CN1917487B - Method for distributing training sequence for time-sharing system - Google Patents

Abstract

A method for distributing training sequence in time-sharing system is suitable for the condition that only one user exists in single time slot of cell, including: (1) determining the number of different cyclic shift versions of a basic training sequence and the length of a channel estimation window according to system configuration; (2) generating cyclic shift versions according to the number of the cyclic shift versions and the length of a channel estimation window determined in the step (1); (3) different cyclic shifted versions of the same basic training sequence are allocated for each cell. The invention applies the distribution method of the training sequence of the single code set to a plurality of cells under the condition that the single time slot of the cell of the time-slot system only has one user, thereby reducing the mutual interference of different training sequences adopted among the cells in the original system and reducing the complexity of a channel estimation algorithm.

Description

A kind of distribution method that is used for the training sequence of sub time slot sytem

Technical field

The present invention relates to the mobile communication technology field, be specifically related to a kind of distribution method that is used for the training sequence of sub time slot sytem.

Background technology

Because the complexity and the time variation of channel in the coherent reception process, need be estimated channel at receiving terminal in the wireless mobile communications, utilize the channel response that obtains that signal is carried out coherent detection then.Except the application in input, channel estimating also plays important fundamental role in correlation subsystem such as physical layer measurement, smart antenna, quick control, switching and RRM.

In time division multiplexing S-CDMA (TD-SCDMA) system; The structure of business time-slot burst is as shown in Figure 1, and burst is respectively by two length that the data block, of 352chips is long to be formed for the guard time of 16chips for the midamble sign indicating number of 144chips and one are long at interval.The intermediate code (midamble) at burst middle part is used for carrying out channel estimating, and said intermediate code also can be described as channel estimating code.The data block on both sides is used for transmit traffic data.

Above-mentioned training sequence is to generate like this: for the same time slot of same sub-district, and a given basic midamble sign indicating number, different users adopts the training sequence of the different cyclic shift versions of this identical basic midamble sign indicating number as it.By same basic code m _PThe K that derives a specific training sequence shifted version m ^(k)K=1...K constitutes set, is called the training sequence collection.In the TD-SCDMA system, the cycle of basic midamble sign indicating number is 128, and the code length of the training sequence shifted version of derivation is Lm=128+16.

Slot CDMA system carries out the method for channel estimating; Be called the Steiner estimator; Detailed content is referring to document B.Steiner, and P.W.Baier, " Low Cost Channel Estimation in the UplinkReceiver of CDMA Mobile Radio Systems; " FREQUENZE, 47 (1993) 11-12).This channel estimation methods can estimate the channel estimation results that belongs to a plurality of users of a basic code collection in this sub-district simultaneously, suppresses to belong to the interference between a plurality of users' the intermediate code of same basic code.This method is the method for a plurality of users that belong to same basic code collection being carried out channel estimating, also claims solid size collection channel estimation methods.

Handle because solid size collection channel estimation methods is all treated as white Gauss noise to other signal except that this yard collection signal response of same time stack, so solid size collection channel estimation methods itself is just perfect inadequately.And, for the slot CDMA system of minizone synchronous working, be in the signal of the adjacent community user of this sub-district intersection byer force to this area interference, and the signal of co-frequency neighbor cell signal and this sub-district is a slot synchronization.That is to say, the interference of the homogeneous frequency adjacent area that intermediate code response in this sub-district is suffered mainly from the response of the intermediate code signal of other yard collection of this cell synchronizing.In addition, in other some are used, the situation of the intermediate code response signal stack of a plurality of yards also might emergent power close collection.

Therefore, slot CDMA system is under the situation of many code channels of homogeneous frequency adjacent area work, and the method for solid size collection channel estimating often can not satisfy the requirement of systematic function, need provide performance better channel estimation methods.

So the inventor has mentioned many yards collection channel estimation methods in the patent of many yards collection of disclosed 03100670 time slot CDMA system on August 11st, 2004 channel estimation methods.Many yards collection channel estimation methods are primarily aimed at the situation that has many yards collection channel estimating code response signals, realize multi-code set associated channel estimating, and improve solid size collection channel estimating performance.

