CN1314216C - Quasi maximum posterior probability detection method for layer space hour code system - Google Patents

Abstract

A quasi-maximum a posteriori probability detection method and system for a layered space-time code system, the detection method uses the QR decomposition-interference deletion structure of the layered space-time code system, from the bottommost transmitting antenna to the topmost transmitting antenna Calculate the posterior probability of partial vectors in turn, when performing breadth-first search in the symbolic vector tree diagram, only keep one or a group of partial vectors with the largest posterior probability of partial vectors, and delete the remaining partial vectors; and in the process of breadth-first search The number of partial vectors for each layer of the selected transmit antenna is different; one or a group of vectors with the largest posterior probability of the vector obtained after the search is completed, and one of the vectors with the largest posterior probability is used as the estimated value of the transmitted signal vector , that is, the hard decision result; or obtain the soft information of the posterior probability of the transmitted symbol according to the set of vector posterior probability. The detection method greatly reduces the complexity of detection while maintaining high performance.

Description

Quasi-Maximum A Posteriori Detection Method and System for Hierarchical Space-Time Code System

technical field

The invention relates to a quasi-maximum a posteriori probability detection method and system for a layered space-time code system, and belongs to the technical field of signal detection and estimation of a wireless communication system.

Background technique

Now, the spectrum resource of the wireless channel is already very limited, but the wireless communication system is required to be able to transmit data at a higher rate. Since the channel capacity of multiple-input multiple-output (MIMO) is linearly related to the minimum number of transmitting and receiving antennas, the academic community generally believes that the multi-antenna technology is the main solution for future broadband wireless communication systems. Among them, the hierarchical space-time code structure has received more attention because it can obtain the highest multiplexing gain.

The existing detection methods of layered space-time codes mainly include linear detection algorithm, suboptimal detection algorithm and optimal detection algorithm. The following are brief descriptions of each:

Linear detection algorithms include QR decomposition-interference deletion algorithm, zero-forcing ZF algorithm, and minimum mean square error MMSE algorithm. The complexity of the detection is linear with the number of transmitting antennas, but the performance deteriorates significantly.

For example QR decomposition-interference removal algorithm. For a multiple-input multiple-output MIMO system with n _T transmit antennas and n _R receive antennas, the channel is an n _R ×n _T -dimensional fading channel H, and the system noise is an n _R- dimensional Gaussian white noise vector v, that is, in v Each component is white Gaussian noise with mean 0 and squared deviation 1. n _T- dimensional transmit signal vector $\mathrm{the\; s}={[{\mathrm{the\; <figure-callout\; id="1"\; label="s"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="s"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T$ with n _R- dimensional received signal vector $\mathrm{the\; y}={[{\mathrm{the\; <figure-callout\; id="1"\; label="y"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">y</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="y"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">y</figure-callout>}}_2,\&CenterDot;\&Center\; Dot;\&CenterDot;,{\mathrm{the\; y}}_{{\mathrm{no}}_R}]}^T$ send and receive between

The signal relationship model is: $\mathrm{the\; y}=\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\mathrm{Hs}+v;$ Among them, ρ is the total energy of the transmitted signal on the n _T transmitting antennas.

The QR decomposition-interference deletion algorithm is to convert the relationship model of the sending and receiving signals into a QR decomposition-interference deletion structure: perform QR decomposition on the fading channel matrix H in the relationship model between the sending and receiving signals, that is, H=QR, where Q is unitary matrix, and R is an upper triangular matrix. After substituting this formula into the above-mentioned transceiver signal relationship model, the obtained formula is: $\mathrm{the\; y}=\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\mathrm{QRs}+v.$

Multiply both sides of this formula by Q ^H to the left to obtain the relationship model of sending and receiving signals after orthogonal transformation:

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; =</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}{\mathrm{<span\; class="notranslate">\; Q</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; QRs</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; Q</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; =</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}\mathrm{<span\; class="notranslate">\; Rs.</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&omega;</span>}<span\; class="notranslate">\; ;</span>$

