CN1194478C - Multi-user test method for radio receiver - Google Patents

Abstract

The present invention relates to a multi-user detection method for a radio receiver which is used for a simplified iteration interference cancellation receiver of a code division multiple access (CDMA) system. The method can effectively suppress multi-user interference and raise system capacity; especially, under the condition of combining with decoding, the method can have better effect. The method comprises: a, at least one soft input soft output (SISO) module and a conversion module are set corresponding to every user; b, an iteration frequency M and an iteration frequency parameter n are set; c, n is set equal to one, the SISO module carries out symbol judgment to result signals combined and output by an RAKE receiver; d, according to the result of the symbol judgment, parallel interference cancellation is carried out to input receiving signals after discretization, and a likelihood ratio is output. E, one is added to n, the conversion module converts the result of the symbol judgment output by the SISO module and then feeds back the result to input in the SISO module. Afterwards, the steps d, e are executed until n is equal to M.

Description

A kind of multi-user test method of wireless receiver

Technical field

The present invention relates to radio communication reception technique field, relate to a kind of multi-user test method in the receiver for spread spectrum communication or rather.

Background technology

After the Turbo code of superior performance is found, iterate treatment technology and extensively caused attention, as seen list of references 1[C.Berrou and A.Glavieux, " Near optimum erro-correcting coding anddecoding:Turbo codes; " IEEE Trans Commun., Vol.44, Oct.1996] and list of references 2[C.Berrou and A.Glavieux, and P.Thitimajshima, " Near Shannon limiterrorcorrection coding and decoding:Turbo codes; " on Proc.1993 Int.Conf.onCommunica-tions, PP.1064-1070,1993.].

Turbo code can be widely used in many detection/decoding problems in principle, for example the Multiuser Detection of serial concatenation of codes, combined signal source and channel-decoding, combined channel decoding etc.At list of references 3[M.Moher, " Aniterative multiuser decoder for near-capacity communications; " IEEETrans.Commun., vol.46, pp.870-880, July 1998.] and list of references 4[M.C.Valenti andB.D.Woerner, " Iterative multiuser detection for convolutionally coded asynchronousDS-CDMA; " in Proc.IEEE Int.Symp.Personal, Indoor, and Mobile Radio Communica-tions (PIMRC), Boston, MA, pp.213-217., 1998] in, obtain a kind of multiuser detection algorithm that iterates at synchronization code multi-address division system of optimum, this algorithm that iterates carries out based on the minimum principle of mutual entropy (Cross-entropy).

Though also obtained the implementation method of corresponding suboptimum, the complexity of its realization remains the user of moderate quatity and is difficult to accept.At list of references 5[H.E.Gamal and E.Geraniotis, " Iterativemultiuser detection for coded CDMA signals in AWGN and fading channels; " IEEETrans.on Selected Areas in Communications, vol.18, no.1, pp.30-41, Jan.2000.] in, provided the receiver that iterates, but the very high problem of computation complexity has been arranged equally based on MMSE criterion and Turbo criterion.

At list of references 6[X.Wang, and H.V.Poor, " Iterative (Turbo) soft inter-ferencecancellation and decoding for coded CDMA ", IEEE Trans.Commun., vol.47, No.7, pp.1046-1061, July 1999] in, a kind of new receiver that iterates has been proposed, this receiver is based on soft inputting and soft output (SISO) and posterior probability channel decoding.Its problem that iterates the algorithm existence comprises: be not to adopt the output of Rake receiver as the input that iterates for the first time; Constant channel when being applied to can not be applied to time varying channel; Adopt complicated matrix inversion technique, and be not that each of multipath is directly handled.

Summary of the invention

The objective of the invention is to design a kind of multi-user test method of wireless receiver, it is a kind of soft interference cancelling method that iterates that is applied to wireless receiver, compare with same class methods in the past, can reduce the complexity of calculating greatly, and can be applicable to chnnel coding simultaneously and not have to use under two kinds of situations of chnnel coding.

The technical scheme that realizes the object of the invention is such: a kind of multi-user test method of wireless receiver is characterized in that comprising following treatment step:

A. corresponding each user is provided with soft inputting and soft output (SISO) module and a conversion module;

B., iterative times M and iterative times parameter n are set, and M, n are positive integer;

C. establish n and equal 1, the consequential signal that merges output through RAKE receiver is carried out symbol judgement by soft inputting and soft output (SISO) module;

D. according to the symbol judgement result received signal after discretization of input is carried out Parallel Interference Cancellation and exported likelihood ratio;

E. make n increase by 1, the symbol judgement result who soft inputting and soft output (SISO) module is exported by the conversion module of soft inputting and soft output (SISO) module carries out feeding back input soft inputting and soft output (SISO) module again after the conversion, execution in step D, E stop when n=M then.

Described steps A, be that a corresponding K user is provided with K soft inputting and soft output (SISO) module and K conversion module, each conversion module is arranged between the feedback input end and extrinsic information output of corresponding soft inputting and soft output (SISO) module, what be used to handle combined channel coding and decoding not iterates soft inputting and soft output, exports described likelihood ratio by soft inputting and soft output (SISO) module.

