CN101616427A - Multi-user MIMO test method between a kind of base station - Google Patents

Abstract

The invention discloses the multi-user MIMO test method between a kind of base station, comprising:, determine in this sub-district and the neighbor cell same frequency user with the identical running time-frequency resource of described arbitrary CU for arbitrary user of each sub-district, place, base station; Between the base station of the base station at described arbitrary user place and this arbitrary user's sub-district, same frequency user place, share with user's identification information frequently, be used for that this cell base station estimates other sub-districts described with user frequently in the characteristic of channel of this base station; Each base station is carried out joint-detection with the user of sub-district, place with this user's same frequency user, obtains this user and with user's testing result frequently.Use the present invention, can realize the multiuser MIMO joint-detection of a plurality of sub-districts, effectively suppress co-channel interference, improve systematic function.

Description

Multi-user MIMO test method between a kind of base station

Technical field

The present invention relates to the technology of multiple-input and multiple-output (MIMO) input, the multi-user MIMO test method between particularly a kind of base station.

Background technology

For the MIMO communication system, optimum detection method is maximum likelihood (ML) method or maximum a posteriori probability (MAP) method on the meaning of associating error probability minimum, but the complexity of these two kinds of detection methods is all very high, is difficult in the real system and uses.And the demand of practical application makes the research of MIMO detection algorithm mainly concentrate on the high performance detector of seeking low complex degree.

At present, the high performance detection method of low complex degree mainly comprises: linear ZF (ZF) algorithm, in conjunction with serial interference elimination (OSIC in turn, Ordered Successive Interference Cancellation) ZF-OSIC algorithm, linear minimum mean-squared error (MMSE, Minimum Mean Square Error) algorithm, the MMSE-OSIC algorithm, QR decomposes (QRD) algorithm, ordering QRD algorithm, ball decoding (Sphere Decoding, SD) algorithm, the QRD-M algorithm, lax (the SDR that detects of semidefinite, SemidefiniteRelaxation) algorithm, probabilistic data association (PDA, Probabilistic Data Association) algorithm or the like.Wherein the SD algorithm be up to now performance near the suboptimum detection algorithm of maximum likelihood algorithm.Promptly search under the condition of not failing at ideal situation, its theoretical performance is identical with maximum likelihood algorithm.Though SD algorithm average computation complexity in the applied environment of most of medium-scale wireless communication systems is the multinomial complexity, remains exponential complexity under worst case.

The SDR algorithm is modeled as a positive semidefinite planning problem with MIMO detection problem, relaxes then and utilizes interior some algorithm of numerical computations to find the solution for protruding optimization problem, can reach the detection performance that approaches maximum likelihood algorithm, and only pay the polynomial computation complexity.Compare with the exponential complexity under the SD algorithm worst case, the worst case complexity of SDR algorithm remains the multinomial complexity, but it is an approximate data that shortcoming is the SDR algorithm, can only be near maximum likelihood algorithm on performance, and the exponent number of its multinomial complexity is still higher.In addition, for different modulation systems, the concrete form of SDR algorithm is also different, the algorithm pattern that neither one is general, and the complexity of practical application is higher.

The PDA algorithm is another kind of important suboptimum detection algorithm.This algorithm utilizes Gaussian approximation, goes to estimate the posterior probability of each transmitting antenna data symbol, can be described as a kind of approximate form of MAP algorithm.It relatively is applicable to larger communication system, and for the less system of scale, performance decreases, but is better than the MMSE-OSIC algorithm, and is worse than ball decoding algorithm and SDR algorithm.From the complexity aspect, the worst case complexity of PDA algorithm also is the multinomial complexity, and is lower than the complexity under the worst case of SDR algorithm.Because the PDA algorithm is an iterative computation in the process of the posterior probability of estimating each symbol, the slower problem of posterior probability convergence rate of individual symbols can appear sometimes inevitably, but under the situation of a given rational maximum iteration time, this problem is very little to detecting Effect on Performance.

Except above suboptimum detection algorithm, the algorithm that utilizes iteration signal processing thought to carry out iterative detection also is a very important research direction.Such algorithm utilizes the character of SISO (soft input soft ourput), provides very excellent detection performance with lower complexity, and for example PDA algorithm itself can just be thought a kind of SISO Iterative detection algorithm.Special needs to be pointed out is, can be simple iterative detection (for example only utilizing the PDA algorithm to carry out MIMO detects) at receiving terminal to the iterative processing of signal, also can be iteration joint-detection decoding (for example Turbo iteration joint-detection decoding algorithm).Latter's complexity is higher, but the performance that provides is better, has been particularly useful for adopting the system of Turbo code, convolution code, LDPC sign indicating number.Consider that the such SISO algorithm of PDA has natural applicability in iteration received signal processing module, basic MIMO detection algorithm related in this patent is the PDA algorithm.

More than the various MIMO detection algorithms of Jie Shaoing are applied in the Single User MIMO usually.At present, because the restriction of user terminal volume and disposal ability, its number of antennas can limit usually to some extent, improves transmission performance for utilizing the MIMO technology better, and MIMO is applied between a plurality of users, forms multiuser MIMO.

