HK1201647A1 - Joint detection method and device - Google Patents

Abstract

The present invention discloses a joint detection method and device, wherein the method comprises: constructing a system equation x=Bd+n according to a received signal in need of joint detection; solving the system equation using an iterative algorithm having convergence characteristics, and acquiring a sent signal according to a solution result. By means of the present invention, to use the technical solution of solving the system equation with the iterative linear equation does not need to directly inverse the matrix, can realize parallel processing of the data, and reduces the complexity of the joint detection.

Description

Joint detection method and device

Technical Field

The invention relates to the field of communication, in particular to a joint detection method and device.

Background

The Time Division-Synchronous Code Division multiple Access (TD-SCDMA) system is one of the standards for the third generation mobile communication system. In the TD-SCDMA system, signals of multiple users are aliased in the time domain and the frequency domain, and when receiving, the signals of each user need to be separated in the digital domain by using a certain signal separation method. The method of signal separation generally includes single user detection and multi-user detection, joint detection (joint detection, abbreviated as JD) is one of the key technologies in TD-SCDMA system, and belongs to one of multi-user detection, which can eliminate the influence of multi-access Interference (MAI) and Inter-Symbol Interference (ISI) caused by multipath on the system performance, improve the anti-Interference capability of the system, and increase the system capacity.

In a TD-SCDMA system, a typical propagation process is an uplink propagation process, and a downlink propagation can be regarded as a special case of the uplink propagation. Fig. 1 is a schematic diagram of an uplink signal transmission model of K users according to the related art, and transmission of uplink signals is described below with reference to fig. 1.

In FIG. 1, two receiving antennas are shown, where c^(k)Is the user spreading code, h^(k,m)Is the channel from the Kth user to the m-th antenna, n^(m)For the noise of the m-th antenna channel, d^(k)For the data transmitted for the k-th user,and (3) for the estimated data signal of the kth user, setting the user number as K, and transmitting N data signals by each user. The vector of data symbols transmitted by the kth user can be represented as:

where the superscript T denotes the transpose operation.

The information symbols pass through a spreading sequence of length Q:

after spreading, the data is scrambled, modulated and transmitted via an antenna.

The mth antenna channel response of the kth user between the receiver antenna and the transmitter antenna can be expressed as:

wherein W represents the maximum window length of the channel response, and W ≠ 1 means intersymbol interference; m denotes the number of antennas of the base station.

The excitation response on the mth antenna of the kth user is then:

the antenna of the base station thus receives a signal of

The received signal can be expressed as:

x=Bd+n （6）

joint detection estimates the user's transmitted signal d from x and B in equation 6, where n represents noise and x represents the received signal. From this equation, it is necessary to know the system matrix B, which includes the spreading code and the radio channel response for each user. There are several methods for obtaining B in the related art, for example, in TD-SCDMA system, the base station knows these spreading codes, and then can obtain the channel response of each user by transmitting a midamble and using a B.

D can be calculated under the condition of B determination and noise n determination, fig. 2 is a schematic position diagram of a joint detection algorithm in a TD-SCDMA system receiver according to the related art, as shown in fig. 2, after receiving signals, channel estimation is carried out, then a system matrix is constructed, and then the system matrix is solved through the joint detection algorithm.

In the related art, a commonly used joint detection algorithm includes: time domain Cholesky Decomposition (CD for short), time domain approximate Cholesky Decomposition, frequency domain equalization solution, and the like. Among them, these solving methods involve the inversion operation of the matrix, and the operation complexity is relatively high.

For example, the time-domain Cholesky decomposition inversion computation complexity is O ((22K)³). Where K represents the number of users, and the method does not benefit much from a powerful vector engine (32 bits, Media Access Control (MAC) widely used in hardware or Digital signal processors (DSP for short), which is a large number of available Media Access controls). Fig. 3 is a schematic diagram of Cholesky decomposition according to the related art, as shown in fig. 3, the CD method can only process a part of signals each time, and needs to perform a switching operation before the next multiplication, so that the time required for calculation is long and complicated, and the calculation formula of the CD method is as follows:

where L represents the lower triangular matrix of the Cholesky decomposition of the a matrix, a is the matrix to be decomposed, and k is the subscript (kth row, or kth column) of the matrix.

