CN100501442C - Multiuser detector based on iterative message transfer algorithm - Google Patents

Abstract

A multiuser detector based on iterative message passing algorithm relates to a rapid pseudo code acquisition method and a multiuser detector in marine radio navigation system, which solves the problems of serious sky-wave interference in the marine radio navigation system and difficulty in eliminating the sky-wave interference by conventional method. The detector comprises an interference cancellation structure with serial-parallel sub-groups, wherein signals are divided into a plurality of groups according to power, a parallel interference cancellation structure is adopted within each group, and a serial interference cancellation structure is adopted between the groups. Based on IMPA algorithm, the invention can acquire pseudo codes, which comprises the following steps: representing the constraint relationship of the pseudo code sequence by a factor graph, carrying out soft message iterative computation to obtain maximum a-posteriori estimation and generating a local sequence according to the maximum a-posteriori estimation and the decision condition. The invention shortens the acquisition time of long pseudo codes and effectively improves the acquisition probability, and can be applied in signal detection of satellite navigation, 4D mobile communication system, ultra-wide band system, etc.

Description

Multi-user detector based on the iterative message pass-algorithm

Technical field

The present invention relates to pseudo-code quick capturing method and multi-user detector in the radio navigation system of ocean.

Background technology

Short-and-medium wave band has that propagation distance is far away at sea, the characteristics that the propagation of ground wave is stable, and be widely used in navigation field.Radio navigation system utilization through earthwave signal in ocean carries out the marine navigation location.Receiver is measured the pseudorange between receiver carrier and the guidance station by catching the earthwave signal of the individual guidance station of N (N 〉=3), thereby determines the position of carrier by positioning calculation.On short-and-medium wave band, the signal of being launched by guidance station arrives receivers through two kinds of paths: a kind of is the earthwave of propagating along ground surface; A kind of is the sky wave that reflection arrives receiver through the ionosphere one or many.On daytime, arrive ionospheric sky wave major part and absorbed by the lower D layer of electron density; Night, the D layer disappears, electromagnetic wave arrives the bigger E layer of electron density, most of energy is reflected, and the earthwave signal is caused stronger interference, and because the decay gradually along with the increase of propagation distance of earthwave signal intensity, so after certain distance, sky wave singal intensity can surpass the earthwave signal intensity, and receiver is locked on the sky wave mistakenly, causes positioning solution to depart from the actual vector position.Therefore in a few thing zone, sky wave can be greater than earthwave 30dB, and at this moment earthwave is submerged in the secondary lobe of sky wave singal fully, if do not eliminate the influence of sky wave, then at a distance, receiver makes the perform region of system diminish cisco unity malfunction.

Why very serious the influence of it wave interference is, main cause is that day wave amplitude and phase change are than very fast, conventional coded sequence is caught and is carried out the variation that Multipath Time Delay Estimation can't be followed the tracks of sky wave singal, promptly can't catch fast and accurately sky wave singal.

Catch (being example with the coherent acquisition) of coded sequence is meant receiver when the spread-spectrum signal of search transmission, and the local phase place that generates coded sequence of adjustment receiver makes the phase place of local generation coded sequence consistent with the phase place of transmission coded sequence.The catching method of existing coded sequence has matched filtering method, sliding correlation method etc., wherein matched filtering method is to mate (relevant) with received signal with the code element of part pseudo-random code, therefore the matched filtering method can reduce the processing gain of system, is not suitable for using under the low signal-to-noise ratio condition; Sliding correlation method is the processing gain of loss system not, and realizes easily, is coded sequence catching method commonly used at present.In slip was caught, the capture time of full parallel capture was the shortest, but the complexity that realizes is the highest; The capture time that full serial is caught is the longest, and the complexity of realization is minimum; It is both compromises that mixing is caught.

Multi-user detector commonly used at present comprises serial and Parallel Interference Cancellation device.The counteracting serial interference device has robustness preferably in multidiameter fading channel, but because it adopts the mode of serial to work, need sort to each multipath signal power, therefore has bigger decoding delay more for a long time the user; The Parallel Interference Cancellation device adopts parallel mode to work, in the multistage processing procedure, has less decoding delay, and do not need user power is sorted, can be when realizing by advantages such as multiprocessors parallel processing, thereby more have practical value, but being not suitable for signal power differs bigger occasion.World wave power differs bigger in radio navigation system, and the signal path that arrives receiver is many, adopts above-mentioned serial or parallel Canceller all can not satisfy in the existing radio-navigation systems separately sky wave is accurately caught, separated.

IMPA of the prior art (Iterative Message Passing Algorithms) algorithm is to be developed by channel coding theorem, utilize the factor graph model, this method can be expanded to many fields, comprise fields such as signal Processing, control automatically, artificial intelligence, neural network, graph theory planning, but be not applied to the acquiring pseudo code field as yet.

The IMPA algorithm is introduced:

(1), the single node situation, referring to Fig. 3.

The node N that degree is K+1, as shown in Figure 3.Limit variable on the K+1 bar limit is respectively X _i(i=0,1 ..., K), they are respectively at alphabet A _i(i=0,1 ..., K) go up value.In the single node situation, the final purpose of message transmission is to calculate the posterior probability of each variable about this node.Suppose the inherent probable value of known this K+1 limit variable about node N

And constraint set $S_N\&Subset;{\mathrm{<figure-callout\; id="0"\; label="A"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">A</figure-callout>}}_0\&times;{\mathrm{<figure-callout\; id="1"\; label="A"\; filenames="C200710144950E00182.png,C200710144950E00193.png"\; state="\{\{state\}\}">A</figure-callout>}}_1\&times;\&CenterDot;\&CenterDot;\&CenterDot;\&times;A_n.$ Limit variable x ₀Posterior probability about node N is:

Outer probability $P_N^{\mathrm{ext}}=(x_0={\mathrm{\&zeta;}}_0)$ With conditional probability P (N|x ₀=ζ ₀) relation as follows:

$P_N^{\mathrm{ext}}(x_0={\mathrm{\&zeta;}}_0)={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">c</figure-callout>}}_0^{\&prime;}P\left(N\right|x_0={\mathrm{\&zeta;}}_0)---\left(2\right)$

Wherein It is normaliztion constant.

