CN1627675A - Interference cancellation method base on soft decision in CDMA system - Google Patents

Abstract

The method includes following steps: (1) receiving demodulated output signal, and making out hard decision; (2) determining prior probability value of symbol based on result information of previous modulation; (3) calculating coefficient of reliability based on prior probability value of symbol; (4) determining result of decision based on result of hard decision from demodulated output signal and coefficient of reliability. The method is more accurate, improves effect and performance of CDMA system.

Description

In the code division multiple access system based on the interference cancellation method of soft-decision

Technical field

The present invention relates to the interference cancellation method in the CDMA mobile communication systems, particularly in the code division multiple access system based on the interference cancellation method of soft-decision.

Background technology

3-G (Generation Three mobile communication system) is to satisfy international mobile communication (the International Mobile Telecommunication 2000 that International Telecommunications Union proposes, be called for short " IMT-2000 ")/following public land mobile (the Future Public Land Mobile Telephone Systems of system, abbreviation " FPLMTS ") third generation mobile communication system of standard, requirement has good network compatibility, can realize the roaming between a plurality of different systems in the global range, not only will be for the mobile subscriber provide speech and low-rate data business, and multimedia service widely will be provided.According to this standard, Wideband Code Division Multiple Access (WCDMA) (Wideband Code Division Multiple Access has been proposed at present in the world, be called for short " WCDMA "), code division multiple access (Code Division Multiple Access, be called for short " CDMA ") 2000, time division CDMA (Time Division Code Division Multiple Access, be called for short " TD-CDMA "), TD SDMA (Time Division Synchronous CodeDivision Multiple Access is called for short " TD-SCDMA ") 3-G (Generation Three mobile communication system) scheme.Though these schemes are very not identical, the whole world adopts CDMA technology to reach common understanding in 3-G (Generation Three mobile communication system).

CDMA mobile communication system has advantages such as high power capacity, high quality-of-service and good confidentiality.But also there are many shortcomings accordingly, disturb (Mutiple Access Interference is called for short " MAI ") such as multiple access.In the cdma communication system of reality, there is certain correlation between each subscriber signal, Here it is, and multiple access disturbs the root that exists.The MAI that is produced by the individual user is no doubt very little, and along with the increase of number of users or the increase of signal power, MAI just becomes main an interference of cdma communication system, has directly limited the raising of cdma system capacity, coverage and performance.

Multiuser detection (Multi-user Detection is called for short " MUD ") is to overcome the key technology that multiple access disturbs in the cdma system, is a kind of Enhanced Technology that improves cdma system capacity, coverage and performance.Traditional detection technique is carried out the spreading code matching treatment according to classical direct sequence spread spectrum theory respectively to each user's signal fully, thereby anti-MAI interference performance is relatively poor; Multiuser detection is on the basis of traditional detection technology, make full use of and cause all subscriber signal information that MAI disturbs that the signal of unique user is detected, thereby has good interference free performance, solved near-far problem, reduced the requirement of system to the power control precision, therefore can effectively utilize the uplink spectrum resource more, significantly improve power system capacity.

Verdu in 1986 proposes to add viterbi algorithm with matched filter and realizes Maximum likelihood sequence detection (Maxinum-Likelihood Sequence Detection, be called for short " MLS detection "), be applicable to the channel that is subjected to intersymbol interference (Inter Symbol Interference is called for short " ISI ") influence.But the complexity of viterbi algorithm remains the index power level of number of users, i.e. 2 k power, and MLS detector need be known the amplitude and the phase place of received signal, and this will obtain by estimation.MLS detects too complicated, impracticable, the suboptimum multiuser detection that everybody is easy to realize in searching.The suboptimum multiuser detection is divided into two classes, and promptly linear multi-user detects and non-linear Multiuser Detection.Decorrelation is carried out in the output of conventional detector for the former or other linear transformation is beneficial to receive judgement, comprises methods such as decorrelation detection, least mean-square error detection, subspace oblique projection detection and multinomial expansion detection; The latter is different from the former, comprise interference cancellation (Interference Cancellation, be called for short " IC ") detection method, the interference cancellation method is considered as useful signal with the signal of desired user, other users' signal is considered as interference signal, from received signal, eliminates other users' interference earlier, obtain the signal of desired user, signal to desired user detects then, thereby improves the performance of system.

