CN1416233A - Recerving method based on time-frequency diversity - Google Patents

Abstract

In the spread spectrum communication system, when the channel of transmitting signal is in quick fading, the performances of the traditional RAKE receiver is declined obviously. Based on the idea of the time and frequency diversity, the new receiving method having lower complexity is put forward in order to utilzie the multiple paths-Doppler combined processing farthest. Based on the invented method, two schemes of the multiple user detection are provided. The method overcomes the influence of the receiving signal caused by the fading channel, especially the quick fading channel. The invention improves the detection performance of the receiver under the quick fading channel so as to improve the detection performance of the multiple-user receiver. Combining with the technique of enhancing the multiple user detection, the better receiving effect is realized.

Description

A kind of method of reseptance based on the T/F diversity

Affiliated field

The present invention relates to a kind of method of reseptance, exactly, relate to the multiuser detection in a kind of spread spectrum communication, belong to the spread spectrum technic field based on the T/F diversity.

Background technology

In CDMA bandspread communication system, existing most of receivers all are to adopt Rake (RAKE) reception technique, the i.e. reception technique of rake usually.This technology can be utilized the mutual uncorrelated characteristic of multipath signal, and diversity merges each footpath information, thereby reaches the multipath fading of restraining wireless channel, finally improves the purpose of the detection performance of receiver.But the channel circumstance that the RAKE reception technique is suitable for declining at a slow speed is in when system under the channel circumstance of quick decline, and owing to increasing sharply of channel estimation errors, the performance of traditional RAKE receiver can obviously descend, and it detects effect and tends to remarkable deterioration.

Document [1] " multipath-Doppler in the mobile radio communication unites diversity " (" Jointmultipath-Doppler diversity in mobile wireless communication ", IEEE Trans.Commun., vol.47, pp.123-132, Jan.1999) and document [2] " Multiuser Detection in the multi-path environment that declines fast " (" Multiuser detection in fast-fading multipath environments ", IEEE J.Selected Areas in Commun., vol.16, pp.1691-1701, Dec.1998) a kind of new thought of Doppler's diversity has been proposed, point out that in fact channel provides the possible of another kind of diversity, i.e. Doppler's diversity (Doppler diversity) in the variation of time domain.Unite in time selectivity (time-selective) signal model of diversity multipath-Doppler, the signal decomposition of process channel becomes the multipath-Doppler's of mutually orthogonal different branch signals, claims that this model is T/F model or TF (Time-Frequency) model.Document [1] [2] has also been introduced the receiver that joined multi-path-Doppler handles in an enterprising step of this TF model basis.

Note T is the cycle of data symbol, T _sBe the duration (spreading duration) of frequency spreading wave, then according to the explanation of document [1] [2], this TF model should comprise two kinds of situations: TF model 1:T=T _sWith TF model 2:T＜T _sAlthough document [1] [2] has been discussed TF RAKE receiver, decorrelator and least mean-square error (MMSE) receiver under TF model 1, this TF RAKE receiver receives at single subscriber signal, and is not suitable for multi-user's reception; Though several receivers in back are to receive at the multi-user, the complexity of these receivers is too high, is difficult to realize.

On TF model 2 times, document [3] " reception of the time selectivity signal in the communication under the multidiameter fading channel " (" Time-selective signaling and reception for communication over multipathfading channels ", IEEE Trans.Commun., vol.48, pp.83-94, Jan.2000) in single user's multipath signal receiver only has been discussed, and be not mentioned to the receiver of corresponding multi-user multi-path.

Summary of the invention

The purpose of this invention is to provide a kind of method of reseptance based on the T/F diversity, just provide a kind of at method of reseptance than realization multipath-Doppler's diversity of receiver end under the low computational complexity, this method can overcome the especially fast fading channel influence that reception causes to signal of fading channel, thereby significantly improves the detection performance of multiuser receiver; And under condition, realize better reception further combined with the Multiuser Detection enhancement techniques.

Method of reseptance based on the T/F diversity of the present invention is achieved in that this method of reseptance includes the following step:

(1) in the band spectrum modulation of signal, with data symbol frequency spreading wave duration T _sBe set to equal the period T of data symbol, i.e. T _s=T;

(2) utilize multipath-Doppler's branch signal of data symbol frequency spreading wave, do the TF matched filtering of T/F to received signal, and export the matched filtering result of each each TF branch of footpath of each user;

(3) all the TF matched filtering results to each user carry out the merging of TF RAKE high specific, are combined operating result then and carry out the symbol hard decision;

(4) on the basis that TF RAKE receives, further adopt parallel interference to eliminate the Multiuser Detection enhancement techniques of (PIC), detect performance so that further improve.

The TF RAKE high specific union operation that described (3) step is carried out is the result who utilizes each TF branch signal matched filtering, and the multipath of respective branches-Doppler's channel parameter, carries out the operation that high specific merges.