Described steps A, be that a corresponding K user is provided with K soft inputting and soft output (SISO) module, a K de-interleaving block, a K soft input soft output decode module, a K interleaving block and K conversion module, between the feedback input end that each conversion module is arranged on corresponding soft inputting and soft output (SISO) module and the extrinsic information output of corresponding interleaving block, what be used to handle the combined channel coding and decoding iterates soft inputting and soft output, exports described likelihood ratio by the soft input soft output decode module.

Described step C is the symbol manipulation of result's value of peek after according to following formula RAKE being merged, the r in the formula _{K, l}(i) be l footpath signal in the L footpath signal of k user among K the user who obtains at i symbol, h _{K, l}(i) be the channel estimation value of i symbol,

${\tilde{b}}_k\left(i\right)=\mathrm{sgn}\left(\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{r}_{k,l}\left(i\right)h_{k,l}^{*}\left(i\right)\right).$

Described step e is to the symbol judgement of soft inputting and soft output (SISO) module output λ as a result according to following formula not uniting when coding and decoding iterates ₁[b _k(i)] carry out conversion,

${\tilde{b}}_k\left(i\right)=\tanh \left(\frac{1}{2}{\mathrm{\λ}}_1\right[b_k\left(i\right)\left]\right).$

Described step e is to the symbol judgement of soft inputting and soft output (SISO) module output λ as a result when iterating according to following formula in the associating coding and decoding ^p ₂[b _k(i)] carry out conversion,

${\tilde{b}}_k\left(i\right)=\tanh \left(\frac{1}{2}{\mathrm{\λ}}_2^p\right[b_k\left(i\right)\left]\right).$

Described λ ^p ₂[b _k(i)] be the soft input information λ that deducts described soft input soft output decode device from the soft output of described soft input soft output decode device ₁[b _k(i)] the likelihood ratio after.

Among the described step D received signal after discretization of input is carried out Parallel Interference Cancellation, it is the non-l footpath signal that removes all K user, the l footpath signal that promptly keeps all K user, and then remove and remove k l footpath signal with other outdoor user, promptly only keep k user's l footpath signal, the processing formula is:

$r_{k,l}\left(i\right)=r\left(i\right)-\underset{\mathrm{la}=1,\mathrm{la}\≠l}{\overset{L}{\mathrm{\Σ}}}{H}_{\mathrm{la}}\left(i\right)\tilde{b}\left(i\right)-H_l\left(i\right){\tilde{b}}_k\left(i\right)$

Wherein, $\tilde{b}\left(i\right)=[{\tilde{b}}_1\left(i\right),{\tilde{b}}_2\left(i\right),...,{\tilde{b}}_K\left(i\right)],{\tilde{b}}_k\left(i\right)$ Represent that k component is changed to zero

H _l(i) be channel state matrix;

Output likelihood ratio among the described step D further comprises following treatment step:

A. calculate the weighted value of k user l footpath signal earlier, the h in the formula _{K, l}(i) be described channel state matrix H _l(i) the k columns certificate in, σ is the background noise variance,

$w_{k,l}\left(i\right)=\frac{h_{k,l}\left(i\right)}{{\mathrm{\σ}}^2+{\left|\right|h_{k,l}\left(i\right)\left|\right|}^2};$

B. the addition of above-mentioned two formulas being multiplied each other respectively obtains

$z_k\left(i\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{w}_{k,l}\left(i\right)r_{k,l}\left(i\right);$

C. utilize channel state matrix H _l(i) carry out delivery again and calculate,

${\mathrm{\μ}}_{k,l}\left(i\right)=\frac{{\left|\right|h_{k,l}\left(i\right)\left|\right|}^2}{{\mathrm{\σ}}^2+{\left|\right|h_{k,l}\left|\right|}^2}$

D. utilize three formula of step b.c. to calculate likelihood ratio λ at last ₁[b _k(i)], Re is realistic computing in the formula,

${\mathrm{\λ}}_1\left[b_k\left(i\right)\right]=\frac{2\mathrm{Re}\left\{{\mathrm{\μ}}_k\left(i\right)z_k\left(i\right)\right\}}{\underset{l=1}{\overset{L}{\mathrm{\Σ}}}[{\mathrm{\μ}}_{k,l}\left(i\right)-{\mathrm{\μ}}_{k,l}{\left(i\right)}^2]}.$

Described background noise variances sigma adopts traditional mean variance method to estimate; Described channel state matrix H _l(i) be that channel impulse response is under the identical condition, to estimate by traditional channel estimation methods in hypothesis symbol i cycle.

Described background noise variances sigma can be set to comprise a constant of zero.

The method of iterating that simplification of the present invention iterates soft interference cancelling method and list of references 6 has difference in essence, is in particular in:

Method of the present invention adopts the output of Rake receiver as the input that iterates for the first time;

Method of the present invention can be applied to time varying channel, rather than constant channel when only being applied to;

Method of the present invention is handled each footpath signal of multipath signal, has therefore adopted different interference cancellation methods and different extrinsic information (extrinsic information) formula;

From implementation complexity, the inventive method is compared with traditional PIC method only a small amount of increase.