In the formula, the vector z=Q ^H y is the signal vector after the orthogonal transformation of the received signal vector y, and the vector $z={[{\mathrm{<figure-callout\; id="1"\; label="z"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">z</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">z</figure-callout>}}_2,\&Center\; Dot;\&CenterDot;\&Center\; Dot;,z_{{\mathrm{no}}_T}]}^T$ The dimensionality of is n _T . Since Q is a unitary matrix, Q ^H Q is equal to a unitary matrix. ω=Q ^H v, $\mathrm{\&omega;}={[{\mathrm{\&omega;}}_1,{\mathrm{\&omega;}}_2,\&CenterDot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{\&omega;}}_{{\mathrm{no}}_T}]}^T$ is an _nT- dimensional Gaussian white noise vector.

Because the n _T components of the transmitted signal vector s are s ₁ , s ₂ , ..., s _nT respectively, the n _T components of the signal vector z are z ₁ , z ₂ , ..., z _nT , and the Gaussian white noise vector ω The n _T components are respectively ω ₁ , ω ₂ ,..., ω _nT , the element of the upper triangular matrix R is r _ij , and i and j are natural numbers within the interval [1, n _T ]. Therefore, the above orthogonally transformed transceiver signal relationship model can be written from vector form to matrix form: This equation is the QR decomposition-interference removal structure.

可以由下面的QR分解-干扰删除检测算法得到。After obtaining the QR decomposition-interference deletion structure, use the upper triangular structure of the matrix R to send the estimated values of the signals s ₁ , s ₂ , ..s _nT (that is, the hard decision result)

It can be obtained by the following QR decomposition-interference deletion detection algorithm.

${\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; (</span>\sqrt{\frac{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}}\frac{{\mathrm{<span\; class="notranslate">\; z</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}{{\mathrm{<span\; class="notranslate">\; r</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}<span\; class="notranslate">\; )</span>$

${\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; r</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; (</span>\sqrt{\frac{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}}{\mathrm{<span\; class="notranslate">\; z</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$

...

${\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="r"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">r</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1,1</span>}}}<span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="no\; T\; r"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">no</figure-callout></span><figure-callout\; id="1"\; label="no\; T\; r"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">\; </figure-callout>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{}_{}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$

In the formula, L(x) represents a symbol in the transmitted symbol set, and the distance between this symbol and x is the smallest among all transmitted symbols.

Although the above-mentioned QR decomposition-interference removal algorithm has lower complexity, its performance is much worse than other algorithms.

The optimal detection algorithm includes maximum a posteriori probability MAP detection and maximum likelihood ML detection. Under the condition of equal probability of the prior probability of the signal source sending symbols, these two detection methods can obtain the optimal performance. However, the complexity of these two detection methods is exponentially related to the number of transmitting antennas. Therefore, the optimal algorithm usually only provides a theoretical limit that the detection performance can achieve, and it is generally not used in the actual communication system. For example, the maximum a posteriori probability MAP detection method uses a greedy search algorithm in the symbol vector tree diagram to retain all possible paths in the tree structure, and its complexity is the exponential power of the number of transmitting antennas. For example, when the number of transmit antennas is 4 and QPSK modulation is performed (the number of transmit symbols is 4), there are 4 nodes on the transmit antennas of the fourth layer, 4×4=16 nodes on the transmit antennas of the third layer, and the transmit antennas of the second layer There are 16×4=64 nodes on the antenna, and there are 64×4=256 nodes on the first-layer transmitting antenna. A total of 4 ⁴ =256 paths need to be searched, and then the vector posterior probabilities of all possible 256 paths need to be calculated: P(s ^j |y), where j is a natural number in the interval [1, 256]. Then select the maximum value from the 256 vector posterior probabilities, and the corresponding vector is the estimated value of the transmitted symbol vector (hard decision). And when the number of transmit antennas is 4 and 16QAM modulation is adopted (the number of transmit symbols is 16), there are 16 nodes on the transmit antennas of the fourth layer, 16×16=256 nodes on the transmit antennas of the third layer, and the second layer There are 256×16=4096 nodes on the transmitting antenna, and 4096×4=65536 nodes on the first layer transmitting antenna. 16QAM modulation needs to search for 65536 paths, and then needs to calculate the vector posterior probability of all possible 65536 paths: P(s ^j |y), j is a natural number in the interval [1, 65536]. Then select the maximum value from the 65536 vector posterior probabilities, and the corresponding vector is the estimated value of the transmitted symbol vector (hard decision). Such computational complexity is too high for the communication system to tolerate. Therefore, suboptimal algorithms are usually used in practical communication systems.