Described (4) step is carried out the Multiuser Detection enhancement techniques operation that parallel interference is eliminated (PIC), and its concrete steps are as follows:

(a) the progression G of multistage parallel interference eliminated is set, a level counting number index is g, g=1 ..., G, and establish initial index g=1;

(b) utilize TF RAKE to receive symbol hard decision and the channel parameter that obtains, carry out the regenerative operation of each TF branch signal, promptly disturb regenerative operation, obtain the regenerated signal of each TF branch;

(c), carry out the interference cancellation operation to each TF branch signal of each user;

(d) each purer TF branch signal is carried out the TF coupling again, obtain the TF matched filtering result of this branch; The operation of this TF matched filtering is identical with described step (2);

(e), carry out TF RAKE and merge, and carry out the symbol hard decision of PIC at the corresponding levels to the TF matched filtering result of each TF branch signal of each user; The operation that this TF RAKE merges is identical with described step (3);

(f) value with g increases by 1,

(g) utilize the symbol hard decision result of upper level, the operation of repeating step (b)～(f) is till finishing the operation that g=G level parallel interference eliminates; At last also will be with carrying out the value output that afterbody g=G level parallel interference is eliminated the symbol hard decision that obtains after the operation, as final data symbol detection result.

The regenerative operation of each TF branch signal in described (b) step is each TF branch signal with each user's frequency spreading wave signal, multiply by the value of this user's symbol hard decision at the corresponding levels; And each TF branch signal of each user's frequency spreading wave then is that multipath-Doppler's channel parameter of utilizing the TF channel estimating to obtain and this known user's frequency spreading wave obtain.

The multipath that the described TF of utilization channel estimating obtains-Doppler's channel parameter can be that each grade PIC is identical, promptly uses same TF channel estimation value, also can reappraise step by step to upgrade to use.

Interference cancellation operation in described (c) step is each TF branch signal for each user, from baseband receiving signals, deduct the TF interference signal of this branch, obtain this purer branch signal, the TF interference signal of this branch is meant: the regenerated signal of all the TF branch signals except this TF branch signal.

In described (4) step, also can adopt the Multiuser Detection enhancement techniques of the soft interference cancellation algorithm of iteration SISO, detect performance so that further improve.

Method of reseptance based on the T/F diversity of the present invention can be achieved in that also this method of reseptance includes the following step at least:

(1) in the band spectrum modulation of signal, with symbols spread waveform duration T _sBe set to period T, i.e. T greater than data symbol _s＞T, and establish I=T _s/ T is a positive integer;

(2) utilize multipath-Doppler's branch signal of data symbol frequency spreading wave, do the TF matched filtering of T/F to received signal, and export the matched filtering result of each each TF branch of footpath of each user; TF matched filtering herein can be provided with the symbol detection window size, so that the TF RAKE union operation of back more effectively overcomes intersymbol interference (ISI);

(3) all the TF matched filtering results to each user carry out the merging of TF RAKE high specific, are combined operating result then and carry out the symbol hard decision; Each user's here TF RAKE high specific union operation, it is the operation of respectively all symbols in the detection window in each footpath of this user being removed intersymbol interference earlier, the TF channel parameter that utilizes the TF channel estimating to obtain then, the high specific union operation is carried out in all each footpath, the value after the high specific that obtains all symbols in this user's detection window at last merges;

The operation of each footpath intersymbol interference of the removal in described (3) step, be meant that the TF matched filtering result to the vector form in this footpath multiply by the inverse matrix of TF correlation matrix, and this TF correlation matrix is the result that the related operation by Dan Jing obtains, the dimension of this matrix reduces greatly, can significantly reduce operand.

After (3) step, can optionally increase following operation: on the basis that TF RAKE receives, further adopt parallel interference to eliminate the Multiuser Detection enhancement techniques of (PIC) or the soft interference cancellation algorithm of iteration SISO, detect performance so that further improve.

Described parallel interference is eliminated the Multiuser Detection enhancement techniques operation of (PIC), and its concrete steps are as follows:

(a) the progression G of multistage parallel interference eliminated is set, a level counting number index is g, g=1 ..., G, and establish initial index g=1;

(b) utilize TF RAKE to receive symbol hard decision and the channel parameter that obtains, carry out the regenerative operation of each TF branch signal, promptly disturb regenerative operation, obtain the regenerated signal of each TF branch, and this interference regenerative operation is to carry out at all symbols in the detection window;

(c), carry out the interference cancellation operation to each TF branch signal of each user;

(d) each purer TF branch signal is carried out the TF coupling again, obtain the TF matched filtering result of this branch; The operation of this TF matched filtering is identical with described step (2);

(e), carry out TF RAKE and merge, and carry out the symbol hard decision of PIC at the corresponding levels to the TF matched filtering result of each TF branch signal of each user; The operation that this TF RAKE merges is identical with described step (3);

(f) value with g increases by 1,

(g) utilize the symbol hard decision result of upper level, the operation of repeating step (b)～(f) is till finishing the operation that g=G level parallel interference eliminates; At last also will be with carrying out the value output that afterbody g=G level parallel interference is eliminated the symbol hard decision that obtains after the operation, as final data symbol detection result.

The regenerative operation of each TF branch signal in described (b) step is the value that each TF branch signal of each user's frequency spreading wave signal be multiply by this user's symbol hard decision at the corresponding levels, at last to all symbols summations in the detection window; And each TF branch signal of each user's frequency spreading wave is that multipath-Doppler's channel parameter of utilizing the TF channel estimating to obtain and this known user's frequency spreading wave obtain.

Interference cancellation operation in described (c) step is each TF branch signal for each user, deducts the TF interference signal of this branch from baseband receiving signals, obtains this purer branch signal, and this operation is between the operating space with whole detection window; The TF interference signal of this branch is meant: the regenerated signal of all the TF branch signals except this TF branch signal.