For another example, the time-domain approximation Cholesky decomposition, which can reduce the complexity of matrix inversion, can only guarantee performance in a matrix with a dominant strict diagonal structure, with an inversion complexity of-O ((2K)³)。

For example, in frequency domain equalization, the inversion of a large matrix is reduced to the inversion of several small matrices by Fast Fourier Transform (FFT), which has a complexity of-O (24 × K)³/2), however, Cholesky decomposition cannot be performed sufficiently in parallel, and all the operation units cannot be fully used, so that extra Inverse Fast Fourier Transform (IFFT) operation complexity of O ((K +1) is brought about²*12log24)。

Therefore, the above methods all require matrix inversion, which results in increased complexity of joint detection, and no effective solution has been proposed to solve the above problems.

Disclosure of Invention

The invention provides a joint detection method and a joint detection device, which at least solve the problem that the joint detection method in the related technology needs to directly invert a matrix, and the computation amount is large, so that the complexity of joint detection is increased.

According to an aspect of the present invention, there is provided a joint detection method, including: constructing a system equation x = Bd + n according to the received signals needing joint detection, wherein x is a received signal matrix, B is a system matrix, d is a transmitted signal matrix, and n is a noise matrix; and solving the system equation by using an iterative method with convergence characteristics, and acquiring a sending signal according to a solving result.

Preferably, solving the system equation using an iterative method having a convergence property includes: solving the system equation by using a Jacobi iteration method; alternatively, the system equations are solved using a Gauss Seidel iterative method with successive super-relaxed SORs.

Preferably, the SOR factor using the Gauss Seidel iteration with SOR is greater than 0 and less than 2.

Preferably, B is a block-shaped toeplitz matrix.

According to another aspect of the present invention, there is also provided a joint detection apparatus, including: the system comprises a construction module, a detection module and a processing module, wherein the construction module is set to construct a system equation x = Bd + n according to received signals needing joint detection, wherein x is a received signal matrix, B is a system matrix, d is a transmitted signal matrix, and n is a noise matrix; and the processing module is used for solving the system equation by using an iterative method with convergence characteristics and acquiring a sending signal according to a solving result.

Preferably, the processing module is configured to solve the system equation by using Jacobi iteration; alternatively, it is arranged to solve the system equations using a Gauss Seidel iterative method with successive super-relaxed SORs.

Preferably, the SOR factor of the gaussian-Seidel iteration of SOR is greater than 0 and less than 2.

Preferably, B is a block-shaped toeplitz matrix.

By adopting the method for solving the system equation by the iterative linear equation set, the parallel processing of data can be realized without directly inverting the matrix, thereby simplifying the complexity of solution and reducing the complexity of realizing joint detection.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:

fig. 1 is a schematic diagram of an uplink signal transmission model of K users according to the related art;

FIG. 2 is a schematic diagram of the location of a joint detection algorithm in a TD-SCDMA system receiver according to the related art;

FIG. 3 is a schematic diagram of Cholesky decomposition according to the related art;

FIG. 4 is a flow chart of a joint detection method according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a block-shaped Toeplitz matrix according to an embodiment of the present invention;

FIG. 6 is a block diagram of a joint detection apparatus according to an embodiment of the present invention;

FIG. 7 is a schematic diagram of a joint detection method according to a preferred embodiment of the present invention;

FIG. 8 is a graphical representation of a comparison of the performance of several joint detection methods according to the preferred embodiment of the present invention.

Detailed Description

It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.

It should be noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system such as a set of computer-executable instructions and that, although a logical order is illustrated in the flowcharts, in some cases, the steps illustrated or described may be performed in an order different than presented herein.

The methods and apparatuses described in the following embodiments may be applied to a receiver, and the methods and apparatuses may be implemented by one or more processors, such as CPUs. The embodiment provides a joint detection method, which can be applied to a TD-SCDMA system, and for systems of other systems, if the same problem exists, the method provided in the embodiment is also applicable. Fig. 4 is a flowchart of a joint detection method according to an embodiment of the present invention, and as shown in fig. 4, the flowchart includes the following steps:

step S402, constructing a system equation x = Bd + n according to the received signals that need to be jointly detected, where x is a received signal matrix, B is a system matrix, d is a transmitted signal matrix, and n is a noise matrix.

And S404, solving the system equation by using an iterative method with convergence characteristics, and acquiring a sending signal according to a solving result.