When only there is a user in single time slot in the sub time slot sytem, only need to distribute a cyclic shift version of training sequence in sub-district.In this case, existing implementation can be divided into two kinds:

(1) adopts solid size collection channel estimation methods.This method is used as the training sequence code of adjacent sub-district as interference, does not do any processing in the channel estimating part of receiving terminal.

(2) adopt many yards collection channel estimation methods.This method need be known the basic training sequences sign indicating number that use adjacent sub-district; Utilize the patent of many yards collection of 03100670 time slot CDMA system channel estimation methods; Each training sequence code to neighbor cell carries out channel estimating respectively; Carry out repeatedly interference counteracting and iteration then,, improve the accuracy of channel estimating to reduce of the interference of adjacent sub-district training sequence code to this sub-district training sequence code.

Yet there is deficiency respectively in above-mentioned two kinds of implementation methods.Solid size collection channel estimation methods amount of calculation is less, but owing to do not consider the interference of adjacent sub-district training sequence, so the detection poor-performing of channel estimating.Especially when the user was in cell edge, the interference that the training sequence of adjacent sub-district causes was very big, may cause the user normally to receive, and formed Zone switched blind area.Make the system can't operate as normal.

Many yards collection channel estimation methods detect the training sequence of this sub-district and adjacent sub-district simultaneously, and utilize repeatedly the interference cancellation method of iteration, can effectively eliminate the interference of adjacent sub-district, improve channel estimated accuracy.But many yards collect channel estimation methods owing to need carry out the channel estimating of many sub-districts simultaneously, and need carry out interference eliminated computing repeatedly, have improved the complexity of calculating greatly.The iterations of interference counteracting is relevant with the multipath distribution of the quantity of each community user and each subscriber channel, and under the prerequisite that can't obtain these message, iterations is difficult to confirm that iterations too much all can influence final precision of channel estimation with very few.Therefore also there is the high and unsettled shortcoming of estimated accuracy of computation complexity in many yards collection channel estimation methods.

Summary of the invention

The object of the present invention is to provide a kind of distribution method that is used for the training sequence of sub time slot sytem, have only the channel estimating that adopts prior art to cause under a user's the situation to detect poor performance and computing complicated technology problem to solve the single time slot in sub-district.

For addressing the above problem, the present invention discloses a kind of distribution method that is used for the training sequence of sub time slot sytem, is applicable to that the single time slot in sub-district has only a user's situation.In the method, system is the different cyclic shift versions of each cell allocation based on identical basic training sequences.

A kind of distribution method that is used for the training sequence of sub time slot sytem comprises: the quantity K and the long W of channel estimation window of the different cyclic shift versions of basic training sequences confirmed in (1) according to system configuration; (2) long according to the quantity and the channel estimation window of the cyclic shift version of confirming in the step (1), generate the cyclic shift version; (3) be the different cyclic shift versions of each cell allocation based on identical basic training sequences.

The cyclic shift version of sub-district and distribution has definite corresponding relation.For the cyclic shift version of each cell allocation can be 1, also can be more than 1.

The length of each channel estimation window can be identical in step (1) and the step (2), also can be different.

When the length of each channel estimation window not simultaneously; Step (1) comprises quantity K and the long W of channel estimation window that confirms the different cyclic shift versions of basic training sequences according to system configuration, and the quantity K of the different cyclic shift versions of the length P of the long W of channel estimation window, basic training sequences and basic training sequences satisfies W _nThe channel estimation window that is n sub-district is long.

