CN104618061A - Detection method for multi-user signal in large-scale multi-antenna system - Google Patents

Abstract

The invention relates to a detection method for multi-user signal in a large-scale multi-antenna system and belongs to the technical field of wireless communication. The method adopts the obtained CSI to detect the multi-user information at the base station side. The method has low complexity and good error bit rate performance, which is close to the error bit rate performance of the optimal MMSE detection within a few iterations. The method adopts the Lanczo iterative algorithm for obtaining the system of linear equations for avoiding the mode of obtaining MMSE orthogonal matrix by directly using large-scale matrix inversion mode, the detection calculating complexity is greatly reduced; in addition, the method can calculate LLR value of each information bit in lower complexity as the input of the decoder and the joint detection decoding performance is obviously improved. The complexity is obviously reduced while the ratio of the user number K and the base station antenna number N is certain and the number tends to infinity, and the error bit rate performance can gradually reach to the maximum likelihood detection (optimal) performance.

Description

The detection method of multiple user signals in a kind of large-scale multi-antenna system

Technical field

The present invention relates to the detection method of multiple user signals in a kind of large-scale multi-antenna system, particularly relate to a kind of based on multi-user test method in the low complex degree large-scale multi-antenna system of Lanczos algorithm, belong to wireless communication technology field.

Background technology

Along with the growth at full speed of mobile terminal number, the development of Internet of Things, and being on the increase of wireless data service kind, user constantly proposes new challenge to existing mobile communication technology to the demand of wireless data service.Have the technology of satisfied following wireless mobile communications to spectrum efficiency, energy efficiency and channel capacity demand potential quality as a kind of, extensive multiple-input and multiple-output (hereinafter referred to as Massive MIMO) technology has caused in academia and industrial circle and has paid close attention to widely.

Massive MIMO technology by the antenna at base station (hereinafter referred to as BS) sidepiece administration One's name is legion, for the communication between communication user provides the abundant degree of freedom to improve diversity gain and spatial multiplexing gain.But when the antenna number of BS is very large, the computation complexity of traditional high-performance MIMO detection technique will become very huge, the computation complexity as the detection technique based on globular decoding (SD) algorithm exponentially increases along with the increase of antenna number.Although and decline to some extent than the detection technique based on SD algorithm in complexity based on the detection technique of neighborhood search, can not ensure that final testing result is not local optimum.Under traditional MIMO frame (non-extensive antenna), the linear detection algorithm that complexity is lower, as ZF (ZF) detection algorithm can not provide enough beam forming gain, matched filtering (MF) detection algorithm fully can not eliminate the mutual interference of signal between each user, has some limitations and detection perform is not very good.But when the antenna number of BS side is tending towards infinite, because the channel between user and the different antennae of BS is progressive orthogonal, linear detection algorithm progressively with lower complexity can reach optimum detection perform.But BS is due to the restriction by spatial dimension and construction cost in actual scene, can not adopt infinite many antenna configurations, therefore, the linear detection algorithm close to optimal detection performance of low complex degree still requires study.

Summary of the invention

The object of the invention is the detection method proposing multiple user signals in a kind of large-scale multi-antenna system, utilize existing channel condition information (hereinafter referred to as the CSI) signal to multiple user to detect and be separated in BS side, the log-likelihood information (hereinafter referred to as LLR) of every bit in each subscriber signal is calculated in testing process, and carry out iterative decoding according to this log-likelihood information, under lower computation complexity, ensure certain bit error rate performance.

The detection method of multiple user signals in the large-scale multi-antenna system that the present invention proposes, comprises the following steps:

(1) set antenna for base station number in large-scale multi-antenna system as N, number of users is K, and each user configures an antenna, then the up channel matrix of large-scale multi-antenna system is wherein represent complex field, K the user of user in T time slot transmit into wherein R is the code check of subscriber signal coding, transmits S after chnnel coding, intertexture and sign map, obtains antenna transmission signal, be designated as the power of antenna transmission signal is antenna transmission signal X is by the antenna transmission of user, and through channel, obtain base station received signal at the N root antenna of base station, be designated as Y=HX+W, wherein, W represents the additive white Gaussian noise in base station received signal, element in matrix W is the signal to noise ratio of base station received signal is

(2) the matched filtering vector of above-mentioned base station received signal is calculated wherein for the up channel matrix H of moment t in T time slot _tconjugate transpose, y _tfor the t row in above-mentioned base station received signal Y, represent the signal vector that base station receives at moment t;

(3) the up channel matrix of above-mentioned large-scale multi-antenna system is calculated gram matrix G _t, wherein ρ is the signal to noise ratio of base station received signal, i _kfor K rank unit matrix;

(4) use Lanczos process, calculate and above-mentioned G _trelevant row orthogonal matrix Q and symmetric triple-diagonal matrix T, detailed process is as follows:

(4-1) calculate at the above-mentioned base station received signal y of moment t _tthe balanced Matrix C of least mean-square error _t, by the balanced Matrix C of this least mean-square error _tto y _tcarry out filtering, obtain the antenna transmission signal x of moment t _testimated value ${\hat{x}}_t=C_t{y}_t;$