The performance of the suboptimal algorithm is very close to that of the optimal algorithm, however, the complexity is greatly reduced. The best suboptimal detection algorithm known so far is the Sphere Decoding (SD, Sphere Decoding,) algorithm. The sphere decoding algorithm is a quasi-maximum likelihood algorithm, which uses the depth-first search method to search in the symbol vector tree graph, and its performance is close to the optimal algorithm. However, there are still deficiencies in the sphere decoding algorithm: (1) the complexity of the sphere decoding has a cubic relationship with the number of transmitting antennas, which is still relatively high; (2) the sphere decoding algorithm uses depth-first search in the symbol vector tree diagram, because There is a backtracking process in the depth-first search, so the nodes in the tree diagram may be visited multiple times, which makes the search process more cumbersome; (3) the sphere decoding algorithm needs to select the search radius before searching. If the search radius is not selected properly, the complexity will be unacceptable. control. For example, if the selected search radius is too large, it is necessary to visit almost all nodes in the symbolic vector tree diagram, and the complexity is too high; if the selected search radius is too small, there may not be a path within the search radius after the search ends , causing the algorithm not to converge, and it is necessary to increase the search radius before re-searching, which will result in multiple repeated searches and higher complexity.

Therefore, how to find a high-performance, low-complexity layered space-time code detection algorithm is still a problem that people in the industry are exploring.

Contents of the invention

The purpose of the present invention is to provide a kind of quasi maximum a posteriori probability detection method for layered space-time code system, this detection method greatly reduces the complexity of searching and calculation when signal detection, while maintaining higher performance, has Good application prospects.

Another object of the present invention is to provide two kinds of layered space-time code systems that use the quasi-maximum a posteriori probability detection method to output hard decision results and soft information respectively as detectors in the receiver, as implementation devices for the detection method.

Yet another object of the present invention is to provide a working method of a layered space-time code system using a quasi-maximum a posteriori probability detection method to output soft information as a detector in an iterative receiver.

In order to achieve the above object, the present invention provides a quasi-maximum a posteriori probability detection method for layered space-time code systems, which is to obtain the estimated value of the transmitted signal vector according to the received signal vector and the MIMO channel matrix As a hard decision result, or obtain the soft information of the posterior probability of the transmitted symbol; it is characterized in that: the detection method is a quasi-maximum posterior probability detection method; the method utilizes the QR decomposition-interference deletion of the layered space-time code system structure, calculate the partial vector posterior probability sequentially from the bottom transmit antenna to the top transmit antenna, when performing breadth-first search in the symbol vector tree diagram, only keep the largest one or a group of partial vectors in the partial vector posterior probability, delete The remaining partial vectors; and the number of partial vectors of each layer of transmit antennas selected in the breadth-first search process is different; one or a group of vectors with the largest vector posterior probability obtained after the search; one of them has the largest posterior probability The vector of the posterior probability is used as the estimated value of the transmitted signal vector, that is, the result of the hard decision; or the soft information of the posterior probability of the transmitted symbol is obtained according to the set of vector posterior probabilities.