The interference cancellation multi-user test method is divided into again: successive interference cancellation (Serial InterferenceCancellation is called for short " SIC ") and parallel interference cancellation (Parallel Interference Cancellation is called for short " PIC ").SIC is made up of multistage, successively all user's sequence signals is adjudicated, reproduces, is eliminated in each level, and to give the following MAI that alleviate at different levels, each user's operating sequence is to determine according to the descending order of signal power.Be treated to example the first time with the first order, and its output is the strongest user's data judgement of signal and removes the later received signal of MAI that this user causes.Each time of first order processing subsequently in like manner and for it.Last result is that the weak more benefit of signal is many more.Subsequent stages is handled identical.SIC improves a lot than conventional detector on performance, and changes not quite on hardware, be easy to realize, but the SIC time-delay is bigger, need carry out the power ordering, and amount of calculation is bigger, and initialize signal is estimated sensitivity.PIC has multilevel hierarchy, and what be different from SIC is that the parallel MAI that estimates and remove that each user causes of its each grade disturbs, and carries out data decision then.Design philosophy and the SIC of PIC are basic identical, but because PIC is parallel processing, overcome the shortcoming of the big time-delay of SIC, and need not when situation changes, to resequence, it is little to have time-delay, the advantage that computational complexity is little has higher utility in various MUD, be the most possible at present method that realizes.

Interference cancellation method as mentioned above, after the restituted signal to any user carries out symbol judgement, all need be by the court verdict of this this symbol of user this subscriber signal of regenerating, and from received signal, eliminate this user's interference, promptly deduct this user's regenerated signal, thereby eliminate the influence that this subscriber signal detects other subscriber signals.

The method that the restituted signal to the user that present interference cancellation method adopts carries out symbol judgement has following several, as: hard decision method, this method are directly adjudicated according to the symbol of receiving terminal user's restituted signal; For another example: decision method and the soft decision method that in US Patent specification US5418814, propose based on fixed threshold; And for example: also have a kind of decision method at present, calculate two threshold values,, then do not participate in offseting processing if the court verdict value between these two threshold values, illustrates that the possibility that court verdict makes a mistake is bigger based on the error probability thresholding.

The purpose of these methods is: the increase of the noise power that the court verdict of minimizing mistake as far as possible causes.Prior probability based on the interference cancellation method conventional letter of soft-decision evenly distributes, and determines court verdict according to the coefficient of reliability and the hard decision result that calculate gained.

Above-mentioned interference cancellation method based on soft-decision comprises following steps:

Subscriber signal behind the receiving demodulation;

Carry out hard decision;

The computed reliability coefficient;

According to coefficient of reliability and the hard decision result subscriber signal of regenerating;

In described multiple user signals, described regeneration subscriber signal is offseted processing.

In actual applications, there is following problem in such scheme: because the prior probability and the uneven distribution of symbol in the actual conditions, cause the decision threshold that calculates reasonable inadequately, the method performance is not ideal enough, has influenced the raising of systematic function.Under the situation of the court verdict mistake of certain symbol of user, traditional soft decision method can cause that this symbol power of this user is not only eliminated in the later signal of interference cancellation, increase to original four times on the contrary, this detection to other subscriber signals is very unfavorable.

Cause the main cause of this situation to be, existing decision method based on soft-decision has supposed that in advance the prior probability of symbol evenly distributes, and its actual distribution is not estimated.

Summary of the invention

The technical problem to be solved in the present invention provides in a kind of code division multiple access system the interference cancellation method based on soft-decision, make more accurate based on soft decision method, thereby improve based on the performance of the interference cancellation method of soft-decision and the performance of code division multiple access system.

In order to solve the problems of the technologies described above, the invention provides in a kind of code division multiple access system based on the interference cancellation method of soft-decision, comprise following steps:

A receiving demodulation output signal is also carried out hard decision;

B determines this demodulated symbols prior probability value according to the demodulation result information before this demodulation;

C is according to described symbol prior probability value computed reliability coefficient;

D determines court verdict according to the hard decision result and the described coefficient of reliability of described demodulated output signal.