(4-2) the equivalent linear equation group model detected in a large-scale multi-antenna system is set up adopt Lanczos process, solve equivalent linear equation group model, obtain the antenna transmission signal x of moment t _testimated value concrete steps are as follows;

(4-2-1) initialization: establish iterations threshold value P, if estimated value initial value be then above-mentioned model initial complement vector be r ₀, complement vector r ₀mould be β, β=|| r ₀|| ₂, set up a row orthogonal matrix Q, during initialization, the 0th row of row orthogonal matrix Q and first row are respectively q ₀=0 He set up a symmetric triple-diagonal matrix T, during initialization, first main diagonal element of symmetric triple-diagonal matrix T is α ₁, wherein represent q ₁conjugate transpose, zero time diagonal element is θ ₀=0, the row orthogonal matrix Q for the first time in iteration ₁=[q ₁], the symmetric triple-diagonal matrix T for the first time in iteration ₁=[α ₁], iteration variable p=1 is set;

(4-2-2) after p iteration, the complement vector r of above-mentioned model is obtained _p, r _p=G _tq _p-α _pq _p-θ _p-1q _p-1, and according to this complement vector, the symmetric triple-diagonal matrix T obtained when calculating the p time iteration _p+1p+1 capable on secondary diagonal element θ _p=|| r _p|| ₂, to θ _pjudge, if θ _p=0, then judge that optimum interruption occurs Lanczos process, stops iteration, and makes P=p, Q=Q _p, T=T _p, carry out step (5), if θ _p≠ 0, then carry out step (4-2-3)-step (4-2-6) successively;

(4-2-3) according to the model complement vector r after above-mentioned p iteration _p, the row orthogonal matrix Q obtained when calculating the p time iteration _p+1p+1 row: $q_{p+1}=\frac{r_p}{{\left|\right|r_p\left|\right|}_2};$

(4-2-4) according to above-mentioned row orthogonal matrix Q _p+1p+1 row q _p+1, calculate symmetric triple-diagonal matrix T _p+1main diagonal element on p+1 is capable: wherein represent q _p+1conjugate transpose;

(4-2-5) according to the result of calculation of step (4-2-2), step (4-2-3) and step (4-2-4), above-mentioned row orthogonal matrix is upgraded respectively and symmetric triple-diagonal matrix is: Q _p+1=[Q _p, q _p+1],

(4-2-6) upgrade iteration variable and make p=p+1, according to above-mentioned iteration threshold P, iteration variable p is judged, if p > is P, then stops iteration, and make Q=Q _p, T=T _p, carry out step (5), if p≤P, then carry out step (4-2-2);

(5) carry out LU decomposition to the symmetric triple-diagonal matrix T that above-mentioned steps (4) obtains, obtain lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T, detailed process is as follows:

(5-1) set up lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T, initialization is carried out to L and U, makes L=I _p, U=I _p, make the element of the first row first row of U be U _1,1=T _1,1, the element of the first row secondary series of U is U _1,2=T _1,2, iteration threshold M is set, makes M=P, iteration from iteration variable is m=2;

(5-2) according to the symmetric triple-diagonal matrix T that step (4) obtains, and the lower dual-diagonal matrix L that obtains of the m-1 time iteration and upper pair angle matrix U, the element of the capable m-1 row of m of the lower dual-diagonal matrix L that the m time iteration obtains is calculated the element U of the capable m row of m of upper two angles matrix U _m,m=T _m,m-L _{m, m-1}u _{m-1, m}with the element U of the capable m+1 row of the m of upper pair angle matrix U _{m, m+1}=T _{m, m+1}-L _{m, m-1}u _{m-1, m};

(5-3) make iteration variable m=m+1, iteration variable m is judged, if m < is M, then carries out step (5-2)-step (5-3), if m >=M, then stop iteration, carry out step (5-4);

(5-4) according to the symmetric triple-diagonal matrix T that step (4) obtains, and the lower dual-diagonal matrix L that obtains of above-mentioned the M-1 time iteration and upper pair angle matrix U, calculate the element of the capable M-1 row of M of the lower dual-diagonal matrix L that the M time iteration obtains with the element U of the capable M row of the M of upper pair angle matrix U _m,M=T _m,M-L _{m, M-1}u _{m-1, M};

(6) according to dual-diagonal matrix L under above-mentioned steps (5) and upper pair angle matrix U, above-mentioned antenna transmission signal x is solved _testimated value coordinate vector z in the linear subspaces that each row of the row orthogonal matrix Q of step (4) are opened, detailed process is as follows:

(6-1) define an intermediate variable b, b=Uz, wherein U is the upper dual-diagonal matrix in above-mentioned steps (5), and z is above-mentioned antenna transmission signal x _testimated value coordinate vector in the linear subspaces that each row of the row orthogonal matrix Q of step (4) are opened;