Inherent probability and posterior probability are respectively

$P_N^{\mathrm{int}}(x_0={\mathrm{\&zeta;}}_0)=P(x_0={\mathrm{\&zeta;}}_0)---\left(4\right)$

$P_N^{\mathrm{post}}(x_0={\mathrm{\&zeta;}}_0)=P(x_0={\mathrm{\&zeta;}}_0|N)---\left(5\right)$

The pairing incident N of node N sets up, and if only if K+1 limit variable x that node N links to each other _i(i=0,1 ..., K) satisfy the local restriction relation.Here, the local restriction relation is by a S set _NRepresent that be referred to as local restriction collection (local constraint set), it is alphabet A _i(i=0,1 ..., the subclass of direct product K).Alphabetic(al) direct product A ₁* A ₂* ... A _KProvided limit variable x _i(i=1,2 ..., K) might value combination, therefore press total probability formula $P\left(E\right)=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}P(E,F_k)=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}P\left(E\right|F_k)P\left(F_k\right)$ Have

$\underset{{\left\{{\text{\&zeta;}}_i\right\}}_{i=1}^K\&Element;{\mathrm{\&Pi;}}_{i=1}^K{A}_i}{\mathrm{\&Sigma;}}P(N,{\{x_i={\mathrm{\&zeta;}}_i\}}_{i=1}^K)=1$

To in the following formula the right summation symbol each, have

$P(N,{\{x_i={\mathrm{\&zeta;}}_i\}}_{i=1}^K|x_0={\mathrm{\&zeta;}}_0)=P\left(N{|\{x_i={\mathrm{\&zeta;}}_i\}}_{i=0}^K\right)P\left({\{x_i={\mathrm{\&zeta;}}_i\}}_{i=1}^K\right|x_0={\mathrm{\&zeta;}}_0)---\left(6\right)$

Because all limit variablees

Separate, thereby second on following formula the right can be written as

$P\left({\{x_i={\mathrm{\&zeta;}}_i\}}_{i=1}^K\right|x_0={\mathrm{\&zeta;}}_0)={\mathrm{\&Pi;}}_{i=1}^K{P}_N^{\mathrm{int}}(x_i={\mathrm{\&zeta;}}_i)---\left(7\right)$

For first on the right of the formula (6), have only and work as ${\{x_i={\mathrm{\&zeta;}}_i\}}_{i=0}^K\&Element;S_N$ The time conditional probability $P\left(N\right|{\{x_i={\mathrm{\&zeta;}}_i\}}_{i=0}^K)=1;$ Otherwise $P\left(N\right|{\{x_i={\mathrm{\&zeta;}}_i\}}_{i=0}^K)=0.$ Therefore $P\left(N\right|{\{x_i={\mathrm{\&zeta;}}_i\}}_{i=0}^K)$ Can be expressed as follows with indicator function (indicator function):

$P\left(N\right|{\{x_i={\mathrm{\&zeta;}}_i\}}_{i=0}^K)=[({\mathrm{\&zeta;}}_0,{\mathrm{\&zeta;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&zeta;}}_K)\&Element;S_N]---\left(8\right)$

Composite type (6), (7) and (8), formula (5) can be write as

$P\left(N\right|{\mathrm{<figure-callout\; id="0"\; label="x"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">x</figure-callout>}}_0={\mathrm{\&zeta;}}_0)=\underset{{\left\{{\mathrm{\&zeta;}}_i\right\}}_{i=1}^K\&Element;{\mathrm{\&Pi;}}_{i=1}^K{A}_i}{\mathrm{\&Sigma;}}[({\mathrm{\&zeta;}}_0,{\mathrm{\&zeta;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&zeta;}}_K)\&Element;S_N]{\mathrm{\&Pi;}}_{i=1}^K{P}_N^{\mathrm{int}}(x_i={\mathrm{\&zeta;}}_i)---\left(9\right)$

In following formula, the summation symbol is only to [(ζ ₀, ζ ₁..., ζ _K) ∈ S _NTherefore the item summation of]=1 can merge to indicator function in the summation scope and go, and so just has

$P\left(N\right|{\mathrm{<figure-callout\; id="0"\; label="x"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">x</figure-callout>}}_0={\mathrm{\&zeta;}}_0)=\underset{({\mathrm{\&zeta;}}_0,{\mathrm{\&zeta;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&zeta;}}_K)\&Element;S_N}{\underset{({\mathrm{\&zeta;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&zeta;}}_K)}{\mathrm{\&Sigma;}}}{\mathrm{\&Pi;}}_{i=1}^K{P}_N^{\mathrm{int}}(x_i={\mathrm{\&zeta;}}_i)---\left(10\right)$

By the definition (seeing formula (2)) of outer probability, x ₀Outer probability about node N is:

$P_N^{\mathrm{ext}}\left(x_0\right)={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">c</figure-callout>}}_0^{\&prime;}P\left(N\right|{\mathrm{<figure-callout\; id="0"\; label="x"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">x</figure-callout>}}_0={\mathrm{\&zeta;}}_0)={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">c</figure-callout>}}_0^{\&prime;}\underset{({\mathrm{\&zeta;}}_0,{\mathrm{\&zeta;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&zeta;}}_K)\&Element;S_N}{\underset{({\mathrm{\&zeta;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&zeta;}}_K)}{\mathrm{\&Sigma;}}}{\mathrm{\&Pi;}}_{i=1}^K{p}_N^{\mathrm{int}}(x_i={\mathrm{\&zeta;}}_i)---\left(11\right)$

So just derived x ₀Calculating formula about the outer probability of node N.Again by formula (11), x ₀Posterior probability about node N can be calculated as follows,

$P_N^{\mathrm{post}}\left(x_0\right)=c_0{P}_N^{\mathrm{ext}}\left(x_0\right)P_N^{\mathrm{int}}\left(x_0\right)=c_0{\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">c</figure-callout>}}_0^{\&prime;}\underset{({\mathrm{\&zeta;}}_0,{\mathrm{\&zeta;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&zeta;}}_K)\&Element;S_N}{\underset{({\mathrm{\&zeta;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&zeta;}}_K)}{\mathrm{\&Sigma;}}}{\mathrm{\&Pi;}}_{i=0}^K{p}_N^{\mathrm{int}}(x_i={\mathrm{\&zeta;}}_i)---\left(12\right)$

So far, the message transmission in the relevant single node situation is very clear.If selecting constraint set SN is the parity checking collection, the message transmission in the single node can be corresponding to single inspection sign indicating number (maximum posteriori decoding of (single paritycheck code) so.