The method comprises the steps of:

(1) After the system is started, initialize the system parameters: initialize the prior probability of each transmit symbol on each layer of transmit antennas to equal probability, and set the number m _k of partial vectors selected by each layer of transmit antennas, where k is The serial number of the transmitting antenna layer, which is a natural number with an interval of [1, n _T ], where n _T is the number of layers of the transmitting antenna, that is, the serial number of the bottom antenna;

(2) Convert the relationship model between the sending and receiving signals into a QR decomposition-interference deletion structure (QR Decomposition and Cancellation Structure) through an orthogonal transformation, where the QR decomposition is an orthogonal-triangular decomposition (orthogonal-triangular decomposition);

For a multiple-input multiple-output MIMO system with n _T transmit antennas and n _R receive antennas, its channel is n _R × n _T dimensional fading channel H, and the system noise is that each component has a mean value of 0 and a variance of Be the _nR- dimensional Gaussian white noise vector v of 1; assume that the sending symbol set is A={a ₁ ,..., a _i ,..., a _L }, where L is the number of symbols in the sending symbol set, a _i is a transmission symbol in the transmission symbol set, and i is the serial number of the transmission symbol; then n _T- dimensional transmission signal vector $\mathrm{the\; s}={[{\mathrm{the\; <figure-callout\; id="1"\; label="s"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="s"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T$ with n _R- dimensional received signal vector $\mathrm{the\; y}{[{\mathrm{the\; <figure-callout\; id="1"\; label="y"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">y</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="y"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">y</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; y}}_{{\mathrm{no}}_R}]}^T$ The relationship model between sending and receiving signals is: $\mathrm{the\; y}=\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\mathrm{Hs}+v,$ In the formula, ρ is the total energy of the transmitted signal on the n _T transmitting antennas;

The relationship model between the receiving and sending signals is transformed into a QR decomposition-interference deletion structure through orthogonal transformation. Correspondingly, the fading channel matrix H is transformed into an upper triangular matrix R; n _R- dimensional received signal vector y is transformed into n _T -dimensional signal vector z composed of n _T components z ₁ , z ₂ ,..., z _nT ;

For n _T- dimensional transmitted signal vector $\mathrm{the\; s}={[{\mathrm{the\; <figure-callout\; id="1"\; label="s"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="s"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_k,{\mathrm{the\; s}}_{k+1},\&Center\; Dot;\&CenterDot;\&CenterDot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T,$ Among them, the last n _T -k+1 components form part of the vector s ${\tilde{S}}_k={[{\mathrm{the\; s}}_k,{\mathrm{the\; s}}_{k+1},\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T;$ Similarly, for n _T- dimensional signal vector $z={[{\mathrm{<figure-callout\; id="1"\; label="z"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">z</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">z</figure-callout>}}_2,\&CenterDot;\&CenterDot;\&CenterDot;,z_k,z_{k+1},\&CenterDot;\&CenterDot;\&CenterDot;,z_{{\mathrm{no}}_T}]}^T,$ Among them, the last n _T -k+1 components form a part vector of vector z ${{\tilde{Z}}_k}^{\&prime;}={[z_k,z_{k+1}^{\&prime;},\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,z_{{\mathrm{no}}_T}]}^T,$ In the above two formulas, k is the serial number of the transmitting antenna layer;

以及通过最底层发送天线上的部分矢量后验概率的计算方法得到其部分矢量的后验概率 其中j是第n_T层发送天线上备选的部分矢量的序号，分别取区间[1，L]内的自然数，L是发送天线上备选的部分矢量个数；从第n_T层发送天线上的L个备选的部分矢量后验概率中选择最大的m_nT个数值作为第n_T层发送天线上的一组部分矢量的后验概率，记为 其对应的部分矢量即为第n_T层发送天线上的m_nT个部分矢量，记为 j是第n_T层发送天线上的部分矢量的序号，分别取区间[1，m_nT]内的自然数，并删除其余的L-m_nT个部分矢量；(3) Start breadth-first search: shilling the sequence number k=n _T of the transmit antenna layer, calculate the n _T layer, that is, the L candidate partial vectors on the bottom transmit antenna