Wherein, described hard decision comprises following steps:

Whether judge described demodulated output signal less than 0, if then the hard decision result is set to-1, otherwise the hard decision result is set to 1.

Coefficient of reliability is calculated as follows described in the described step C:

$f(m,k)=1-2P_e(m,k)$

$=1-\frac{2}{1+\frac{P_1}{P_{-1}}{e}^{\frac{2A(m,k)Y(m,k)}{{\mathrm{\σ}}^2(m,k)}}}$ When Y (m, k) 〉=0 o'clock;

$f(m,k)=1-2P_e(m,k)$

$=1-\frac{2}{1+\frac{P_{-1}}{P_1}{e}^{-\frac{2A(m,k)Y(m,k)}{{\mathrm{\σ}}^2(m,k)}}}$ When Y (m, k)＜0 o'clock;

Wherein, for the k time demodulation of m symbol, f (m, k) coefficient of reliability, P for adjudicating _e(m, k) error probability, σ for adjudicating ²(m k) is the variance of noise in the signal after the demodulation, and ((m k) is the demodulated output signal value to Y to A, P for m, the k) distortion factor that described symbol is caused for channel fading ₁, P _-1Be respectively the prior probability value of separating the described symbol value of timing+1 ,-1 for the k time.

The method of described definite symbol prior probability value is: get the symbol posterior probability values that this demodulated symbols prior probability value equals a preceding demodulation result, and described symbol posterior probability values is calculated gained by supposing that a preceding demodulated symbols prior probability evenly distributes, and the ratio of m symbol being separated the described symbol prior probability of timing value for the k time is expressed as:

Wherein, P ₁, P _-1Be respectively the prior probability value of separating the described symbol value of timing+1 ,-1 for the k time, (m k-1) is the distortion factor that channel fading causes described symbol in the k-1 time demodulation of m symbol, Y (m to A, k-1) be the k-1 time demodulated output signal value of m symbol, σ ²(m k-1) is the variance of noise in the signal after the k-1 time demodulation of m symbol.

The method of described definite symbol prior probability value is taken as the ratio of the symbol posterior probability values of a preceding demodulation result for the ratio of the required symbol prior probability of this demodulation value, and described posterior probability values is formulated as by iterative computation gained repeatedly:

Wherein, P ₁, P _-1Be respectively the prior probability value of separating the described symbol value of timing+1 ,-1 for the k time, P ₁(m, k-2), P _-1(m k-2) is respectively the posterior probability values of separating the described symbol value of timing+1 ,-1 for the k-2 time, A (m, k-1) be the distortion factor that channel fading causes described symbol in the k-1 time demodulation of m symbol, (m k-1) is the k-1 time demodulated output signal value of m symbol, σ to Y ²(m k-1) is the variance of noise in the signal after the k-1 demodulation of m symbol.

Among the described step D, described court verdict is the hard decision result of described demodulated output signal and the product of described coefficient of reliability.

By relatively finding, technical scheme difference with the prior art of the present invention is, on U.S. Pat 5418814 disclosed interference cancellation method bases based on soft-decision, the present invention determines this demodulated symbols prior probability according to the demodulation result information before this demodulation in the demodulation repeatedly, and carry out coefficient of reliability according to the prior probability value and calculate, in conjunction with hard decision result and the coefficient of reliability that calculates, adjudicate final symbol again.

Difference on this technical scheme, brought comparatively significantly beneficial effect, promptly by accurate estimation to the symbol prior probability, improve the accuracy of coefficient of reliability calculating and the reliability of decision method, further improve the validity of interference cancellation method, thereby improved the performance of whole CDMA mobile communication systems.

Description of drawings

Fig. 1 is based on the decision method flow chart of the interference cancellation method of soft-decision in the code division multiple access system according to an embodiment of the invention.

Embodiment

For making the purpose, technical solutions and advantages of the present invention clearer, the present invention is described in further detail below in conjunction with accompanying drawing.

Soft decision method according to an embodiment of the invention is described earlier.