In multi-user MIMO system, if the detection method of directly using under the above-mentioned Single User MIMO is carried out input, so for user to be detected, can only with in other neighbor cells outside this sub-district, user place with this user with other users frequently as interference, detect and carry out independent signal at this user itself, like this, in testing process, can't rationally utilize other signal messages, make that final testing result accuracy is not high with the frequency user.

Summary of the invention

In view of this, the invention provides the multi-user MIMO test method between a kind of base station, can realize the multiuser MIMO joint-detection of a plurality of sub-districts, effectively suppress co-channel interference, improve systematic function.

For achieving the above object, the present invention adopts following technical scheme:

Multi-user MIMO test method between a kind of base station comprises:

For arbitrary user of each sub-district, place, base station, determine in this sub-district and the neighbor cell same frequency user with the identical running time-frequency resource of described arbitrary CU;

Between the base station of the base station at described arbitrary user place and this arbitrary user's sub-district, same frequency user place, share with user's identification information frequently, be used for that this cell base station estimates other sub-districts described with user frequently in the characteristic of channel of this base station; Each base station is carried out joint-detection with the user of sub-district, place with this user's same frequency user, obtains this user and with user's testing result frequently.

Preferably, after obtaining described user and the testing result with the frequency user thereof, this method further comprises:

Determine described user and the same probabilistic information of user's testing result correspondence frequently thereof;

With detected neighbor cell with frequently user's testing result and the base station that corresponding probabilistic information sends to corresponding neighbor cell thereof;

Each base station is to arbitrary user of sub-district, place, the corresponding probabilistic information of testing result of this subscriber signal that the probabilistic information of detected this user's of this cell base station signal detecting result correspondence and other neighbor cells are sent merges, and determines the final detection result of this subscriber signal.

Preferably, the corresponding probabilistic information of signal detecting result that merges this user that probabilistic information and other neighbor cells of detected this user's of this cell base station signal detecting result correspondence send by high specific.

Preferably, a user being carried out joint-detection with this user's same frequency user comprises: utilize iterative detection to carry out joint-detection to described user and with user frequently;

In each iterative detection, detect to described user and with the signal on each transmitting antenna of frequency user successively; And when the signal on described transmitting antenna to each detects, the signal on other transmitting antennas outside the transmitting antenna to be detected as interference, is determined up-to-date information symbol probability on the described transmitting antenna to be detected;

Up-to-date information symbol probability is used for next iteration and detects on each described transmitting antenna that each iterative detection is obtained, information symbol probability on all described transmitting antennas is all restrained, perhaps, reach the default iterations upper limit up to, then finishing iteration detects, and utilize information symbol probability up-to-date on each described transmitting antenna, determine the subscriber signal testing result on each described transmitting antenna, with the testing result of the subscriber signal testing result on all transmitting antennas of each user as the relative users signal.

Preferably, the mode that the signal on the transmitting antenna to be detected is detected comprises:

Construct the channel matrix between described user and same user frequently and described user place cell base station $H=[H_{1s},H_{2s},\·\·\·,H_{N_{U}s}],$ Wherein, H _{I, s}Be the characteristic of channel between i user among described user's the same frequency user and s base station;

The received signal of utilization on each described transmitting antenna calculated $\tilde{y}={\left(H^{H}H\right)}^{-1}{H}^H{y}_s,$ Wherein, () ^HConjugate transpose is got in expression;

Go up up-to-date information symbol probability P according to other the described transmitting antenna k except that described transmitting antenna j to be detected _m(x _k| y), determine the up-to-date information symbol probability on the described transmitting antenna to be detected $P_m\left(x_j\right|y)=\frac{{\mathrm{\φ}}_m\left(x_j\right)}{\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{\mathrm{\φ}}_m\left(x_j\right)},$