(6-2) dual-diagonal matrix L under above-mentioned steps (5) is utilized, by Lb=β e ₁solve intermediate variable b, β is the initial complement vector r in above-mentioned steps (4-2-1) ₀mould, e ₁first element be 1, other elements are 0, and solution procedure is: to first element of b, according to first diagonal element L of lower dual-diagonal matrix L _1,1, solve b ₁=β/L _1,1, to a jth element b of b _j, j=2 ..., P, according to jth-1 element b of b _j-1with the element L on the j row of lower dual-diagonal matrix L _{j, j-1}and L _j,j, solve successively

(6-3) according to P the element b of above-mentioned b _pwith P the diagonal element U of above-mentioned upper dual-diagonal matrix U _p,P, P the element solving z is z _p=b _p/ U _p,P, to i-th element z of z _i, i=P-1 ..., 1, according to i-th element b of above-mentioned b _i, the i-th+1 element z of z _i+1with the element diagonal element U on i-th row of upper dual-diagonal matrix U _i,iand U _{i, i+1}solve i-th element of z successively wherein P is the iteration threshold in above-mentioned steps (4);

(7) each row of above-mentioned row orthogonal matrix Q are carried out linear combination according to each element of z, obtain above-mentioned antenna transmission signal x _testimated value ${\hat{x}}_t=\mathrm{Qz};$

(8) according to the definition of log-likelihood ratio, respectively calculate with in the corresponding log-likelihood ratio of each bit of each subscriber signal, detailed process is as follows:

(8-1) utilize lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T in step (5), obtain the inverse T of symmetric triple-diagonal matrix T ^-1: T ^-1=[a ₁a _p], wherein, to T ^-1c row a _c, 1≤c≤P, by solving LUa _c=e _cobtain, and except c element be except 1, all the other elements are all 0, and detailed process is as follows:

(8-1-1) an intermediate variable g is defined, g=Ua _c, wherein U is the upper dual-diagonal matrix in above-mentioned steps (5), a _cfor the inverse T of symmetric triple-diagonal matrix T ^-1in c row;

(8-1-2) utilize the lower dual-diagonal matrix L of above-mentioned steps (5), pass through Lg=e _csolve intermediate variable g, wherein e _cvalue except c element be except 1, all the other elements are all 0, and solution procedure is: to first element of g, according to first diagonal element L of lower dual-diagonal matrix L _1,1and e _cfirst element e _{c, 1}, first element of g is: g ₁=e _{c, 1}/ L _1,1, to l the element g of g _l, l=2 ..., P, according to l-1 the element g of g _l-1, e _cl element e _c,lwith the element L on the l row of lower dual-diagonal matrix L _{l, l-1}and L _l,l, solve successively

(8-1-3) according to P the element g of above-mentioned g _pwith P the diagonal element U of above-mentioned upper dual-diagonal matrix U _p,P, solve a _cp element be a _c,P=g _p/ U _p,P, to a _cthe n-th element a _c,n, n=P-1 ..., 1, according to the n-th element g of above-mentioned g _n, a _c(n+1)th element a _{c, n+1}with the element diagonal element U in the n-th line of upper dual-diagonal matrix U _n,nand U _{n, n+1}, solve a successively _cthe n-th element wherein P is the iteration threshold in above-mentioned steps (4);

(8-2) according to the up channel matrix H of the moment t of the row orthogonal matrix Q of above-mentioned steps (4), symmetrical three diagonal angle T and above-mentioned steps (2) _t, calculate each user in large-scale multi-antenna system and to transmit x _twith the estimated value that transmits between equivalent channel gain matrix B, and calculate the equivalent received noise of multiple user signals in large-scale multi-antenna system calculate covariance matrix Θ, $\mathrm{\&Theta;}=Q{T}^{-1}{Q}^H{H}_t^H{H}_{t}Q{\left(T^{-1}\right)}^H{Q}^H,$ Wherein T ^-1the inverse of T;

(8-3) according to the covariance matrix Θ of above-mentioned equivalent channel gain matrix B and equivalent noise, the equivalent channel gain μ of a kth user in large-scale multi-antenna system is obtained _k, μ _kfor a kth diagonal element B of above-mentioned equivalent channel gain matrix B _k,k, calculate transmitting of a kth user and be subject to the interference strength In that other users transmit _k, In _k=Σ _{u, u ≠ k}| B _k,u| ²e _s, calculate the intensity N of when being received by the base station noise that base station receives of transmitting of a kth user _k, N _k=Θ _k,kn ₀, and obtain above-mentioned interference strength and noise intensity sum wherein 1≤k≤K, u representative removes kth and uses other outdoor users;

(8-4) according to the definition of log-likelihood ratio, the log-likelihood ratio L of each bit in each subscriber signal is calculated _k,b, wherein, L _k,brepresent the antenna transmission signal x of a kth user _t,kin the log-likelihood ratio of b bit, with represent that b bit is the set of all available symbols of 0 and 1 respectively, a and a' represents respectively with in symbol;

(9) deinterleaving and channel decoding are carried out to the log-likelihood ratio obtained in step (8), obtain the testing result of the bit sequence that each user of t launches realize the detection to multiple user signals in large-scale multi-antenna system.