In the related art, the joint detection method needs to directly invert the matrix, and the computation amount is large, which results in increased complexity. Through the steps, the method for solving the system equation by adopting the iterative linear equation set does not need to directly invert the matrix, and can realize the parallel processing of data, thereby simplifying the complexity of solution and reducing the complexity of joint detection.

There are many kinds of iterative methods with convergence characteristics, and two kinds are preferred in the embodiment: firstly, solving a system equation by using a Jacobi iterative method; second, the system equation is solved by using a gaussian-Seidel (Gauss-Seidel) iterative method with successive super relaxation (SOR) and an SOR factor greater than 0 and smaller than 2 may be preferably used. Both methods do not require matrix inversion and both converge. The convergence rate of the Gauss-Seidel iterative method with the SOR is high, and although the convergence rate of solving the equation by adopting the Jacobi iterative method is lower than that of the Gauss-Seidel iterative method with the SOR, the problem of complexity increase caused by direct matrix inversion in the related technology can be solved. Other iterative methods with convergence characteristics can also solve the problems of the related art, but the performance is different, and the iterative methods can be comprehensively considered and determined by comparison in implementation.

Preferably, the system matrix B may be a block-shaped Toeplitz matrix, fig. 5 is a schematic structural diagram of the block-shaped Toeplitz matrix according to the embodiment of the present invention, as shown in fig. 5, the block-shaped Toeplitz matrix has a good structure, and the row of the first block and the row of the last block have at most 2K (assuming that each user has only one spreading code, K represents the number of users; if a user can allocate multiple codewords, in an actual system, K is the number of codewords of all active users in the time slot at that time) nonzero elements; the middle block has a maximum of 3K non-zero elements in the row (as shown in the middle block of fig. 3).

The approximate Cholesky decomposition can only be guaranteed for performance in a matrix where absolute diagonal dominance occurs, such as shown in the upper left box of fig. 5. For the matrix structure shown in the middle box of fig. 5, the row has at most 3K non-zero elements, and therefore, the SOR-based Gauss-Seidel iterative linear equation solver can be effectively applied. In order to perform the FDE method based on the block FFT operation for the matrix within the range of the dotted line in fig. 5, one row and one column are added to the original matrix to realize the block round, but it is generally difficult to perform the FFT operation for the number of singular points, and therefore, two rows and two columns are added to use the 24-pt FFT.

The embodiment also provides a joint detection device, which can be used for realizing the joint detection method. Fig. 6 is a block diagram of a joint detection apparatus according to an embodiment of the present invention, as shown in fig. 6, the apparatus includes: a construction module 62, and a processing module 64. These two modules are explained below.

A constructing module 62, configured to construct a system equation x = Bd + n according to the received signal that needs to be jointly detected, where x is a received signal matrix, B is a system matrix, d is a transmitted signal matrix, and n is a noise matrix;

a processing module 64, coupled to the construction module 62, is configured to solve the system equation by using an iterative method with a convergence property, and obtain the transmission signal according to the solution result.

Preferably, the processing module 64 is arranged to solve the system equations using Jacobi iteration; alternatively, it is arranged to solve the system equations using a Gauss Seidel iterative method with successive super-relaxed SORs.

It should be noted that: the modules and sub-units related in this embodiment may be implemented in a software manner, or may be implemented in a hardware manner. The modules and sub-units described therein may also be in a processor, such as a processor, including a configuration module 62, a processing module 64. Where the names of these modules, sub-units in some cases do not constitute a limitation of the module itself, for example, the construction module 62 may also be described as a "module configured to construct a system equation from received signals that require joint detection".

Fig. 5 is a schematic diagram of a joint detection method according to a preferred embodiment of the present invention, and as compared with fig. 1, it is obvious that the signals shown in fig. 5 are processed simultaneously, and as long as there is enough powerful hardware (e.g., a multiplier or an adder), all the arithmetic units can calculate simultaneously, all the data (or called signals) can be calculated at one time, and it is not necessary to process a part of the signals and then process another part of the signals, thereby saving the time for joint detection. It can be seen that, as long as the hardware is sufficiently powerful, the joint detection method shown in the above embodiments can be used to perform sufficiently parallel processing.

In order to make the technical solution and implementation method of the present invention clearer, the following describes the implementation process in detail with reference to the preferred embodiments.

In the preferred embodiment, a single antenna system is taken as an example for explanation.

