CN101147349B - A pre-filtering SDMA transmitting and receiving method, system and a pre-filter thereof - Google Patents

Abstract

A pre-filtering SDMA transmitting and receiving method and system, and a filter thereof and its generating method, the SDMA transmitting system calculates the complete diagonalization pre-filter matrix composed of U sub-blocks satisfying block diagonalization condition and complete diagonalization condition; at the same time, after the data stream which the said transmitting system sends to the users passes the pre-filter, it is transmitted through numbers of antennas synchronously, and reaches each user terminal via air channel; after the terminal receives the transmitting signal, judges the received signal directly, and recovers the data stream sent to itself, at the same time, feeds the measured channel characteristic back to the said transmitting system ,the channel is the sub-channel from the said transmitting system to the terminal. In addition to eliminate the multi-user interference, the present invention can also eliminate the interference between various data streams of the user, then simplifies the design of the receiver.

Description

The space division multiple access method of sending and receiving of pre-filtering, system and used prefilter

Technical field

The present invention relates to space division multiple access (the Space Division MultipleAccess in the mobile communication system, SDMA) system, the space division multiple access that particularly is directed to multi-user's multiple-input, multiple-output (MIMO:Multiple Inputand Multiple Output) system down link pre-filtering sends (prefilter) method, system, prefilter and generation method thereof.

Background technology

Compare with a single aerial system, the MIMO technology has been brought significant capacity or diversity gain owing to utilized a plurality of parallel space channels, therefore has been subjected to general attention in recent years.Up to the present, the research in this field also mainly is conceived to the point-to-point multiple antenna communication of single user, and does not consider the common-channel interference between the multi-user.Until recently, the attentiveness of people's research begins to transfer to multi-user MIMO system.Multi-user MIMO system exists bigger difference for the Single User MIMO system, the common-channel interference that need not consider in the single user system if do not obtain good treatment, will cause the severe exacerbation of systematic function.The method that TDMA technology and CDMA technology all once were used as the antagonism common-channel interference was suggested in multi-user MIMO system, but the two all exists significantly deficiency.The spectrum efficiency of TDMA technology is too low, so system is not high with capacity, can not embody the jumbo advantage of mimo system.The problem that CDMA technology exists is that multi-user MIMO system need consume a large amount of sign indicating number resources, and system is difficult to and can meets this requirement.

Compare with CDMA with TDMA, SDMA (Spatial Division Multiple Access, space division multiple access) technology is because its intrinsic high spectrum utilization, and do not consume a yard resource, is to eliminate a up-and-coming developing direction of disturbing in the multi-user MIMO system.In addition, if SDMA is combined with TDMA or CDMA, can further improve the flexibility of system applies.The key technology of SDMA is the interference of how eliminating between MUI (Multi-user Interference, multi-user interference) and inner each the circuit-switched data stream of each user.Because the asymmetry of data service itself, generally descending often more urgent than up to two-forty, jumbo demand, so this technology also is mainly used in down link.

Below three pieces of lists of references proposed to eliminate the method for MUI:

《Joint?Tx-Rx?optimization?for?MIMO-SDMA?based?on?a?null-spaceconstraint》，A.Bourdoux，N.Khaled，Joint?Tx-Rx?optimization?forMIMO-SDMA?based?on?a?null-space?constraint，VTC?2002-Fall，Vancouver，Canada，Sept.2002：171-174。

《A?transmit?preprocessing?technique?for?Multi-user?MIMOsystems?using?a?decomposition?approach》，L.Choi，R.D.Murch，A?transmit?preprocessing?technique?for?Multi-user?MIMO?systemsusing?a?decomposition?approach，IEEE?Transactions?On?WirelessCommunications，Jan.2004，3(1)：20-24。And

《Zero-forcing?methods?for?downlink?spatial?multiplexing?inmulti-user?MIMO?channels》，Q.H.Spencer，A.L.Swindlehurst，M.haardt，Zero-forcing?methods?for?downlink?spatial?multiplexing?inmulti-user?MIMO?channels，IEEE?Transactions?On?Signal?Processing，Feb.2004，52(2)：461-471。

The method that is proposed is by designing a prefilter at transmitting terminal, make equivalent channel by block diagonalization (Block Diagonalization), equivalent channel matrix after just channel matrix and prefilter multiply each other is a block diagonalization, hereinafter abbreviates this algorithm as the BD algorithm.Under ideal conditions, such processing can make MUI be eliminated fully at transmitting terminal, the interference between inner each the circuit-switched data stream of only remaining user, and this interference need be handled at user terminal by the detection algorithm of Single User MIMO system.

Fig. 1 shows and adopts the space division multiple access of part block diagonalization prefilter to send and receiving system, in the space division multiple access emission between signal source C ¹～C ^UAnd be exactly part block diagonalization matrix filter between the channel matrix H.If part block diagonalization matrix filter K is U pre-filtering matrix K ^uLevel splicing, and select K ^uRow be positioned at H _c ^uKernel (null-space), then part block diagonalization condition just can satisfy.The channel conditions that needs in this case to know all users simultaneously could receive self signal.