Description of drawings

Fig. 1 is in the code division multiple access system, multiple user signals transmission flow schematic diagram.

Fig. 2 be the present invention do not unite coding and decoding iterate soft inputting and soft output (SISO) method flow block diagram.

Fig. 3 is that the soft inputting and soft that iterates of combined channel decoding of the present invention is exported (SISO) method flow block diagram.

Coding schematic flow sheet when Fig. 4 is transmission.

Adopt RAKE to receive when Fig. 5 is no chnnel coding and mix signal to noise ratio (snr), the error rate (BER) the performance comparative graph that iterates method of reseptance with Multiuser Detection of the present invention.

Adopt RAKE to receive when Fig. 6 is no chnnel coding and mix number of users and the error rate (BER) the performance comparative graph that iterates method of reseptance with Multiuser Detection of the present invention.

Fig. 7 adopts Multiuser Detection of the present invention to mix signal to noise ratio (snr), the error rate (BER) the performance comparative graph that iterates method of reseptance when being the combined channel coding.

Embodiment

Also further specify technical scheme of the present invention in conjunction with the accompanying drawings below by embodiment.

Referring to Fig. 1, illustrate code division multiple access system signal transmission flow among the figure.User's bit signal b of K user ₁, b ₂... b _kCorresponding respectively 1 to K sign map module (Symbol Mapper) 1, the 2...3 of importing, through 0,1 signal being become after the sign map signal after the modulation, correspondence is sent into spread spectrum module (Spreading) 4,5...6 again, become the signal behind the spread spectrum, K user's spread-spectrum signal sends after the adder addition, and sending signal is r (t) through the signal that fading channel 7 (Fading Channel) arrives receiving terminal 8.

The technical field that the inventive method relates to is code division multiple access (CDMA) system, when sending subscriber signal, must " 0 ", " 1 " signal transformation be become the signal after the modulation through the sign map module.Modulation system can be varied, for example exactly " 0 " is mapped to-1 when adopting BPSK (biphase phase shift keying) modulation, " 1 " is mapped to+1; And then through spread spectrum module, become the signal behind the spread spectrum.Suppose that spreading factor is N, i.e. the signal of preceding 1 bit of spread spectrum is through then becoming a frequency spreading wave of N chip behind the spread spectrum.

Utilize mathematical formulae to represent that the signal of k user's transmission in [iT, i (the T+1)] time period among K the user is:

$\underset{j=0,...,N-1}{\mathrm{\Σ}}{A}_k\left(i\right)b_k\left(i\right)c_k\left(j\right)\mathrm{\ψ}(t-\mathrm{iT}-\mathrm{jTc}).$

Wherein, A _k(i) be k user's energy factors, express the difference of user's energy, b _k(i) be the later bit of modulation that k user sends at moment i, for example under the situation of BPSK modulation be+1 or-1, c _k(j) be spread spectrum code sequence, j=0 ..., N-1, ψ (t-iT-jT _c) be the waveform of a chip after the normalization, wherein T _cBe the time span of a chip, the time span of T is-symbol, T _c=T/N, i wherein, j represent j chip of i symbol.Make s _k(t)=∑ _{J=0 ..., N-1}c _k(j) ψ (t-jT _c), then it is to be used for k user's bit is carried out modulated waveform.In each user's spread spectrum module 4,5...6, utilize different spreading codes to carry out spread spectrum, addition sends then, in the transmission through fading channel 7.

Each subscriber signal all passes through chnnel coding (41) and interweave (42) as last sending, and its flow process as shown in Figure 4; When the multi-user test method of receiver is not considered to iterate mutually with channel decoding, then can not consider chnnel coding.Thereby will produce two kinds of methods for technical scheme of the present invention: a kind of be not with the method for reseptance that iterates of channel decoding associating, flow process is as shown in Figure 2; Another is and the method for reseptance that iterates of channel decoding associating that flow process as shown in Figure 3.

Referring to Fig. 2, the signal model of the signal that RAKE receiver 20 receives is r (i).In Fig. 1 explanation, k user's signal is through fading channel, and signal can be subjected to the change of two aspects: one is the influence that is subjected to multipath effect, has generated L waveform by a waveform; Another one is that waveform itself has been subjected to distortion, and therefore the waveform of j chip of i bit can be expressed as through having become the stack of a plurality of waveforms behind the channel:

$\underset{l=1,...,L}{\mathrm{\Σ}}{g}_{k,l}\left(t\right)\mathrm{\ψ}(t-\mathrm{iT}-j{T}_c-{\mathrm{\τ}}_{k,l})$

G wherein _{K, l}(t) be a plural number, the impulse response (waveform distortion) of expression time varying channel, τ _{K, l}The expression delayed action, the signal of the user k that then receives is:

$y_k\left(t\right)=\underset{i}{\mathrm{\Σ}}{A}_k\left(i\right)b_k\left(i\right)\underset{j=0,...,N-1}{\mathrm{\Σ}}{c}_k\left(j\right)\underset{t=1,...,L}{\mathrm{\Σ}}{g}_{k,l}\left(t\right)\mathrm{\ψ}(t-\mathrm{iT}-j{T}_c-{\mathrm{\τ}}_{k,l})$

Therefore, the signal at receiver side user k is:

r(t)＝∑ _ky _k(t)+σn(t)

Wherein, n (t) is that average is zero, variance is the standard Gaussian noise of l, and σ represents the variance of background noise.