Wherein, ${\mathrm{\φ}}_m\left(x_j\right)\stackrel{\mathrm{\Δ}}{=}\exp \left(\begin{array}{cc}-{\left(\begin{array}{c}\mathrm{Re}\left(w\right)\\ \mathrm{Im}\left(w\right)\end{array}\right)}^T& {\mathrm{\Λ}}_j\left(\begin{array}{c}\mathrm{Re}\left(w\right)\\ \mathrm{Im}\left(w\right)\end{array}\right)\end{array}\right),$ ${\mathrm{\Λ}}_j={(\left(\begin{array}{cc}\mathrm{Re}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)& -\mathrm{Im}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)\\ \mathrm{Im}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)& \mathrm{Re}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)\end{array}\right)+Q)}^{-1},$ $Q=\underset{k\≠j}{\mathrm{\Σ}}({\mathrm{\ω}}_k+\mathrm{Re}\left({\mathrm{\ξ}}_k\right))e_k{e}_k^T+\underset{k\≠j}{\mathrm{\Σ}}({\mathrm{\ω}}_k-\mathrm{Re}\left({\mathrm{\ξ}}_k\right))e_{k+N_T}{e}_{k+N_T}^T+\underset{k\≠j}{\mathrm{\Σ}}\mathrm{Im}\left({\mathrm{\ξ}}_k\right)e_k{e}_{k+N_T}^T+\underset{k\≠j}{\mathrm{\Σ}}\mathrm{Im}\left({\mathrm{\ξ}}_k\right)e_{k+N_T}{e}_k^T,$ e _kBe that k element is 1, other elements are zero unit vector, $w=\tilde{y}-x_j{e}_j-\underset{k\≠j}{\mathrm{\Σ}}{\tilde{x}}_k{e}_k,$ $\mathrm{Cov}\left(\tilde{n}\right)={\mathrm{\σ}}^2{\left(H^{H}H\right)}^{-1},$ ${\mathrm{\ω}}_k=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}(a_m-{\tilde{x}}_k){(a_m-{\tilde{x}}_k)}^{*}{P}_m\left(x_k\right|y),$ ${\mathrm{\ξ}}_k=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}(a_m-{\tilde{x}}_k){(a_m-{\tilde{x}}_k)}^T{P}_m\left(x_k\right|y),$ ${\tilde{x}}_k=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{a}_m{P}_m\left(x_k\right|y),$ Real part is got in Re () expression, and imaginary part, σ are got in Im () expression ²Be noise power, a _mBe the value of m constellation point of information symbol, M is the information symbol constellation point number on the transmitting antenna k.

Preferably, in first time during iteration, the posterior probability of each described transmitting antenna is initialized as even distribution in the constellation point set.

As seen from the above technical solution, among the present invention,, determine in this sub-district and the neighbor cell same frequency user with the identical running time-frequency resource of this arbitrary CU for arbitrary user of each sub-district, place, base station; Between the base station of the base station at this arbitrary user place and this arbitrary user's sub-district, same frequency user place, share with user's identification information frequently; Each base station is carried out joint-detection with the user of sub-district, place with this user's same frequency user, obtains this user and with user's testing result frequently.By the way, can be converted to useful signal, carry out each joint-detection, thereby effectively suppress co-channel interference, improve systematic function with the frequency user with the phase mutual interference between the user frequently.

Description of drawings

Fig. 1 is the network model of seven cell.

Fig. 2 is the network interferences model of 19 sub-districts.

Fig. 3 is the simulation result schematic diagram of three kinds of MIMO detection methods.

Embodiment

For making purpose of the present invention, technological means and advantage clearer, the present invention is described in further details below in conjunction with accompanying drawing.

The user who takies identical running time-frequency resource is called with user frequently, with frequently between the user owing to take identical running time-frequency resource, so the phase mutual interference is bigger, and for the alien frequencies user who takies different running time-frequency resources, interference is less each other, can ignore in detection, can not influence signal detecting result.Therefore, when carrying out the input of multiuser MIMO, consider together the phase mutual interference between the user frequently among the present invention, and to being used with the interference between the user frequently, to carrying out joint-detection with the frequency user.

Further, the a plurality of users that participate in multiuser MIMO may be arranged in different sub-districts, these users and the characteristic of channel between the cell base station of place is different separately, therefore, if the same frequency user between a plurality of base stations is carried out joint-detection, need be between with the different districts at user place frequently the sharing users identification information, be used to calculate this cell base station and other sub-districts with the channel information between the user frequently.

Based on above-mentioned analysis, basic thought of the present invention is: be adjacent the identification information of sharing between cell base station with the frequency user in the base station, and user in arbitrary base station is carried out joint-detection with the same frequency user of this user in this sub-district and neighbor cell.

Simultaneously, above-mentioned to after frequently the user carries out joint-detection, in arbitrary base station side, can access each testing result with the frequency user, can further between each base station, share testing result with the frequency user with frequency user place, like this, arbitrary base station, except that the testing result of detected community user own, can also receive the testing result of other base stations for this user, the testing result of each base station for this same user can be merged, thereby further improve the accuracy of testing result.As above, by between a plurality of base stations, sharing identification information and user's testing result with the frequency user, minizone frequency user's interference signal together is converted into useful information, realizes the allied signal detection of multiuser MIMO by the cooperation of many base stations, and can improve the detection performance greatly.

Next many cooperation between base stations of the application being carried out the method that multiuser MIMO detects is described in detail.

Consider the VBLAST model.Suppose N is arranged _TIndividual transmitting antenna, N _RIndividual reception antenna, N _T≤ N _ROriginal bit stream at first serial to parallel conversion is a N _TParallel sub-bit stream is encoded respectively, is modulated and be mapped as data symbol every way bit stream then, delivers on the transmitting antenna separately and launches.Suppose that the sub-bit stream on each transmitting antenna adopts same modulation system, the constellation point number is M.Data symbol on each antenna is that unit carries out burst transmissions with the burst frame.Each burst frame is become by L data symbols.Be easy analysis, suppose L=1, modulation system is QPSK, then certain constantly the signal phasor received of receiver can be expressed as

y＝Hx+n????(1)

Wherein $y={[{\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="A2009100901400017H1.png,A2009100901400018H1.png"\; state="\{\{state\}\}">y</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="A2009100901400019H1.png"\; state="\{\{state\}\}">y</figure-callout>}}_2,\&CenterDot;\&CenterDot;\&CenterDot;,y_{N_R}]}^T$ Be N _RDimension received signal vector.H is N _R* N _TThe dimension channel matrix, its element h _IjThe channel fading coefficient of expression from transmitting antenna j to reception antenna i.X is that dimension is N _T* 1 multi-user multi-antenna sends signal phasor.