Therefore, although above-mentioned block diagonalization method can be eliminated MUI fully, very effective to overcoming near-far interference, but it exists following defective: (every circuit-switched data flow point is not during corresponding to an antenna in order to eliminate interference between inner each circuit-switched data stream of user, be equal to the interference that will eliminate between inner each the root antenna of user), each travelling carriage need be known the channel conditions of all other travelling carriages at any time.Because the time-varying characteristics of wireless channel, this is not real aspect the realization of user terminal, and more common situation is that the decline between each travelling carriage of arbitrary moment is independently, each travelling carriage can only be known the channel conditions of oneself, and can't know the channel status of other travelling carriages.In fact the block diagonalization method is difficult to eliminate the interference between inner each the circuit-switched data stream of user.

In addition, user terminal needs extra receiving filter when adopting the block diagonalization method, also makes its structure relative complex.

Summary of the invention

The technical problem to be solved in the present invention provides a kind of space division multiple access method of sending and receiving that is used for the pre-filtering of multi-user MIMO system down link, this method can be eliminated the interference between multi-user interference and inner each the circuit-switched data stream of user, be easy to realize, and can simplify the design of user terminal.The present invention also will provide a kind of space division multiple access of this method of realizing to send and receiving system.

In order to realize above-mentioned technical problem, the invention provides a kind of space division multiple access method of sending and receiving of pre-filtering, the signal that is used for the multi-user MIMO system down link transmits and receives, and may further comprise the steps:

(A) the space division multiple access emission system calculates in real time by satisfying $H_c^u\·F^u=O$ And H ^uF ^uThe U of=I sub-piece F ^uThe complete diagonalization pre-filtering matrix F that forms, F=[F ¹F ^U], wherein, I is a unit matrix, o is a null matrix, H ^uBe the channel characteristics of user terminal u feedback, H _c ^uBe the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u, u=1,2 ..., U;

Simultaneously, described emission system sends information source data flow C to U user ¹～C ^U, the data flow that sends to user u is designated as C ^u, described data flow promptly multiplies each other with described complete diagonalization pre-filtering matrix through pre-filtering, obtains by U group data stream C ^uF ^uThat forms transmits, u=1, and 2 ..., U; And transmit by sending simultaneously from many antennas described, arrive each user through air traffic channel;

(B) after each user terminal is received described transmitting, directly adjudicate to received signal, recover the data flow that sends to oneself, simultaneously, the channel characteristics H that user terminal u will measure from described emission system to this null terminator Null channel ^uFeed back in the described emission system, u=1,2 ..., U.

Further, above-mentioned space division multiple access method of sending and receiving can have following characteristics: described space division multiple access emission system is when determining the transmitting power of subscriber signal, and the employing water flood carries out the power unequal distribution between the user.

Further, above-mentioned space division multiple access method of sending and receiving can have following characteristics: it is characterized in that described sub-piece F ^u=N ^uE ^u, N wherein ^uBe H _c ^uThe orthogonal basis of corresponding kernel, E ^uThen calculate by following formula:

$E^u=\frac{\sqrt{P^u}}{{\left|\right|{(H^u\·N^u)}^{+}\left|\right|}_F}{(H^u\·N^u)}^{+}$

In the formula, || || _FFrobenius norm, P are asked in expression ^uBe transmitting power at user u, u=1,2 ..., U, " () ⁺" expression is to the pseudo-inverse operation of matrix.

The space division multiple access of pre-filtering provided by the invention sends and receiving system, comprise a space division multiple access emission system and a U user terminal that includes U user, this emission system comprises U user source generator, many antennas, signal transmission power distribution module, prefilter and generation modules thereof, each user terminal comprises receiver, wireless channel matrix measurement mechanism and Duo Gen reception antenna, it is characterized in that:

Described prefilter is complete diagonalization prefilter, and its matrix F is made of U sub-piece, i.e. F=[F ¹F ^U], F ^uSatisfy the block diagonalization condition $H_c^u\·F^u=O$ With complete diagonalization condition H ^uF ^u=I,

Wherein, I is a unit matrix, and 0 is null matrix, H ^uBe the matrix that the row of respective user u constitutes in the whole channel matrix H, H _c ^uBe the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u, u=1,2 ..., U;

Described prefilter generation module is used for the channel characteristics H that obtains according to each user terminal measurement of receiving ^u, calculate the described complete diagonalization pre-filtering matrix F that satisfies block diagonalization condition and complete diagonalization condition in real time;

The receiver of described user terminal is used for directly adjudicating to received signal, recovers the data flow that sends to oneself.

Further, above-mentioned space division multiple access transmission and receiving system also can have following characteristics: described signal transmission power distribution module employing water flood carries out the power unequal distribution between the user.

Further, above-mentioned space division multiple access transmission and receiving system also can have following characteristics: the sub-piece F in the described pre-filtering matrix ^u=N ^uE ^u, N wherein ^uBe H _c ^uThe orthogonal basis of corresponding kernel, E ^uBe H ^u* N ^uPseudoinverse.