When the length of each channel estimation window not simultaneously, the process that step (2) generates the cyclically shifted sequences version comprises: A definition length is the basic training sequences of P m _P=( m ₁, m ₂..., m _P), with vector m _pCyclic extensions is to length i _Max,

L in the formula _mThe length of each the cyclic shift version that sends for need, W _nThe channel estimation window that is n sub-district is long; B obtains a new vector m, Work as i=1 ..., during P, m _i= m _p, as i=(P+1) ..., i _MaxThe time, m _i= m _I-PC uses the new vector that obtains among the step B mGenerate each cyclic shift version, k length is L _mThe cyclic shift version do Wherein work as k=1,2 ..., during K-1, the cyclic shift version m ^(k)In each element do

I=1 ..., L _m, K length is L _mThe cyclic shift version in each element

I=1 ..., L _m

Each sub-district described in the present invention can be the sub-district in after the cell group in the sub time slot sytem every group.

Compared with prior art; The present invention has the following advantages: have only under a user's the situation at sub time slot sytem list time slot; The method of salary distribution of the solid size training being practiced sequence is applied to a plurality of sub-districts, compares with single cell channel estimation method, has eliminated the interference of each minizone training sequence; Compare with many yards collection channel estimation methods, reduced the complexity of channel estimation calculation, and accuracy of channel estimation increases all than these two kinds of methods.

Description of drawings

Fig. 1 is existing TD-SCDMA system business time slot burst structure sketch map;

Fig. 2 is the sketch map of the distribution method of a kind of training sequence that is used for sub time slot sytem disclosed by the invention;

Fig. 3 is a kind of sketch map that generates the cyclic shift version method disclosed by the invention.

Embodiment

Below in conjunction with accompanying drawing 2 and accompanying drawing 3, specify the present invention.

According to preceding surface analysis, have only under the situation that a user transmits at each sub-district synchronization, we can stipulate that a plurality of sub-districts only use a basic training sequences, the user of each sub-district uses the different cyclic shift versions of this basic training sequences.When receiving terminal carries out channel estimating, can just can estimate required multi-cell channel estimated result according to the channel estimation method of existing single sub-district.Concrete exercise sequence distributing method is following:

Step S101: the quantity K and the long W of channel estimation window that confirm the different cyclic shift versions of basic training sequences according to system configuration.W can be according to the length P of basic training sequences, confirm based on the quantity K of the different cyclic shift versions of basic training sequences.Confirm that criterion is following:

In the formula; expression rounds downwards x, promptly finds the maximum integer that is less than or equal to x.For example, for the TD-SCDMA system, P=128, K gets 4, and W=32 then representes that one can generate 4 cyclic shift versions by basic training sequences, supplies 4 neighbor cells to use simultaneously, and each channel estimation window length is 32.

Step S102: quantity and channel estimation window according to the definite cyclic shift version of step S101 are long, generate each cyclic shift version.

According to cyclic shift version quantity K and the long W of channel estimation window that step S101 confirms, the method that generates each cyclic shift version is following:

With length is the vector that the basic training sequences of P is defined as following formula m _p

m _P＝( m ₁， m ₂，...， m _P)

In order to generate the cyclically shifted sequences of different editions, with vector m _pCyclic extensions is length i _Max, i _MaxValue be i _Max=L _m+ (K-1) W, L in the formula _mThe length of each the cyclic shift version that sends for need.

Like this, our can obtain constituting one new vector by basic training sequences m, define formula as follows:

$\underset{\‾}{m}=({\underset{\‾}{m}}_1,{\underset{\‾}{m}}_2,\·\·\·,{\underset{\‾}{m}}_{i_{\max }})=({\underset{\‾}{m}}_1,{\underset{\‾}{m}}_2,\·\·\·,{\underset{\‾}{m}}_{L_m+(K-1)W}),$

Vector mPreceding P element and vector m _pPreceding P element identical, the element of back is more successively from vector m _pFirst order of elements is taken out and is arranged:

m _i= m _pI=1 wherein ..., P

m _i= m _I-PI=(P+1) wherein ..., i _Max

The cycle training sequence of using said method to obtain mJust can generate each cyclic shift version.To k length is L _mThe cyclic shift version m ^(k)The generation method following:

${\underset{\‾}{m}}^{\left(k\right)}=({\underset{\‾}{m}}_1^{\left(k\right)},{\underset{\‾}{m}}_2^{\left(k\right)},\·\·\·,{\underset{\‾}{m}}_{L_m}^{\left(k\right)}),$

Wherein each element is:

${\underset{\‾}{m}}_i^{\left(k\right)}={\underset{\‾}{m}}_{i+(K-k)W}$ I=1 wherein ..., L _mK=1 ..., K

K based on the basic training sequences sign indicating number m _pThe cyclic shift version m ^(k)K=1...K has constituted a training sequence collection.