(2) binode situation is referring to Fig. 4.

The binodal point diagram that constitutes by node L and node R shown in Figure 4.Node L and node R are by common edge (or being called inner edge) x ₀Link together.In Fig. 4, except x ₀Outside, node L also links to each other with other K limit variable, and node R also links to each other with the individual limit of other K ' variable.Remember that binodal point diagram shown in Figure 4 is G, it comprises two nodes, K+K ' individual outside variable and an inner edge variable.Supposing the inherent probability of known all outside variablees, is that equiprobability distributes with seasonal inner edge variable about the inherent probability of scheming G, promptly

$P_G^{\mathrm{int}}\left(x_i^L\right)=P\left(x_i^L\right),i=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,K;$

$P_G^{\mathrm{int}}\left(x_i^R\right)=P\left(x_i^R\right),i=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,K\&prime;;$

$P_G^{\mathrm{int}}\left(x_0\right)=1/\left|A_0\right|$

The constraint set of note node L (R) is S _L(S _R), the constraint set of figure G is S _G, S then _GSatisfy S simultaneously by all _LAnd S _RThe value combination (x of K+K '+1 a limit variable ₀,

...,

...,

) constitute.Judge the S that whether belongs to of K+K '+1 a limit variable _GCan represent with indicator function,

$[(x_0,x_1^L,x_2^L,...,x_K^L,x_1^R,x_2^R,...,x_{K\&prime;}^R)\&Element;S_G]$

$=[(x_0,x_1^L,x_2^L,...,x_K^L)\&Element;S_L][(x_0,x_1^R,x_2^R,...,x_{K\&prime;}^R)\&Element;S_R]---\left(13\right)$

For convenience, all limit variablees have been renumberd,

$(x_0,x_1,x_2,...,x_{K+K\&prime;})=(x_0,x_1^L,x_2^L,...,x_K^L,x_1^R,x_2^R,...,x_{K\&prime;}^R)$

Like this, the derivation of similar formula (11) can obtain any limit variable about figure G outer probability be

$P_G^{\mathrm{ext}}\left(x_i\right)=c_i^{\&prime;}\underset{\underset{~\left\{x_i\right\}}{(x_0,x_1,\&CenterDot;\&CenterDot;\&CenterDot;,x_{K+K\&prime;})\&Element;S_G}}{\mathrm{\&Sigma;}}\underset{J\&NotEqual;i}{\overset{\mathrm{}}{\underset{J=0}{\overset{K+K\&prime;}{\mathrm{\&Pi;}}}}}{P}_G^{\mathrm{int}}\left(x_j\right)---\left(14\right)$

Wherein～{ x _iRepresent except variable x _iOutside other all variablees, the summation symbolic representation in the following formula is satisfied S to all when variable xi gets certain fixed value _GValue combination summation.Because the summation symbol relates to the individual variable of K+K ' in the following formula, if each variable all is a binary, so always has 2 ^{K+K '}Individual sum term.And have only 2 during single node as can be known by formula (11) ^KIndividual sum term will reduce complexity greatly if the outer probability calculation therefore can be with binode the time converts the outer probability calculation of single node to.

Easy proof incident L, { x ₀=ζ ₀And Markov chain of incident R formation, promptly

P(L，R|x ₀)＝P(L|x ₀)P(R|x ₀) (15)

In fact, have by total probability formula

$P(L,R|x_0)=\underset{{\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;}}{\mathrm{\&Sigma;}}P(L,R,{\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;}|x_0)$

$=\underset{{\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;}}{\mathrm{\&Sigma;}}P(L,R,{\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;}|x_0)P({\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;}|x_0)$ (16)

For first on following formula the right, when $({\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;},x_0)\&Element;S_G$ The time value be 1, otherwise value is 0.

Therefore can be expressed as follows with indicator function (formula (13)),

$P(L,R{\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;},x_0)=\left[\right({x_0,\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;})\&Element;S_G]$

$=\left[\right({x_0,\left\{x_i^L\right\}}_{i=1}^K)\&Element;S_L]\left[\right({x_0,\left\{x_i^R\right\}}_{i=1}^{K\&prime;})\&Element;S_R]=P\left(L\right|x_0,{\left\{x_i^L\right\}}_{i=1}^K)P\left(R\right|x_0,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;\&prime;})$ (17)

Again by

Separate, so formula (16) can be decomposed into

$P({\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;}|x_0)=P\left({\left\{x_i^L\right\}}_{i=1}^{K}P\left({\left\{x_i^R\right\}}_{i=1}^{K\&prime;}\right)\right)---\left(18\right)$

With formula (17) and (18) substitution formula (16),

$P(L,R|x_0)=\underset{{\left\{x_i^L\right\}}_{i=1}^K,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;}}{\mathrm{\&Sigma;}}P\left(L\right|{\left\{x_i^L\right\}}_{i=1}^K,x_0)P(R|{\left\{x_i^R\right\}}_{i=1}^{K\&prime;},x_0)P\left({\left\{x_i^L\right\}}_{i=1}^K{|x}_0\right)P\left({\left\{x_i^R\right\}}_{i=1}^{K\&prime;}\right|x_0)$

$=\left(\underset{{\left\{x_i^L\right\}}_{i=1}^K}{\mathrm{\&Sigma;}}P\left(L\right|{\left\{x_i^L\right\}}_{i=1}^K,x_0)P\left({\left\{x_i^L\right\}}_{i=1}^K\right|x_0)\right)\left(\underset{{\left\{x_i^R\right\}}_{i=1}^{K\&prime;}}{\mathrm{\&Sigma;}}P\left(R\right|{\left\{x_i^R\right\}}_{i=1}^{K\&prime;},x_0)P\left({\left\{x_i^R\right\}}_{i=1}^{K\&prime;}\right|x_0)\right)$

$=P\left(L\right|x_0)P\left(R\right|x_0)$

So just obtain formula (15), thereby proved incident L, { x ₀=ζ ₀And Markov chain of incident R formation.