Further, above-mentioned space division multiple access transmission and receiving system also can have following characteristics: described prefilter generation module is to calculate E in the following manner ^u:

$E^u=\frac{\sqrt{P^u}}{{\left|\right|{(H^u\·N^u)}^{+}\left|\right|}_F}{(H^u\·N^u)}^{+}$

In the formula, || || _FFrobenius norm, P are asked in expression ^uBe transmitting power at user u, u=1,2 ..., U, " () ⁺" expression is to the pseudo-inverse operation of matrix.

Another technical problem that the present invention will solve provides a kind of space division multiple access emission system that is used for the pre-filtering of multi-user MIMO system down link, by complete diagonalization prefilter, eliminate the interference between inner each the circuit-switched data stream of multi-user interference and user at network side.

In order to solve the problems of the technologies described above, the present invention provides a kind of space division multiple access emission system of pre-filtering again, comprise a space division multiple access emission system that includes U user, this system comprises U user source generator, many antennas, signal transmission power distribution module, prefilter and generation modules thereof, it is characterized in that:

Described prefilter is complete diagonalization prefilter, and its matrix F is made of U sub-piece, i.e. F=[F ¹F ^U], F ^uSatisfy the block diagonalization condition $H_c^u\·F^u=O$ With complete diagonalization condition H ^uF ^u=I,

Wherein, I is a unit matrix, and o is a null matrix, H ^uBe the matrix that the row of respective user u constitutes in the whole channel matrix H, H _c ^uBe the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u, u=1,2 ..., U;

Described prefilter generation module is used for the channel characteristics H that obtains according to each user terminal measurement of receiving ^u, calculate the described complete diagonalization pre-filtering matrix F that satisfies block diagonalization condition and complete diagonalization condition in real time.

The technical problem to be solved in the present invention provides a kind of generation method that is used for the complete diagonalization pre-filtering matrix of multi-user MIMO system, and the prefilter that adopts its generator matrix, with by in the pre-filtering of transmitting terminal, eliminate the interference between inner each the circuit-switched data stream of multi-user interference and user to transmitting.

In order to realize above-mentioned technical problem, the invention provides the generation method of complete diagonalization pre-filtering matrix in a kind of multi-user MIMO system, may further comprise the steps:

(a) the space division multiple access emission system that comprises U user receives U user terminal measurement feedback and next channel characteristics H in the time interval of setting ^uAfter, its vertical splicing is obtained channel matrix H, u=1 wherein, 2 ..., U;

(b) for the arbitrary u value in the set [1..U], all row corresponding to user u of deletion from matrix H obtain a matrix H respectively _c ^u

(c) by U H _c ^uObtain the orthogonal basis N of its corresponding kernel respectively ^u, then with U N ^uLevel is spliced to form matrix N, N=[N ¹N ^U];

(d) according to the H that obtains ^uAnd N ^u, obtain to satisfy H ^u* N ^uPseudoinverse E ^u, u=1,2 ..., U;

(e) make F ^u=N ^uE ^u, again with U pre-filtering matrix F ^uThe level splicing obtains complete diagonalization

The matrix F of prefilter=[F ¹F ^U], when next time interval arrives, return step (a).

The present invention also provides the complete diagonalization prefilter in a kind of multi-user MIMO system, comprise a N matrix filter that is used to eliminate inter-user interference, it is characterized in that, comprise that also is used to eliminate an E matrix filter of disturbing between the inner multiplex data stream of user, wherein:

The matrix N of described N matrix filter is spliced N=[N by U sub-piece level ¹N ^U], N ^uBe H _c ^uThe orthogonal basis of corresponding kernel, and H _c ^uBe the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u, u=1,2 ..., U;

The matrix of described E matrix filter

E ^uBe H ^u* N ^uContrary pseudo-;

Matrix F=the N of described complete diagonalization prefilter * E.

As from the foregoing, based on for the transmission pre-filtering system of block diagonalization unified algorithm, the present invention has the following advantages with respect to existing:

1), can also eliminate the interference between inner each the circuit-switched data stream of user except eliminating the multi-user interference.

2) real more feasible, each receives the user only need know that the channel conditions of oneself is just much of that, does not need to know simultaneously all users' channel conditions, and this point is vital to SDMA MIMO The Application of Technology.

3) owing to simultaneously inner each the root interference between antennas of MUI and user has been eliminated fully at transmitting terminal, so user terminal just do not need more the receiving filter of complexity to handle, and greatly simplified the receiver design.

Summary of drawings

Fig. 1 adopts the space division multiple access sending part to divide the space division multiple access transmission of block diagonalization prefilter and the schematic diagram of receiving system.

Fig. 2 is that the embodiment of the invention adopts the space division multiple access transmission of complete diagonalization prefilter and the schematic diagram of receiving system.

Fig. 3 is the flow chart of the complete diagonalization prefilter of present embodiment F Matrix Solving method.

Fig. 4 is the flow chart of the space division multiple access method of sending and receiving of present embodiment pre-filtering.

Fig. 5 is BD algorithm and the BER performance comparison diagram of CD algorithm under independent mimo channel.

Fig. 6 is BD algorithm and the BER performance comparison diagram of CD algorithm under Nokia spacing related MIMO channel.