From certain user's of receiving terminal demodulation output signal with Y (m represents it is that the corresponding demodulation of m symbol is exported for m, k) expression, and k represents it is the output result of the k time demodulation, then Y (m k) can be expressed from the next:

Formula one Y (m, k)=A (m, k) d (m)+n (m, k)

Wherein, a (m) is this user's a m symbol, and value is+1 or-1; A (m k) is the property the taken advantage of distortion factor that channel fading causes symbol, and A (m, k)＞0; N (m k) is additive white Gaussian noise, obey the N that distributes just very much (0, σ ²(m, k)), average is 0, σ ²(m k) is variance.

Here, the n of normal distribution (m, probability density function k) is:

$f\left(x\right)=\frac{1}{\sqrt{2\mathrm{\π}}\mathrm{\σ}(m,k)}\exp (-\frac{x^2}{2{\mathrm{\σ}}^2(m,k)})$

Wherein, f is the probability density value, and x is the value variable, and exp () represents the exponent arithmetic of natural constant e.

Formula one can physical interpretation be: arrive receiving terminal when symbol a (m) sends through channel from transmitting terminal, channel makes this symbol produce the property the taken advantage of distortion (first) of channel fading and the additivity distortion (second) of white Gaussian noise.This is a public channel model.

The demodulation first time of subscriber signal is that baseband signal is carried out demodulation, and each demodulation after this all is to carry out after the interference cancellation of a preceding demodulation is handled, and also promptly the later baseband signal of a preceding interference cancellation is carried out demodulation.

To Y (m k) carries out soft-decision, the soft-decision result be ρ (m, k)=f (m, k) d (m, k).Wherein, (m k) is the coefficient of reliability of judgement to f, and (m k) is the hard decision result to d.The formula of hard decision is as follows:

Formula two

Above-mentioned soft-decision is the court verdict that satisfies bayesian criterion.Shown in three,

Formula three

E(C)＝A ²(m，k){(1-f(m，k)) ²(1-P _e(m，k))+(1+f(m，k)) ²P _e(m，k)}

＝MINIMUN

(m k) makes wrong cost C=A to soft-decision ρ ²(m, k) (a (m)-ρ (m, k)) ²The average minimum.Wherein, P _e(m, k) error probability of the k time judgement of m symbol of expression.

With following formula to f (m, k) differentiate, and make that derivative is 0 can obtain:

Formula four f (m, k)=1-2P _e(m, k)

In U.S. Pat 5418814 disclosed interference cancellation methods, directly suppose a (m) value for+1 and-1 prior probability equate, be all P{a (m)=1}=P{a (m)=-1}=.And under this hypothesis, obtain coefficient of reliability f (m, computing formula k):

Formula five $f(m,k)=1-2P_e(m,k)=\tanh \left\{\frac{A(m,k)\left|Y(m,k)\right|}{{\mathrm{\σ}}^2(m,k)}\right\}$

Under the situation that can't obtain any prior information, be merely able to make a (m) value and be the equal hypothesis of+1 and-1 prior probability.The least favorable situation that judgement under this hypothesis comes down to adjudicate.Therefore, be+1 and-1 prior probability if can estimate to obtain a (m) value, and (m, k), then (m, it is more accurate that value k) is bound to for f to utilize this information to obtain f.

The present invention is at priori probability P { a (m)=1}=P ₁, P{a (m)=-1}=P _-1It is as follows to obtain the coefficient of reliability formula under the condition:

Formula six a $f(m,k)=1-2P_e(m,k)$

$=1-\frac{2}{1+\frac{P_1}{P_{-1}}{e}^{\frac{2A(m,k)Y(m,k)}{{\mathrm{\σ}}^2(m,k)}}}$ When Y (m, k) 〉=0 o'clock;

Formula six b $f(m,k)=1-2P_e(m,k)$

$=1-\frac{2}{1+\frac{P_{-1}}{P_1}{e}^{-\frac{2A(m,k)Y(m,k)}{{\mathrm{\σ}}^2(m,k)}}}$ When Y (m, k)＜0 o'clock;

