CN102281091A - Reception method for multi-antenna communication system - Google Patents

Abstract

The invention discloses a reception method for a multi-antenna communication system and belongs to the field of the transmission of radio signals. The reception method is characterized by comprising the following steps of: ordering the Euclidean distance increments of only the top two layers of signals in a search tree of a detection algorithm so that only K sub-nodes with smaller Euclidean distance increments are kept in each two layers; then calculating the accumulated Euclidean distances of the (2N-1) layer in the top two layers, wherein N represents the number of the antenna; and sequentially selecting the (2N-2) layer and the (2N-3) layer from the (2N-2) layer, calculating the Euclidean distance increment of each layer and adding the calculated Euclidean distance increments with K accumulated Euclidean distances in the first period to obtain the accumulated Euclidean distance of the (2N-3) layer, and performing the calculation in the same manner till the accumulated Euclidean distance of the first layer is calculated. According to the invention, two adjacent layers of signals can be simultaneously treated so that the ordering, selecting, calculating and searching times are reduced, the system time relay and the calculation complexity are decreased, and lots of hardware resources are saved.

Description

A kind of method of reseptance that is used for multiple antenna communication

Technical field

The invention belongs to the wireless messages transmission field, particularly adopt the system and the standard such as WLAN (wireless local area network), broadband wireless access, mobile communication of multi-antenna technology.

Background technology

Multiple-input and multiple-output (MIMO) technology is all used many antennas at signal sending end and receiving terminal, in identical frequency band, transmit a plurality of data flow simultaneously by the space-domain multiplexing mode, can provide than the higher spectrum efficiency of a single aerial system (SISO), be applied to new generation broadband wireless communication standards such as 3GPPLTE, LTE-Advanced, IEEE802.11n at present.The best hard decision detection mode of MIMO wireless system is maximum likelihood (ML) detection method, its complexity of directly implementing makes ASIC or FPGA only can be used for the low-order modulation scheme of a few antennas along with the increase of number of antennas and order of modulation grows exponentially.Subsequently, people have proposed the K-Best detection algorithm, and it can keep bit error rate (BER) performance that compares favourably with best ML detector, can significantly reduce computation complexity again, is subjected to extensive concern.

General K-Best algorithm divides the system model real number to be separated, and reduces the hardware implementation complexity by deciphering in real number field, introduces the principle of this algorithm below.Consider that transmitting antenna number and reception antenna number are the MIMO communication system baseband signal equivalent model of N

y＝Hx+n (1)

Wherein, H is N * N channel matrix; X=[x ₁, x ₂..., x _N] be that N ties up the transmission signal; Y=[y ₁, y ₂..., y _N] be that N ties up received signal; N is a N dimension additive white Gaussian noise.For avoiding complex operation to bring extra hardware expense, system model (1) real numberization can be decomposed into:

$\left[\begin{array}{c}\mathrm{Re}\left(y\right)\\ \mathrm{Im}\left(y\right)\end{array}\right]=\left[\begin{array}{cc}\mathrm{Re}\left(H\right)& -\mathrm{Im}\left(H\right)\\ \mathrm{Im}\left(H\right)& \mathrm{Re}\left(H\right)\end{array}\right]\left[\begin{array}{c}\mathrm{Re}\left(x\right)\\ \mathrm{Im}\left(x\right)\end{array}\right]+\left[\begin{array}{c}\mathrm{Re}\left(n\right)\\ \mathrm{Im}\left(n\right)\end{array}\right]---\left(2\right)$

Wherein Re (*) and Im (*) represent the real part and the imaginary part of plural number (*) respectively.The channel matrix of real numberization

Decomposition obtains through QR Adopting maximum-likelihood criterion to find the solution (1) can get:

$\hat{x}=\underset{x\∈{\mathrm{\Ω}}^{2N}}{\arg \min }\underset{i=1}{\overset{2N}{\mathrm{\Σ}}}\left|\right|{\tilde{y}}_i^T-\underset{j=i+1}{\overset{2N}{\mathrm{\Σ}}}{R}_{\mathrm{ij}}{x}_j-R_{\mathrm{ii}}{x}_i\left|\right|---\left(3\right)$

Wherein,

The x estimated value of European metric range minimum is satisfied in expression; Arg min represents the mathematical formulae of minimizing;

The European metric range of representing the i layer signal; What Q represented is: 2N * 2N orthogonal dimension matrix; What R represented is: 2N * 2N ties up upper triangular matrix;

Be the 2N dimensional vector, Be the real number field received signal.