Fig. 7 is BD algorithm and the Shannon capacity comparison diagram of CD algorithm under independent mimo channel.

Fig. 8 is BD algorithm and the interruption capacity (interruption rate: 0.1) comparison diagram of CD algorithm under independent mimo channel.

Fig. 9 is BD algorithm and the interruption capacity (interruption rate: 0.1) comparison diagram of CD algorithm under Nokia spacing related MIMO channel.

Preferred forms of the present invention

The present invention proposes a kind of space division multiple access method of sending and receiving, system of new pre-filtering, diagonalization prefilter and its implementation fully in this system abbreviate the CD algorithm as.Its core concept is that the part block diagonalization condition that will mention in aforementioned three pieces of lists of references is further strengthened being complete diagonalization (Complete diagonalization), and the interference that is about between inner each the circuit-switched data stream of user also eliminates at transmitting terminal fully.

The present invention is described in further detail below in conjunction with drawings and Examples.

Fig. 2 is that the embodiment of the invention adopts the space division multiple access of complete diagonalization prefilter to send and receiving system, comprises a space division multiple access emission system and a U user terminal that includes U user.The space division multiple access emission system comprises U user source generator, the antenna sets that comprises A root antenna, signal transmission power distribution module (not shown), complete diagonalization prefilter and generation module thereof.

The user source generator is used to generate the information source data flow that will send to each user, and each user's corresponding data flow comprises the multipath symbols stream that obtains behind the demultiplexing.

Antenna sets is used for the multiplex data stream signal through pre-filtering is sent simultaneously by A root antenna.

The diagonalization prefilter is used for described information source data stream is carried out filtering fully, and its matrix F is made of U sub-piece, and F=[F is arranged ¹F ^U], F ^uSatisfy the block diagonalization condition $H_c^u\·F^u=O$ With complete diagonalization condition H ^uF ^u=I.Wherein, I is a unit matrix, and 0 is null matrix, H ^uBe the matrix that the row of respective user u constitutes in the whole channel matrix H, H _c ^uIt is the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u.

The prefilter generation module is used for the channel characteristics H that obtains according to each user terminal measurement of receiving ^u, calculate the complete diagonalization pre-filtering matrix F that satisfies block diagonalization condition and complete diagonalization condition in real time.

The signal transmission power distribution module is used for the power that each user transmits is distributed, and present embodiment is that the employing water flood carries out the power unequal distribution between the user.

Each user terminal comprises many reception antennas, MMSE (least mean-square error) receiver and a wireless channel matrix measurement mechanism.Wherein:

Many reception antennas are used to receive the signal of network side emission, and its content is the stack of issuing all user's data.

The MMSE receiver is used for adjudicating to received signal, recovers the data flow that sends to this user terminal;

Wireless channel matrix measurement mechanism is used to measure the MIMO subchannel from the space division multiple access emission system to this user terminal, obtains channel characteristics H ^u, and feed back to the space division multiple access emission system.

Compared to Figure 1, present embodiment has replaced part block diagonalization matrix filter of the prior art with complete diagonalization prefilter F.This complete diagonalization prefilter F comprises N matrix filter and E matrix filter again, and the N matrix filter also can be called the inter-user interference Canceller, and the process of its execution is for offsetting inter-user interference; The E matrix also can be named with interference offset device between indoor multiplex data stream, and the process of its execution is for offsetting the interference of the most data flow of single user between mutually.

These two filters all are to follow the real-time change of space channel feature and change, and its variation is dependent on the channel characteristics H that measures the MIMO subchannel from the space division multiple access emission system to u user of acquisition from u user ^u, u=1,2 ..., U.

The method that present embodiment generates complete diagonalization prefilter F is carried out following steps by the prefilter generation module in the space division multiple access emission system as shown in Figure 3:

Step 110 in the time interval of setting, receives U user terminal measurement feedback and next channel characteristics H ^u, channel characteristics H ^uOnly at MIMO subchannel from the space division multiple access emission system to u user, u=1,2 ..., U;

H ^uBe the B of respective user u in the whole channel matrix H ^uOK, i.e. H ^uIt is MIMO subchannel from the base station to u user.

Step 120 is with U user's channel matrix H ^uVertical splicing obtains channel matrix H, for the arbitrary u value in the set [1..U], deletes all B corresponding to user u respectively from matrix H ^uOK, obtain a matrix H _c ^u

Channel matrix H=[(H ¹) ^T(H ^U) ^T] ^T, its dimension is

H _c ^uHave

OK.

Step 130 is by U H _c ^uObtain the orthogonal basis N of its corresponding kernel respectively ^u, u=1,2 ..., U is spliced to form matrix N, N=[N with U sub-piece level then ¹N ^U], its u sub-piece N ^uDimension be A * D ^u

Step 140 is according to the H that obtains ^uAnd N ^u, obtain simultaneously and satisfy condition $H_c^u\·N^u\·E^u=O$ With condition H ^uN ^uE ^uThe E of=I ^u, wherein I is a unit matrix, o is a null matrix, this E ^uBe H ^u* N ^uPseudoinverse; Present embodiment calculates by Frobenius norm normalization algorithm, be shown below, but the present invention is not limited thereto:

$E^u=\frac{\sqrt{P^u}}{{\left|\right|{(H^u\·N^u)}^{+}\left|\right|}_F}{(H^u\·N^u)}^{+}$

In the formula, || || _FFrobenius norm, P are asked in expression ^uBe transmitting power at user u, u=1,2 ..., U, " () ⁺" expression is to the pseudo-inverse operation of matrix.