Characteristics of the present invention are based on the thought of time and frequency diversity, farthest utilize the diversity performance of channel, adopt multipath-Doppler to unite the processing procedure of diversity and the TF method of reseptance of the low complex degree of a kind of simplification of proposing; And two kinds of Multiuser Detection schemes based on this method have been proposed.The receiver that this method of reseptance can significantly improve under fast fading channel detects performance.Basis of the present invention is that multipath-Doppler unites diversity, realizes the selection of time characteristic (time-selective) of radio propagation channel more than or equal to the data symbol cycle by the symbols spread waveform duration.For single user receives, can adopt the low complex degree TF RAKE method of reseptance among the present invention, for the multi-user, then on the basis that the TFRAKE of low complex degree receives, further adopted multiuser detection, each user profile of Combined Treatment reaches the purpose that detects multiple user signals more accurately in the time of can receiving a plurality of subscriber signals (promptly detecting the user) at the same time; And all reception processing procedures all are based on multipath-Doppler and unite that diversity carries out.

Description of drawings

Fig. 1 is the schematic diagram (number of users=10) of the Multiuser Detection performance under the TF1 model.

Fig. 2 is the schematic diagram (SNR=5dB) of the Multiuser Detection result under the TF1 model.

Fig. 3 is the schematic diagram (number of users=4) of the Multiuser Detection result under the TF2 model.

Fig. 4 is the schematic diagram (SNR=5dB) of the Multiuser Detection result under the TF2 model.

Embodiment

The present invention can be divided into two kinds of situations and discuss respectively, and one of them is that promptly the band spectrum modulation process satisfies T on TF model 1 time _sDuring the condition of=T, the operating procedure of its specific embodiments is described below:

(1) in the band spectrum modulation of signal, with data symbol frequency spreading wave duration T _sBe set to equal the period T of data symbol, i.e. T _s=T;

Signal with the relevant modulation of binary phase shift keying (BPSK) is an example earlier.Radio propagation channel still adopts the incoherent scattering WSSUS wireless channel model under the wide stable meaning that proposes in the document [1], and the multipath branch of promptly supposing same transmission signal is incoherent mutually.Be located at receiving terminal, continuous time, the baseband signal model of variable was: $r\left(t\right)=\sqrt{E_s}\underset{i}{\mathrm{\Σ}}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{b}_k\left(i\right)x_k^i(t-{\mathrm{\τ}}_k)+n\left(t\right)----\left(1\right)$

B wherein _k(i) { 1,1} is k user's an i bit (or being called data symbol) to ∈, x _k ⁱ(t) be the baseband receiving signals of k user's i bit, τ _kBe the transmission lag time of k subscriber signal, Es is the energy of each symbol, and n (t) is that power spectral density is the additive Gaussian noise of N, and Tc is the chip period of spreading code.According to multipath-Doppler's canonical (multipath-Doppler canonical representation) expression of the radio propagation channel that provides in the document [1], the baseband receiving signals x of k user's i bit in above-mentioned (1) formula _k ⁱCan be expressed as: $x_k^i\left(t\right)=\frac{T_c}{T}\underset{l=0}{\overset{L}{\mathrm{\&Sigma;}}}\underset{m=-M}{\overset{M}{\mathrm{\&Sigma;}}}{H}_k^{m,l}\left(i\right)s_k(t-\mathrm{iT}-l{T}_c){\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="A0113442800201.png"\; state="\{\{state\}\}">e</figure-callout>}}^j----\left(2\right)$

H wherein _k ^{M, l}(i) be channel coefficients (by estimating that with conventional channel similar T/F channel estimating obtains) corresponding to k user's i symbol, total (L+1) is (2M+1) individual, and s _k(t) be k user's frequency spreading wave, satisfy $s_k\left(t\right)=0\mathrm{ift}\&NotElement;[0,T].$ In addition, in order to express easily, expression below introducing: $u_k^{m,l}\left(t\right)=s_k(t-{\mathrm{lT}}_c){\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="A0113442800201.png"\; state="\{\{state\}\}">e</figure-callout>}}^j,----\left(3\right)$

The frequency spreading wave s that represents k user _k(t) through m the Doppler branch in the footpath of the l behind multipath-Doppler's channel, or abbreviate as (m, l)-TF branch, $j=\sqrt{-1}.$

(2) do the TF matched filtering of T/F, and export the matched filtering result of each each TF branch of footpath of each user; The TF matched filtering operation of T/F is the multipath-Doppler's branch signal that utilizes the data symbol frequency spreading wave, carries out matched filtering to received signal.For k user's i symbol, its corresponding to (m, l)-the TF matched filtering of TF branch signal is output as: $z_k^{m,l}\left(i\right)\text{=}{\&Integral;}_{{\mathrm{\&tau;}}_{,}}^{T+{\mathrm{\&tau;}}_k}r\left(t\right)s_k^{*}(t-\mathrm{iT}-l{T}_c)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A0113442800201.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;m}(t-\mathrm{iT})}\mathrm{dt}.----\left(4\right)$

(3) all the TF matched filtering results to each user carry out the merging of TF RAKE high specific, are combined operating result then and carry out the symbol hard decision; TF RAKE high specific union operation is the result who utilizes each TF branch signal matched filtering, and the multipath of respective branches-Doppler's channel parameter, carries out the operation that high specific merges.Particularly, be to utilize TF matched filtering output result, the merging of TF RAKE high specific and the symbol hard decision result that obtain k user's i symbol are: $\widehat{b_k}\left(i\right)=\mathrm{sgn}\left\{\mathrm{Re}\left(\underset{l=0}{\overset{L}{\mathrm{\&Sigma;}}}\underset{m=-M}{\overset{M}{\mathrm{\&Sigma;}}}{H}_k^{m,l*}\left(i\right)z_k^{m,l}\left(i\right)\right)\right\},----\left(5\right)$