Among Fig. 2, what import RAKE receiver should be the discrete form of received signal, and promptly r (i) the following describes how to obtain r (i) from r (t).At first channel impulse response is carried out the chip-level matched filtering, filter function is ψ (t), is { f after the signal pulse response process chip-level matched filtering in k user l footpath then _{K, l}(m) }, be formulated as:

∫g _k，l(t)ψ(t-iT-jT _c-τ _k，l)ψ(t-mT _c)dt

When having only t within the specific limits, following formula is just non-vanishing.

Because to different chips, therefore the value difference of spreading code, considers spreading code, the chip-level matched filtering output of each chip waveform is:

{h _k，l(m)}＝A _k{c _k(m)}·{f _k，l(m)}，l＝1，2，...，L

Channel status can be organized into matrix form, i.e. channel state matrix H _l(i):

$H_{l\left(i\right)}=\left[\begin{array}{cccc}{\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="C0112049500181.png,C0112049500231.png"\; state="\{\{state\}\}">h</figure-callout>}}_{1,l}\left(\mathrm{iN}\right)& {\mathrm{<figure-callout\; id="2"\; label="h"\; filenames="C0112049500181.png,C0112049500211.png"\; state="\{\{state\}\}">h</figure-callout>}}_{2,l}\left(\mathrm{iN}\right)& \&CenterDot;\&CenterDot;\&CenterDot;& h_{K,l}\left(\mathrm{iN}\right)\\ {\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="C0112049500181.png,C0112049500231.png"\; state="\{\{state\}\}">h</figure-callout>}}_{1,l}(\mathrm{iN}+1)& {\mathrm{<figure-callout\; id="2"\; label="h"\; filenames="C0112049500181.png,C0112049500211.png"\; state="\{\{state\}\}">h</figure-callout>}}_{2,l}(\mathrm{iN}+1)& \&CenterDot;\&CenterDot;\&CenterDot;& h_{K,l}(\mathrm{iN}+1)\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ {\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="C0112049500181.png,C0112049500231.png"\; state="\{\{state\}\}">h</figure-callout>}}_{1,l}(\mathrm{iN}+N-1)& {\mathrm{<figure-callout\; id="2"\; label="h"\; filenames="C0112049500181.png,C0112049500211.png"\; state="\{\{state\}\}">h</figure-callout>}}_{2,l}(\mathrm{iN}+N-1)& \&CenterDot;\&CenterDot;\&CenterDot;& h_{K,l}(\mathrm{iN}+N-1)\end{array}\right]$

In general this matrix can not accurately obtain, but can estimate by traditional channel estimation methods, for example can adopt the WMSA channel estimation methods to estimate.

Utilize this channel state matrix H _l(i), we can be expressed as the signal r (t) that receives following discrete form r (i).

After all signals in l footpath have passed through the matched filtering of chip-level, for:

$r_{n,l}\left(i\right)={\&Integral;}_{\mathrm{iT}+n{T}_c}^{\mathrm{iT}+(n-1)T_c}{r}_l\left(t\right)\mathrm{\&psi;}(t-\mathrm{iT}-n{T}_c)\mathrm{dt},0\&le;n<N.$

Wherein, r _l(t) be all users l footpath signal and, total N the chip of l footpath signal in a symbol period can be expressed as vector form:

According to the signal that sends and the channel of process thereof, arbitrary user's of reception signal can be expressed as the product of this subscriber signal of chip-level channel matrix and transmission and add background noise, promptly is expressed as:

r _l(i)＝H _l(i)b(i)+σn _l(i)，l＝1，2，...，L.

Then the result of the chip-level matched filtering of this subscriber signal of receiving at receiver is:

With under the situation about combining of encoding do not iterating the handling process that soft inputting and soft is exported (IterativeSISO) method below in conjunction with Fig. 2 explanation.

Receiver is except that former RAKE receiving unit 20, also set up soft inputting and soft output module 211 respectively for K user, 212, ... 21K, with be connected each soft inputting and soft output module 211,212, ... the conversion module 221 between 21K input and output, 222, ... 22K, received signal r after discretization (i) input RAKE receiving unit 20 is also imported each soft inputting and soft output module 211 simultaneously, 212, ... the first input end of 21K, K subscriber signal of RAKE receiving unit 20 outputs imported each soft inputting and soft output module 211,212, ... second input of 21K, by K conversion module 221,222, ... corresponding respectively each soft inputting and soft output module 211 of input of signal after the conversion of 22K feedback, 212, ... the 3rd input of 21K, K soft inputting and soft output module 211,212, ... the corresponding K of extrinsic information (Extrinsic Information) feedback input conversion module 221 of 21K output, 222, ... 22K, by K soft inputting and soft output module 211,212, ... 21K output information bit (Information bits).