The detection method of multiple user signals in the large-scale multi-antenna system that the present invention proposes, its advantage is, complexity is low (to be about wherein I is iterations); Performance of BER is good, just can approach the performance of BER that optimum MMSE detects within a few step iteration.This method solves system of linear equations by Lanczos iterative algorithm, avoids and obtains the balanced matrix of MMSE by direct mode of inverting to large-scale matrix, can greatly reduce the computation complexity of detection; In addition, this detection method can calculate the input of LLR value as decoder of each information bit in lower complexity, significantly can promote the performance of joint-detection decoding.In ratio one timing of number of users K (K<N) and antenna for base station number N and the two number is all tending towards infinite time, the reduction of this complexity is particularly evident, and performance of BER progressively can reach (optimum) performance of Maximum Likelihood Detection simultaneously.

Accompanying drawing explanation

Fig. 1 is the Massive MIMO multi-user uplink scene schematic diagram that the inventive method relates to.

Fig. 2 is the FB(flow block) of the inventive method.

Embodiment

The detection method of multiple user signals in the large-scale multi-antenna system that the present invention proposes, as shown in Figure 2, this detection method comprises the following steps its FB(flow block):

(1) set antenna for base station number in large-scale multi-antenna system as N, number of users is K, and each user configures an antenna, then the up channel matrix of large-scale multi-antenna system is wherein represent complex field, K the user of user in T time slot transmit into wherein R is the code check of subscriber signal coding, transmits S after chnnel coding, intertexture and sign map, obtains antenna transmission signal, be designated as the power of antenna transmission signal is in one embodiment of the present of invention, code check can be Turbo code, and coded system can be convolution code or low density parity check code, and sign map mode can be quadrature amplitude modulation; Antenna transmission signal X is by the antenna transmission of user, and through channel, obtain base station received signal at the N root antenna of base station, be designated as Y=HX+W, wherein, W represents the additive white Gaussian noise in base station received signal, element in matrix W is the signal to noise ratio of base station received signal is the Massive MIMO multi-user uplink scene schematic diagram that the inventive method relates to as shown in Figure 1.

(2) the matched filtering vector of above-mentioned base station received signal is calculated wherein for the up channel matrix H of moment t in T time slot _tconjugate transpose, y _tfor the t row in above-mentioned base station received signal Y, represent the signal vector that base station receives at moment t;

(3) the up channel matrix of above-mentioned large-scale multi-antenna system is calculated gram matrix G _t, wherein ρ is the signal to noise ratio of base station received signal, i _kfor K rank unit matrix;

(4) use Lanczos process, calculate and above-mentioned G _trelevant row orthogonal matrix Q and symmetric triple-diagonal matrix T, detailed process is as follows:

(4-1) calculate at the above-mentioned base station received signal y of moment t _tthe balanced Matrix C of least mean-square error _t, by the balanced Matrix C of this least mean-square error _tto y _tcarry out filtering, obtain the antenna transmission signal x of moment t _testimated value ${\hat{x}}_t=C_t{y}_t;$

(4-2) the equivalent linear equation group model detected in a large-scale multi-antenna system is set up adopt Lanczos process, solve equivalent linear equation group model, obtain the antenna transmission signal x of moment t _testimated value concrete steps are as follows;

(4-2-1) initialization: establish iterations threshold value P, if estimated value initial value be then above-mentioned model initial complement vector be r ₀, complement vector r ₀mould be β, β=|| r ₀|| ₂, set up a row orthogonal matrix Q, during initialization, the 0th row of row orthogonal matrix Q and first row are respectively q ₀=0 He set up a symmetric triple-diagonal matrix T, during initialization, first main diagonal element of symmetric triple-diagonal matrix T is α ₁, wherein represent q ₁conjugate transpose, zero time diagonal element is θ ₀=0, the row orthogonal matrix Q for the first time in iteration ₁=[q ₁], the symmetric triple-diagonal matrix T for the first time in iteration ₁=[α ₁], iteration variable p=1 is set;

(4-2-2) after p iteration, the complement vector r of above-mentioned model is obtained _p, r _p=G _tq _p-α _pq _p-θ _p-1q _p-1, and according to this complement vector, the symmetric triple-diagonal matrix T obtained when calculating the p time iteration _p+1p+1 capable on secondary diagonal element θ _p=|| r _p|| ₂, to θ _pjudge, if θ _p=0, then judge that optimum interruption occurs Lanczos process, stops iteration, and makes P=p, Q=Q _p, T=T _p, carry out step (5), if θ _p≠ 0, then carry out step (4-2-3)-step (4-2-6) successively;

(4-2-3) according to the model complement vector r after above-mentioned p iteration _p, the row orthogonal matrix Q obtained when calculating the p time iteration _p+1p+1 row: $q_{p+1}=\frac{r_p}{{\left|\right|r_p\left|\right|}_2};$