(4) on the basis that TF RAKE receives, further adopt parallel interference to eliminate the Multiuser Detection enhancement techniques of (PIC), further improve and detect performance; Its concrete operations step is as follows:

(a) the progression G of multistage parallel interference eliminated is set, a level counting number index is g, g=1 ..., G, and establish initial index g=1;

(b) utilize TF RAKE to receive symbol hard decision and the channel parameter that obtains, carry out the regenerative operation of each TF branch signal, promptly disturb regenerative operation, obtain the regenerated signal of each TF branch; The regenerative operation of a TF signal branch described here is each TF branch signal with each user's frequency spreading wave signal, multiply by that the value of this user's symbol hard decision at the corresponding levels obtains.The TF interference signal that obtains in each grade PIC generally all is inequality.

I symbol b corresponding to k user _k(i), its (m, l)-regenerated signal of TF Doppler signal branch is: ${\hat{s}}_k^{m,l}\left(t\right)=\sqrt{E_s}{u}_k^{m,l}(t-\mathrm{iT})H_k^{m,l}\left(i\right){\hat{b}}_k\left(i\right),t\&Element;[\mathrm{iT}+{\mathrm{lT}}_c,(i+1)T+{\mathrm{lT}}_c],----\left(6\right)$

Wherein, u _k ^{M, l}(t) as the formula (3), and  _k(i) obtain by the formula (5) of step (3).

Each TF branch signal of each user's frequency spreading wave is that multipath-Doppler's channel parameter of utilizing the TF channel estimating to obtain and this known user's frequency spreading wave obtain.This multipath-Doppler's channel parameter can be that each grade PIC is identical, promptly uses same TF channel estimation value, also can reappraise step by step to upgrade to use.

(c), carry out the interference cancellation operation to each TF branch signal of each user; Promptly, from baseband receiving signals, deduct the TF interference signal of this branch, obtain this purer branch signal for each TF branch signal of each user; The TF interference signal of this branch is meant: the regenerated signal of all the TF branch signals except this TF branch signal.

To k user (m, l)-TF branch, interference cancellation operation is: ${\hat{r}}_k^{m,l}\left(t\right)=r\left(t\right)-\underset{(\stackrel{\&OverBar;}{k},\stackrel{\&OverBar;}{m},\stackrel{\&OverBar;}{l})\&NotEqual;(k,m,l)}{\mathrm{\&Sigma;}}{\hat{s}}_{\stackrel{\&OverBar;}{k}}^{\stackrel{\&OverBar;}{m},\stackrel{\&OverBar;}{l}}(t-{\mathrm{\&tau;}}_{\stackrel{\&OverBar;}{k}}),----\left(7\right)$

Attention: herein interference cancellation operation in realization, can deduct earlier all TF branch signals regenerated signal and, add the regenerated signal of that TF branch of oneself

Like this, will significantly reduce operand;

(d) each purer TF branch signal is carried out the TF coupling again, obtain the TF matched filtering result of this branch; Same step (2) is operated in the TF matched filtering;

To carrying out the TF matched filtering again through the purer TF branch signal after the interference eliminated, obtain i symbol of k user corresponding to (m, l) the TF matched filtering of TF branch is output as: $z_k^{m,l}\left(i\right)={\&Integral;}_{\mathrm{iT}+{\mathrm{\&tau;}}_{k,l}}^{(i+1)T+{\mathrm{\&tau;}}_{k,l}}{\hat{r}}_k^{m,l}\left(t\right)u_k^{m,l*}(t-\mathrm{iT}-{\mathrm{lT}}_c)\mathrm{dt},----\left(8\right)$

τ wherein _{K, l}=τ _k+ lT _cIt is the time delay in k user's l footpath;

(e), carry out TF RAKE and merge, and carry out the symbol hard decision of PIC at the corresponding levels to the TF matched filtering result of each TF branch signal of each user; The same step of operation (3) that TF RAKE merges;

Above-mentioned union operation adopts simple T F RAKE high specific to merge, and the TF RAKE of k user's i symbol merges and the symbol hard decision is: ${\hat{b}}_k\left(i\right)=\mathrm{sgn}\left\{\mathrm{Re}\left(\underset{l=0}{\overset{L}{\mathrm{\&Sigma;}}}\underset{m=-M}{\overset{M}{\mathrm{\&Sigma;}}}{H}_k^{m,l*}\left(i\right)z_k^{m,l}\left(i\right)\right)\right\},----\left(9\right)$

Note z in the formula (9) _k ^Ml(i) and  _k(i) (if channel coefficients difference not at the same level then also has H _k ^{M, l}(i)), generally all be inequality in fact at each grade PIC;

(f) value with g increases by 1,

(g) utilize the symbol hard decision result of upper level, the operation of repeating step (b)～(f) is till finishing the operation that g=G level parallel interference eliminates; At last also will be with the value output of the symbol hard decision that obtains behind the afterbody g=G level PIC, as final data symbol detection result.

