}---[\mathrm{Equ}.7]$

δ wherein _FBe suitable delay parameter, and vector feedback unit 470 is calculated feedback filtering vector 471 according to following formula according to delay judgment vector 461:

$V_i=\underset{m=1}{\overset{M_B}{\mathrm{\&Sigma;}}}{B}_m{\hat{X}}_{i-m}.---[\mathrm{Equ}.8]$

In feedback path, codeword generator 460 is selected delay judgment vector 461 according to time-delay decision index 451, and time-delay decision index 451 is obtained according to decision index 449 by delay cell 450.

Differential vector combiner 420 calculates estimated signal vector 421 according to following formula:

${\hat{D}}_i\underset{\&OverBar;}{\mathrm{\&Delta;}}{Z}_i-V_i=Z_i-\underset{m=1}{\overset{M_B}{\mathrm{\&Sigma;}}}{B}_m{\hat{X}}_{i-m},---[\mathrm{Equ}.9]$

It represents the estimation of following desired signal vector:

$D_i{\underset{\&OverBar;}{\mathrm{\&Delta;}}B}_0{X}_i,---[\mathrm{Equ}.10]$

Here be also referred to as equalizer target vector.The desired signal vector is conversely by N * N matrix B ₀Specify, can be to matrix B ₀Select, to optimize performance.According to a preferred embodiment of the present invention, select to have the matrix B of three-legged structure ₀, alternatively it is carried out standardization, make to have the identity element that is positioned on the leading diagonal.

Compare estimated signal vector now according to the VDFE of [Equ.9]

With signal according to [Equ.3] Wherein the latter is by the prior art calculation of equipments among Fig. 2.Estimated signal vector

And do not rely on q, because it does not use possible current code word Elements C _{0, q}... C _{N-1, q}, q=0...Q-1.Therefore, different with Fig. 2,470 couples of VDFE of single vector feedback unit are just enough.In addition, VDFE does not generally have and does not need to suppress ICI, because signal Approximate with required signal phasor [Equ.10], its element advantageously comprises the ICI component.On the other hand, prior art estimates that [Equ.3] attempts to estimate to suppress ICI (except suppressing ISI) by the ICI that deducts to each possible current transmission code word hypothesis.

Can calculate and/or adjust matrix value feedforward and back feedforward coefficient and the matrix B of VDFE according to other optimisation criteria that widely adopt in least mean-square error (MMSE), ZF standard or the balancing technique field ₀Under the channel condition that changes along with the time, the coefficient of VDFE can be adjusted, so that the MMSE minimum.In a kind of preferred embodiment, select the VDFE coefficient, make that when not having last decision error estimated signal vector 421 is provided by following formula:

${\hat{D}}_i\underset{\&OverBar;}{\mathrm{\&Delta;}}{B}_0{X}_i+N_i,---[\mathrm{Equ}.11]$

X wherein _iBe the code word of current transmission, N _iIt is noise vector with irrelevant component.Adopt quick euclidean distance minimizer 440 to select decision index 449, $q={\hat{q}}_i$ The time, minimize Euclidean distance square:

${\mathrm{\&Delta;}}_{i,\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="A20048001495700211.png"\; state="\{\{state\}\}">q</figure-callout>}}^2\underset{\&OverBar;}{\mathrm{\&Delta;}}{\left|\right|{\hat{D}}_i-B_0{C}_q\left|\right|}^2={\left|\right|{\hat{D}}_i\left|\right|}^2-2\mathrm{Re}\left\{{\mathrm{\&rho;}}_{i,q}\right\}+{\mathrm{\&epsiv;}}_q,---[\mathrm{Equ}.12]$

Wherein defined code word vector C _q=[C _{N-1, q},, C _{1, q}, C _{0, q}] ^T, relevant

${\mathrm{\&rho;}}_{i,q}\underset{\&OverBar;}{\mathrm{\&Delta;}}{C}_q^{*}{\mathrm{<figure-callout\; id="0"\; label="B"\; filenames="A20048001495700191.png,A20048001495700221.png"\; state="\{\{state\}\}">B</figure-callout>}}_0^{*}{\hat{D}}_i,---[\mathrm{Equ}.13]$

And energy term:

${\mathrm{\&epsiv;}}_q\underset{\&OverBar;}{\mathrm{\&Delta;}}{\left|\right|B_0{C}_q\left|\right|}^2,---[\mathrm{Equ}.14]$