Utilize user's spreading code to carry out a yard matched filtering by RAKE receiving unit 20,, obtain the signal r in k user's l footpath for i symbol _{K, l}(i), utilize the channel estimating h of i the symbol that obtains _{K, l}(i) carry out Rake and merge, channel estimating can utilize common channel estimation methods to obtain, for example the WMSA method.

Following steps are finished by soft inputting and soft output module and conversion module.The length of the given signal r (i) after the chip-level matched filtering that receives is N, and given iterative times M and to establish the iterative times parameter be n, and n is since 1, and incremental change is 1, until n during for M end iterate.

Step 1 is estimated channel state matrix H by channel estimation methods _l(i).

Step 2, from n=1, result after merging through Rake is carried out symbol judgement (only just the result after merging through Rake being carried out symbol judgement under the situation at n=1), obtains the symbol judgement result of k user, an i symbol, l footpath received signal and channel estimating:

${\tilde{b}}_k\left(i\right)=\mathrm{sgn}\left(\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{r}_{k,l}\left(i\right)h_{k,l}^{*}\left(i\right)\right)$

Wherein, sgn represents the symbol (+1 or-1) of peek value.

If n=2,3 ... (promptly under the situation of n＞1), then carry out following computing:

${\tilde{b}}_k\left(i\right)=\tanh \left(\frac{1}{2}{\mathrm{\&lambda;}}_1\right[b_k\left(i\right)\left]\right)$

This step conversion is the function finished by the conversion among Fig. 2 (transform) module (promptly is not equal at 1 o'clock at n, SISO resume module be the transformation results of conversion module).Tan in the formula gets tangent (tangent):

$\tanh \left(x\right)=\frac{e^x-e^{-x}}{e^x+e^{-x}}$

Therefore for step 2, when n=1, be directly with the amalgamation result of Rake part 20 outputs, be input in the SISO module and handle, when n＞1, then to the λ as a result of SISO module output ₁[b _k(i)] carrying out conversion obtains

And then be input to the SISO module.

Step 3 comprises step a and b;

Step a carries out Parallel Interference Cancellation, removes all K user's non-l footpath signal, promptly keeps all K user's l footpath signal, and then removes and remove k the l footpath signal with other outdoor user, promptly only keeps k user's l footpath signal,

$r_{k,l}\left(i\right)=r\left(i\right)-\underset{\mathrm{la}=1,\mathrm{la}\&NotEqual;l}{\overset{L}{\mathrm{\&Sigma;}}}{H}_{\mathrm{la}}\left(i\right)\tilde{b}\left(i\right)-H_l\left(i\right){\tilde{b}}_k\left(i\right)---\left(1\right)$

Wherein, $\tilde{b}\left(i\right)=[{\tilde{b}}_1\left(i\right),{\tilde{b}}_2\left(i\right),...,{\tilde{b}}_K\left(i\right)],{\tilde{b}}_k\left(i\right)$ Represent that k component is changed to zero

Step b, output likelihood ratio,

Calculate the weighted value of k user l footpath signal earlier,

$w_{k,l}\left(i\right)=\frac{h_{k,l}\left(i\right)}{{\mathrm{\&sigma;}}^2+{\left|\right|h_{k,l}\left(i\right)\left|\right|}^2}---\left(2\right)$

H in the formula _{K, l}(i) (black matrix) is the row of one in the aforementioned channels state matrix (k row).

The addition of multiplying each other respectively of (1) formula and (2) formula is obtained,

$z_k\left(i\right)=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{w}_{k,l}\left(i\right)r_{k,l}\left(i\right)---\left(3\right)$

Utilize channel state matrix H _l(i) carry out delivery again and calculate,

${\mathrm{\&mu;}}_{k,l}\left(i\right)=\frac{{\left|\right|h_{k,l}\left(i\right)\left|\right|}^2}{{\mathrm{\&sigma;}}^2+{\left|\right|h_{k,l}\left|\right|}^2}\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\left(4\right)$

${\mathrm{\&mu;}}_k\left(i\right)=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{k,l}\left(i\right)\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\left(5\right)$

Utilize three formulas in front (3), (4), (5) to obtain following formula at last, Re is realistic computing in the formula:

${\mathrm{\&lambda;}}_1\left[b_k\left(i\right)\right]=\frac{2\mathrm{Re}\left\{{\mathrm{\&mu;}}_k\left(i\right)z_k\left(i\right)\right\}}{\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}[{\mathrm{\&mu;}}_{k,l}\left(i\right)-{\mathrm{\&mu;}}_{k,l}{\left(i\right)}^2]}$

Step 4 makes n increase by 1, repeats above-mentioned steps 1 to step 3, stops iterative process when n=M, and this moment just can be from the λ of output ₁[b _k(i)] obtain information bit in, need to prove, in the step 1 that repeat this moment,, therefore just reappraise channel state matrix H because of n no longer is 1 _l(i).