Another kind of situation of the present invention is that promptly the band spectrum modulation process satisfies T on TF model 2 times _sDuring the condition of＞T, at this moment, because symbols spread wave period T _sGreater than the data symbol period T, therefore can bring intersymbol interference ISI (Inter-Symbol Interference), can bring certain amount of calculation to increase in order to remove ISI, but T _sSuitably increase can maximally utilise Doppler's diversity of channel, overcomes quick decline to detecting the influence that performance causes.In order to remove ISI, receive when handling a detection window must be set, the size of this detection window can be made as a frequency spreading wave the twice of topped scope.In addition, for the detection window edge symbol being done more accurate estimation, the level and smooth amplitude of each detection window can be less than a detection window width, but also can obtain sign estimation more accurately to the testing result integrated treatment of adjacent detection window.In the following description, for simplicity, suppose that detection window smoothly is a detection window size, and, the testing result of each detection window directly exported.

In a word, satisfy T when the band spectrum modulation process _sDuring＞T condition, its concrete operating procedure of implementing is as follows:

(1) in the band spectrum modulation of signal, with data symbol frequency spreading wave duration T _sBe set to period T, i.e. T greater than data symbol _s＞T also establishes I=T _s/ T is a positive integer.

Still be modulated to example with relevant BPSK below, corresponding symbol is i=-I in the detection window of Kao Lving here ... I symbol, and i=0 the center that symbol is a detection window: $r\left(t\right)=\sqrt{E_s}\underset{i=-I}{\overset{I}{\mathrm{\&Sigma;}}}\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{b}_k\left(i\right)x_k^i(t-{\mathrm{\&tau;}}_k)+n\left(t\right),----{\left(1\right)}^{\&prime;}$

The baseband receiving signals x of k user's i bit in above-mentioned (1) ' formula ^l _k(t) can be expressed as: $x_k^i\left(t\right)=\frac{T_c}{T_s}\underset{l=0}{\overset{L}{\mathrm{\&Sigma;}}}\underset{m=-M}{\overset{M}{\mathrm{\&Sigma;}}}{H}_k^{m,l}\left(i\right)s_k(t-\mathrm{iT}-{\mathrm{lT}}_c){\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="A0113442800201.png"\; state="\{\{state\}\}">e</figure-callout>}}^j,----{\left(2\right)}^{\&prime;}$

The meaning of all the other each marks is identical with the explanation of (1) formula and (2) formula.In addition, in order to express easily, introduce following formula: $u_k^{m,l}\left(t\right)=s_k(t-l{T}_c){\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="A0113442800201.png"\; state="\{\{state\}\}">e</figure-callout>}}^j,----{\left(3\right)}^{\&prime;}$

The frequency spreading wave s that represents k user _k(t) through m the Doppler branch in the footpath of the l behind multipath-Doppler's channel, or (m, l)-TF branch.

(2) do the TF matched filtering of T/F, and export the matched filtering result of each TF branch in each footpath of each user; TF matched filtering herein can be provided with the symbol detection window size, so that the TF RAKE union operation of back more effectively overcomes intersymbol interference (ISI), but operand can increase to some extent.The TF matched filtering operation of T/F is the multipath-Doppler's branch signal that utilizes the data symbol frequency spreading wave, carries out matched filtering to received signal.

The detection window size is set is 2I+1 symbol (size of smoothing windows also can be taken as and is not more than 2I+1 symbol), at first level and smooth detection window makes the corresponding symbol index set of detection window { I....I}.

For k user's i ∈ I ... ,+I} 2I+1 symbol altogether, its corresponding to (m, l) the TF matched filtering of branch signal is output as: $z_k^{m,l}\left(i\right)={\&Integral;}_{-\mathrm{IT}}^{\mathrm{iT}+T_s}r\left(t\right)s_k^{*}(t-\mathrm{iT}-{\mathrm{lT}}_c)e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="A0113442800201.png"\; state="\{\{state\}\}">j</figure-callout>}2\mathrm{\&pi;m}(t-\mathrm{iT})/T_s}\mathrm{dt},----{\left(4\right)}^{\&prime;}$

(3) all the TF matched filtering results to each user carry out the merging of TF RAKE high specific, are combined operating result then and carry out the symbol hard decision; A user's TF RAKE high specific union operation, be earlier respectively all symbols in the detection window in each footpath of this user to be removed the intersymbol interference operation (note: the present invention directly is the operation of removing intersymbol interference respectively to each, Zhong Gejing compares with document [1] [2], greatly reduce computation complexity), the TF channel parameter that utilizes the TF channel estimating to obtain then, all are directly carried out high specific merge, the value after the TF RAKE that obtains all symbols in this user's detection window at last merges.Wherein, the operation that each directly removes intersymbol interference is meant that the TF matched filtering result to the vector form in this footpath multiply by the inverse matrix of TF correlation matrix.Since between each footpath and uncorrelated, the obtaining of this TF correlation matrix, the just result of the related operation of Dan Jing, so the dimension of matrix can reduce greatly, and operand also significantly reduces.