(4-2-4) according to above-mentioned row orthogonal matrix Q _p+1p+1 row q _p+1, calculate symmetric triple-diagonal matrix T _p+1main diagonal element on p+1 is capable: wherein represent q _p+1conjugate transpose;

(4-2-5) according to the result of calculation of step (4-2-2), step (4-2-3) and step (4-2-4), above-mentioned row orthogonal matrix is upgraded respectively and symmetric triple-diagonal matrix is:

(4-2-6) upgrade iteration variable and make p=p+1, according to above-mentioned iteration threshold P, iteration variable p is judged, if p > is P, then stops iteration, and make Q=Q _p, T=T _p, carry out step (5), if p≤P, then carry out step (4-2-2);

(5) the symmetric triple-diagonal matrix T obtained above-mentioned steps (4) carries out LU decomposition (resolving into the product of an a lower dual-diagonal matrix L and upper dual-diagonal matrix U by symmetric triple-diagonal matrix T), obtain lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T, detailed process is as follows:

(5-1) set up lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T, initialization is carried out to L and U, makes L=I _p, U=I _p, make the element of the first row first row of U be U _1,1=T _1,1, the element of the first row secondary series of U is U _1,2=T _1,2, iteration threshold M is set, makes M=P, iteration from iteration variable is m=2;

(5-2) according to the symmetric triple-diagonal matrix T that step (4) obtains, and the lower dual-diagonal matrix L that obtains of the m-1 time iteration and upper pair angle matrix U, the element of the capable m-1 row of m of the lower dual-diagonal matrix L that the m time iteration obtains is calculated the element U of the capable m row of m of upper two angles matrix U _m,m=T _m,m-L _{m, m-1}u _{m-1, m}with the element U of the capable m+1 row of the m of upper pair angle matrix U _{m, m+1}=T _{m, m+1}-L _{m, m-1}u _{m-1, m};

(5-3) make iteration variable m=m+1, iteration variable m is judged, if m < is M, then carries out step (5-2)-step (5-3), if m >=M, then stop iteration, carry out step (5-4);

(5-4) according to the symmetric triple-diagonal matrix T that step (4) obtains, and the lower dual-diagonal matrix L that obtains of above-mentioned the M-1 time iteration and upper pair angle matrix U, calculate the element of the capable M-1 row of M of the lower dual-diagonal matrix L that the M time iteration obtains with the element U of the capable M row of the M of upper pair angle matrix U _m,M=T _m,M-L _{m, M-1}u _{m-1, M};

(6) according to dual-diagonal matrix L under above-mentioned steps (5) and upper pair angle matrix U, above-mentioned antenna transmission signal x is solved _testimated value coordinate vector z in the linear subspaces that each row of the row orthogonal matrix Q of step (4) are opened, detailed process is as follows:

(6-1) define an intermediate variable b, b=Uz, wherein U is the upper dual-diagonal matrix in above-mentioned steps (5), and z is above-mentioned antenna transmission signal x _testimated value coordinate vector in the linear subspaces that each row of the row orthogonal matrix Q of step (4) are opened;

(6-2) dual-diagonal matrix L under above-mentioned steps (5) is utilized, by Lb=β e ₁solve intermediate variable b, β is the initial complement vector r in above-mentioned steps (4-2-1) ₀mould, e ₁first element be 1, other elements are 0, and solution procedure is: to first element of b, according to first diagonal element L of lower dual-diagonal matrix L _1,1, solve b ₁=β/L _1,1, to a jth element b of b _j, j=2 ..., P, according to jth-1 element b of b _j-1with the element L on the j row of lower dual-diagonal matrix L _{j, j-1}and L _j,j, solve successively

(6-3) according to P the element b of above-mentioned b _pwith P the diagonal element U of above-mentioned upper dual-diagonal matrix U _p,P, P the element solving z is z _p=b _p/ U _p,P, to i-th element z of z _i, i=P-1 ..., 1, according to i-th element b of above-mentioned b _i, the i-th+1 element z of z _i+1with the element diagonal element U on i-th row of upper dual-diagonal matrix U _i,iand U _{i, i+1}solve i-th element of z successively wherein P is the iteration threshold in above-mentioned steps (4);

(7) each row of above-mentioned row orthogonal matrix Q are carried out linear combination according to each element of z, obtain above-mentioned antenna transmission signal x _testimated value ${\hat{x}}_t=\mathrm{Qz};$

(8) according to the definition of log-likelihood ratio, respectively calculate with in the corresponding log-likelihood ratio of each bit of each subscriber signal, detailed process is as follows:

(8-1) utilize lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T in step (5), obtain the inverse T of symmetric triple-diagonal matrix T ^-1: T ^-1=[a ₁a _p], wherein, to T ^-1c row a _c, 1≤c≤P, by solving LUa _c=e _cobtain, and except c element be except 1, all the other elements are all 0, and detailed process is as follows:

(8-1-1) an intermediate variable g is defined, g=Ua _c, wherein U is the upper dual-diagonal matrix in above-mentioned steps (5), a _cfor the inverse T of symmetric triple-diagonal matrix T ^-1in c row;