Referring to Fig. 3, the SISO method that iterates of uniting with decoding shown in the figure.And the difference of flow process shown in Figure 2 is: increased de-interleaving block 33, SISO decoding module 34 and interleaving block 35 between each SISO module 31 and corresponding conversion module 32, export extrinsic information and feed back to conversion module 32 by interleaving block 35, by SISO decoding module 34 output information bits.Wherein the operation method of SISO module 31 and conversion module 32 and shown in Figure 2 be not consistent with the SISO method that iterates of decoding associating.

Still utilize user's spreading code to carry out a yard matched filtering,, obtain the signal r in k user's l footpath for i symbol by the RAKE receiving unit _{K, l}(i), utilize the channel estimating h of i the symbol that obtains _{K, l}(i) carry out Rake and merge, channel estimating can utilize common channel estimation methods to obtain, for example the WMSA method.

Following steps are finished jointly by soft input/output module, de-interleaving block, SISO decoding module, interleaving block and conversion module.The length of the given signal r (i) after the chip-level matched filtering that receives is N, and given iterative times is M, and to establish the iterative times parameter be n, and n is since 1, and incremental change is 1, finishes to iterate during for M until n.

Step 1: estimate channel state matrix H by channel estimation methods _l(i).

Step 2: from n=1, to the result after RAKE merges carry out symbol judgement (only just the result after merging through RAKE is carried out symbol judgement under the situation at n=1, obtains the directly symbol judgement result of received signal and channel estimating of k user, an i symbol, l:

${\tilde{b}}_k\left(i\right)=\mathrm{sgn}\left(\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{r}_{k,l}\left(i\right)h_{k,l}^{*}\left(i\right)\right)$

Wherein, sgn represents the symbol (+1 or-1) of peek value.

If n=2,3 ... (promptly under the situation of n＞1), then carry out following computing

${\tilde{b}}_k\left(i\right)=\tanh \left(\frac{1}{2}{\mathrm{\&lambda;}}_2^p\right[b_k\left(i\right)\left]\right)$

This step conversion be the function finished by conversion module among Fig. 3 32 (promptly be not equal at 1 o'clock at n, the SISO resume module be the transformation results of conversion module.Tan in the formula gets tangent (tangent):

$\tanh \left(x\right)=\frac{e^x-e^{-x}}{e^x+e^{-x}}$

Therefore for step 2, when n=1, be directly the amalgamation result of Rake output to be input in the SISO module 31 to handle, when n＞1, then to the λ as a result of SISO module 31 outputs ² _p[b _k(i)] (this value with embodiment illustrated in fig. 2 different) carried out conversion and obtained

And then feedback is input to SISO module 31.

Step 3, form by step a and step b:

Step a carries out Parallel Interference Cancellation, remove all K user non-l footpath signal and and remove and remove k l footpath signal with other outdoor user, the l that promptly only keeps k user is signal directly,

$r_{k,l}\left(i\right)=r\left(i\right)-\underset{\mathrm{la}=1,\mathrm{la}\&NotEqual;l}{\overset{L}{\mathrm{\&Sigma;}}}{H}_{\mathrm{la}}\left(i\right)\tilde{b}\left(i\right)-H_l\left(i\right){\tilde{b}}_k\left(i\right)............\left(11\right)$

Wherein, $\tilde{b}\left(i\right)=[{\tilde{b}}_1\left(i\right),{\tilde{b}}_2\left(i\right),...,{\tilde{b}}_K\left(i\right)],{\tilde{b}}_k\left(i\right)$ Represent that k component is changed to zero

Step b calculates likelihood ratio λ ₁[b _k(i)]:

Calculate the weighted value of k user l footpath signal earlier:

$w_{k,l}\left(i\right)=\frac{h_{k,l}\left(i\right)}{{\mathrm{\&sigma;}}^2+{\left|\right|h_{k,l}\left(i\right)\left|\right|}^2}............\left(22\right)$

H in the formula _{K, l}(i) (represent) it is the row of one in the aforementioned channels state matrix (k row) with black matrix.

Therefore, with the addition of multiplying each other respectively of (11) formula and (22) formula, can calculate:

$z_k\left(i\right)=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{w}_{k,l}\left(i\right)r_{k,l}\left(i\right)............\left(33\right)$

Utilize channel state matrix H _l(i) carrying out delivery again calculates:

${\mathrm{\&mu;}}_{k,l}\left(i\right)=\frac{{\left|\right|h_{k,l}\left(i\right)\left|\right|}^2}{{\mathrm{\&sigma;}}^2+{\left|\right|h_{k,l}\left|\right|}^2}................\left(44\right)$

${\mathrm{\&mu;}}_k\left(i\right)=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{k,l}\left(i\right)............\left(55\right)$

Utilize aforementioned three formulas (33), (44), (55) to obtain following formula at last, Re is realistic portion in the formula:

${\mathrm{\&lambda;}}_1\left[b_k\left(i\right)\right]=\frac{2\mathrm{Re}\left\{{\mathrm{\&mu;}}_k\left(i\right)z_k\left(i\right)\right\}}{\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}[{\mathrm{\&mu;}}_{k,l}\left(i\right)-{\mathrm{\&mu;}}_{k,l}{\left(i\right)}^2]}$

Step 4 is utilized the channel decoding of soft output earlier, from λ ₁[b _k(i)] calculate λ ₂ ^p[b _k(i)], again n is increased by 1, repeat above-mentioned steps 1 to 3, when n reaches the value of the iterating M that sets in advance, finish, and by decoding module 34 output information bits (bit of judgement).Need to prove, in the step 1 that repeat this moment,, therefore just reappraise channel state matrix H because of n no longer is 1 _l(i).

Wherein the channel decoding output valve of soft output is a maximum likelihood ratio, and is identical with general soft input soft output decode device, for example Turbo code SOVA decoding.λ ₂ ^p[b _k(i)] be the information λ that the output of soft-output decoder has cut the soft decoding input ₁[b _k(i)] the likelihood ratio after, the i.e. prior information that increases by channel decoding.

According to Fig. 4 flow process, because when subscriber signal sends, passed through channel decoding and interweaved (finishing) by channel coding module 41 and interleaving block 42, therefore in Fig. 3 decoding and feedback, also need respectively to carry out anti-(separating) and interweave (Deinterlever) and again by interleaving block 35 interweave (interlever) by anti-(separating) interleaving block 33.

Under Fig. 2 and two kinds of situations of Fig. 3, it calculates all needs some estimators such as σ, H _l(i), its method of estimation is:

The estimation of background noise variances sigma can adopt common mean variance method to estimate; Also can simply be set to a constant, for example zero.

Channel state matrix H _l(i) estimation: to calculate simply in order making, can to suppose that channel impulse response is identical in a symbol period, just can simplify the operand of channel estimating greatly.This can utilize common channel estimation method to estimate.

Referring to Fig. 5, adopt RAKE to receive (top curve among the figure) and Multiuser Detection of the present invention when being no chnnel coding shown in the figure and mix signal to noise ratio (snr), the error rate (BER) the performance comparative graph that iterates reception (first kind not with the situation of encoding and combining) method (lower curve among the figure), under identical signal to noise ratio (snr), the former error rate (BER) is higher than the latter.

Referring to Fig. 6, adopt RAKE to receive (top curve among the figure) and Multiuser Detection of the present invention when being no chnnel coding shown in the figure and mix number of users and the error rate (BER) the performance comparative graph that iterates reception (first kind not with the situation of encoding and combining) method (lower curve among the figure), under identical number of users (Users), the former error rate (BER) is higher than the latter.

Referring to Fig. 7, adopt Multiuser Detection mixing of the present invention to iterate signal to noise ratio (snr), the error rate (BER) performance chart of method of reseptance (second kind of situation about combining) when shown in the figure being the combined channel coding with coding, the performance that is better than lower curve among Fig. 5 is arranged.

Method of the present invention adopts the output of Rake receiver as the input that iterates for the first time (n=1);

Method of the present invention can be applied to time varying channel, and the method for document 6 constant channel can only be used for the time;

Method of the present invention is all handled each footpath signal of multipath signal, has therefore adopted different interference cancellation methods and different extrinsic information (extrinsic information) formula;

The complexity of the inventive method is compared with traditional PIC only a small amount of increase.

Claims (10)

1. the multi-user test method of a wireless receiver is characterized in that comprising following treatment step:

A. corresponding each user is provided with a soft inputting and soft output SISO module and a conversion module;

B., iterative times M and iterative times parameter n are set, and M, n are positive integer;

C. establish n and equal 1, the consequential signal that merges output through RAKE receiver is carried out symbol judgement by soft inputting and soft output SISO module;

D. according to the symbol judgement result received signal after discretization of input is carried out Parallel Interference Cancellation and exported likelihood ratio;

E. make n increase by 1, the symbol judgement result who soft inputting and soft output SISO module is exported by the conversion module of soft inputting and soft output SISO module carries out feeding back input soft inputting and soft output SISO module again after the conversion, and execution in step D, E stop when n=M then.

2. the multi-user test method of a kind of wireless receiver according to claim 1, it is characterized in that: described steps A, be that a corresponding K user is provided with K soft inputting and soft output SISO module and K conversion module, each conversion module is arranged between the feedback input end and extrinsic information output of corresponding soft inputting and soft output SISO module, what be used to handle combined channel coding and decoding not iterates soft inputting and soft output, exports described likelihood ratio by soft inputting and soft output SISO module.

3. the multi-user test method of a kind of wireless receiver according to claim 1, it is characterized in that: described steps A, be that a corresponding K user is provided with K soft inputting and soft output SISO module, K de-interleaving block, K soft input soft output decode module, K interleaving block and K conversion module, between the feedback input end that each conversion module is arranged on corresponding soft inputting and soft output SISO module and the extrinsic information output of corresponding interleaving block, what be used to handle the combined channel coding and decoding iterates soft inputting and soft output, exports described likelihood ratio by the soft input soft output decode module.