K=1 ..., k user's i=-I ,-I+1 ..., the TF RAKE high specific merging of I symbol is operating as with declaring firmly: ${\hat{b}}_k=\mathrm{sgn}\{\mathrm{Re}\left(\underset{l=0}{\overset{L}{\mathrm{\&Sigma;}}}{h}_{k,l}^H{F}_{k,l}{z}_{k,l}\right)\},----{\left(5\right)}^{\&prime;}$

Z wherein _{K, l}Be the output result who represents with matrix form of TF matched filter, F _{K, l}Be the matrix of intersymbol interference ISI, h _{K, l}Be multipath-Doppler's channel coefficients, their concrete definition is as follows respectively:

z _{K, l}=[z _{K, l}(-I) ..., z _{K, l}(I)] ^TBe (2I+1) (2M+1) dimensional vector, wherein $z_{k,l}\left(i\right)=[z_k^{-M,1}\left(i\right),...,z_k^{M,1}\left(i\right),...,z_k^{-M,L}\left(i\right),...,z_k^{M,L}\left(i\right){]}^T,$ And i=-I ..., I; $F_{k,l}=R_{k,l}^{-1}$ (2I+1) that be k user l footpath (2M+1) ties up TF correlation matrix R _kInverse matrix, play the effect of intersymbol interference ISI, wherein: $R_{k,l}=\left[\begin{array}{ccc}{R}_{k,l}^T(-I,-I)& ...& R_{k,l}^T(-I,I)\\ ...& ...& ...\\ R_{k,l}^T(I,-I)& ...& R_{k,l}^T(I,I)\end{array}\right],$ The transposition of the subscript T representing matrix in the formula, and right-I≤i, j≤I has $R_{k,l}(i,j)={\&Integral;}_{-\mathrm{IT}}^{\mathrm{IT}+T_s}{u}_k^{l^{*}}(t-\mathrm{iT})u_k^{l^T}(t-\mathrm{jT})\mathrm{dt},$ Again wherein $u_k^l\left(t\right)=[u_k^{-M,l}\left(t\right),...,u_k^{M,l}\left(t\right){]}^T$ Be the 2M+1 dimension Doppler vector in the l footpath of k user's frequency spreading wave, u _k ^{M, l}(t) definition see also formula (3) '; $h_{k,l}=[H_k^{-M,l}(-I),...,H_k^{M,l}(-I),...,H_k^{-M,l}\left(I\right),...,H_k^{M,l}\left(I\right){]}^T$ It is TF channel estimating vector.

The present invention can adopt data symbol hard decision result in (5) ' formula as the last output of receiver, also can be on the basis that above TF RAKE receives, adopt parallel interference to eliminate the Multiuser Detection enhancement techniques of (PIC) again, can further improve the detection performance; Certainly, this step is to belong to selectable content of operation.Its concrete operations step is as follows:

(a) the progression G of multistage parallel interference eliminated is set, a level counting number index is g, g=1 ..., G, and establish g=1;

(b) utilize TF RAKE to receive symbol hard decision value and the channel parameter that obtains, carry out the regenerative operation of each TF branch signal, promptly disturb regenerative operation, obtain the regenerated signal of each TF branch; Disturb regenerative operation to carry out at all symbols in the detection window.The regenerative operation of a TF signal branch is each TF branch signal with each user's frequency spreading wave signal, multiply by that the value of this user's symbol hard decision at the corresponding levels obtains.The TF interference signal that obtains in each grade PIC generally all is inequality.

Corresponding to k user's i=-I ..., I is 2I+1 the symbol that detection window is interior altogether, its (m, l)-regenerated signal of TF branch signal is: ${\hat{s}}_k^{m,l}\left(t\right)=\underset{i=-I}{\overset{I}{\mathrm{\&Sigma;}}}\sqrt{E_b}{u}_k^{m,l}(t-\mathrm{iT})H_k^{m,l}\left(i\right){\hat{b}}_k\left(i\right),t\&Element;[-\mathrm{IT},\mathrm{IT}+T_s],----{\left(6\right)}^{\&prime;}$

Wherein, u _k ^Ml(t) suc as formula (3) ' shown in, and  _k(i) by the formula (5) of step (3) ' obtain.Note, in the interference elimination treatment process of g at the same level not, TF channel parameter H _k ^Ml(t), symbol judgement value  _k(i) etc. generally be inequality.

Each TF branch signal of each user's frequency spreading wave is that multipath-Doppler's channel parameter of utilizing the TF channel estimating to obtain and this known user's frequency spreading wave obtain.This multipath-Doppler's channel parameter is can each grade identical, promptly uses same TF channel estimation value, also can reappraise step by step to upgrade to use.

(c), carry out the interference cancellation operation for each TF branch signal of each user.Promptly, from baseband receiving signals, deduct the TF interference signal of this branch, promptly carry out interference cancellation, obtain this purer branch signal for each TF branch signal of each user; The TF interference signal of this branch is meant, the regenerated signal of all the TF branch signals except this TF branch signal.

To k user (m, l)-TF branch, interference cancellation operation is: ${\hat{r}}_k^{m,l}\left(t\right)=r\left(t\right)-\underset{(\stackrel{\&OverBar;}{k},\stackrel{\&OverBar;}{m},\stackrel{\&OverBar;}{l})\&NotEqual;(k,m,l)}{\mathrm{\&Sigma;}}{\hat{s}}_{\stackrel{\&OverBar;}{k}}^{\stackrel{\&OverBar;}{m},\stackrel{\&OverBar;}{l}}(t-{\mathrm{\&tau;}}_{\stackrel{\&OverBar;}{k}}),----{\left(7\right)}^{\&prime;}$

H wherein _k ^{M, l}(i) be that the TF channel estimating obtains, can each grade identical, also can estimate respectively in each level.

Attention: interference cancellation operation herein in the realization of real system, can deduct earlier all regenerated signals and, add oneself that TF branch regenerated signal Will significantly reduce operand like this.