Consider the limit variable now Outer probability about figure G has

$P_G^{\mathrm{ext}}\left(x_1^R\right)=c_{1R}^{\&prime;}R(L,R|x_1^R)$

Wherein Be normaliztion constant.By total probability formula, the conditional probability in the following formula can be write as again

$P(L,R|x_1^R)=\underset{x_0,{\left\{x_i^R\right\}}_{i=2}^R}{\mathrm{\&Sigma;}}P(L,R,{\mathrm{<figure-callout\; id="0"\; label="x"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">x</figure-callout>}}_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}|x_1^R)$

$=\underset{x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}}{\mathrm{\&Sigma;}}P\left(R\right|L,x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}{,x}_1^R)P(L,x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}|x_1^R)---\left(19\right)$

$=\underset{x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}}{\mathrm{\&Sigma;}}P\left(R\right|L,x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}{,x}_1^R)P\left(L\right|x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}{,x}_1^R)P(x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}{|x}_1^R)$

By incident L, { x ₀=ζ ₀And incident R constitute Markov chain, first on following formula the right can become by abbreviation with second difference

$P\left(R\right|L,x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}{,x}_1^R)=P\left(R\right|x_0,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;})---\left(20\right)$

$P\left(L\right|x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}{,x}_1^R)=P\left(L\right|x_0)---\left(21\right)$

Simultaneously, separate by the limit variable, the 3rd of formula (19) the right can be decomposed into

$P(x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}{|x}_1^R)=P_G^{\mathrm{int}}\left(x_0\right){\mathrm{\&Pi;}}_{i=2}^{K\&prime;}{P}_G^{\mathrm{int}}\left(x_i^R\right)---\left(22\right)$

Formula (20)-(22) substitution formula (19) is got

$P(L,R|x_1^R)=\underset{x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}}{\mathrm{\&Sigma;}}P\left(R\right|x_0,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;})P\left(L\right|x_0)P_G^{\mathrm{int}}\left(x_0\right){\mathrm{\&Pi;}}_{i=2}^{K\&prime;}{P}_G^{\mathrm{int}}\left(x_i^R\right)---\left(23\right)$

Therefore, the limit variable R about the outer probability of figure G is

$P_G^{\mathrm{ext}}\left(x_1^R\right)=c_{1R}^{\&prime;\&prime;}P(L,R|x_1^R)=c_{1R}^{\&prime;\&prime;}\underset{x_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;}}{\mathrm{\&Sigma;}}P\left(R\right|x_0,{\left\{x_i^R\right\}}_{i=1}^{K\&prime;})P\left(L\right|x_0)P_G^{\mathrm{int}}\left(x_0\right){\mathrm{\&Pi;}}_{i=2}^{K\&prime;}{P}_G^{\mathrm{int}}\left(x_i^R\right)---\left(24\right)$

In addition, because $P_G^{\mathrm{int}}\left(x_0\right)(=1/|A_0\left|\right)$ Be a constant, therefore it can be integrated in the normaliztion constant and go, simultaneously with conditional probability Be integrated in the summation scope, so have

$P_G^{\mathrm{ext}}\left(x_1^R\right)=c_{1R}^{\&prime;\&prime;}P(L,R|x_1^R)=c_{1R}^{\&prime;}\underset{\underset{~\left\{x_1^R\right\}}{({\mathrm{<figure-callout\; id="0"\; label="x"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">x</figure-callout>}}_0,{\left\{x_i^R\right\}}_{i=2}^{K\&prime;})\&Element;S_R}}{\mathrm{\&Sigma;}}P\left(L\right|x_0){\mathrm{\&Pi;}}_{i=2}^{K\&prime;}{P}_G^{\mathrm{int}}\left(x_i^R\right)---\left(25\right)$

Contrast formula (11) and (25) are not difficult to find: if order

$P_R^{\mathrm{int}}\left(x_0\right)={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">c</figure-callout>}}_0^{\&prime;}P\left(L\right|x_0)=P_L^{\mathrm{ext}}\left(x_0\right)---\left(26\right)$

$P_R^{\mathrm{int}}\left(x_i^R\right)=P_G^{\mathrm{int}}\left(x_i^R\right)$

Calculate the limit variable of gained so by formula (11)

Outer probability about node R

Equal the limit variable

Outer probability about G

Therefore, if whole figure is adopted the assignment relation shown in the formula (27), the limit variable just can obtain by the outer probability calculation of limit variable about node about the outer probability calculation of figure G so, thereby greatly reduces computation complexity.

$P_R^{\mathrm{int}}\left(x_i^R\right)=P_G^{\mathrm{int}}\left(x_i^R\right),$ $P_L^{\mathrm{int}}\left(x_i^L\right)=P_G^{\mathrm{int}}\left(x_i^L\right)$

$P_R^{\mathrm{int}}\left(x_0\right)={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">c</figure-callout>}}_0^{\&prime;}P\left(L\right|x_0)=P_L^{\mathrm{ext}}\left(x_0\right),$ $P_L^{\mathrm{int}}\left(x_0\right)={\mathrm{<figure-callout\; id="0"\; label="c"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">c</figure-callout>}}_0^{\&prime;\&prime;}P\left(R\right|x_0)=P_R^{\mathrm{ext}}\left(x_0\right)---\left(27\right)$

When beginning to calculate, at first give its inherent probability about the inherent probability assignment of figure about the company node with the outside variable.The outside variable only links to each other with a node, therefore only need with it about the inherent probability of figure compose to it about company node inherent probability get final product.Inner edge variable x has been arranged after these inherent probability ₀Outer probability about two nodes that it connected just can calculate by through type (11).Then with inner edge variable x ₀Outer probability about the company node R

Assignment is given

Equally with inner edge variable x ₀Outer probability about company node L

Assignment is given

Such two nodes the inherent probability of limit variable all known, still adopt formula (11) to calculate the outer probability of each outside variable, and then obtain their posterior probability about whole figure.