Suppose that each user has only a transmitting antenna, then transmitted data symbols vector $x={[x_1,x_2,\&CenterDot;\&CenterDot;\&CenterDot;,x_{N_T}]}^T$ Can equivalence be N _TIndividual user's transmission signal phasor; Suppose that each user has many transmit antennas, then the transmitted data symbols vector $x={[x_1^T,x_2^T,\&CenterDot;\&CenterDot;\&CenterDot;,x_{N_U}^T]}^T\stackrel{\mathrm{\&Delta;}}{=}{[x_1,x_2,\&CenterDot;\&CenterDot;\&CenterDot;,x_{N_T}]}^T$ Can equivalence be N _UIndividual user's transmission signal phasor, always total N _TIndividual transmission symbol.

Suppose that channel is a systems of quasi-static flat Rayleigh fading channels, the quasistatic here is meant in a burst frame, the channel fading coefficient is constant, and the channel fading coefficient is a change at random between each burst frame, and the channel of at present general communication system can satisfy above-mentioned condition.In addition, suppose that receiving terminal has precise channels to estimate with synchronously, so CSI is accurately as can be known.Under rich scatter channel environment, h _IjObedience independent same distribution, average are zero multiple Gaussian Profile. $n={[{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="A2009100901400017H1.png,A2009100901400018H1.png"\; state="\{\{state\}\}">n</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A2009100901400019H1.png"\; state="\{\{state\}\}">n</figure-callout>}}_2,\&CenterDot;\&CenterDot;\&CenterDot;,n_{N_R}]}^T$ Be N _RDimension AWGN noise vector, its each component all are that average is zero, and variance is 2 σ ²Separate multiple normally distributed random variable.The channel of at present general communication system can satisfy above-mentioned condition.Because the PDA algorithm can detect a plurality of users simultaneously, this provides convenience for multi-cell cooperating.Suppose that each sub-district uses the PDA algorithm respectively, then for single subdistrict, the Mathematical Modeling of its Multiuser Detection is identical with formula (1).

The base station collaboration MIMO system model is further expanding (1) formula among the present invention.Suppose in the sub-district and to adopt the quadrature multiple access between the user, therefore can think not have the phase mutual interference between the user in the sub-district; Have only and produce the phase mutual interference between neighbor cell.With simple sub-district linear array model is example, and then the received signal at place, s base station can be expressed as:

$y_s=\underset{i=s-1}{\overset{s+1}{\mathrm{\&Sigma;}}}{H}_{\mathrm{is}}{x}_i+n_s,s=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;N---\left(2\right)$

Y wherein _sBe N _RDimensional vector, X _iBe user's transmission signal vector of i sub-district, dimension N _Ti* 1, N _TiBe the transmitting antenna (supposing that the inherent synchronization in each sub-district has only a user and neighbor cell user to form co-channel interference) of i community user, H _IsBe the same frequency user of i sub-district and the channel response vector between s base station.The linear array of network has N base station.For each base station, can be with various detectors with X _iDetect respectively, for example PDA algorithm etc.When considering more actual network model, the interference signal number in (2) can be more, and this depends on concrete network configuration.

Carrying out input with employing PDA algorithm in the following examples is example, and the specific implementation of multi-user MIMO test method among the present invention is described.

As shown in Figure 1, the position polar coordinate representation of network element is an initial point with the center cell center, is 0 degree angle with horizontal right.As representing the base station of sub-district, the upper right corner, can use eNB1 (R, π/6) to represent its coordinate.Among Fig. 1 one group use shadow representation with user frequently, be that this can be planned definite by network with user frequently as UE0, UE1, UE2, UE3, UE04, UE5 and UE6.Suppose that a certain frequency is dispensed on a sector, lower right in the sub-district.

Area interference model such as Fig. 2 only consider co-channel interference here.If the base station eNB 0 of sub-district Cell0 is wanted the information of process user UE0, at this moment eNB0 has received interference signal simultaneously, these interference signals are to be that the center of circle is that the same frequency user who comprises in the circle of radius sends with R with eNB0, i.e. the signal x that sends of UE2, UE3 and UE4 user ₂, x ₃And x ₄So signal y that base station eNB 0 receives ₀For

y ₀＝H ₀₀x ₀+H ₂₀x ₂+H ₃₀x ₃+H ₄₀x ₄+n ₀????(3)

Suppose that base station eNB 0 carries out multiuser MIMO and detect, concrete grammar comprises:

Step 1 is determined the same frequency user of user UE0 in base station eNB 0 coverage cell.