For raising the efficiency, can calculate in advance them, and be stored in the question blank.Under the channel condition that changes along with the time, energy term is colluded and can be adjusted.Because And do not rely on codewords indexes q, so it is irrelevant with minimizing of [Equ.12].Therefore, also can minimize equally and measure:

$M_{i,q}\underset{\&OverBar;}{\mathrm{\&Delta;}}{\mathrm{\&epsiv;}}_q-2\mathrm{Re}\left\{{\mathrm{\&rho;}}_{i,q}\right\},---[\mathrm{Equ}.15]$ ]

Shown in the detail drawing of the quick euclidean distance minimizer 440 among Fig. 4.Can be according to the relevant of [Equ.13] by estimated signal vector 421 is multiplied by matrix B ₀ Linear transform unit 430, be that correlator bank 441 realizes afterwards, under GRM block code structure, correlator bank can be used as FWT equipment and effectively realizes.Perhaps, linear transformation B ₀ ^*Can absorb by filtered device 410 and 470.

Below, identical reference symbol or numeral are used to represent same or analogous parts.

Fig. 5 has provided the schematic block diagram of minimizing state vector sequential estimator (RS-VSE) equipment 500, and state set wherein is corresponding to one group of K most probable previous code word 0≤k＜K.For exemplary embodiment shown in Figure 5, K=4.RS-VSE equipment 500 comprises vector feedforward unit 410, and a plurality of status quantitavie processor 520.0...520.K-1 and Viterbi residue is measured selector 550, aforesaid state measure processor each all with a most probable previous code word Be associated, this code word is also referred to as a first leading decision code word Each status quantitavie processor 520.k comprises a fast metric calculator 540.k and vector feedback unit 470.k, is used for according to delay judgment vector 461.k, calculates feedback filtering vector 471.k.At delay judgment vector 461.k by code word Under the hypothesis that provides, status quantitavie processor 520.k calculates estimated signal vector 421.k.Fast metric calculator 540.k, be also referred to as quick euclidean measuring calculator, corresponding to the quick euclidean distance minimizer of describing in conjunction with Fig. 4 440, and be used to calculate the set that Q euclidean measured, from this set, select P minimum to measure M then _i(k, 0) ... M _i(k, 3), wherein for the embodiment among Fig. 5, P=4, and P is corresponding to decision index

Minimum is measured M _i(k, 0) ... M _i(k, 3) and corresponding decision index Can think optimum state, be forwarded to the Viterbi residue and measure selector 550 that the latter passes through decision index

(wherein 0≤k＜K) selects new most probable code word set.On the feedback path of each status quantitavie processor 520.k inside, the delay cell that is designated as D is equipped with, be used to generate the codeword generator of delay judgment vector 461.k, and the vector feedback unit 470.k that calculates feedback filtering vector 471.k according to delay judgment vector 461.k.

For the technical staff in Veterbi decoding field, obviously RS-VSE equipment 500 can reduce any error propagation effect relevant with DFE, and realizes the performance of error rate performance near maximum-likelihood sequence estimation.In addition, compare with traditional maximum-likelihood sequence estimator, RS-VSE equipment 500 utilizes fast metric calculator 540.0...540.k, has reduced and has calculated the complexity of measuring that is associated with the Q that sends from each state lattice branch.

Any one disclosed embodiment can make up with one or more other embodiment that illustrate and/or describe.One or more features among the embodiment also can make up.

The present invention can be contained in the computer program, and this product comprises the feature that allows realization described here, and when packing this product in computer system, this product can be realized this method.

Computer program device herein or computer program mean any expression with the instruction set of any language, representative and mark, it be intended to make the system with information processing capability can be directly or below both or one of them carry out specific function after finishing: a) transform to another language, code or mark; B) reappear with different material form.

Claims (12)

1. vector decision-feedback equalizer (VDFE) equipment (400) that in adopting the system of block code, is used to detect the code word that sends by dispersive channel, this equipment comprises:

Vector feedforward unit (410), wherein the signal phasor of Jie Shouing (401) is processed into feedforward filtered vector (411);

Vector feedback unit (470), wherein delay judgment vector (461) is processed into feedback filtering vector (471);

Differential vector combiner (420), it receives feedforward filtered vector (411) and feedback filtering vector (471), and transmits expression desired signal vector (D _i=B ₀X _i) the estimated signal vector (421) of estimation;

Euclidean distance minimizer unit (440) wherein is each the possible code word (C as the block code element _q), corresponding to estimated signal vector (421) and required differential vector combiner output (B ₀C _q) between square (Δ of Euclidean distance _{I, q} ²), calculate set of metrics (M _{I, q)}, and select to have corresponding nearest code word Decision index (449), make nearest code word Be implemented in set of metrics (M _{I, q}) in minimum euclid distance square; And

Codeword generator (460), it selects delay judgment vector (461) based on the time-delay decision index (451) that obtains according to decision index (449).