In the present invention, when carrying out the k time judgement, the posterior probability P that obtains during with (k-1) inferior judgement ₁(m, k-1)=P{a (m)=1|Y (m, k-1) }, P _-1(m, k-1)=P{a (m)=-1|Y (m, k-1) } when replacing the k time judgement, the prior probability P that uses in the formula six ₁And P _-1

Calculate posterior probability P according to following two formulas then ₁(m, k-1), P _-1(m, k-1):

Formula seven $P_1(m,k-1)=P\{a\left(m\right)=1|Y(m,k-1)>0\}$

$=\frac{P_1{f}_1\left(Y(m,k-1)\right)}{P_1{f}_1\left(Y(m,k-1)\right)+P_{-1}{f}_{-1}\left(Y(m,k-1)\right)}$

Formula eight $P_{-1}(m,k-1)=P\{a\left(m\right)=-1|Y(m,k-1)<0\}$

$=\frac{P_{-1}{f}_{-1}\left(Y(m,k-1)\right)}{P_1{f}_1\left(Y(m,k-1)\right)+P_{-1}{f}_{-1}\left(Y(m,k-1)\right)}$

Wherein,

$f_1\left(y\right)=\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}(m,k)}\exp (-\frac{{(y-A(m,k))}^2}{2{\mathrm{\&sigma;}}^2(m,k)})$

$f_2\left(y\right)=\frac{1}{\sqrt{2\mathrm{\&pi;}}\mathrm{\&sigma;}(m,k)}\exp (-\frac{{(y+A(m,k))}^2}{2{\mathrm{\&sigma;}}^2(m,k)})$

By formula seven and formula eight, the ratio of posterior probability is:

Formula nine $\frac{P_{-1}(m,k-1)}{P_1(m,k-1)}=\frac{P_{-1}}{P_1}\&CenterDot;\frac{f_{-1}\left(Y(m,k-1)\right)}{f_1\left(Y(m,k-1)\right)}$

When formula was calculated the ratio of posterior probability in the use, we can adopt two kinds of specific practices:

First kind of way, suppose: $P_1=\frac{1}{2},$ $P_{-1}=\frac{1}{2},$ The ratio that then can obtain posterior probability is:

Formula ten a $\frac{P_{-1}(m,k-1)}{P_1(m,k-1)}=\frac{f_{-1}\left(Y(m,k-1)\right)}{f_1\left(Y(m,k-1)\right)},$ When k＞1.

$=e^{-\frac{2A(m,k-1)Y(m,k-1)}{{\mathrm{\&sigma;}}^2(m,k-1)}}$

When k=1, without any posterior probability information.For consistent with formula ten a of k＞1 o'clock, order

Formula ten b $\frac{P_{-1}(m,0)}{P_1(m,0)}=1$

Second kind of way replaces the ratio of the prior probability in the formula nine with the ratio of the posterior probability of the k-2 time judgement, even $\frac{P_{-1}}{P_1}=\frac{P_{-1}(m,k-2)}{P_1(m,k-2)},$ And substitution formula nine, then can obtain the recurrence formula of the ratio of following posterior probability:

Formula 11 a $\frac{P_{-1}(m,k-1)}{P_1(m,k-1)}=\frac{P_{-1}(m,k-2)}{P_1(m,k-2)}\&CenterDot;\frac{f_{-1}\left(Y(m,k-1)\right)}{f_1\left(Y(m,k-1)\right)},$ When

$=\frac{P_{-1}(m,k-2)}{P_1(m,k-2)}\&CenterDot;e^{\frac{2A(m,k-1)Y(m,k-1)}{{\mathrm{\&sigma;}}^2(m,k-1)}}$

k＞1。

When k=1, can utilize order without any posterior probability information:

Formula 11 b $\frac{P_{-1}(m,0)}{P_1(m,0)}=1$

Value with formula ten or formula 11

In the replacement formula six Just can obtain as shown in the formula the k time the judgement coefficient of reliability:

Formula 12 a

$f(m,k)=1-2P_e(m,k)$

$=1-\frac{\text{2}}{1+\frac{P_1(m,k-1)}{P_{-1}(m,k-1)}{e}^{\frac{2A(m,k)Y(m,k)}{{\mathrm{\&sigma;}}^2(m,k)}}}$ When Y (m, k) 〉=0 o'clock