By above argumentation, can draw such as drawing a conclusion: the inner edge variable about the outer probability of a node of company be it about the inherent probability of another node of company.Acyclic figure The above results is generalized among the general figure, is not difficult to find, if can accurately calculate the posterior probability of each outside variable about whole figure according to the method described above.If when having ring, proved that in theory the message pass-algorithm converges on the equilibrium point of the free energy of bass (Bethe); In practice, a large amount of simulation results show that all the message pass-algorithm can extraordinaryly approach maximum posteriori decoding, and computation complexity is significantly smaller than real maximum posteriori decoding.

Summary of the invention

Wave interference is big over the ground in order to solve in the radio navigation system of ocean sky wave, and classic method can't be eliminated the problem of day wave interference, and the present invention proposes a kind of based on the multi-user detector based on the iterative message pass-algorithm.

Multi-user detector based on the iterative message pass-algorithm comprises: the signal that is used for receiving carries out demodulation and then obtains the demodulating equipment 20 of restituted signal, A signal capture in parallel and reconfiguration device 10 that is used for multiple signals are carried out acquisition and tracking and signal reconstruction, individual signal capture in parallel of described A and reconfiguration device 10 are respectively one-level from big to small according to power, secondary ... up to final stage, wherein one-level is identical to the structure of the signal capture in parallel of A-1 level and reconfiguration device 10, all form by b signal capture and reconstructed module 11 and an interference cancellation module, described interference cancellation module with the stack of b signal capture and reconstructed module 11 output signals and, after from the input signal of signal capture in parallel and reconfiguration device 10, deducting, export to next stage signal capture in parallel and reconfiguration device 10, demodulating equipment 20 output restituted signals are given one-level signal capture in parallel and reconfiguration device 10, final stage signal capture in parallel and reconfiguration device 10 are made up of b signal capture and reconstructed module 11 and parallel interference cancellation module 13, parallel interference cancellation module 13 is by time delay module 15, accumulator module 14 an and b subtraction block 16 is formed, and described accumulator module 14 is with the output signal u of b signal capture and reconstructed module 11 ₁, u ₂... u _bDeduct the output signal u of b signal capture and reconstructed module 11 after adding up again respectively ₁, u ₂... u _bObtaining b adds up and signal

...

Signal after b subtraction block 16 delayed time through time delay module 15 with final stage signal capture in parallel and reconfiguration device 10 input signals respectively deducts described b and adds up and signal

...

B output signal of final acquisition.

Described b equals the signal way that multi-user detector can receive.

Described interference cancellation module comprises Postponement module and subtraction block, and subtraction block is used for exporting to next stage signal capture in parallel and reconfiguration device 10 after input signal with the signal of b signal capture and reconstructed module 11 outputs and signal capture in parallel after the Postponement module time-delay and reconfiguration device 10 deducts.

The acquisition procedure of the described signal capture of present embodiment and 11 pairs of pseudo-codes of reconstructed module is:

Signal segment after step 1, one section demodulation of intercepting is obtained symbol level information;

Step 2, the symbol level information of obtaining is applied to the IMPA algorithm, and then obtains the local pseudo-code sequence of generation;

Step 3, local pseudo-code sequence and the receiving sequence that will obtain are carried out relevant, judge whether to have realized to the catching of pseudo-code, and have surpassed threshold value as the correlation of infructescence, acquisition success then, execution in step four; Otherwise, catch failure, intercept another segment signal fragment, and obtain new symbol level information, return the local pseudo-code sequence that step 2 is obtained generation again;

Step 4, the signal of successfully catching is carried out channel estimating and signal reconstruction, catch and finish.

The present invention has adopted the interference cancellation method of packet chain and combination, it has eliminated the shortcoming of existing independent counteracting serial interference method and Parallel Interference Cancellation method, and had the advantage of existing independent counteracting serial interference method and Parallel Interference Cancellation method simultaneously, it can keep less data delay, has reduced the memory space of intermediate data.

The acquisition procedure of the signal capture described in the present invention and 11 pairs of pseudo-codes of reconstructed module is to adopt the thought based on the IMPA algorithm of introducing in the background technology to realize, the restriction relation of pseudo-code sequence is represented with factor graph, and on factor graph, carry out the iterative computation of soft information, drawing maximum a posteriori at last estimates, then, estimate and judgment condition obtains the local sequence of a generation according to maximum a posteriori, of the present inventionly can effectively solve the capture time problem of long acquiring pseudo code, also effectively raise the acquisition probability of catching simultaneously based on the pseudo-code quick capturing method of IMPA algorithm.Be illustrated in figure 8 as signal to noise ratio (S/N ratio) under-15db condition, adopt quick capturing method of the present invention to carry out the curve of the acquisition probability of acquiring pseudo code with the iterations variation of IMPA algorithm, can effectively improve acquisition probability by increasing iterations as seen from the figure, but be subjected to the restriction of signal to noise ratio (S/N ratio) signal.The curve that the acquisition probability of quick capturing method of the present invention changed with signal to noise ratio (S/N ratio) when curve that the acquisition probability that is illustrated in figure 9 as the curve that acquisition probability that full serial catches changes with signal to noise ratio (S/N ratio), full parallel capture changes with signal to noise ratio (S/N ratio) and iterations were 20 times, pseudo-code based on the IMPA algorithm is captured under the lower signal to noise ratio (S/N ratio) condition fast as seen from Figure 9, acquisition probability is all the time more than or equal to full parallel mode, when signal to noise ratio (S/N ratio) greater than after-the 13.5db, the acquisition probability that the acquisition probability of catching fast based on the pseudo-code of IMPA algorithm is caught more than or equal to serial all the time.When signal to noise ratio (S/N ratio) greater than after-the 11db, the acquisition probability of catching fast based on the pseudo-code of IMPA algorithm remains at 100%, acquisition performance is very stable.

Figure 10 is under the radio conditions, has strong signal and weak signal simultaneously, the design sketch after the strong signal cancellation.Wherein (a) is Interference Cancellation correlation peak figure, (b) is the partial enlarged drawing of (a).From figure (b) as can be seen, after the strong signal cancellation, the auto-correlation secondary lobe of strong signal reduces, the signal-to-noise performance of weak signal be improved significantly.