(d) each purer TF branch signal is carried out the TF matched filtering again, obtain the TF matched filtering result of this branch; The same step of the operation of TF matched filtering (2);

To carry out the TF coupling again through the purer signal after the interference eliminated, obtain k user i=-I ..., common 2I+1 the symbol of I corresponding to (m, l)-the TF matched filtering of TF branch is output as: $z_k^{m,l}\left(i\right)={\&Integral;}_{\mathrm{iT}+{\mathrm{\&tau;}}_{k,l}}^{(i+1)T+{\mathrm{\&tau;}}_{k,l}}{\hat{r}}_k^{m,l}\left(t\right)u_k^{m,l*}(t-\mathrm{iT}-{\mathrm{lT}}_c)\mathrm{dt},----{\left(8\right)}^{\&prime;}$

τ wherein _{K, l}=τ _k+ lT _cIt is the time delay in k user's the 1st footpath.

(e), carry out TF RAKE and merge, and carry out the symbol hard decision of PIC at the corresponding levels to the TF matched filtering result of each TF branch signal of each user; Wherein the operation of TF RAKE merging is identical with step (3), is the TF RAKE union operation of low complex degree.

On TF model 2 times, need to remove ISI, therefore adopt the TF RAKE method of reseptance of the TF model 2 of low complex degree, concrete steps are identical with above-mentioned step (3), so obtain the symbol hard decision of the corresponding levels: ${\hat{b}}_k=\mathrm{sgn}\left\{\mathrm{Re}\left(\underset{l=0}{\overset{L}{\mathrm{\&Sigma;}}}{h}_{k,l}^H{F}_{k,l}{z}_{k,l}\right)\right\},----{\left(9\right)}^{\&prime;}$

Note matched filtering output vector z wherein _{K, l}With symbol transient state hard decision value  _k(i) (if TF channel parameter difference not at the same level then also includes variable h _{K, l}) in fact generally all inequality at each grade PIC.In addition, attention the court verdict  here _kBe temporary intermediate object program, be used for following interference regeneration and elimination, rather than final detection result.

(f) value with g increases by 1;

(g) utilize the symbol of upper level to declare the result firmly, the operation of repeating step (b)-(f) is till the operation of finishing the elimination of g=G level parallel interference.At last also will be with the data symbol value of the declaring output firmly that obtains behind the afterbody g=G level PIC, as final symbol detection result.

The present invention is the emulation enforcement test that TF model 1 and TF model have all carried out multi-user detector for 2 times at two kinds of TF models, and in all l-G simulation tests, mobile station speed all is set at 120km/h.

Referring to Fig. 1 and Fig. 2, it is based on two kinds of error rate BER result schematic diagrams in conjunction with PIC Multiuser Detection and no PIC that TF model 1 provides.As seen from the figure, there is the TF multi-user detector performance of PIC to be better than not having the TF detector performance of PIC, although it is also fine not have TF multi-user's receptivity of PIC.

Referring to Fig. 3 and Fig. 4, this two figure has provided the simulation result under TF model 2.Can see: PIC still has preferable performance to improve.

In addition, please note: under the Multiuser Detection condition, the testing result under the TF model 2 is not as TF model 1, and this result is different with single user's TF method of reseptance.Receive situation at single user TF, the detection performance under two kinds of TF models almost is the same.

Claims (13)

1, a kind of method of reseptance based on the T/F diversity, it is characterized in that: this method of reseptance includes the following step:

(1) in the band spectrum modulation of signal, with data symbol frequency spreading wave duration T _sBe set to equal the period T of data symbol, i.e. T _s=T;

(2) utilize multipath-Doppler's branch signal of data symbol frequency spreading wave, do TF (Time-Frequency) matched filtering of T/F to received signal, and export the matched filtering result of each each TF branch of footpath of each user;

(3) all the TF matched filtering results to each user carry out the merging of TF RAKE high specific, are combined operating result then and carry out the symbol hard decision;

(4) on the basis that TF RAKE receives, further adopt parallel interference to eliminate the Multiuser Detection enhancement techniques of (PIC), detect performance so that further improve.

2, the method for reseptance based on the T/F diversity according to claim 1, it is characterized in that: the TF RAKE high specific union operation that described (3) step is carried out, be the result who utilizes each TF branch signal matched filtering, and the multipath of respective branches-Doppler's channel parameter, carry out the operation that high specific merges.

3, the method for reseptance based on the T/F diversity according to claim 1 is characterized in that: described (4) step is carried out the Multiuser Detection enhancement techniques operation that parallel interference is eliminated (PIC), and its concrete steps are as follows:

(a) the progression G of multistage parallel interference eliminated is set, a level counting number index is g, g=1 ..., G, and establish initial index g=1;

(b) utilize TF RAKE to receive symbol hard decision and the channel parameter that obtains, carry out the regenerative operation of each TF branch signal, promptly disturb regenerative operation, obtain the regenerated signal of each TF branch;

(c), carry out the interference cancellation operation to each TF branch signal of each user;

(d) each purer TF branch signal is carried out the TF coupling again, obtain the TF matched filtering result of this branch; The operation of this TF matched filtering is identical with described step (2);

(e), carry out TF RAKE and merge, and carry out the symbol hard decision of PIC at the corresponding levels to the TF matched filtering result of each TF branch signal of each user; The operation that this TF RAKE merges is identical with described step (3);

(f) value with g increases by 1,

(g) utilize the symbol hard decision result of upper level, the operation of repeating step (b)～(f), till the operation of finishing the elimination of g=G level parallel interference, at last also will be with carrying out the value output that afterbody g=G level parallel interference is eliminated the symbol hard decision that obtains after the operation, as final data symbol detection result.