Formula 12 b

$f(m,k)=1-2P_e(m,k)$

$=1-\frac{2}{1+\frac{P_{-1}(m,k-1)}{P_1(m,k-1)}{e}^{-\frac{2A(m,k)Y(m,k)}{{\mathrm{\&sigma;}}^2(m,k)}}}$ When Y (m, k)＜0 o'clock

According to hard decision result and formula 12 computation's reliability coefficients, it is as follows to obtain the soft-decision result:

Formula 13 a

$\mathrm{\&rho;}(m,k)=f(m,k)d(m,k)$

$=1-\frac{2}{1+\frac{P_1(m,k-1)}{P_{-1}(m,k-1)}{e}^{\frac{2A(m,k)Y(m,k)}{{\mathrm{\&sigma;}}^2(m,k)}}}$ When Y (m, k) 〉=0 o'clock

Formula 13 b

$\mathrm{\&rho;}(m,k)=f(m,k)d(m,k)$

$=-1+\frac{2}{1+\frac{P_{-1}(m,k-1)}{P_1(m,k-1)}{e}^{-\frac{2A(m,k)Y(m,k)}{{\mathrm{\&sigma;}}^2(m,k)}}}$ When Y (m, k)＜0 o'clock

The soft-decision result here can be used for the regeneration of signal, and concrete interference cancellation method can adopt the structure of parallel interference cancellation or the structure of successive interference cancellation, adopts formula 13 to calculate the soft-decision result when symbol judgement.Posterior probability among the soft-decision result can adopt formula ten or formula 11 to calculate.

In a preferred embodiment of the present invention, the user has finished demodulation for the first time, and when the first time, demodulation result was carried out the symbol soft-decision, owing to do not have any prior information of a (m) value, according to the soft-decision result of formula ten b and formula 13 compute signs.Calculate according to formula ten a then

This value is used for symbol judgement soft-decision result's calculating for the second time.After obtaining the court verdict of user symbol, carry out this user's signal regeneration and follow-up processing such as interference cancellation by this court verdict.

Finished the user and to have separated timing for the second time, judgement has for the first time been obtained Bring formula 13 into, calculate the soft-decision result of this symbol.Calculate according to formula ten a then Be used for the calculating of decision threshold for the third time.

Other successive character judging process to the user carry out identical processing.

In another preferred embodiment of the present invention, the user at first finishes demodulation for the first time, when the first time, demodulation result was carried out the symbol soft-decision, owing to do not have any prior information of a (m) value, according to the soft-decision result of formula 11 b and formula 13 compute signs.And calculate according to formula 11 a This value is used for symbol judgement for the second time.After obtaining the court verdict of user symbol, carry out this user's signal regeneration and follow-up processing such as interference cancellation by this court verdict.

Finished the user and to have separated timing for the second time, judgement has for the first time been obtained Bring formula 13 into, calculate the soft-decision result of this symbol.Calculate according to formula 11 a then This value is used for the calculating of soft-decision for the third time.

Other successive character judging process to the user carry out identical processing.

Below according to this decision method, and with reference to Fig. 1, describe the step of decision method of the k time demodulation of the interference cancellation method of corresponding m subscriber signal in detail.

As shown in Figure 1, at first enter step 101, receive this demodulated output signal Y (m, k).The expression formula of described demodulated output signal is a formula one.

Then enter step 102, (m k) carries out hard decision to Y.The formula of hard decision is seen formula two.

Then enter step 103, determine this demodulated symbols prior probability value according to the demodulation result information before this demodulation, computational methods see that the difference front of the computational methods that formula ten or formula 11, two formulas are represented addresses; When calculating the prior probability value here, need use a preceding demodulation result parameter, suppose that then prior probability evenly distributes in the time of for the first time.

Then enter step 104, according to symbol prior probability computed reliability coefficient.Computational methods are seen formula 12.

Then enter step 105, carry out symbol judgement according to coefficient of reliability and hard decision result.Concrete computational methods be ρ (m, k)=(m, k) (m k), can use directly also that final computing formula---formula 13 is calculated to d to f.