2. according to the equipment of claim 1, desired signal vector (D wherein _i=B ₀X _i) comprise a filter factor matrix (B ₀), this matrix obtains according to the impulse response that dispersive channel records.

3. according to the equipment of arbitrary aforementioned claim, wherein euclidean distance minimizer unit (440) comprises linear transform unit (430), wherein estimated signal vector (421) is multiplied by filter factor matrix (B ₀) the complex conjugate transposed matrix, to obtain correlator bank input vector (431).

4. according to the equipment of claim 3, wherein correlator bank input vector (431) is corresponding to the output of differential vector combiner (420), and vector feedforward unit (410) and vector feedback unit (470) are carried out and filter factor matrix (B ₀) the multiplication of complex conjugate transposed matrix.

5. according to the equipment of arbitrary aforementioned claim, wherein euclidean distance minimizer unit (440) is passed through from calculating good energy term ε in advance _qIn deduct continuous item (ρ _{I, q}) the multiple of real part, calculate set of metrics (M _{I, q}).

6. according to the equipment of claim 3 or 4 and 5, each continuous item (ρ wherein _{I, q}) by with correlator bank input vector (431) with from each code word (C of block code _q) relevant correlator bank (441) calculating.

7. according to the equipment of claim 6, wherein correlator bank (441) realizes with the Fast transforms unit form of carrying out fast Walsh transform (FWT), thereby has caused the quick realization of euclidean distance minimizer unit (440).

8. according to the equipment of arbitrary aforementioned claim, wherein vector feedforward unit (410) adopts the fractional tap spacing.

9. according to the equipment of arbitrary aforementioned claim, wherein vector feedforward unit (410), vector feedback unit (470) and differential vector combiner (420) are exported the decimal system version of feedforward filtered vector (411), feedback filtering vector (471) and estimated signal vector (421) respectively, are used for eliminating the redundancy of code word.

10. method that in adopting the system of block code, detects the code word that sends by dispersive channel, this method may further comprise the steps:

In vector feedforward unit (410), the signal phasor (401) that receives is processed into the filtered vector (411) that feedovers;

In vector feedback unit (470), delay judgment vector (461) is processed into feedback filtering vector (471);

By differential vector combiner (420),, draw expression desired signal vector (D according to feedforward filtered vector (411) and feedback filtering vector (471) _i=B ₀X _i) the estimated signal vector (421) of estimation;

By euclidean distance minimizer unit (440), be each possible code word (C as the block code element _q), corresponding to estimated signal vector (421) and required differential vector combiner output (B ₀C _q) between square (Δ of Euclidean distance _{I, q} ²), calculate set of metrics (M _{I, q}), and select to have corresponding nearest code word

Decision index (449), make nearest code word Be implemented in set of metrics (M _{I, q}) in minimum euclid distance square; And

Based on the time-delay decision index (451) that obtains according to decision index (449), select delay judgment vector (461).

11. a computer program that is stored on the computer usable medium comprises computer-readable program means, is used to make the method for computer execution according to claim 10.

12. one kind is reduced state vector sequential estimator (RS-VSE) equipment (500), is used for detecting the code word that sends by dispersive channel in the system that adopts block code, this equipment comprises:

Vector feedforward unit (410), wherein the signal phasor of Jie Shouing (401) is processed into feedforward filtered vector (411);

A plurality of status quantitavie processors (520.0...520.K-1), its each all comprise: according to the euclidean distance minimizer unit (540.k) of claim 1 to 9, be used to export corresponding to minimum euclid distance square set of metrics; And vector feedback unit (470.k), be used for according to code word corresponding to first leading decision

Delay judgment vector (461.k), calculate feedback filtering vector (471.k); And Viterbi residue measures selector (550), and it receives set of metrics from a plurality of status quantitavie processors (520.0...520.K-1), and therefrom selects final minimum set of metrics.

Images