(8-1-2) utilize the lower dual-diagonal matrix L of above-mentioned steps (5), pass through Lg=e _csolve intermediate variable g, wherein e _cvalue except c element be except 1, all the other elements are all 0, and solution procedure is: to first element of g, according to first diagonal element L of lower dual-diagonal matrix L _1,1and e _cfirst element e _{c, 1}, first element of g is: g ₁=e _{c, 1}/ L _1,1, to l the element g of g _l, l=2 ..., P, according to l-1 the element g of g _l-1, e _cl element e _c,lwith the element L on the l row of lower dual-diagonal matrix L _{l, l-1}and L _l,l, solve successively

(8-1-3) according to P the element g of above-mentioned g _pwith P the diagonal element U of above-mentioned upper dual-diagonal matrix U _p,P, solve a _cp element be a _c,P=g _p/ U _p,P, to a _cthe n-th element a _c,n, n=P-1 ..., 1, according to the n-th element g of above-mentioned g _n, a _c(n+1)th element a _{c, n+1}with the element diagonal element U in the n-th line of upper dual-diagonal matrix U _n,nand U _{n, n+1}, solve a successively _cthe n-th element wherein P is the iteration threshold in above-mentioned steps (4);

(8-2) according to the up channel matrix H of the moment t of the row orthogonal matrix Q of above-mentioned steps (4), symmetrical three diagonal angle T and above-mentioned steps (2) _t, calculate each user in large-scale multi-antenna system and to transmit x _twith the estimated value that transmits between equivalent channel gain matrix B, and calculate the equivalent received noise of multiple user signals in large-scale multi-antenna system calculate covariance matrix Θ, $\mathrm{\&Theta;}=Q{T}^{-1}{Q}^H{H}_t^H{H}_{t}Q{\left(T^{-1}\right)}^H{Q}^H,$ Wherein T ^-1the inverse of T;

(8-3) according to the covariance matrix Θ of above-mentioned equivalent channel gain matrix B and equivalent noise, the equivalent channel gain μ of a kth user in large-scale multi-antenna system is obtained _k, μ _kfor a kth diagonal element B of above-mentioned equivalent channel gain matrix B _k,k, calculate transmitting of a kth user and be subject to the interference strength In that other users transmit _k, In _k=Σ _{u, u ≠ k}| B _k,u| ²e _s, calculate the intensity N of when being received by the base station noise that base station receives of transmitting of a kth user _k, N _k=Θ _k,kn ₀, and obtain above-mentioned interference strength and noise intensity sum wherein 1≤k≤K, u representative removes kth and uses other outdoor users;

(8-4) according to the definition of log-likelihood ratio, the log-likelihood ratio L of each bit in each subscriber signal is calculated _k,b, wherein, L _k,brepresent the antenna transmission signal x of a kth user _t,kin the log-likelihood ratio of b bit, with represent that b bit is the set of all available symbols of 0 and 1 respectively, a and a' represents respectively with in symbol;

(9) deinterleaving and channel decoding are carried out to the log-likelihood ratio obtained in step (8), obtain the testing result of the bit sequence that each user of t launches realize the detection to multiple user signals in large-scale multi-antenna system.

Claims (1)

1. the detection method of multiple user signals in large-scale multi-antenna system, is characterized in that this detection method comprises the following steps:

(1) set antenna for base station number in large-scale multi-antenna system as N, number of users is K, and each user configures an antenna, then the up channel matrix of large-scale multi-antenna system is wherein represent complex field, K the user of user in T time slot transmit into wherein R is the code check of subscriber signal coding, transmits S after chnnel coding, intertexture and sign map, obtains antenna transmission signal, be designated as the power of antenna transmission signal is antenna transmission signal X is by the antenna transmission of user, and through channel, obtain base station received signal at the N root antenna of base station, be designated as Y=HX+W, wherein, W represents the additive white Gaussian noise in base station received signal, element in matrix W is the signal to noise ratio of base station received signal is

(2) the matched filtering vector of above-mentioned base station received signal is calculated wherein for the up channel matrix H of moment t in T time slot _tconjugate transpose, y _tfor the t row in above-mentioned base station received signal Y, represent the signal vector that base station receives at moment t;

(3) the up channel matrix of above-mentioned large-scale multi-antenna system is calculated gram matrix G _t, wherein ρ is the signal to noise ratio of base station received signal, i _kfor K rank unit matrix;

(4) use Lanczos process, calculate and above-mentioned G _trelevant row orthogonal matrix Q and symmetric triple-diagonal matrix T, detailed process is as follows:

(4-1) calculate at the above-mentioned base station received signal y of moment t _tthe balanced Matrix C of least mean-square error _t, by the balanced Matrix C of this least mean-square error _tto y _tcarry out filtering, obtain the antenna transmission signal x of moment t _testimated value

(4-2) the equivalent linear equation group model detected in a large-scale multi-antenna system is set up adopt Lanczos process, solve equivalent linear equation group model, obtain the antenna transmission signal x of moment t _testimated value concrete steps are as follows;