And the posterior probability of its partial vector is obtained through the calculation method of the posterior probability of the partial vector on the bottom transmitting antenna Wherein j is the sequence number of the optional partial vector on the nth _T layer transmitting antenna, which is a natural number in the interval [1, L] respectively, and L is the number of optional partial vectors on the transmitting antenna; from the n _T layer transmitting antenna Among the L alternative partial vector posterior probabilities on , select the largest m _nT values as the posterior probability of a group of partial vectors on the nth _T layer transmitting antenna, denoted as The corresponding partial vectors are the m _nT partial vectors on the transmit antenna of the nth _T layer, denoted as j is the sequence number of the partial vectors on the nth _T layer transmitting antenna, respectively take natural numbers in the interval [1, m _nT ], and delete the remaining Lm _nT partial vectors;

(4) Let the serial number k=n _T -1 of the transmitting antenna layer;

j是第k+1层发送天线上的部分矢量的序号，分别取区间[1，m_k+1]内的自然数；其中有一个部分矢量

向第k层发送天线产生L个分支，形成L个维数为n_T-k+1的部分矢量 该L个维数为n_T-k+1的部分矢量的后n_T-k维分量都是 且它们的第1维分量分别为a₁，a₂，...，a_L，该L个在信号矢量 条件下的部分矢量后验概率为 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k),$ 上述两式中，i是发送符号的序号，分别取区间[1，L]内的自然数；再通过相邻两层发送天线上的部分矢量后验概率间的递推关系式计算第k+1层发送天线上的部分矢量 向第k层发送天线产生的L个备选的部分矢量的后验概率 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k)$ 这样，第k+1层发送天线上的m_k+1个部分矢量向第k层发送天线生成了L·m_k+1个备选的部分矢量，该部分矢量的后验概率为 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k)$ i是发送符号的序号，分别取区间[1，L]内的自然数，j是第k+1层发送天线上的部分矢量的序号，分别取区间[1，m_k+1]内的自然数；从第k层发送天线上的L·m_k+1个备选的部分矢量的部分矢量后验概率中选择最大的m_k个数值作为第k层发送天线上的一组部分矢量的后验概率，记为

其对应的部分矢量即为第k层发送天线上的m_k个部分矢量，记为

j是第k层发送天线上的部分矢量的序号，分别取区间[1，m_k]内的自然数；删除其余的L·m_k+1-m_k个部分矢量；(5) There are m _k+1 partial vectors with dimension n _T -k on the transmitting antenna of layer k+1 Its partial vector posterior probability is

j is the serial number of the partial vector on the transmitting antenna of the k+1 layer, which is a natural number in the interval [1, m _k+1 ]; there is a partial vector

Generate L branches to the k-th layer transmit antenna to form L partial vectors with dimension n _T -k+1 The last n _T -k dimensional components of the L partial vectors with dimensions n _T -k+1 are all And their first dimension components are respectively a ₁ , a ₂ ,..., a _L , the L ones in the signal vector The partial vector posterior probability under the condition is $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ In the above two formulas, i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L] respectively; then the k+1th is calculated by the recursive relationship between the partial vector posterior probabilities on the adjacent two layers of transmitting antennas Partial vectors on layer transmit antennas The posterior probability of the L candidate partial vectors generated by the transmitting antenna to the kth layer $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k)$ In this way, the m _k+1 partial vectors on the k+1th layer transmit antenna generate L m _k+1 candidate partial vectors to the kth layer transmit antenna, and the posterior probability of this partial vector is $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k)$ i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L], and j is the serial number of the partial vector on the k+1th layer transmitting antenna, and is a natural number in the interval [1, m _k+1 ]; Select the largest m _k values from the partial vector posterior probabilities of L m _k+1 candidate partial vectors on the k-th layer transmitting antenna as the posterior probability of a group of partial vectors on the k-th layer transmitting antenna , denoted as