According to formula (3), the tree searching structure of general K-Best MIMO detection algorithm is (2 * 2 antennas, 16-QAM modulation, K=4) as shown in Figure 1, the MIMO detector detects from top the beginning, need to calculate every layer of Euclidean distance increment (INC), and obtain the PED of this layer, and select the individual minimum PED of K and the corresponding node (survival node) thereof of this layer reservation by sorting operation with K accumulation Euclidean distance (PED) addition that the upper strata keeps.For order of modulation is the K-Best detector tree searching structure of M, and every layer needs the quantity of the PED of calculating, ordering to be

When K and M increase, the complexity of calculating is exponential increase, becomes hard-wired bottleneck.

Adopt above-mentioned K-Best algorithm 2 * 2MIMO detector hardware configuration as shown in Figure 2.Detector is made up of QR decomposing module, balance module, generation module to be selected, 4 layers of K-Best module and conclusive judgement module.Wherein, each K-Best module comprises that the Euclidean distance computing unit (DCM) of node layer Euclidean distance is worked as in calculating, the sequencing unit (SSM) of K node is selected in ordering.The number of plies to be detected is separated along with the channel model real number divides and is doubled in this K-Best detector, and detector detects the 2N layer signal respectively 2N different phase.

Can see that the sequencing selection unit of traditional structure K-Best MIMO detector has also taken great amount of hardware resources.For this reason, the present invention proposes a kind of novel K-Best MIMO detector arrangement, can detect two layer signals simultaneously, reduced ordering and tree searching times, and have the BER performance slightly good with traditional K-best detector.

Summary of the invention

The object of the present invention is to provide a kind of detection method of mimo system of low complex degree.

Thought of the present invention is: adopt new tree searching structure, only the Euclidean distance increment to the highest two layer signals sorts, and the adjacent two layers of parallel processing simultaneously signal has reduced sequencing selection computing and tree searching times.

Performing step of the present invention is as follows:

Step (1), transmitting antenna number of initialization and reception antenna number are the multiple-input and multiple-output MIMO communication system of N, and its baseband signal equivalent model is expressed as: y=Hx+n, wherein, y=[y ₁, y ₂..., y _N] be that N ties up received signal, x=[x ₁, x ₂..., x _N] for the N dimension sends signal, H is N * N channel matrix, n is a N dimension additive white Gaussian noise;

Step (2) utilizes the QR decomposing module to the N * N dimension real number channel matrix of described channel matrix through obtaining after the real numberization Decompose: Wherein, Q is a unitary matrice, and R is a upper triangular matrix;

Step (3), balance processing module are utilized the real domain part of described unitary matrice Q to described N dimension received signal

Carry out equilibrium treatment, obtain the 2N dimensional vector T is the transposition symbol;

Step (4) utilizes generation module to be selected to produce to be used for to be selected R of the Euclidean distance increment that calculates each layer signal of detection algorithm search tree from described upper triangular matrix R _{2N, 2N}X, n=1,2 ..., 2N-1,2N;

Step (5), first order K-Best module are calculated K less accumulation Euclidean distance PED of 2N-1 layer successively according to following steps _2N-1(x);

Step (5.1) is calculated as follows the Euclidean distance increment INC as top 2N layer _2N(x), the start node in the search tree is a lowermost layer in the described K-Best pattern

Wherein,

Be the real number field received signal of 2N layer, R _{2N, 2N}Be 2N layer upper triangular matrix R _2NMiddle sequence number is the element of 2N;

Step (5.2) is calculated as time Euclidean distance increment INC of the 2N-1 layer of high level according to following formula _2N-1(x), ${\mathrm{INC}}_{2N-1}\left(x\right)=|{\tilde{y}}_{2N-1}^T-R_{2N-\mathrm{1,2}N-1}x|;$

Step (5.3) is calculated M accumulation of 2N-1 layer Euclidean distance PED according to following formula _2N-1(x), M is an order of modulation, PED _2N-1(x)=INC _2N(x)+INC _2N-1(x);

According to sorting from small to large, the highest in two-layer every layer only keeps the less L of Euclidean distance increment accumulation Euclidean distance increment to step (5.4) Euclidean distance increments all in the step (5.3), L＜K, and every layer of L is identical.Make that K accumulation Euclidean distance of 2N-1 layer is the permutation and combination of 2N layer, each L of 2N-1 layer Euclidean distance increment, use PED _2N-1(x) expression simultaneously, makes that each father node respectively keeps a child node that has the minimum euclidean distance increment in the 2N-1 layer, amounts to K altogether;

Step (6), second level K-Best module is calculated 2N-2 earlier, and K is individual altogether to the two-layer altogether Euclidean distance increment of 2N-3, K the accumulation Euclidean distance addition of the 2N-1 layer that calculates of the previous stage K-Best module that this K Euclidean distance increment and step (5) are obtained again, obtain 2N-2 layer and 2N-3 layer altogether K of the two-layer 2N-3 layer that remains accumulate Euclidean distance PED _2N-3(x);

Step (7), repeating step (6) calculate till the 1st layer always;

Step (8) is selected a group node that has the European distance of increment of minimum accumulation with a judging module and is exported as detector from K the node that remains.