Step 150 makes F ^u=N ^uE ^u, u=1,2 ..., U is again with U pre-filtering matrix F ^uLevel is spliced, and promptly obtains the matrix F of complete diagonalization prefilter, when the next time interval arrives, returns step 110 then.

Matrix F=[the F of complete diagonalization prefilter ¹F ^U], also can be expressed as:

Wherein, visible matrix E is made of the sub-piece E of u block diagonalization on its diagonal U block diagonalization matrix ^uDimension be D ^u* C ^u

After obtaining matrix F, be ∑ C with total line number ^U(C ^uExpression sends to the data flow of user u), columns is infinitely-great data source matrix multiple process, is exactly the pre-filtering processing procedure.

Top method for solving is according to two constraintss $\left\{\begin{array}{cc}{H}_c^u\·F^u=O& \left(1\right)\\ H^u\·F^u=I& \left(2\right)\end{array}\right.$ Derivation is come out, and the 0th, null matrix also can be expressed as condition one: $H_c^u\·N^u\·E^u=O$ With condition two: H ^uN ^uE ^u=I.Condition one is the block diagonalization condition, can eliminate MUI fully after satisfying.Condition two is complete diagonalization conditions, just can eliminate the interference between inner each circuit-switched data stream of user after satisfying.This will further illustrate hereinafter.

As shown in Figure 4, the present embodiment space division multiple access method of sending and receiving that is used for the pre-filtering of multi-user MIMO system down link may further comprise the steps:

Step 210, the space division multiple access emission system sends multiplex data stream to U user, and the data flow that sends to user u is C ^u(having comprised the multiplex data stream that demultiplexing obtains), u=1,2,, U, simultaneously, this emission system calculates the complete diagonalization pre-filtering matrix F that satisfies block diagonalization condition and complete diagonalization condition in real time according to each channel characteristics of user terminal feedback, and this method is stated hereinbefore;

Step 220, information source data flow C ¹～C ^UPass through the corresponding sub-block matrix E of block diagonalization respectively ^uFiltering, the process of filtering is data flow C ^uWith the sub-piece E of block diagonalization ^uThe process of matrix multiple;

Step 230, each C ^uAnd E ^uMultiplied result C ^uE ^uContinue by the filtering of N matrix filter, promptly with matrix N in N ^uMultiply each other, output A road is through the C that transmits of pre-filtering ^uE ^uN ^u, u=1,2 ..., U is sent to A root antenna respectively;

Step 240 will transmit through the A road of pre-filtering and be sent simultaneously by A root antenna, arrive each user through air traffic channel;

Step 250, each user terminal is received the stack of issuing all user's data simultaneously, uses the MMSE receiver directly to adjudicate to received signal, recovers the data flow C that sends to oneself ^uSimultaneously, the MIMO subchannel of u user terminal measurement from the space division multiple access emission system to u user obtains channel characteristics H ^u, and feed back in the space division multiple access emission system.

Form with the symbolic vector computing illustrates why said method can eliminate the interference between MUI and inner each the circuit-switched data stream of user below:

The symbolic vector of supposing arbitrary moment user is x (k), and x (k)=[x is arranged ¹(k) ^TX ^U(k) ^T] ^T, wherein: symbolic vector x ^u(k) and data flow C ^uAnother expression mode, $x^u\left(k\right)={[x_1^u\left(k\right)\·\·\·x_{C^u}^u\left(k\right)]}^T,$ C ^uBe meant the symbols streams way that one road incoming bit stream demultiplexing is obtained.

The signal that u user side of present embodiment received can be expressed as: ${\hat{x}}^u\left(k\right)=H^u\·F\·x\left(k\right)+n^u\left(k\right),$ Satisfy for all users $H_c^u\·F^u=O,$ Just mean H ^uF _c ^u=0,0 is null matrix, promptly so-called quadrature assurance, therefore ${\hat{x}}^u\left(k\right)=H^u\·F^u\·x^u\left(k\right)+n^u\left(k\right),$ So far, MUI is eliminated fully.

But, the C of user u ^uCircuit-switched data stream

Between interference also do not eliminate.By satisfying H ^uF ^u=I has then further eliminated the interference between inner each the circuit-switched data stream of user u, and this is because the I matrix is a unit matrix, its diagonal element is 1, and other elements are 0, satisfies this condition, mean that certain circuit-switched data stream is only with own relevant, with the complete quadrature of other data flow.

Further specify the realization flow that adopts space division multiple access to send the multi-user MIMO system down link of prefilter with an application example below.

Suppose that the base station has 6 antennas, there are 3 users in system, and each user has 2 antennas.Therefore, such system configuration has 3 * 2=6 parallel data flow to send simultaneously.3 road incoming bit streams demultiplex into 2 tunnel symbols streams at the base station end after the QPSK modulation.Every road symbols streams is divided into the Frame that comprises 480 symbols again.