Step S103: be the different cyclic shift versions of each cell allocation based on identical basic training sequences.

In sub time slot sytem; Because the number of the cyclic shift version of training sequence is limited; When sub-district in the system reach a certain number of after, during for example greater than cyclic shift version number, we need divide into groups intrasystem sub-district; Grouping can be adopted multiple mode, as according to the geographical position, have only a user's neighbor cell to divide at one group single time slot as far as possible.After the cell group the different cyclic shift versions of every group of interior cell allocation based on basic training sequences.

To collect by the sign indicating number that step S102 obtains m ^(k)K=1...K distributes to k sub-district according to certain distribution principle, and one or more cyclic shift version is used in each sub-district.The cyclic shift version of sub-district and distribution has definite corresponding relation.The distribution principle here can be the distribution method of order assignment, also can be the distribution principle according to the sub-district distributes or other principles are confirmed.

When practical application, each sub-district is sent the cyclic shift version of distributing to this sub-district simultaneously.According to the algorithm principle of Steiner estimator, behind the Steiner of receiving terminal estimator, obtaining total length by these cyclic shift versions is i _Max, by total channel estimation results that K the channel estimation window length channel estimating that is W is formed, the channel estimating of a corresponding k sub-district, thus realize being undertaken the channel estimating of a plurality of sub-districts by a sign indicating number collection.

Have only a user as long as satisfy the single time slot in sub-district, cyclic shift version and sub-district are the corresponding relations of confirming, and the corresponding a plurality of sub-districts of basic training sequences, and said method all is suitable for.

Be generation and the distribution method that example has been explained training sequence cyclic shift version with the isometric situation of each cell channel estimating window above, utilize similar mode, also can this method be expanded to the long unequal situation of each cell channel estimating window.

When each cell channel estimating window length was unequal, the step S101 of distribution method and S102 need make following modification:

Confirm the quantity K and the long W of channel estimation window of the different cyclic shift versions of basic training sequences according to system configuration.The quantity K of the different cyclic shift versions that W can generate according to the length P of basic training sequences, based on basic training sequences confirms.Confirm that criterion is following:

When each cell channel estimating window is long when not waiting, channel estimation window is long, the quantity K of the different cyclic shift versions of basic training sequences and The relation of basic training sequences length P satisfies following formula:

$P\≥\underset{n=1}{\overset{K}{\mathrm{\Σ}}}{W}_n$

In the formula, W _nThe channel estimation window that is the n sub-district is long, n=1, and 2 ..., K-1, K are the quantity of the different cyclic shift versions of basic training sequences.

In order to generate the cyclically shifted sequences of different editions, be the basic training sequences of P with definition length m _p, m _P=( m ₁, m ₂..., m _P), with vector m _pCyclic extensions is length i _Max, i _MaxValue do L in the formula _mThe length of each the cyclic shift version that sends for need.

Like this, we can obtain a new vector that is made up of basic training sequences m, define formula as follows:

$\underset{\‾}{m}=({\underset{\‾}{m}}_{1,}{\underset{\‾}{m}}_2...,{\underset{\‾}{m}}_{i_{\max }})=({\underset{\‾}{m}}_{1,}{\underset{\‾}{m}}_2...,{\underset{\‾}{m}}_{L_m}+\underset{n=1}{\overset{K-1}{\mathrm{\Σ}}}{W}_n)$

Vector mPreceding P element and vector m _pPreceding P element identical, the element of back is more successively from vector m _pFirst order of elements is taken out and is arranged:

m _i= m _pI=1 wherein ..., P

m _i= m _I-PI=(P+1) wherein ..., i _Max

The cycle training sequence of using said method to obtain mJust can generate each cyclic shift version.To k length is L _mThe cyclic shift version m ^(k)The generation method following:

${\underset{\‾}{m}}^{\left(k\right)}=({\underset{\‾}{m}}_1^{\left(k\right)},{\underset{\‾}{m}}_2^{\left(k\right)},...,{\underset{\‾}{m}}_{L_m}^{\left(k\right)})$

Cyclic shift version wherein m ^(k)Preceding k (k=1,2, ..., K-1) each element of individual cyclic shift version is:

${\underset{\‾}{m}}_i^{\left(k\right)}={\underset{\‾}{m}}_{i+\underset{n=k}{\overset{K-1}{\mathrm{\Σ}}}{W}_n}$

I=1 wherein ..., L _mK=1,2 ..., K-1.

Each element of K cyclic shift version is:

${\underset{\‾}{m}}_i^{\left(K\right)}={\underset{\‾}{m}}_i,i=1,...,L_m$

K based on basic training sequences sign indicating number m _PThe cyclic shift version m ^(k)K=1...K has constituted a training sequence collection.

More than disclosedly be merely several specific embodiment of the present invention, but the present invention is not limited thereto, any those skilled in the art can think variation, all should drop in protection scope of the present invention.

Claims (7)

1. a distribution method that is used for the training sequence of sub time slot sytem is applicable to that the single time slot in sub-district has only a user's situation, it is characterized in that, is the different cyclic shift versions of each cell allocation based on identical basic training sequences,

Comprise the steps:

(1) confirms the quantity K and the long W of channel estimation window of the different cyclic shift versions of basic training sequences according to system configuration;

(2) long according to the quantity and the channel estimation window of the cyclic shift version of confirming in the step (1), generate the cyclic shift version;

(3) be the different cyclic shift versions of each cell allocation based on identical basic training sequences,

Wherein, Step (1) comprises quantity K and the long W of channel estimation window that confirms the different cyclic shift versions of basic training sequences according to system configuration, and the quantity K of the different cyclic shift versions of the length P of the long W of channel estimation window, basic training sequences and basic training sequences satisfies

W _nThe channel estimation window that is n sub-district is long.

2. a kind of distribution method that is used for the training sequence of sub time slot sytem as claimed in claim 1 is characterized in that the cyclic shift version of sub-district and distribution has definite corresponding relation.

3. a kind of distribution method that is used for the training sequence of sub time slot sytem as claimed in claim 1 is characterized in that, described each sub-district is the sub-district in every group after the cell group in the sub time slot sytem.

4. a kind of distribution method that is used for the training sequence of sub time slot sytem as claimed in claim 1 is characterized in that, step (1) is identical with the length of each channel estimation window in the step (2).

5. a kind of distribution method that is used for the training sequence of sub time slot sytem as claimed in claim 1 is characterized in that, step (1) is different with the length of each channel estimation window in the step (2).

6. a kind of distribution method that is used for the training sequence of sub time slot sytem as claimed in claim 5 is characterized in that, the process that step (2) generates the cyclically shifted sequences version comprises:

A definition length is the basic training sequences of P m _P=( m ₁, m ₂..., m _P), with vector m _pCyclic extensions is to length i _Max,

L in the formula _mThe length of each the cyclic shift version that sends for need, W _nThe channel estimation window that is n sub-district is long;

B obtains a new vector m, Work as i=1 ..., during P, m _i= m _p, as i=(P+1) ..., i _MaxThe time, m _i= m _I-P

C uses the new vector that obtains among the step B mGenerate each cyclic shift version, k length is L _mThe cyclic shift version do

Wherein work as k=1,2 ..., during K-1, the cyclic shift version m ^(k)In each element do

I=1 ..., L _m, K length is L _mThe cyclic shift version in each element I=1 ..., L _m

7. a kind of distribution method that is used for the training sequence of sub time slot sytem as claimed in claim 1 is characterized in that, for the cyclic shift version of each cell allocation is 1, or more than 1.

Images