As previously mentioned, the same frequency user of user UE0 be neighbor cell UE2, UE3 and UE4.

Step 2 is shared the identification information with the frequency user between base station eNB 0 and eNB2, eNB3, the eNB4.

Same frequency user (being UE0, UE2, UE3 and the UE4) identification information that this step is shared between a plurality of base stations (being base station eNB 0 and eNB2, eNB3, eNB4) is used for each base station and estimates that other community users are to the characteristic of channel between this base station.

Step 3, base station eNB 0 utilize the PDA algorithm that UE0, UE2, UE3 and UE4 are carried out joint-detection.

The PDA algorithm under the multiuser MIMO of at first deriving is determined the mode of each user's information symbol probabilistic information in each iteration.Particularly, when utilizing the PDA algorithm that the multiuser MIMO signal is carried out joint-detection, successively the signal on every transmitting antenna among each user being detected, is that example describes with the input on one of them transmitting antenna, and the signal detecting mode on other transmitting antennas is identical.

The system model of (2) formula is expanded, and the received signal of s base station is y _s, dimension N _R* 1, N _UThe same frequency total number of users that detects simultaneously for the base station.Order $x={[x_1^T,x_2^T,\&CenterDot;\&CenterDot;\&CenterDot;,x_{N_U}^T]}^T\stackrel{\mathrm{\&Delta;}}{=}{[x_1,x_2,\&CenterDot;\&CenterDot;\&CenterDot;,x_{N_T}]}^T,$ $H=[H_{1s},H_{2s},\&CenterDot;\&CenterDot;\&CenterDot;,H_{N_{U}s}],$ H _{I, s}Be i with the characteristic of channel between user frequently and s base station, N _TBe the transmission total number of symbols that the base station is detected simultaneously, then formula (2) can expand to:

$y_s=H_{1s}{x}_1+H_{2s}{x}_2+...+H_{N_{U}s}{x}_{N_U}+n_s=\mathrm{Hx}+n_s$

(H is taken advantage of on the following formula both sides together ^HH) ^-1H ^H, obtain following formula:

$\tilde{y}=x+\tilde{n}=x_j{e}_j+\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}{x}_k{e}_k+\tilde{n}\stackrel{\mathrm{\&Delta;}}{=}{x}_j{e}_j+{\tilde{N}}_j---\left(4\right)$

Wherein, $\tilde{y}={\left(H^{H}H\right)}^{-1}{H}^H{y}_s,$ j，k∈{1，2，...，N _T}。 Be that average is zero, its covariance is $\mathrm{Cov}\left(\tilde{n}\right)={\mathrm{\&sigma;}}^2{\left(H^{H}H\right)}^{-1}$ Multiple Gaussian noise, σ ²Be noise power, e _kBe that k element is 1, other elements are zero unit vector.

The key of PDA algorithm is exactly disturbing and noise item

The part that on (being called pseudo noise) this stricti jurise is not Gaussian noise forces to be used as Gaussian noise, estimates { a in the planisphere then in each iteration ₁, a ₂..., a _m..., a _MEach data symbol a _mPosterior probability P (x _j=a _m| y)=P _m(x _j| y), each received signal is entered a judgement according to these posterior probability.In detecting specific to the multiuser MIMO among the present invention, when promptly detecting the data symbol on a certain transmitting antenna j, with all transmitting antenna k of other transmitting antennas of this user and this user's same frequency user (k=1,2 ..., N _T, the signal on and the k ≠ j) is used as and is disturbed and noise.

Concrete pseudo noise

With average be

Covariance matrix is ${\mathrm{\&Omega;}}_j\stackrel{\mathrm{\&Delta;}}{=}\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}{\mathrm{\&omega;}}_k{e}_k{e}_k^T+\mathrm{Cov}\left(\tilde{n}\right),$ Pseudo-covariance matrix is ${\mathrm{\&Xi;}}_j\stackrel{\mathrm{\&Delta;}}{=}\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}{\mathrm{\&xi;}}_k{e}_k{e}_k^T$ Gaussian noise go to be similar to.Wherein

${\tilde{x}}_k=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{a}_m{P}_m\left(x_k\right|y)---\left(5\right)$

${\mathrm{\&omega;}}_k=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}(a_m-{\tilde{x}}_k){(a_m-{\tilde{x}}_k)}^{*}{P}_m\left(x_k\right|y)---\left(6\right)$

${\mathrm{\&xi;}}_k=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}(a_m-{\tilde{x}}_k){(a_m-{\tilde{x}}_k)}^T{P}_m\left(x_k\right|y)---\left(7\right)$

Order $w=\tilde{y}-x_j{e}_j-\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}{\tilde{x}}_k{e}_k,$ And