At any one time, send the symbols streams of user u $x^u\left(k\right)={\left[x_1^u\left(k\right)x_2^u\left(k\right)\right]}^T,$ U=1,2,3, x wherein ₁ ^u(k) and x ₂ ^u(k) be 2 tunnel symbols streams that obtain by one road incoming bit stream demultiplexing.User symbol vector x (k)=[x of input prefilter ¹(k) ^Tx ³(k) ^Tx ³(k) ^T] ^T

The structure of the matrix F of diagonalization prefilter is above described in detail fully, has in this example:

$F=\left[\begin{array}{ccc}{N}_1& N_2& N_3\end{array}\right]\left[\begin{array}{ccc}{E}_1& & \\ & E_2& \\ & & {\mathrm{<figure-callout\; id="3"\; label="E"\; filenames="H70546691150143000E000042.png"\; state="\{\{state\}\}">E</figure-callout>}}_3\end{array}\right]$

User terminal is by to received signal

Judgement, ${\hat{x}}^u\left(k\right)=\frac{\sqrt{P^u}}{{\left|\right|{(H^u\&CenterDot;N^u)}^{+}\left|\right|}_F}{x}^u\left(k\right)+n^u\left(k\right)$ Just can recover original transmission signal, judgement be exactly one by the process of fuzzy continuous quantity to the digital quantity of quantification.Wherein:

Be the signal that terminal is received, x ^u(k) be original transmitted signal, n ^u(k) be aerial thermal noise or external interference, N ^uIt is the Nature thermal noise.Recover original transmission signal and be exactly from

The middle x that recovers ^u(k).

The present invention is owing to just eliminated the interference between inner each circuit-switched data stream of MUI and user fully, so just no longer need any additional treatments at user terminal at transmitting terminal.This also is one of characteristics of the present invention: simplified the receiver design.Through the preliminary treatment of transmitting filter, make through the signal after the channel just just there has not been interference, can directly adjudicate.

The Shannon capacity of the complete diagonalizable matrix of the present invention (CD) algorithm is as follows:

$C=\underset{u=1}{\overset{U}{\mathrm{\&Sigma;}}}{\log }_2\left[\det (I+\frac{P_u}{{\mathrm{\&sigma;}}^2}\&CenterDot;\frac{I}{{\left|\right|{(H^u\&CenterDot;N^u)}^{+}\left|\right|}_F^2})\right]$ (u＝1，2，3)

In the formula, I is a unit matrix, and U is number of users=3 of transmitting simultaneously, P ^uIt is the transmitted power of user u.σ ²Refer to such an extent that be the white Gaussian noise power of u user's receiving terminal.

Under the constant power distributive condition, as if the total transmitted power that is illustrated in space division multiple access transmitter in the symbol period with P, the Shannon capacity of CD algorithm becomes:

$C=\underset{u=1}{\overset{U}{\mathrm{\&Sigma;}}}{\log }_2\left[\det (I+\frac{P}{{U\&CenterDot;\mathrm{\&sigma;}}^2}\&CenterDot;\frac{I}{{\left|\right|{(H^u\&CenterDot;N^u)}^{+}\left|\right|}_F^2})\right]$ (u＝1，2，3)

In fact, for this CD algorithm, we can also adopt water flood to carry out power unequal distribution between the user in the signal transmission power distribution module of system, make performance be assigned further improvement than constant power, and this improvement is more obvious than low signal-to-noise ratio the time.

The exponent number of noticing unit matrix I in the following formula is H ^uN ^uOrder, be made as L ^u, then following formula becomes

$C=\underset{u=1}{\overset{U}{\mathrm{\&Sigma;}}}{L_u\log }_2\left[1+\frac{P_u}{{\mathrm{\&sigma;}}^2}\&CenterDot;\frac{1}{{\left|\right|{(H^u\&CenterDot;N^u)}^{+}\left|\right|}_F^2}\right]$ (u＝1，2，3)

Water flood will be found the solution following formula exactly and be existed $\underset{u=1}{\overset{U}{\mathrm{\&Sigma;}}}{P}_u=P$ The maximum of C under the power constraint condition.

By utilization Lagrange multiplier λ, the power that can obtain each user is:

$P_u=\frac{L_u}{\mathrm{\&lambda;}\ln 2}-{\mathrm{\&sigma;}}^2{\left|\right|{(H^u\&CenterDot;N^u)}^{+}\left|\right|}_F^2$ (u＝1，2，3)

The value of λ is passed through $\underset{u=1}{\overset{U}{\mathrm{\&Sigma;}}}{P}_u=P$ Obtain.

Fig. 5 shows the BD algorithm and the BER performance of CD algorithm under independent mimo channel compares.Simulation result has enumerated the BD algorithm respectively and the CD algorithm is respectively 6,7,8 at the antenna for base station number, and the UE number is 3, and antenna number all is the BER performance under 2 the condition.As can be seen, under same antenna configurations, the BD algorithm slightly is better than the CD algorithm.