The corresponding partial vectors are the m _k partial vectors on the transmitting antenna of the kth layer, denoted as

j is the serial number of the partial vector on the kth layer transmitting antenna, which is a natural number in the interval [1, m _k ]; delete the remaining L·m _k+1 -m _k partial vectors;

(6) After the serial number k of the sending antenna layer is subtracted by 1, it is judged whether k>0 is established, if yes, then return to the execution step (5); otherwise, end the breadth-first search process, and perform subsequent steps;

(7) Get a set of m ₁ vectors from the breadth-first search process ${\tilde{S}}_1^j={\mathrm{the\; s}}^j$ and its corresponding vector posterior probability $P\left({\tilde{S}}_1^j\right|{\tilde{Z}}_1)=P\left({\mathrm{the\; s}}^j\right|z),$ Wherein j is the sequence number of the vector on the top-level transmit antenna, which is a natural number in the interval [1, m ₁ ]; if it is necessary to output a hard decision, then perform the subsequent steps; if it is necessary to output soft information, then jump to the execution step (9 );

(8) The number of vectors m ₁ on the topmost antenna = 1, this vector is the estimated value of the transmitted signal vector, that is, the result of the hard decision, and the result of the hard decision is output; the detection method ends;

(9) The number of vectors on the topmost antenna is m ₁ , using the m ₁ vectors and their vector posterior probabilities, according to the total probability formula, the soft information of the symbol posterior probability of each transmitted symbol on each layer of transmitting antennas is obtained, and output The soft message; detection method ends.

和 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k)$ 它们之间的后验概率的递推计算公式为：In the step (6), the recursive calculation formula of the partial vector posterior probability on the adjacent two layers of transmitting antennas, that is, the partial vector on the k+1th layer of transmitting antennas Generate L candidate partial vectors to the k-th layer transmit antenna The posterior probabilities of the two partial vectors are

and $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k)$ The recursive calculation formula of the posterior probability between them is:

$\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; Z</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Proportional;</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; Z</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

s_k＝a_i的条件下z_k的概率，它满足均值为 $\sqrt{\frac{\mathrm{\&rho;}}{n_T}}\underset{l=k+1}{\overset{n_T}{\mathrm{\&Sigma;}}}{r}_{k,l}{s}_l^j+\sqrt{\frac{\mathrm{\&rho;}}{n_T}}{r}_{k,k}{a}_i,$ 方差为1的高斯分布：In the formula, j is the serial number of the partial vector on the transmitting antenna on the k+1th layer, any natural number in the interval [1, m _k+1 ], i is the serial number of the transmitted symbol, and i is in the interval [1, L] A natural number in ; P(s _k =a _i ) is the symbol prior probability of the symbol a _i on the transmitting antenna of the kth layer; $P\left(z_k\right|{\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i)$ is known

The probability of z _k under the condition of s _k ＝a _i , which satisfies the mean $\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\underset{l=k+1}{\overset{{\mathrm{no}}_T}{\mathrm{\&Sigma;}}}{r}_{k,l}{\mathrm{the\; s}}_l^j+\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}{r}_{k,k}{a}_i,$ Gaussian distribution with variance 1:

$\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; \{</span><span\; class="notranslate">\; -</span>{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ;</span>$

为第k+1层发送天线上的第j个部分矢量；s_l ^j为k+1层发送天线上的第j个部分矢量的第l层发送天线上的符号；r_k，l为矩阵H进行QR分解后得到的上三角矩阵R中的第k行第l列的元素；r_k，k为矩阵H进行QR分解后得到的上三角矩阵R中的第k行第k列的元素。In the formula, n _T is the number of transmitting antennas; ρ is the total energy of the transmitted signal on n _T transmitting antennas; a _i is the transmitted symbol, i is the serial number of the transmitted symbol, and i is selected as a natural number in the interval [1, L];

is the j-th partial vector on the k+1th layer transmit antenna; s _l ^j is the symbol on the l-th layer transmit antenna of the j-th partial vector on the k+1 layer transmit antenna; r _k,l is the matrix H The element of row k and column l in the upper triangular matrix R obtained after QR decomposition; r _{k, k} is the element of row k and column k in the upper triangular matrix R obtained after QR decomposition of matrix H.