(4-2-1) initialization: establish iterations threshold value P, if estimated value initial value be then above-mentioned model initial complement vector be r ₀, complement vector r ₀mould be β, β=|| r ₀|| ₂, set up a row orthogonal matrix Q, during initialization, the 0th row of row orthogonal matrix Q and first row are respectively q ₀=0 He set up a symmetric triple-diagonal matrix T, during initialization, first main diagonal element of symmetric triple-diagonal matrix T is α ₁, wherein represent q ₁conjugate transpose, zero time diagonal element is θ ₀=0, the row orthogonal matrix Q for the first time in iteration ₁=[q ₁], the symmetric triple-diagonal matrix T for the first time in iteration ₁=[α ₁], iteration variable p=1 is set;

(4-2-2) after p iteration, the complement vector r of above-mentioned model is obtained _p, r _p=G _tq _p-α _pq _p-θ _p-1q _p-1, and according to this complement vector, the symmetric triple-diagonal matrix T obtained when calculating the p time iteration _p+1p+1 capable on secondary diagonal element θ _p=|| r _p|| ₂, to θ _pjudge, if θ _p=0, then judge that optimum interruption occurs Lanczos process, stops iteration, and makes P=p, Q=Q _p, T=T _p, carry out step (5), if θ _p≠ 0, then carry out step (4-2-3)-step (4-2-6) successively;

(4-2-3) according to the model complement vector r after above-mentioned p iteration _p, the row orthogonal matrix Q obtained when calculating the p time iteration _p+1p+1 row:

(4-2-4) according to above-mentioned row orthogonal matrix Q _p+1p+1 row q _p+1, calculate symmetric triple-diagonal matrix T _p+1main diagonal element on p+1 is capable: wherein represent q _p+1conjugate transpose;

(4-2-5) according to the result of calculation of step (4-2-2), step (4-2-3) and step (4-2-4), above-mentioned row orthogonal matrix is upgraded respectively and symmetric triple-diagonal matrix is: Q _p+1=[Q _p, q _p+1], $T_{p+1}=\left[\begin{array}{cccc}& & & 0\\ & & & .\\ & T_p& & .\\ & & & .\\ & & & {\mathrm{\&theta;}}_p\\ 0& ...& {\mathrm{\&theta;}}_p& {\mathrm{\&alpha;}}_{p+1}\end{array}\right];$

(4-2-6) upgrade iteration variable and make p=p+1, according to above-mentioned iteration threshold P, iteration variable p is judged, if p > is P, then stops iteration, and make Q=Q _p, T=T _p, carry out step (5), if p≤P, then carry out step (4-2-2);

(5) carry out LU decomposition to the symmetric triple-diagonal matrix T that above-mentioned steps (4) obtains, obtain lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T, detailed process is as follows:

(5-1) set up lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T, initialization is carried out to L and U, makes L=I _p, U=I _p, make the element of the first row first row of U be U _1,1=T _1,1, the element of the first row secondary series of U is U _1,2=T _1,2, iteration threshold M is set, makes M=P, iteration from iteration variable is m=2;

(5-2) according to the symmetric triple-diagonal matrix T that step (4) obtains, and the lower dual-diagonal matrix L that obtains of the m-1 time iteration and upper pair angle matrix U, the element of the capable m-1 row of m of the lower dual-diagonal matrix L that the m time iteration obtains is calculated the element U of the capable m row of m of upper two angles matrix U _m,m=T _m,m-L _{m, m-1}u _{m-1, m}with the element U of the capable m+1 row of the m of upper pair angle matrix U _{m, m+1}=T _{m, m+1}-L _{m, m-1}u _{m-1, m};

(5-3) make iteration variable m=m+1, iteration variable m is judged, if m < is M, then carries out step (5-2)-step (5-3), if m >=M, then stop iteration, carry out step (5-4);

(5-4) according to the symmetric triple-diagonal matrix T that step (4) obtains, and the lower dual-diagonal matrix L that obtains of above-mentioned the M-1 time iteration and upper pair angle matrix U, calculate the element of the capable M-1 row of M of the lower dual-diagonal matrix L that the M time iteration obtains with the element U of the capable M row of the M of upper pair angle matrix U _m,M=T _m,M-L _{m, M-1}u _{m-1, M};

(6) according to dual-diagonal matrix L under above-mentioned steps (5) and upper pair angle matrix U, above-mentioned antenna transmission signal x is solved _testimated value coordinate vector z in the linear subspaces that each row of the row orthogonal matrix Q of step (4) are opened, detailed process is as follows:

(6-1) define an intermediate variable b, b=Uz, wherein U is the upper dual-diagonal matrix in above-mentioned steps (5), and z is above-mentioned antenna transmission signal x _testimated value coordinate vector in the linear subspaces that each row of the row orthogonal matrix Q of step (4) are opened;