Multi-user detector based on the iterative message pass-algorithm of the present invention can satisfy Marine Radio Navigation system remote guidance location requirement, can be applied in the existing Marine Radio Navigation system, can also be generalized to other association area, in satellite navigation, the 4th third-generation mobile communication system and radio ultra wide band system, solve catching fast and the multiple access interference cancellation problem of long pseudo-random code.

Description of drawings

Fig. 1 is the fundamental diagram of navigation positioning system, wherein 50 is the multi-user detectors that are used for signal capture and Interference Cancellation, Fig. 2 is the structural representation of the multi-user detector based on the iterative message pass-algorithm of the present invention, Fig. 3 is the single node information transitive graph in the existing IMPA algorithm, Fig. 4 is the binode information transitive graph in the existing IMPA algorithm, Fig. 5 is the factor graph of 15 rank pseudo-codes described in the embodiment two, Fig. 6 is the partial enlarged drawing of check-node 2 and variable node 1 on the factor graph among Fig. 4, Fig. 7 is a structural drawing of realizing the described multi-user detector based on the iterative message pass-algorithm of embodiment two, graph of a relation between Fig. 8 acquisition probability and the iterations, graph of a relation between Fig. 9 acquisition probability and the signal to noise ratio (S/N ratio), Figure 10 are sky wave Interference Cancellation design sketch.

Embodiment

Embodiment one: the multi-user detector based on the iterative message pass-algorithm of present embodiment comprises: the signal that is used for receiving carries out demodulation and then obtains the demodulating equipment 20 of restituted signal, A signal capture in parallel and reconfiguration device 10 that is used for multiple signals are carried out acquisition and tracking and signal reconstruction, individual signal capture in parallel of described A and reconfiguration device 10 are respectively one-level from big to small according to power, secondary ... up to final stage, wherein one-level is identical to the structure of the signal capture in parallel of A-1 level and reconfiguration device 10, all form by b signal capture and reconstructed module 11 and an interference cancellation module, described interference cancellation module with the stack of b signal capture and reconstructed module 11 output signals and, after from the input signal of signal capture in parallel and reconfiguration device 10, deducting, export to next stage signal capture in parallel and reconfiguration device 10, demodulating equipment 20 output restituted signals are given one-level signal capture in parallel and reconfiguration device 10, final stage signal capture in parallel and reconfiguration device 10 are made up of b signal capture and reconstructed module 11 and parallel interference cancellation module 13, parallel interference cancellation module 13 is by time delay module 15, accumulator module 14 an and b subtraction block 16 is formed, and described accumulator module 14 is with the output signal u of b signal capture and reconstructed module 11 ₁, u ₂... u _bDeduct the output signal u of b signal capture and reconstructed module 11 after adding up again respectively ₁, u ₂... u _bObtaining b adds up and signal

Signal after b subtraction block 16 delayed time through time delay module 15 with final stage signal capture in parallel and reconfiguration device 10 input signals respectively deducts described b and adds up and signal

B output signal of final acquisition.

The described interference cancellation module of present embodiment comprises Postponement module and subtraction block, and subtraction block is used for exporting to next stage signal capture in parallel and reconfiguration device 10 after input signal with the signal of b signal capture and reconstructed module 11 outputs and signal capture in parallel after the Postponement module time-delay and reconfiguration device 10 deducts.

The acquisition procedure of the described signal capture of present embodiment and 11 pairs of pseudo-codes of reconstructed module is:

Signal segment after step 1, one section demodulation of intercepting is obtained symbol level information;

Step 2, the symbol level information of obtaining is applied to the IMPA algorithm, and then obtains the local pseudo-code sequence of generation;

Step 3, local pseudo-code sequence and the receiving sequence that will obtain are carried out relevant, judge whether to have realized to the catching of pseudo-code, and have surpassed threshold value as the correlation of infructescence, acquisition success then, execution in step four; Otherwise, catch failure, intercept another segment signal fragment, and obtain new symbol level information, return the local pseudo-code sequence that step 2 is obtained generation again;

Step 4, the signal of successfully catching is carried out channel estimating and signal reconstruction, catch and finish.

Contain a plurality of code elements in the signal segment described in the step 1;

The process of obtaining the local pseudo-code sequence of generation in step 2 is: described symbol level information is applied in the IMPA algorithm as the soft channel initial information of symbol level, after the computing that iterates by algorithm, can obtain by 0 according to judgment criterion at every turn, the 1 N group state estimation value of forming (value of N can be determined and modification voluntarily by the user), from one group of maximum state estimation value of occurrence number, the original state of selected state estimation value, according to the state estimation value of selecting as the shift register that generates local pseudo-code sequence, the local pseudo-code sequence that the generator polynomial of position that the state estimation value occurs in sequence and m sequence and converse polynomial expression thereof just can obtain generating;

Embodiment two: present embodiment is described referring to Fig. 7.The described multi-user detector based on the iterative message pass-algorithm of present embodiment comprises two-stage parallel connection signal capture and reconfiguration device 10, every grade of signal capture in parallel and reconfiguration device have three road signals, one-level signal capture in parallel and reconfiguration device are used to catch the earthwave signal bigger with the reconstruct power ratio, and final stage signal capture in parallel and reconfiguration device are used to catch the earthwave signal less with the reconstruct power ratio.

Radio signal is divided into two groups according to watt level, and the sky wave that power is bigger is divided into one group, and the earthwave that power is lower is divided into one group, and every group has N road signal, and every group of interior N road signal adopts the Parallel Interference Cancellation mode, adopts the counteracting serial interference mode between group; In first group, respectively the sky wave singal in the signal of every road is caught, after then the sky wave singal that captures being carried out channel estimating and signal reconstruction, N road sky wave singal is superimposed exports to interference cancellation module, described sky wave singal is deducted from input signal, and the input signal of deduction sky wave singal is input to next group and carries out the earthwave signal capture.