In the method, when selecting partial vectors for each layer of transmitting antennas, only select the partial vectors that contribute greatly to the output of the detection method, that is, select one or a group of partial vectors with the largest posterior probability of the partial vectors; The number of partial vectors is not fixed, it is different with the decrease of the number of transmitting antennas; in the hard decision output, the topmost transmitting antenna selects one partial vector, and the number of partial vectors selected by the transmitting antennas of other layers varies with The number of transmit antennas decreases but does not decrease monotonically, that is, $m_{{\mathrm{no}}_T}\&le;m_{{\mathrm{no}}_T-1}\&le;\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;\&le;{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">m</figure-callout>}}_2,$ m ₁ =1; when outputting soft information, the number of partial vectors selected for each layer of transmit antennas does not decrease monotonically with the decrease of the number of transmit antennas, namely $m_{{\mathrm{no}}_T}\&le;m_{{\mathrm{no}}_T-1}\&le;\&Center\; Dot;\&Center\; Dot;\&CenterDot;\&le;{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">m</figure-callout>}}_2\&le;{\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">m</figure-callout>}}_1.$

The method is used for a receiver detector in a layered space-time code system or a super three-generation mobile communication system.

即硬判决结果，然后将该硬判决结果送入星座点逆映射器得到二进制数据的估计值

最后送入信宿。In order to achieve the above object, the present invention provides a layered space-time code system detected according to the above detection method, which is composed of a transmitter and a receiver, wherein the transmitter includes: a signal source, a constellation point mapper, a serial-to-parallel conversion and n _T transmitting antennas; the receiver includes: n _R receiving antennas, a detector, a constellation point inverse mapper and a sink; where the binary data {b _i } generated by the source of the transmitter is sent to the constellation point mapper The transmission signal {s ₁ , ..., s _nT } is obtained, and the transmission signal {s ₁ , ..., s _nT } is sent to the serial-to-parallel converter to convert the serial signal into n _T parallel signals, and then Send out through n _T root transmitting antenna; It is characterized in that: the detector in the described receiver is the quasi maximum a posteriori probability detector that adopts quasi maximum a posteriori probability detection method and output hard decision; This quasi maximum a posteriori probability detection After receiving the n _R channels of parallel receiving signals {y ₁ ,...,y _nR } from the n _R receiving antennas in the receiver, the device performs quasi-maximum a posteriori probability detection to obtain the estimated value of the transmitted signal

That is, the hard decision result, and then send the hard decision result to the constellation point inverse mapper to obtain the estimated value of the binary data

Finally sent to the destination.

In order to achieve the above object, the present invention also provides a layered space-time code system that detects according to the above detection method, the system is composed of a transmitter and an iterative receiver, wherein the transmitter includes: a source, a channel encoder, an interleaving device, constellation point mapper, serial-to-parallel converter, and n _T transmitting antennas; the iterative receiver includes: n _R receiving antennas, detectors, bit likelihood ratio calculators, symbol probability calculators, deinterleavers, and interleavers , a channel decoder, two subtractors and a sink; where the binary data { _ci } generated by the source of the transmitter is sent to the channel encoder to obtain encoded binary data {bi _} , which is then sent to the interleaver After processing, the encoded randomly interleaved binary data {b _π(i) } is sent to the constellation point mapper, and the obtained transmitted signal {s ₁ ,...,s _nT } is sent to the serial-to-parallel conversion The serial signal is converted into parallel signals of n _T paths, and then sent out through n _T root transmitting antennas; it is characterized in that: the detector in the iterative receiver adopts the quasi-maximum a posteriori probability detection method and outputs each transmission A quasi-maximum a posteriori probability detector for sending symbol posterior probability soft information over an antenna.