4. the multi-user test method of a kind of wireless receiver according to claim 1 is characterized in that: described step C is the symbol manipulation of result's value of peek after according to following formula RAKE being merged, the r in the formula _{K, l}(i) be l footpath signal in the L footpath signal of k user among K the user who obtains at i symbol, h _{K, l}(i) be the channel estimation value of i symbol,

${\tilde{b}}_k\left(i\right)=\mathrm{sgn}\left(\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{r}_{k,l}\left(i\right)h_{k,l}^{*}\left(i\right)\right).$

5. the multi-user test method of a kind of wireless receiver according to claim 1 and 2 is characterized in that: described step e is to the symbol judgement of soft inputting and soft output SISO module output λ as a result according to following formula not uniting when coding and decoding iterates ₁[b _k(i)] carry out conversion,

${\tilde{b}}_k\left(i\right)=\tanh \left(\frac{1}{2}{\mathrm{\&lambda;}}_1\right[b_k\left(i\right)\left]\right).$

6. according to the multi-user test method of claim 1 or 3 described a kind of wireless receivers, it is characterized in that: described step e is to the symbol judgement of soft inputting and soft output (SISO) module output λ as a result when iterating according to following formula in the associating coding and decoding ^P ₂[b _k(i)] carry out conversion,

${\tilde{b}}_k\left(i\right)=\tanh \left(\frac{1}{2}{\mathrm{\&lambda;}}_2^p\right[b_k\left(i\right)\left]\right).$

7. the multi-user test method of a kind of wireless receiver according to claim 6 is characterized in that: described λ ^P ₂[b _k(i)] be the soft input information λ that deducts described soft input soft output decode device from the soft output of described soft input soft output decode device ₁[b _k(i)] the likelihood ratio after.

8. the multi-user test method of a kind of wireless receiver according to claim 1 is characterized in that:

Among the described step D received signal after discretization of input is carried out Parallel Interference Cancellation, it is the non-l footpath signal that removes all K user, the l footpath signal that promptly keeps all K user, and then remove and remove k l footpath signal with other outdoor user, promptly only keep k user's l footpath signal, the processing formula is:

$r_{k,l}\left(i\right)=r\left(i\right)-\underset{\mathrm{la}=1,\mathrm{la}\&NotEqual;l}{\overset{L}{\mathrm{\&Sigma;}}}{H}_{\mathrm{la}}\left(i\right)\tilde{b}\left(i\right)-H_l\left(i\right){\tilde{b}}_k\left(i\right)$

Wherein, ${\tilde{b}}_{\left(i\right)}=[{\tilde{b}}_1\left(i\right),{\tilde{b}}_2\left(i\right),...,{\tilde{b}}_K\left(i\right)],{\tilde{b}}_k\left(i\right)$ Represent that k component is changed to zero H _l(i) be channel state matrix;

Output likelihood ratio among the described step D further comprises following treatment step:

A. calculate the weighted value of k user l footpath signal earlier, the h in the formula _{K, l}(i) be described channel state matrix H _l(i) the k columns certificate in, σ is the background noise variance,

$w_{k,l}\left(i\right)=\frac{k_{k,l}\left(i\right)}{{\mathrm{\&sigma;}}^2+{\left|\right|k_{k,l}\left(i\right)\left|\right|}^2};$

B. the addition of above-mentioned two formulas being multiplied each other respectively obtains

$z_k\left(i\right)=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{w}_{k,l}\left(i\right)r_{k,l}\left(i\right);$

C. utilize channel state matrix H _l(i) carry out delivery again and calculate,

${\mathrm{\&mu;}}_{k,l}\left(i\right)=\frac{{\left|\right|h_{k,l}\left(i\right)\left|\right|}^2}{{\mathrm{\&sigma;}}^2+{\left|\right|h_{k,l}\left|\right|}^2}$

${\mathrm{\&mu;}}_k\left(i\right)=\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{k,l}\left(i\right)$

D. utilize three formula of step b.c. to calculate likelihood ratio λ at last ₁[b _k(i)], Re is realistic computing in the formula,

${\mathrm{\&lambda;}}_1\left[b_k\left(i\right)\right]=\frac{2\mathrm{Re}\left\{{\mathrm{\&mu;}}_k\left(i\right)z_k\left(i\right)\right\}}{\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}[{\mathrm{\&mu;}}_{k,l}\left(i\right)-{\mathrm{\&mu;}}_{k,l}{\left(i\right)}^2]}$

9. the multi-user test method of a kind of wireless receiver according to claim 8 is characterized in that: described background noise variances sigma, adopt traditional mean variance method to estimate; Described channel state matrix H _l(i) be that channel impulse response is under the identical condition, to estimate by traditional channel estimation methods in hypothesis symbol i cycle.

10. the multi-user test method of a kind of wireless receiver according to claim 8 is characterized in that: described background noise variances sigma can be set to comprise a constant of zero.

Images