Catching polynomial expression in the present embodiment is g (D)=1+D+D ¹⁵([100003] ₈) the detailed process of quick capturing method of 15 grades of m sequence pseudo-codes be:

Generator polynomial is g (D)=1+D+D ¹⁵([100003] ₈) the generating mode of 15 grades of m sequences can represent with factor graph as shown in Figure 4, by m sequence generator polynomial g (D)=1+D+D ¹⁵There is restriction relation for each check-node 2 as can be known $x_k\&CirclePlus;x_{k-1}\&CirclePlus;x_{k-15}=0,$ So, have four kinds of array mode (x corresponding to the value of each check-node 2 _k, x _K-1, x _K-15) ∈ { (0,0,0), (0,1,1), (1,0,1), (1,1,0) }.

In information exchanging process, all information are two-way on the limit of connected node to be transmitted repeatedly, with variable node 1 that check-node 2 is connected on the IMPA algorithm will carry out iterative computation to information according to polynomial restriction relation.

For the process that explain information clearly transmits on factor graph, a check-node 2 in the factor graph shown in Figure 4 to be amplified, the partial structurtes of check-node 2 are amplified as shown in Figure 5.

Suppose that k PN chip signal constantly is x _k=0,1, the signal waveform after the BPSK modulation is $y_k={(-1)}^{x_k},$ Then the received signal through additive white Gaussian noise channel is:

$z_k=\sqrt{E_c}{y}_k{e}^{{\mathrm{j\&theta;}}_c}+n_k=\sqrt{E_c}{(-1)}^{x_k}{e}^{j{\mathrm{\&theta;}}_c}+n_k,0\&le;k\&le;M-1---\left(28\right)$

Wherein, n _kBe that average is 0, variance is the white noise sample value of σ, and its monolateral band power spectrum density is N ₀M is the length of receiving sequence, θ _cBe carrier phase offset, E _cBe chip energy, for convenience, suppose θ _c=0.As seen from the figure, through white noise (Normal Distribution) channel, the reception likelihood probability of k chip is shown in formula (29):

$P\left(z_k\right|x_k)={\frac{1}{\sqrt{2\mathrm{\&pi;}{\mathrm{\&sigma;}}^2}}e}^{-\frac{{[z_k-{(-1)}^{x^k}\sqrt{E_c}]}^2}{2{\mathrm{\&sigma;}}^2}}---\left(29\right)$

For convenience of calculation, general normal employing log-likelihood probability, that is:

$\ln P\left(z_k\right|x_k)=C+\frac{\sqrt{E_c}{z}_k{(-1)}^{x^l}}{{\mathrm{\&sigma;}}^2}---\left(30\right)$

Wherein C is and x _kIrrelevant constant deserves to be called in the formula and x _kRelevant amount is the soft channel information M of chip-level _Ch[x _k] be:

$M_{\mathrm{ch}}\left[x_k\right]=\frac{2\sqrt{E_c}{z}_k{(-1)}^{{\mathrm{<figure-callout\; id="0"\; label="x\; k\; N"\; filenames="C200710144950E00211.png,C200710144950E00212.png"\; state="\{\{state\}\}">x</figure-callout>}}_k}}{N_{}},x_k=\mathrm{0,1}---\left(31\right)$

Z wherein _kIt is corresponding real part in (28) formula.

The tolerance M[τ of check-node 2 _k] be:

$M\left[{\mathrm{\&tau;}}_k\right]={\mathrm{LI}}_k\left[x_{k-15}\right]+L{I}_k^{\&prime;}\left[x_{k-1}\right]+\mathrm{RI}\left[x_k\right]---\left(32\right)$

The information updating formula that can be obtained each node by formula (32) is as follows:

${\mathrm{LO}}_k\left[x_{k-15}\right]=\underset{{\mathrm{\&tau;}}_k:x_{k-15}}{\min }M\left[{\mathrm{\&tau;}}_k\right]-{\mathrm{LI}}_k\left[x_{k-15}\right],$ ?x _k-15＝0，1 (33)

${\mathrm{LO}}_k^{\&prime;}\left[x_{k-1}\right]=\underset{{\mathrm{\&tau;}}_k:x_{k-1}}{\min }M\left[{\mathrm{\&tau;}}_k\right]-{\mathrm{LI}}_k^{\&prime;}\left[x_{k-1}\right],$ x _k-1＝0，1 (34)

${\mathrm{RO}}_k\left[x_k\right]=\underset{{\mathrm{\&tau;}}_k:x_k}{\min }M\left[{\mathrm{\&tau;}}_k\right]-R{I}_k\left[x_k\right],$ x _k＝0，1 (35)

$\mathrm{MO}\left[x_k\right]={\mathrm{LO}}_{k+1}^{\&prime;}\left[x_k\right]+{\mathrm{LO}}_{k+15}\left[x_k\right]+{\mathrm{RO}}_k\left[x_k\right],$ ?x _k＝0，1 (36)

${\mathrm{LI}}_k^{\&prime;}\left[x_{k-1}\right]={\mathrm{RO}}_{k-1}\left[x_{k-1}\right]+{\mathrm{LO}}_{k+14}\left[x_{k-1}\right]+M_{\mathrm{ch}}\left[x_{k-1}\right],$ x _k-1＝0，1 (37)

${\mathrm{LI}}_k\left[x_{k-15}\right]={\mathrm{RO}}_{k-15}\left[x_{k-15}\right]+M_{\mathrm{ch}}\left[x_{k-15}\right]+{\mathrm{LO}}_{k-14}\&prime;\left[x_{k-15}\right],$ x _k-15＝0，1?(38)

${\mathrm{RI}}_k\left[x_k\right]={\mathrm{LO}}_{k+1}^{\&prime;}\left[x_k\right]+{\mathrm{LO}}_{k+15}\left[x_k\right]+M_{\mathrm{ch}}\left[x_k\right],$ x _k＝0，1 (39)

Wherein, LO _k[x _K-i] be k the check node information of k-i variable node output to the left; RO _k[x _k] be k the check node information of k variable node output to the right; LI _k[x _K-i] be k-i the information that variable node is imported to k check node from the left side; RI _k[x _k] be k the information that variable node is imported to k check node from the left side.