Different with traditional K-Best detector, the detector of proposition only comprises two K-Best modules.Each K-Best module is calculated the Euclidean distance increment of adjacent two layers signal simultaneously, and with each K the Euclidean distance increment and the addition of K accumulation previous stage Euclidean distance of two-layer reservation, just can obtain K that this stage keeps and accumulate Euclidean distance and corresponding node thereof.Judging module is selected a group node that has minimum accumulation Euclidean distance path and is exported as decoding from the K group node of each layer reservation.The detector of the present invention design has reduced sequencing selection computing and tree searching times, and two layer signals of parallel detection simultaneously, system's time delay and hardware resource take to be compared traditional K-Best structure and reduce greatly.

Description of drawings

Fig. 12 * 2 16-QAM tradition K-Best (K=4) detection algorithm tree searching structure.

Fig. 2 tradition K-Best signal detector structure.

The novel K-Best signal detector of Fig. 3 structure.

Fig. 4 detection algorithm flow process.

Figure 52 * 2 antennas, QPSK modulation K-Best algorithm BER performance.

Embodiment

In this detector, treatment step to received signal such as Fig. 4 are described below:

The first step is to channel matrix

Carry out the QR decomposition and obtain unitary matrice Q and upper triangular matrix R.

In second step, utilize unitary matrice Q to carry out equilibrium treatment to received signal.

The 3rd step, utilize R compute euclidian distances increment to produce generation module to be selected, can reduce Euclidean distance incremental computations hardware resource and take.

The 4th step, in first K-Best module, at first calculate the (2N, 2N-1) Ceng Euclidean distance increment, and every layer of Euclidean distance increment ordering back is only kept best L, and (L＜K) individual calculates then and accumulates Euclidean distance PED _2N-1(x).

In the 5th step, other K-Best module is calculated the Euclidean distance increment of 1 layer of 2N-2 to the earlier, each this K Euclidean distance increment and the addition of K accumulation previous stage Euclidean distance, obtains K the accumulation Euclidean distance that this stage keeps then.

In the 6th step, judging module is selected a group node that has minimum accumulation Euclidean distance path and is exported as detector from the K group node that keeps.

The MIMO detector that the present invention proposes mainly is made up of QR decomposing module, balance module, generation module to be selected, two-layer K-Best module and judging module.Wherein, each K-Best module comprises that the Euclidean distance computing unit (DCM) of node layer Euclidean distance is worked as in calculating, the sequencing unit (SSM) of K node is selected in ordering.For example to 2 * 2 systems, hardware configuration such as Fig. 3 of the K-Best MIMO detector of proposition.

For analyzing the performance of the MIMO detector that proposes, under the flat fading channel model, point-to-point MIMO link to 2 * 2 antennas, QPSK, 16/64-QAM modulation, no chnnel coding has carried out emulation, and compare with the performance of traditional K-Best algorithm, maximum likelihood (ML) algorithm respectively, its bit error rate (BER) performance is as shown in Figure 5.Simulation result shows that along with the increase of K, the structure of proposition has slightly good BER performance than traditional K-Best detector.

The mimo system of forming with two width of cloth transmitting antennas and two amplitude receiver antennas is an example explanation input implementation step below.

The first step: channel matrix is carried out OR decompose, obtain

$Q=\left[\begin{array}{cccc}{Q}_{11}& Q_{12}& Q_{13}& Q_{14}\\ Q_{21}& Q_{22}& Q_{23}& Q_{24}\\ Q_{31}& Q_{32}& Q_{33}& Q_{34}\\ Q_{41}& Q_{42}& Q_{43}& Q_{44}\end{array}\right]R=\left[\begin{array}{cccc}{R}_{11}& R_{12}& R_{13}& R_{14}\\ 0& R_{22}& R_{23}& R_{24}\\ 0& 0& R_{33}& R_{34}\\ 0& 0& 0& R_{44}\end{array}\right]$

Second step: utilize unitary matrice Q to carry out equilibrium treatment to received signal, obtain

The 3rd step: utilize R to generate generation module to be selected, for example for the 16QAM modulation, generation module is R * (± 3), R * (± 1).