Assuming that there are K users in each slot, spread with K different orthogonal codes, respectively, each user transmits N symbols in one data block, and the N symbols for the K-th user are represented as:

each symbol sequence is spread with a sequence of spreading factor Q, the spreading sequence being:

the spreading sequence is combined with an NxN identity matrix h^(k,m)Performing Kronecker multiplication to obtain a block diagonal spreading matrix corresponding to the k spreading code,

suppose the receiving end has M antennas, the k user (corresponding to the k spreading code), and the channel impulse response vector h on the M antenna^(k,m)If the length of the antenna is W, the data sequence with the length of NQ + W-1 of the K users arrives at the receiving end synchronously and is interfered by a static white gaussian noise sequence, and the white noise sequence corresponding to the mth antenna can be represented as:

thus, the reception amount at the mth antenna can be expressed as:

wherein the content of the first and second substances,

i.e. H^(k,m)∈C^(NQ+W-1)×NQIs formed by a channel impulse response vector h^(k,m)Composed Toeplitz matrix.

B^(k,m)=H^(k,m)C^(k)∈C^(NQ+W-1)×N （6）

B^(k,m)Is formed by a mixed response vector b^(k,m)A block-Toeplitz matrix formed by mixing response vectors b^(k,m)Can be expressed as a channel impulse response vector h^(k,m)With corresponding spreading code c^(k)Is as follows:

the received vector can be expressed as:

the total received vector can be expressed as:

x=Bd+n （9）

wherein the content of the first and second substances,

wherein V = [ b =⁽¹⁾,b⁽²⁾,…b^(k)]∈C^(NQ+W-1)M×K （10）

And, b^(k)=[b^(k,1)T,b^(k,2)T,…,b^(k,M)T]^TA column vector consisting of the mixed channel impulse response vectors on the M antennas.

Equation (9) is a system equation, x represents a received data vector, B represents a system matrix, d represents a data vector of K users in one time slot, and n represents additive noise. The purpose of joint detection is to estimate the original signal d sent by the user according to B and x in the above formula, where B is determined by the spreading codes and channel impulse responses of all users, so that the premise of the joint detection algorithm is to obtain the spreading codes and channel impulse responses of all users. In fact, the TD-SCDMA system sets a training sequence Midamble used for channel estimation in the frame structure, and can estimate the channel impulse response according to the received training sequence partial signal and the known training sequence, and the spreading code is known, so as to achieve the purpose of estimating the original signal of the user.

In a preferred embodiment, the solution of the system equation shown in equation (9) can be considered as solving for A_m,nx=b（A_m,nBlock Toeplitz matrix of order m × n), Jacobi and Gauss-Seidel are described separately below.

(1) Jacobi iterative method

Let the coefficient matrix A of the system of linear equations Ax = b be invertible and the main diagonal element a be₁₁,a₂₂,…a_nnAre all not zero, and are all not zero,

let D = diag (a)₁₁,a₂₂,…a_nn) And decompose a into a = (a-D) + D, so Ax = b can be written as follows:

Dx=(D-A)x+b

let x = B₁x+f₁

Wherein, B₁=I-D^-1A,f₁=D^-1b。

With B₁The Jacobi iteration formula for the iteration matrix (i.e., the system matrix) is as follows:

x^(k+1)=B₁x^(k)+f₁

expressing the formula with the components of the vector is:

wherein the content of the first and second substances,is the initial vector.

Therefore, the Jacobi iteration method is simple in formula, and only multiplication of a matrix and a vector needs to be calculated once per iteration. Two sets of storage units are needed to store x in the calculation^(k)And x^(k+1)。

(2) Gauss-Seidel iterative solution method

As known from the Jacobi iterative formula, x is used in each step of the iteration calculation process^(k)All components of (a) to calculate x^(k+1)Is obviously calculating the ith componentThe latest component that has been calculatedNot utilized, the most recently computed component may be intuitively better than the old component.

Therefore, these newly calculated k +1 th order approximations x^(k+1)Component (b) ofAnd the method is utilized to obtain a Gauss-Seidel iteration method for solving the equation set.