Though by reference some preferred embodiment of the present invention, the present invention is illustrated and describes, but those of ordinary skill in the art should be understood that, can do various changes to it in the form and details, and the spirit and scope of the present invention that do not depart from appended claims and limited.

Claims (6)

1. In the code division multiple access system based on the interference cancellation method of soft-decision, it is characterized in that, comprise following steps:

   A receiving demodulation output signal is also carried out hard decision;

   B determines this demodulated symbols prior probability value according to the demodulation result information before this demodulation;

   C is according to described symbol prior probability value computed reliability coefficient;

   D determines court verdict according to the hard decision result and the described coefficient of reliability of described demodulated output signal.
2. 2. based on the interference cancellation method of soft-decision, it is characterized in that in the code division multiple access system according to claim 1 that described hard decision comprises following steps:

   Whether judge described demodulated output signal less than 0, if then the hard decision result is set to-1, otherwise the hard decision result is set to 1.
3. 3. based on the interference cancellation method of soft-decision, it is characterized in that in the code division multiple access system according to claim 2 that coefficient of reliability is calculated as follows described in the described step C:

   $f(m,k)=1-2P_e(m,k)$

   $=1-\frac{2}{1+\frac{P_1}{P_{-1}}{e}^{\frac{2A(m,k)Y(m,k)}{{\mathrm{\&sigma;}}^2(m,k)}}}$ When Y (m, k) 〉=0 o'clock;

   $f(m,k)=1-2P_e(m,k)$

   $=1-\frac{2}{1+\frac{P_{-1}}{P_1}{e}^{-\frac{2A(m,k)Y(m,k)}{{\mathrm{\&sigma;}}^2(m,k)}}}$ When Y (m, k)＜0 o'clock;

   Wherein, for the k time demodulation of m symbol, f (m, k) coefficient of reliability, P for adjudicating _e(m, k) error probability, σ for adjudicating ²(m k) is the variance of noise in the signal after the demodulation, and ((m k) is the demodulated output signal value to Y to A, P for m, the k) distortion factor that described symbol is caused for channel fading ₁, P _-1Be respectively the prior probability value of separating the described symbol value of timing+1 ,-1 for the k time.
4. 4. in the code division multiple access system according to claim 3 based on the interference cancellation method of soft-decision, it is characterized in that, the method of described definite symbol prior probability value is: get the symbol posterior probability values that this demodulated symbols prior probability value equals a preceding demodulation result, and described symbol posterior probability values is calculated gained by supposing that a preceding demodulated symbols prior probability evenly distributes, and the ratio of m symbol being separated the described symbol prior probability of timing value for the k time is expressed as:

   Wherein, P ₁, P _-1Be respectively the prior probability value of separating the described symbol value of timing+1 ,-1 for the k time, (m k-1) is the distortion factor that channel fading causes described symbol in the k-1 time demodulation of m symbol, Y (m to A, k-1) be the k-1 demodulated output signal value of m symbol, σ ²(m k-1) is the variance of noise in the signal after the k-1 demodulation of m symbol.
5. 5. in the code division multiple access system according to claim 3 based on the interference cancellation method of soft-decision, it is characterized in that, the method of described definite symbol prior probability value is taken as the ratio of the symbol posterior probability values of a preceding demodulation result for the ratio of the required symbol prior probability of this demodulation value, and described posterior probability values is passed through repeatedly iterative computation gained, is formulated as:

   

   Wherein, P ₁, P _-1Be respectively the prior probability value of separating the described symbol value of timing+1 ,-1 for the k time, P ₁(m, k-2), P _-1(m k-2) is respectively the posterior probability values of separating the described symbol value of timing+1 ,-1 for the k-2 time, A (m, k-1) be the distortion factor that channel fading causes described symbol in the k-1 time demodulation of m symbol, (m k-1) is the k-1 time demodulated output signal value of m symbol, σ to Y ²(m k-1) is the variance of noise in the signal after the k-1 demodulation of m symbol.
6. 6. based on the interference cancellation method of soft-decision, it is characterized in that in the code division multiple access system according to claim 1 that among the described step D, described court verdict is the hard decision result of described demodulated output signal and the product of described coefficient of reliability.

Images