The 4th step: in first K-Best module, at first calculate (three, four) layer Euclidean distance increment, and every layer of Euclidean distance increment ordering back is only kept best L, and (L＜K) individual, two-layer L is identical, calculates accumulation Euclidean distance PED then _2N-1(x), it is 2N-1, N (N=2) layer Euclidean distance increment sum.As y among Fig. 3 ₁～y ₄Be the real part imaginary part of two groups of data of two groups of antennas, z ₃～z ₄Be the node of three or four layers of reservation, through summation, sequencing selection, summation output again.

The 5th step: in other K-Best module, calculate the Euclidean distance increment of 1 layer of 2N-2 to the earlier, each this K Euclidean distance increment and the addition of K accumulation previous stage Euclidean distance, obtain K the accumulation Euclidean distance that this stage keeps then.As z among Fig. 3 ₁～z ₄Represent that four layers keep node.

The 6th step: judging module is selected a group node that has minimum accumulation Euclidean distance path and is exported as detector from the K group node that keeps.

For the K-Best detector arrangement of verifying proposition has lower hardware implementation complexity, utilize Xilinx Virtex-4 (XC4VLX200) platform that the structure that proposes has been carried out comprehensive simulating.Table 1 has provided 2 * 2 antennas, 16-QAM modulation, and the hardware resource of two kinds of configuration detector takies contrast.Can see that the K-Best detector of proposition has been saved great amount of hardware resources than traditional K-Best detector.

Table 1 detector resource occupation

Claims (1)

1. a method of reseptance that is used for multiple antenna communication is characterized in that, has following performing step successively:

Step (1), transmitting antenna number of initialization and reception antenna number are the multiple-input and multiple-output MIMO communication system of N, and its baseband signal equivalent model is expressed as:

Y=Hx+n, wherein, y=[y ₁, y ₂..., y _N] be that N ties up received signal, x=[x ₁, x ₂..., x _N] for the N dimension sends signal, H is N * N channel matrix, n is a N dimension additive white Gaussian noise;

Step (2) utilizes the QR decomposing module to the channel matrix of described N * N channel matrix through obtaining after the real numberization Decompose:

Wherein, Q is a unitary matrice, and R is a upper triangular matrix;

Step (3), balance processing module are utilized the real domain part of described unitary matrice Q to described N dimension received signal

Carry out equilibrium treatment, obtain the 2N dimensional vector

T is the transposition symbol;

Step (4) is utilized generation module to be selected to produce from described upper triangular matrix R and is used for calculating the to be selected R that the detection algorithm tree is searched for the Euclidean distance increment of each layer signal _{2N, 2N}X, n=1,2 ..., 2N-1,2N;

Step (5), first order K-Best module are calculated K less accumulation Euclidean distance PED of 2N-1 layer successively according to following steps _2N-1(x);

Step (5.1): the Euclidean distance that is calculated as follows as top 2N layer increases INC _2N(x), the start node of search tree is a lowermost layer in the described K-Best algorithm

Wherein

Be the real number field received signal of 2N layer, R _{2N, 2N}Be 2N layer upper triangular matrix R _2NMiddle sequence number is the element of 2N;

Step (5.2): calculate as time Euclidean distance increment of the 2N-1 layer of high level according to following formula,

${\mathrm{INC}}_{2N-1}\left(x\right)=|{\tilde{y}}_{2N-1}^T-R_{2N-\mathrm{1,2}N-1}x|;$

Step (5.3): calculate M accumulation of 2N-1 layer Euclidean distance PED according to following formula _2N-1(x), M is an order of modulation, PED _2N-1(x)=INC _2N(x)+INC _2N-1(x);

Step (5.4): according to sorting from small to large, the highest in two-layer every layer only keeps the less L of Euclidean distance increment accumulation Euclidean distance increment Euclidean distance increments all in the step (5.3), L＜K, and every layer of L is identical.Make that K accumulation Euclidean distance of 2N-1 layer is the permutation and combination of 2N layer, each L of 2N-1 layer Euclidean distance increment, use PED _2N-1(x) expression simultaneously, makes that each father node respectively keeps a child node that has the minimum Eustachian distance increment in the 2N-1 layer, amounts to K altogether;

Step (6), second level K-Best module is calculated 2N-2 earlier, and K is individual altogether to the two-layer altogether Euclidean distance increment of 2N-3, K the accumulation Euclidean distance addition of the 2N-1 layer that calculates of the previous stage K-Best module that this K Euclidean distance increment and step (5) are obtained again, obtain 2N-2 layer and 2N-3 layer altogether K of the two-layer 2N-3 layer that remains accumulate Euclidean distance PED _2N-3(x);

Step (7), repeating step (6) calculate till the 1st layer always;

Step (8) is selected a group node that has minimum accumulation Euclidean distance increment with judging module and is exported as detector from K the node that remains.

Images