(6-2) dual-diagonal matrix L under above-mentioned steps (5) is utilized, by Lb=β e ₁solve intermediate variable b, β is the initial complement vector r in above-mentioned steps (4-2-1) ₀mould, e ₁first element be 1, other elements are 0, and solution procedure is: to first element of b, according to first diagonal element L of lower dual-diagonal matrix L _1,1, solve b ₁=β/L _1,1, to a jth element b of b _j, j=2 ..., P, according to jth-1 element b of b _j-1with the element L on the j row of lower dual-diagonal matrix L _{j, j-1}and L _j,j, solve successively

(6-3) according to P the element b of above-mentioned b _pwith P the diagonal element U of above-mentioned upper dual-diagonal matrix U _p,P, P the element solving z is z _p=b _p/ U _p,P, to i-th element z of z _i, i=P-1 ..., 1, according to i-th element b of above-mentioned b _i, the i-th+1 element z of z _i+1with the element diagonal element U on i-th row of upper dual-diagonal matrix U _i,iand U _{i, i+1}solve i-th element of z successively wherein P is the iteration threshold in above-mentioned steps (4);

(7) each row of above-mentioned row orthogonal matrix Q are carried out linear combination according to each element of z, obtain above-mentioned antenna transmission signal x _testimated value

(8) according to the definition of log-likelihood ratio, respectively calculate with in the corresponding log-likelihood ratio of each bit of each subscriber signal, detailed process is as follows:

(8-1) utilize lower dual-diagonal matrix L and upper pair angle matrix U of symmetric triple-diagonal matrix T in step (5), obtain the inverse T of symmetric triple-diagonal matrix T ^-1: T ^-1=[a ₁a _p], wherein, to T ^-1c row a _c, 1≤c≤P, by solving LUa _c=e _cobtain, and except c element be except 1, all the other elements are all 0, and detailed process is as follows:

(8-1-1) an intermediate variable g is defined, g=Ua _c, wherein U is the upper dual-diagonal matrix in above-mentioned steps (5), a _cfor the inverse T of symmetric triple-diagonal matrix T ^-1in c row;

(8-1-2) utilize the lower dual-diagonal matrix L of above-mentioned steps (5), pass through Lg=e _csolve intermediate variable g, wherein e _cvalue except c element be except 1, all the other elements are all 0, and solution procedure is: to first element of g, according to first diagonal element L of lower dual-diagonal matrix L _1,1and e _cfirst element e _{c, 1}, first element of g is: g ₁=e _{c, 1}/ L _1,1, to l the element g of g _l, l=2 ..., P, according to l-1 the element g of g _l-1, e _cl element e _c,lwith the element L on the l row of lower dual-diagonal matrix L _{l, l-1}and L _l,l, solve successively

(8-1-3) according to P the element g of above-mentioned g _pwith P the diagonal element U of above-mentioned upper dual-diagonal matrix U _p,P, solve a _cp element be a _c,P=g _p/ U _p,P, to a _cthe n-th element a _c,n, n=P-1 ..., 1, according to the n-th element g of above-mentioned g _n, a _c(n+1)th element a _{c, n+1}with the element diagonal element U in the n-th line of upper dual-diagonal matrix U _n,nand U _{n, n+1}, solve a successively _cthe n-th element wherein P is the iteration threshold in above-mentioned steps (4);

(8-2) according to the up channel matrix H of the moment t of the row orthogonal matrix Q of above-mentioned steps (4), symmetrical three diagonal angle T and above-mentioned steps (2) _t, calculate each user in large-scale multi-antenna system and to transmit x _twith the estimated value that transmits between equivalent channel gain matrix B, and calculate the equivalent received noise of multiple user signals in large-scale multi-antenna system calculate covariance matrix Θ, $\mathrm{\&Theta;}={\mathrm{QT}}^{-1}{Q}^H{H}_t^H{H}_{t}Q{\left(T^{-1}\right)}^H{Q}^H,$ Wherein T ^-1the inverse of T;

(8-3) according to the covariance matrix Θ of above-mentioned equivalent channel gain matrix B and equivalent noise, the equivalent channel gain μ of a kth user in large-scale multi-antenna system is obtained _k, μ _kfor a kth diagonal element B of above-mentioned equivalent channel gain matrix B _k,k, calculate transmitting of a kth user and be subject to the interference strength In that other users transmit _k, In _k=Σ _{u, u ≠ k}| B _k,u| ²e _s, calculate the intensity N of when being received by the base station noise that base station receives of transmitting of a kth user _k, N _k=Θ _k,kn ₀, and obtain above-mentioned interference strength and noise intensity sum wherein 1≤k≤K, u representative removes kth and uses other outdoor users;

(8-4) according to the definition of log-likelihood ratio, the log-likelihood ratio L of each bit in each subscriber signal is calculated _k,b, wherein, L _k,brepresent the antenna transmission signal x of a kth user _t,kin the log-likelihood ratio of b bit, with represent that b bit is the set of all available symbols of 0 and 1 respectively, a and a' represents respectively with in symbol;

(9) deinterleaving and channel decoding are carried out to the log-likelihood ratio obtained in step (8), obtain the testing result of the bit sequence that each user of t launches realize the detection to multiple user signals in large-scale multi-antenna system.