The BER performance that Fig. 6 shows under BD algorithm and the CD algorithm Nokia spacing related MIMO channel compares.Simulation result has enumerated the BD algorithm respectively and the CD algorithm is respectively 6,7,8 at the antenna for base station number, and the UE number is 3, and antenna number all is the BER performance under 2 the condition.As can be seen, under same antenna configurations, the BD algorithm slightly is better than the CD algorithm.

Fig. 7 shows BD algorithm and the Shannon Capacity Ratio of CD algorithm under independent mimo channel.Simulation result has enumerated the BD algorithm respectively and the CD algorithm is respectively 6,7,8 at the antenna for base station number, and the UE number is 3, and antenna number all is the Shannon volumetric properties under 2 the condition.As can be seen, under same antenna configurations, the BD algorithm slightly is better than the CD algorithm.

Fig. 8 shows BD algorithm and the interruption capacity (interruption rate: 0.1) relatively of CD algorithm under independent mimo channel.Simulation result has enumerated the BD algorithm respectively and the CD algorithm is respectively 6,7,8 at the antenna for base station number, and the UE number is 3, and antenna number all is the interruption volumetric properties under 2 the condition.As can be seen, under same antenna configurations, the BD algorithm slightly is better than the CD algorithm.

Fig. 9 shows the interruption capacity (interruption rate: 0.1) relatively under BD algorithm and the CD algorithm Nokia spacing related MIMO channel.Simulation result has enumerated the BD algorithm respectively and the CD algorithm is respectively 6,7,8 at the antenna for base station number, and the UE number is 3, and antenna number all is the interruption volumetric properties under 2 the condition.As can be seen, under same antenna configurations, the BD algorithm slightly is better than the CD algorithm.

The analysis of simulation result of complex chart 5～9, the performance of CD (all diagonalization) algorithm and BD (part block diagonalization) algorithm are consistent at independent mimo channel with variation tendency under the Nokia spacing related MIMO channel, along with the increase of correlation between the antenna, poor system performance.Difference is that the performance of CD algorithm is a little less than the BD algorithm in both cases.This is understandable, and the BD algorithm is under the condition that it is supposed, performance is optimum, but this hypothesis is unpractical, can't put into effect.Though CD algorithm performance is a little less than the BD algorithm, but one of feasible algorithm.And under the situation of number of users and every user's reception antenna invariable number, along with the number of transmit antennas purpose increases, the performance difference of CD algorithm and BD algorithm can dwindle gradually.

Industrial applicibility

The inventive method can be applicable to the transmitting and receiving of multi-user MIMO system down link signal in the space division multiple access communication system, eliminates the interference between inner each circuit-switched data stream of multi-user interference and user, and the simplified receiver design, possesses industrial applicibility.

Claims (14)

1. the space division multiple access method of sending and receiving of a pre-filtering, the signal that is used for the multi-user MIMO system down link transmits and receives, and may further comprise the steps:

(A) the space division multiple access emission system calculates in real time by satisfying

And H ^uF ^uThe U of=I sub-piece F ^uThe complete diagonalization pre-filtering matrix F that forms, F=[F ¹F ^u], wherein, I is a unit

Battle array, O is a null matrix, H ^uBe the channel characteristics of user terminal u feedback, H _c ^uBe the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u, u=1,2 ..., U;

Simultaneously, described emission system sends information source data flow C to U user ¹～C ^U, the data flow that sends to user u is designated as C ^u, described data flow promptly multiplies each other with described complete diagonalization pre-filtering matrix through pre-filtering, obtains by U group data stream C ^uF ^uThat forms transmits, u=1, and 2 ..., U; And transmit by sending simultaneously from many antennas described, arrive each user through air traffic channel;

(B) after each user terminal is received described transmitting, directly adjudicate to received signal, recover the data flow that sends to oneself, simultaneously, the channel characteristics H that user terminal u will measure from described emission system to this null terminator Null channel ^uFeed back in the described emission system, u=1,2 ..., U.

2. space division multiple access method of sending and receiving as claimed in claim 1 is characterized in that, described sub-piece F ^u=N ^uE ^u, N wherein ^uBe H _c ^uThe orthogonal basis of corresponding kernel, E ^uThen calculate by following formula:

In the formula, || || _FFrobenius norm, P are asked in expression ^uBe transmitting power at user u, u=1,2 ..., U, " () ⁺" expression is to the pseudo-inverse operation of matrix.

3. space division multiple access method of sending and receiving as claimed in claim 1,

It is characterized in that described space division multiple access emission system is when determining the transmitting power of subscriber signal, the employing water flood carries out the power unequal distribution between the user.