The coefficient matrix a is decomposed into a = D-L-U,

wherein, D = diag (a)₁₁,a₂₂,…a_nn) -L, -U are respectively the lower and upper triangular parts excluding the main diagonal elements of a, so Ax = b can be written as follows:

(D-L)x=Ux+b

i.e. x = B₂x+f₂

Wherein, B₂=(D-L)^-1U，f₂=(D-L)^-1b

With B₂The Gauss-Seidel iterative formula formed for the iterative matrix is:

x^(k+1)=B₂x^(k)+f₂

expressed in the form of components of a vector

As can be seen from the two iterative solutions, the Gauss-Seidel iterative method converges faster than the Jacobi iterative method (i.e., the number of iterations required to achieve the same accuracy is small).

In another preferred embodiment, the solution of the system equation shown in equation (9) can be viewed as solving for Cb = z (where C is a matrix of NxN, z is a vector of Nx1, and b is an unknown vector of Nx1, i.e., the quantity to be solved, and in this preferred embodiment b is solved using Gauss-Seidel and Jacobi.

（1）Gauss-Seidel

The nth iteration formula is as follows:

wherein c is_ijIs the element in the ith row and the jth column in the C matrix,is the value of the (n +1) th iteration of the ith element.

The SOR part is to weight and sum the results of the (n +1) th iteration and the nth iteration based on the above formula as the final solution of the (n +1) th iteration. Wherein the content of the first and second substances,

and the result after weighting is taken as input for the next iteration, alpha being the SOR factor.

（2）Jacobi

Assume that Cb = z is solved as well. The Jacobi iterative formula is as follows:

fig. 8 is a performance comparison diagram of several joint detection methods according to an embodiment of the present invention, and although the CD-approximating algorithm is not described in detail in this embodiment, the performance difference between the CD-approximating algorithm and the Gauss-Seidel iteration method with SOR can be visually seen through fig. 8, as shown in fig. 8, the solid line represents the performance of the CD-approximating method, and the dotted line represents the performance of the joint detection method provided by the embodiment of the present invention. As can be seen from fig. 8, the near CD method performs less well.

Compared with the method with the lowest complexity in the related art, the joint detection method of the embodiment does not need to directly invert the matrix and FFT operation, reduces the operation time of joint detection by at least four times, and the saved time can be reflected to software (code size, execution time and power consumption) or hardware (circuit area, timing budget and power consumption). Thus, the joint detection method described above can be probed by detecting the power profile of the SW or HW or competitor's baseband modem.

It will be apparent to those skilled in the art that the modules or steps of the present invention described above may be implemented by a general purpose computing device, they may be centralized on a single computing device or distributed across a network of multiple computing devices, and they may alternatively be implemented by program code executable by a computing device, such that they may be stored in a storage device and executed by a computing device, or fabricated separately as individual integrated circuit modules, or fabricated as a single integrated circuit module from multiple modules or steps. Thus, the present invention is not limited to any specific combination of hardware and software.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (8)

1. A joint detection method, comprising:

constructing a system equation x = Bd + n according to the received signals needing joint detection, wherein x is a received signal matrix, B is a system matrix, d is a transmitted signal matrix, and n is a noise matrix;

and solving the system equation by using an iterative method with convergence characteristics, and acquiring a sending signal according to a solving result.

2. The method of claim 1, wherein solving the system of equations using an iterative method with convergence properties comprises:

solving the system equation by using a Jacobi iteration method; alternatively, the first and second electrodes may be,

the system equations are solved using a Gauss Seidel iterative method with successive super-relaxed SORs.

3. The method of claim 2, wherein the SOR factor is greater than 0 and less than 2 using a gaussian-Seidel iteration with SOR.

4. The process of any one of claims 1 to 3, wherein B is a block-shaped Toeplitz matrix.

5. A joint detection device, comprising:

the system comprises a construction module, a detection module and a processing module, wherein the construction module is set to construct a system equation x = Bd + n according to received signals needing joint detection, wherein x is a received signal matrix, B is a system matrix, d is a transmitted signal matrix, and n is a noise matrix;

and the processing module is used for solving the system equation by using an iterative method with convergence characteristics and acquiring a sending signal according to a solving result.

6. The apparatus of claim 5, wherein the processing module is configured to solve the system of equations using Jacobi iterative; alternatively, it is arranged to solve the system equations using a Gauss Seidel iterative method with successive super-relaxed SORs.

7. The apparatus of claim 6, wherein the SOR factor of the Gaussian-Seidel Gauss-Seidel iteration of SOR is greater than 0 and less than 2.

8. The device according to any one of claims 5 to 7, wherein B is a block-shaped Toeplitz matrix.