${\mathrm{\&phi;}}_m\left(x_j\right)\stackrel{\mathrm{\&Delta;}}{=}\exp \left(\begin{array}{cc}-{\left(\begin{array}{c}\mathrm{Re}\left(w\right)\\ \mathrm{Im}\left(w\right)\end{array}\right)}^T& {\mathrm{\&Lambda;}}_j\left(\begin{array}{c}\mathrm{Re}\left(w\right)\\ \mathrm{Im}\left(w\right)\end{array}\right)\end{array}\right)---\left(8\right)$

${\mathrm{\&Lambda;}}_j\stackrel{\mathrm{\&Delta;}}{=}{\left(\begin{array}{cc}\mathrm{Re}({\mathrm{\&Omega;}}_j+{\mathrm{\&Xi;}}_j)& -\mathrm{Im}({\mathrm{\&Omega;}}_j-{\mathrm{\&Xi;}}_j)\\ \mathrm{Im}({\mathrm{\&Omega;}}_j+{\mathrm{\&Xi;}}_j)& \mathrm{Re}({\mathrm{\&Omega;}}_j-{\mathrm{\&Xi;}}_j)\end{array}\right)}^{-1}{=(\left(\begin{array}{cc}\mathrm{Re}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)& -\mathrm{Im}\left(\mathrm{Cov}\tilde{n}\right)\\ \mathrm{Im}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)& \mathrm{Re}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)\end{array}\right)+Q)}^{-1}---\left(9\right)$

$Q\stackrel{\mathrm{\&Delta;}}{=}\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}({\mathrm{\&omega;}}_k+\mathrm{Re}\left({\mathrm{\&xi;}}_k\right))e_k{e}_k^T+\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}({\mathrm{\&omega;}}_k-\mathrm{Re}\left({\mathrm{\&xi;}}_k\right))e_{k+N_T}{e}_{k+N_T}^T+\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}\mathrm{Im}\left({\mathrm{\&xi;}}_k\right)e_k{e}_{k+N_T}^T+\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}\mathrm{Im}\left({\mathrm{\&xi;}}_k\right)e_{k+N_T}{e}_k^T$

Wherein, real part is got in Re () expression, and imaginary part, a are got in Im () expression _mValue for m constellation point of information symbol.Because supposed that emission symbol priori etc. is general, then the posterior probability of each symbol is

$P_m\left(x_j\right|y)=\frac{p_m\left(y\right|x_j)P(x_j=a_m)}{p\left(y\right)}=\frac{p_m\left(y\right|x_j)P(x_j=a_m)}{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{p}_m\left(y\right|x_j)P(x_j=a_m)}=\frac{{\mathrm{\&phi;}}_m\left(x_j\right)}{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{\&phi;}}_m\left(x_j\right)}---\left(10\right)$

As above as seen, can be according to the transmission symbol posterior probability on other transmitting antennas k except that transmitting antenna j, through type (5)～(10) calculate the posterior probability that transmitting antenna j goes up information symbol.

Based on the above-mentioned detection that signal on arbitrary transmitting antenna is carried out, the whole process of in multiuser MIMO each subscriber signal being carried out joint-detection comprises:

1) initialization: construct the channel matrix between described user and same user frequently and described user place cell base station $H=[H_{1s},H_{2s},\&CenterDot;\&CenterDot;\&CenterDot;,H_{N_{U}s}],$ And calculate

With P _m(x _j| y) be initialized as the even distribution that constellation point is gathered, j=1,2 ..., N _T, m=1,2 ..., M, and establish iterations counter z=1;

2) detect to described user and with the signal on each transmitting antenna of frequency user successively, and when the signal on described transmitting antenna to each detects, signal on other transmitting antennas outside the transmitting antenna to be detected as interference, is determined up-to-date information symbol probability on the described transmitting antenna to be detected.

Particularly, for

, j ∈ 1,2 ..., N _T, based on { P (k) } _{K ≠ j}Probability vector, calculate P by (5)～(10) _m(x _j| y), and upgrade original P (j) with this result, wherein, each P (k)={ P _m(x _k| y) } be the probability vector that constitutes by M probable value, M is the constellation set element number;

3) if for

P (j) has restrained, and has perhaps arrived the upper limit of given iterations, forwards 4 to), otherwise, make z=z+1, return 2);

4) to x _jProvide testing result

Promptly ${\hat{x}}_j=a_i,$ $l=\arg \underset{m^{\&prime;}=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,M}{\max }\left\{P_{m^{\&prime;}}\left(x_j\right|y)\right\},$ Then $\hat{x}=\left\{{\hat{x}}_j\right|j=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,N_T\}.$

By top step, we can detect each same subscriber signal frequently of a plurality of sub-districts simultaneously.

So far, the most basic multiuser MIMO joint-detection flow process finishes among the present invention.In this basic procedure, will be converted into useful information with frequency user's interference signal, be used for this user's input.Detect performance for further improving, can also comprise the steps:

Step 4, base station eNB 0 are determined this community user and each probabilistic information with frequency user's testing result correspondence thereof, and user's testing result and corresponding probabilistic information thereof send to corresponding neighbor cell base station frequently together with neighbor cell.