The treatment scheme of IMPA algorithm is as follows:

(1) initialization:

(2) updating message:

Utilize formula (33), (34) to LI _k[x _K-15] and

Upgrade, utilize formula (35) to upgrade RO _k[x _k], utilize formula (37), (38), (39) to upgrade

LI _k[x _K-15] and RI _k[x _k], wherein 15≤k≤M-15, i=i+1 then.Renewal process is as follows: LI ₁₅[x ₀] → ... → LI _k[x _K-15] → ... → LI _M-15[x _M-30] and ${\mathrm{<figure-callout\; id="15"\; label="LI"\; filenames="C200710144950E00182.png,C200710144950E00211.png"\; state="\{\{state\}\}">LI</figure-callout>}}_{15}^{\&prime;}\left[x_{14}\right]\&RightArrow;\&CenterDot;\&CenterDot;\&CenterDot;\&RightArrow;{\mathrm{LI}}_k^{\&prime;}\left[x_{k-1}\right]\&RightArrow;\&CenterDot;\&CenterDot;\&CenterDot;\&RightArrow;{\mathrm{LI}}_{M-15}^{\&prime;}\left[x_{M-16}\right].$ When 0≤k≤14 and M-15≤k≤M-1, because the special construction of factor graph, all there are the situation that lacks corresponding limit in check-node 2 and variable node 1, the method that solves is that the quantity of information on the limit that correspondence is lacked regards that 0 gets final product as, being without loss of generality, is that example is discussed with the center section of 15≤k≤M-15.

(3) the shift register state is selected:

It is one group the state estimation vector that does not overlap mutually that the burst that receives is divided into per 15 chips, that is:

M _k[x _k]＝M _ch[x _k]+MO[x _k] (41)

15i≤k≤15i+14 wherein, i=0,1 ... [M/15]

The decision rule of chip estimated value is:

$\left\{\begin{array}{cc}{\hat{x}}_k=0& M_k[x_k=0]<M_k[x_k=1]\\ {\hat{x}}_k=1& M_k[x_k=0]>M_k[x_k=1]\end{array}\right.---\left(42\right)$

(4) if i＜I then returns step (2), otherwise iteration finishes.N group state estimation value that occurrence number is maximum in the statistic procedure (4) and the position that in sequence, occurs thereof.Suppose to use variable k (k=1,2 ..., N) represent number of state indexes, this N group state will be as the original state of the shift register that generates local sequence.Above process is all processes of iteration information pass-algorithm, by said process as can be known, judgment criterion by the IMPA algorithm can obtain organizing the state estimation value by 0,1 N that forms at every turn, from the maximum one group of state estimation value (k=1) of occurrence number, the original state of selected state estimation value, according to the state estimation value of selecting, position that the state estimation value occurs in sequence and generator polynomial g (the D)=1+D+D of m sequence as the shift register that generates local sequence ¹⁵The local sequence that just can obtain generating carries out related calculation local sequence that generates and the sequence that receives, and is used for judging whether to have realized catching pseudo-random code.Surpassed threshold value as the correlation of infructescence and thought that then what capture is correct phase, acquisition success; Otherwise, think and catch failure, select another segment signal to use above-mentioned acquisition procedure again again.

Claims (2)

1, multi-user detector based on the iterative message pass-algorithm, it comprises that the signal that is used for receiving carries out the demodulating equipment (20) of demodulation and then acquisition restituted signal, A signal capture in parallel and reconfiguration device (10) that is used for multiple signals are carried out acquisition and tracking and signal reconstruction, it is characterized in that described A signal capture in parallel and reconfiguration device (10) are respectively one-level from big to small according to power, secondary ... up to final stage, wherein one-level is identical to the structure of the signal capture in parallel of A-1 level and reconfiguration device (10), all form by b signal capture and reconstructed module (11) and an interference cancellation module, described interference cancellation module with the stack of b signal capture and reconstructed module (11) output signal and, after from the input signal of signal capture in parallel and reconfiguration device (10), deducting, export to next stage signal capture in parallel and reconfiguration device (10), demodulating equipment (20) output restituted signal is given one-level signal capture in parallel and reconfiguration device (10), final stage signal capture in parallel and reconfiguration device (10) are made up of b signal capture and reconstructed module (11) and parallel interference cancellation module (13), parallel interference cancellation module (13) is by time delay module (15), accumulator module (14) and b subtraction block (16) are formed, and described accumulator module (14) is with the output signal u of b signal capture and reconstructed module (11) ₁, u ₂... u _bDeduct the output signal u of b signal capture and reconstructed module (11) after adding up again respectively ₁, u ₂... u _bObtaining b adds up and signal

, the signal after b subtraction block (16) uses final stage signal capture in parallel and reconfiguration device (10) input signal through time delay module (15) time-delay respectively deducts described b and adds up and signal

B output signal of final acquisition.

2, the multi-user detector based on the iterative message pass-algorithm according to claim 1 is characterized in that described signal capture and reconstructed module (11) to the acquisition procedure of pseudo-code are:

Signal segment after step 1, one section demodulation of intercepting is obtained symbol level information;

Step 2, the symbol level information of obtaining is applied to the IMPA algorithm, and then obtains the local pseudo-code sequence of generation;

Step 3, local pseudo-code sequence and the receiving sequence that will obtain are carried out relevant, judge whether to have realized to the catching of pseudo-code, and have surpassed threshold value as the correlation of infructescence, acquisition success then, execution in step four; Otherwise, intercept another segment signal fragment, and obtain new symbol level information, return the local pseudo-code sequence that step 2 is obtained generation again;

Step 4, the signal of successfully catching is carried out channel estimating and signal reconstruction, catch and finish;

Contain a plurality of code elements in the signal segment described in the step 1;

The process of obtaining the local pseudo-code sequence of generation in step 2 is: described symbol level information is applied in the IMPA algorithm as the soft channel initial information of symbol level, after the computing that iterates by algorithm, can obtain by 0 according to judgment criterion at every turn, the 1 N group state estimation value of forming, from one group of maximum state estimation value of occurrence number, the original state of selected state estimation value, according to the state estimation value of selecting as the shift register that generates local pseudo-code sequence, the proper polynomial of position that the state estimation value occurs in sequence and m sequence generates local pseudo-code sequence.

Images