4, the method for reseptance based on the T/F diversity according to claim 3, it is characterized in that: the regenerative operation of each TF branch signal in described (b) step, be each TF branch signal, multiply by the value of this user's symbol hard decision at the corresponding levels each user's frequency spreading wave signal; And each TF branch signal of each user's frequency spreading wave then is that multipath-Doppler's channel parameter of utilizing the TF channel estimating to obtain and this known user's frequency spreading wave obtain.

5, the method for reseptance based on the T/F diversity according to claim 4, it is characterized in that: the multipath that the described TF of utilization channel estimating obtains-Doppler's channel parameter, can be that each grade PIC is identical, promptly use same TF channel estimation value, also can reappraise step by step to upgrade and use.

6, the method for reseptance based on the T/F diversity according to claim 3, it is characterized in that: the interference cancellation operation in described (c) step is each TF branch signal for each user, from baseband receiving signals, deduct the TF interference signal of this branch, obtain this purer branch signal, the TF interference signal of this branch is meant: the regenerated signal of all the TF branch signals except this TF branch signal.

7, the method for reseptance based on the T/F diversity according to claim 1 is characterized in that: in described (4) step, also can adopt the Multiuser Detection enhancement techniques of the soft interference cancellation algorithm of iteration SISO, detect performance so that further improve.

8, a kind of method of reseptance based on the T/F diversity, it is characterized in that: this method of reseptance includes the following step at least:

(1) in the band spectrum modulation of signal, with data symbol frequency spreading wave duration T _sBe set to period T, i.e. T greater than data symbol _s＞T, and establish I=T _s/ T is a positive integer;

(2) utilize multipath-Doppler's branch signal of data symbol frequency spreading wave, do the TF matched filtering of T/F to received signal, and export the matched filtering result of each each TF branch of footpath of each user; TF matched filtering herein can be provided with the symbol detection window size, so that the TF RAKE union operation of back more effectively overcomes intersymbol interference (ISI);

(3) all the TF matched filtering results to each user carry out the merging of TF RAKE high specific, are combined operating result then and carry out the symbol hard decision; Each user's here TF RAKE high specific union operation, it is the operation of respectively all symbols in the detection window in each footpath of this user being removed intersymbol interference earlier, the TF channel parameter that utilizes the TF channel estimating to obtain then, the high specific union operation is carried out in all each footpath, the value after the high specific that obtains all symbols in this user's detection window at last merges.

9, the method for reseptance based on the T/F diversity according to claim 8, it is characterized in that: the operation of each footpath intersymbol interference of the removal in described (3) step, be meant that the TF matched filtering result to the vector form in this footpath multiply by the inverse matrix of TF correlation matrix, and this TF correlation matrix is the result that the related operation by Dan Jing obtains, the dimension of this matrix reduces greatly, can significantly reduce operand.

10, the method for reseptance based on the T/F diversity according to claim 8, it is characterized in that: after (3) step, can optionally increase following operation: on the basis that TF RAKE receives, further adopt parallel interference to eliminate the Multiuser Detection enhancement techniques of (PIC) or the soft interference cancellation algorithm of iteration SISO, detect performance so that further improve.

11, the method for reseptance based on the T/F diversity according to claim 10 is characterized in that: described parallel interference is eliminated the Multiuser Detection enhancement techniques operation of (PIC), and its concrete steps are as follows:

(a) the progression G of multistage parallel interference eliminated is set, a level counting number index is g, g=1 ..., G, and establish initial index g=1;

(b) utilize TF RAKE to receive symbol hard decision and the channel parameter that obtains, carry out the regenerative operation of each TF branch signal, promptly disturb regenerative operation, obtain the regenerated signal of each TF branch, and this interference regenerative operation is to carry out at all symbols in the detection window;

(c), carry out the interference cancellation operation to each TF branch signal of each user;

(d) each purer TF branch signal is carried out the TF coupling again, obtain the TF matched filtering result of this branch; The operation of this TF matched filtering is identical with described step (2);

(e), carry out TF RAKE and merge, and carry out the symbol hard decision of PIC at the corresponding levels to the TF matched filtering result of each TF branch signal of each user; The operation that this TF RAKE merges is identical with described step (3);

(f) value with g increases by 1,

(g) utilize the symbol hard decision result of upper level, the operation of repeating step (b)～(f) is till finishing the operation that g=G level parallel interference eliminates; At last also will be with carrying out the value output that afterbody g=G level parallel interference is eliminated the symbol hard decision that obtains after the operation, as final data symbol detection result.

12, the method for reseptance based on the T/F diversity according to claim 11, it is characterized in that: the regenerative operation of each TF branch signal in described (b) step, be the value that each TF branch signal of each user's frequency spreading wave signal be multiply by this user's symbol hard decision at the corresponding levels, at last to all symbols summations in the detection window; And each TF branch signal of each user's frequency spreading wave is that multipath-Doppler's channel parameter of utilizing the TF channel estimating to obtain and this known user's frequency spreading wave obtain.

13, the method for reseptance based on the T/F diversity according to claim 11, it is characterized in that: the interference cancellation operation in described (c) step is each TF branch signal for each user, from baseband receiving signals, deduct the TF interference signal of this branch, obtain this purer branch signal, this operation is between the operating space with whole detection window; The TF interference signal of this branch signal is meant: the regenerated signal of all the TF branch signals except this TF branch signal.

Images