Each base station can be determined the same frequency user's of this community user and neighbor cell thereof information by step 3, and the testing result that comprises this user (promptly

) and corresponding probabilistic information

Like this, the information that each base station detects had both comprised that this sub-district also comprised the information of adjacent sub-district with the frequency user with frequency user profile, at this moment the adjacent sub-district that will detect, this base station passes to adjacent cell base station with user profile frequently, will receive simultaneously adjacent cell base station detected about this sub-district with user's information frequently.In the present embodiment, eNB0 will obtain 4 information of UE0

With

Three neighbor cell base station eNB1, the eNB5 and the eNB6 that come from this eNB0 and other respectively.

This user's that step 5, each base station are sent the probabilistic information of this base station cell user's testing result correspondence and other neighbor cells the corresponding probabilistic information of testing result carries out soft merging to be handled, and determines this base station cell user's final detection result.

After the information transmission finished between the base station, the probabilistic information about this user data that detects comprised the probabilistic information of this cell base station detection and the probabilistic information that adjacent cell base station detects, and at last these probabilistic informations is done total supple-settlement and is merged.Specifically can merge all probabilistic informations of user by high specific, for example,, can obtain x for the UE0 under the eNB0 ₀Conclusive judgement information: ${\hat{x}}_0=f({\hat{x}}_{01},{\hat{x}}_{02},{\hat{x}}_{03},{\hat{x}}_{04}).$

For the performance of the multi-user MIMO test method among explanation the present invention, emulation has been carried out in the detection of three kinds of modes, concrete simulated conditions is referring to table 1, and simulation result is referring to Fig. 3.

Table 1 simulated conditions

Parameter	Numerical value and mode
		Modulation system	??QPSK
Every user antenna number	??2
		The reception antenna number	??8
Adjacent sub-district co-channel interference user and this sub-district	??0.5

Power ratio p with the frequency user
		Channel type	The flat Rayleigh fading channel

In Fig. 3, when single sub-district list user independent detection is adopted in curve 301 expressions, the corresponding relation curve of the signal to noise ratio and the error rate; Curve 302 expression is adopted multi-cell cooperating multiuser MIMO joint-detection and when testing result not being merged, the corresponding relation curve of the signal to noise ratio and the error rate; Curve 301 expression is adopted multi-cell cooperating multiuser MIMO joint-detection and when testing result merged, the corresponding relation curve of the signal to noise ratio and the error rate.Can see that by above-mentioned three curves during the list user independent detection of single sub-district, system detects poor performance; The multi-cell cooperating iterative detection, but the information of not carrying out is when merging, and it is more a lot of than the performance improvement of the single user's independent detection in single sub-district to detect performance; The multi-cell cooperating iterative detection, and the information of carrying out is when merging, system detects performance and further improves, in the error rate 10 ^-4The time obtained the performance gain of about 3dB.

This shows,, abandoned the thought that single user detects respectively on the one hand, in a base station, simultaneously the same frequency user in a plurality of sub-districts is carried out multi-user association and detect by the multiuser MIMO associated detecting method among the present invention; Simultaneously, abandon the thought of simple linearity test, in detection, carried out soft information iteration, made full use of soft go into soft the go out character of algorithm and the gain that iteration is brought; Further, broken the thought of single sub-district independent detection, under the situation of the identification information of sharing same user frequently between neighbor cell and user data information, the co-channel interference that can suppress the minizone effectively, made full use of the space diversity gain that the MIMO model brings, particularly when interference is very serious, improved systematic function greatly.

In addition, since cooperation between base stations MIMO multi-user Iterative detection algorithm only in the end a step to carry out soft information and between the base station reciprocal process that soft information is transmitted takes place when merging, extra information loads is very little, algorithm complex is moderate, data quantity transmitted is also moderate between neighbor cell, and is practical.

Determine in the foregoing description can be applicable to the hexagon cellular network model that present industry is generally used, very big real specific aim is arranged with frequency user's mode.

Being preferred embodiment of the present invention only below, is not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (6)

1, the multi-user MIMO test method between a kind of base station is characterized in that, this method comprises:

For arbitrary user of each sub-district, place, base station, determine in this sub-district and the neighbor cell same frequency user with the identical running time-frequency resource of described arbitrary CU;

Between the base station of the base station at described arbitrary user place and this arbitrary user's sub-district, same frequency user place, share with user's identification information frequently, be used for that this cell base station estimates other sub-districts described with user frequently in the characteristic of channel of this base station; Each base station is carried out joint-detection with the user of sub-district, place with this user's same frequency user, obtains this user and with user's testing result frequently.

2, method according to claim 1 is characterized in that, after obtaining described user and the testing result with the frequency user thereof, this method further comprises:

Determine described user and the same probabilistic information of user's testing result correspondence frequently thereof;

With detected neighbor cell with frequently user's testing result and the base station that corresponding probabilistic information sends to corresponding neighbor cell thereof;

Each base station is to arbitrary user of sub-district, place, the corresponding probabilistic information of testing result of this subscriber signal that the probabilistic information of detected this user's of this cell base station signal detecting result correspondence and other neighbor cells are sent merges, and determines the final detection result of this subscriber signal.