4. the space division multiple access of a pre-filtering sends and receiving system, comprise a space division multiple access emission system and a U user terminal that includes U user, this emission system comprises U user source generator, many antennas, signal transmission power distribution module, prefilter and generation modules thereof, each user terminal comprises receiver, wireless channel matrix measurement mechanism and Duo Gen reception antenna, it is characterized in that:

Described prefilter is complete diagonalization prefilter, and its matrix F is made of U sub-piece, i.e. F=[F ¹F ^U], F ^uSatisfy the block diagonalization condition With complete diagonalization condition H ^uF ^u=I,

Wherein, I is a unit matrix, and O is a null matrix, H ^uBe the matrix that the row of respective user u constitutes in the whole channel matrix H, H _c ^uBe the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u, u=1,2 ..., U;

Described prefilter generation module is used for the channel characteristics H that obtains according to each user terminal measurement of receiving ^u, calculate the described complete diagonalization pre-filtering matrix F that satisfies block diagonalization condition and complete diagonalization condition in real time;

The receiver of described user terminal is used for directly adjudicating to received signal, recovers the data flow that sends to oneself.

5. space division multiple access as claimed in claim 4 sends and receiving system,

It is characterized in that described signal transmission power distribution module employing water flood carries out the power unequal distribution between the user.

6. space division multiple access as claimed in claim 4 sends and receiving system,

It is characterized in that the sub-piece F in the described pre-filtering matrix ^u=N ^uE ^u, N wherein ^uBe H _c ^uThe orthogonal basis of corresponding kernel, E ^uBe H ^u* N ^uPseudoinverse.

7. space division multiple access as claimed in claim 6 sends and receiving system,

It is characterized in that described prefilter generation module is to calculate E in the following manner ^u:

In the formula, || || _FFrobenius norm, P are asked in expression ^uBe transmitting power at user u, u=1,2 ..., U, " () ⁺" expression is to the pseudo-inverse operation of matrix.

8. the space division multiple access emission system of a pre-filtering comprises a space division multiple access emission system that includes U user, and this system comprises U user source generator, many antennas, signal transmission power distribution module, prefilter and generation modules thereof,

It is characterized in that:

Described prefilter is complete diagonalization prefilter, and its matrix F is made of U sub-piece, i.e. F=[F ¹... F ^U], F ^uSatisfy the block diagonalization condition With complete diagonalization condition H ^uF ^u=I,

Wherein, I is a unit matrix, and O is a null matrix, H ^uBe the matrix that the row of respective user u constitutes in the whole channel matrix H, H _c ^uBe the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u, u=1,2 ..., U;

Described prefilter generation module is used for the channel characteristics H that obtains according to each user terminal measurement of receiving ^u, calculate the described complete diagonalization pre-filtering matrix F that satisfies block diagonalization condition and complete diagonalization condition in real time.

9. space division multiple access emission system as claimed in claim 8,

It is characterized in that described signal transmission power distribution module employing water flood carries out the power unequal distribution between the user.

10. space division multiple access emission system as claimed in claim 8,

It is characterized in that the sub-piece F in the described pre-filtering matrix ^u=N ^uE ^u, N wherein ^uBe H _c ^uThe orthogonal basis of corresponding kernel, E ^uBe H ^u* N ^uPseudoinverse.

11. space division multiple access emission system as claimed in claim 10,

It is characterized in that described prefilter generation module is to calculate E in the following manner ^u:

In the formula, ‖ ‖ _FFrobenius norm, P are asked in expression ^uBe transmitting power at user u, u=1,2 ..., U, " () ⁺" expression is to the pseudo-inverse operation of matrix.

12. the generation method of diagonalization pre-filtering matrix fully in the multi-user MIMO system may further comprise the steps:

(a) the space division multiple access emission system that comprises U user receives U user terminal measurement feedback and next channel characteristics H in the time interval of setting ^uAfter, its vertical splicing is obtained channel matrix H, u=1 wherein, 2 ..., U;

(b) for the arbitrary u value in the set [1..U], all row corresponding to user u of deletion from matrix H obtain a matrix H respectively _c ^u

(c) by U H _c ^uObtain the orthogonal basis N of its corresponding kernel respectively ^u, then with U N ^uLevel is spliced to form matrix N, N=[N ¹... H ^U];

(d) according to the H that obtains ^uAnd N ^u, obtain to satisfy H ^u* N ^uPseudoinverse E ^u, u=1,2 ..., U;

(e) make F ^u=N ^uE ^u, again with U pre-filtering matrix F ^uThe level splicing obtains complete diagonalization

The matrix F of prefilter=[F ¹... F ^U], when next time interval arrives, return step (a).

13. the generation method of complete diagonalization pre-filtering matrix as claimed in claim 12 is characterized in that, is to calculate E by following formula in the described step (d) ^u:

In the formula, || || _FFrobenius norm, P are asked in expression ^uBe transmitting power at user u, u=1,2 ..., U, " () ⁺" expression is to the pseudo-inverse operation of matrix.

14. the complete diagonalization prefilter in the multi-user MIMO system, comprise a N matrix filter that is used to eliminate inter-user interference, it is characterized in that, comprise that also is used to eliminate an E matrix filter of disturbing between the inner multiplex data stream of user, wherein:

The matrix N of described N matrix filter is spliced N=[N by U sub-piece level ¹N ^U], N ^uBe

The orthogonal basis of corresponding kernel, and

Be the matrix that from channel matrix H, obtains behind all row of deletion corresponding to user u, u=1,2 ..., U;

The matrix of described E matrix filter

E ^uBe H ^u* N ^uPseudoinverse.

Images