3, method according to claim 2, it is characterized in that, merge this user's that probabilistic information and other neighbor cells of detected this user's of this cell base station signal detecting result correspondence send the corresponding probabilistic information of signal detecting result by high specific.

4, according to arbitrary described method in the claim 1 to 3, it is characterized in that, a user is carried out joint-detection with this user's same frequency user comprise: utilize iterative detection to carry out joint-detection to described user and with user frequently;

In each iterative detection, detect to described user and with the signal on each transmitting antenna of frequency user successively; And when the signal on described transmitting antenna to each detects, the signal on other transmitting antennas outside the transmitting antenna to be detected as interference, is determined up-to-date information symbol probability on the described transmitting antenna to be detected;

Up-to-date information symbol probability is used for next iteration and detects on each described transmitting antenna that each iterative detection is obtained, information symbol probability on all described transmitting antennas is all restrained, perhaps, reach the default iterations upper limit up to, then finishing iteration detects, and utilize information symbol probability up-to-date on each described transmitting antenna, determine the subscriber signal testing result on each described transmitting antenna, with the testing result of the subscriber signal testing result on all transmitting antennas of each user as the relative users signal.

5, method according to claim 4 is characterized in that, the mode that the signal on the transmitting antenna to be detected is detected comprises:

Construct the channel matrix between described user and same user frequently and described user place cell base station $H=[H_{1s},H_{2s},\&CenterDot;\&CenterDot;\&CenterDot;,H_{N_{U}S}],$ Wherein, H _{I, s}Be the characteristic of channel between i user among described user's the same frequency user and s base station;

The received signal of utilization on each described transmitting antenna calculated $\tilde{y}={\left(H^{H}H\right)}^{-1}{H}^H{y}_s,$ Wherein, () H represents to get conjugate transpose;

Go up up-to-date information symbol probability P according to other the described transmitting antenna k except that described transmitting antenna j to be detected _m(x _k| y), determine the up-to-date information symbol probability on the described transmitting antenna to be detected $P_m\left(x_j\right|y)=\frac{{\mathrm{\&phi;}}_m\left(x_j\right)}{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{\&phi;}}_m\left(x_j\right)},$

Wherein, ${\mathrm{\&phi;}}_m\left(x_j\right)\stackrel{\mathrm{\&Delta;}}{=}\exp \left({-\left(\begin{array}{c}\mathrm{Re}\left(w\right)\\ \mathrm{Im}\left(w\right)\end{array}\right)}^T{\mathrm{\&Lambda;}}_j\left(\begin{array}{c}\mathrm{Re}\left(w\right)\\ \mathrm{Im}\left(w\right)\end{array}\right)\right),$ ${\mathrm{\&Lambda;}}_j={\left(\begin{array}{c}\left(\begin{array}{cc}\mathrm{Re}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)& -\mathrm{Im}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)\\ \mathrm{Im}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)& \mathrm{Re}\left(\mathrm{Cov}\left(\tilde{n}\right)\right)\end{array}\right)+Q\end{array}\right)}^{-1},$ $Q=\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}({\mathrm{\&omega;}}_k+\mathrm{Re}\left({\mathrm{\&xi;}}_k\right))e_k{e}_k^T+\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}({\mathrm{\&omega;}}_k-\mathrm{Re}\left({\mathrm{\&xi;}}_k\right))e_{k+N_T}{e}_{k+N_T}^T+\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}\mathrm{Im}\left({\mathrm{\&xi;}}_k\right)e_k{e}_{k+N_T}^T+\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}\mathrm{Im}\left({\mathrm{\&xi;}}_k\right)e_{k+N_T}{e}_k^T,$ e _kBe that k element is 1, other elements are zero unit vector, $w=\tilde{y}-x_j{e}_j-\underset{k\&NotEqual;j}{\mathrm{\&Sigma;}}{\tilde{x}}_k{e}_k,$ $\mathrm{Cov}\left(\tilde{n}\right)={\mathrm{\&sigma;}}^2{\left(H^{H}H\right)}^{-1},$ ${\mathrm{\&omega;}}_k=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}(a_m-{\tilde{x}}_k){(a_m-{\tilde{x}}_k)}^{*}{P}_m\left(x_k\right|y),$ ${\mathrm{\&xi;}}_k=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}(a_m-{\tilde{x}}_k){(a_m-{\tilde{x}}_k)}^T{P}_m\left(x_k\right|y),$ ${\tilde{x}}_k=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{a}_m{P}_m\left(x_k\right|y),$ Real part is got in Re () expression, and imaginary part, σ are got in Im () expression ²Be noise power, a _mBe the value of m constellation point of information symbol, M is the information symbol constellation point number on the transmitting antenna k.

6, method according to claim 5 is characterized in that, in first time during iteration, the posterior probability of each described transmitting antenna is initialized as even distribution in the constellation point set.

Images