In order to achieve the above object, the present invention further provides a working method of an iterative receiver in a layered space-time code system employing the above detection method, which is characterized in that: comprising the following steps:

(1) After the system is started, initialize the system parameters: initialize the prior probability of each transmitted symbol on each layer of transmitting antennas to equal probability, and send the initialized prior probability P(s) of the transmitted symbol into the quasi-maximum posterior probability Detector; initialize the bit likelihood ratio {λ ₂ [b _π(i) ]} output by the local interleaver to be 0, and set the bit likelihood ratio {λ ₂ [b _π(i )] of the initialized local interleaver output ₎ ]} into the first subtractor; then set the number of partial vectors m _k selected by each layer of transmit antenna, where k is the serial number of the transmit antenna layer, which is a natural number with an interval of [1, n _T ], n _T is The number of layers of the transmitting antenna, that is, the serial number of the bottom antenna;

(2) Convert the relationship model between the sending and receiving signals into a QR decomposition-interference deletion structure through an orthogonal transformation, wherein the QR decomposition is an orthogonal-triangulation decomposition;

(3) The detector processes the received information obtained from n _R receiving antennas according to the process of breadth-first search and soft information calculation in the quasi-maximum a posteriori probability detection method, and obtains the soft information p(s| y); then send the soft information p(s|y) of the posterior probability of the transmitted symbol output by the detector to the bit likelihood ratio calculator to obtain the bit likelihood ratio information {Λ ₁ [b _π(i) ]}, Then send the bit likelihood ratio information {Λ ₁ [b _π(i) ]} into the first subtractor;

(4) In the first subtractor, the bit likelihood ratio information {Λ ₁ [b _π(i) ]} is subtracted from the bit likelihood ratio information {λ ₂ [b _π(i) ]} output by the local interleaver to obtain The extrinsic information {λ ₁ [b _π(i) ]}, the extrinsic information {λ ₁ [b _π(i) ]} is sent to the deinterleaver for deinterleaving, and the deinterleaved extrinsic information {λ ₁ [b _i ]};

(5) Divide the deinterleaved external information {λ ₁ [ _bi ]} into two paths: one path is sent to the channel decoder for decoding, and the other path is sent to the second subtractor as prior information;

(6) The deinterleaved external information {λ ₁ [ _bi ]} is sent to the channel decoder for decoding, and the log likelihood ratio {Λ ₂ [ _bi ]} of the symbol soft information is output, and then the symbol soft information The log likelihood ratio {Λ ₂ [b _i ]} of the information is sent to the second subtractor;

(7) In the second subtractor, the log-likelihood ratio {Λ ₂ [b _i ]} of the symbol soft information output by the channel decoder is subtracted from the prior information {λ ₁ [b _i ]} to obtain the extrinsic information { λ ₂ [b _i ]}; after the external information {λ ₂ [b _i ]} is sent to the local interleaver for interleaving, it outputs a new bit likelihood ratio, that is, the new {λ ₂ [b _π(i) ]};

(8) The new bit likelihood ratio {λ ₂ [b _π(i) ]} is sent to the symbol probability calculator to obtain new prior information P(s); the new prior information P(s) is Send it to the quasi-maximum a posteriori probability detector to update the prior information; at the same time, in the process of sequentially executing the above steps (3) to (8) once, complete an iterative operation;

(9) cyclically execute the above steps (3)-(8), that is, after multiple iterations, stop decoding;

(10) After the iteration ends, the hard-decision binary data {Λ ₁ [a ₁ ]} output by the channel decoder is sent to the sink, and the iterative receiver ends its detection and decoding operations.

The number of loop iteration operations performed in the operation steps of the iterative receiver is at least one, and the number of iterations depends on different requirements of detection and decoding time delay and hardware complexity.

At present, the best detection methods for layered space-time codes include optimal detection algorithms and suboptimal detection algorithms. Among them, the optimal algorithm can achieve the best performance, but the complexity is too high, so the optimal algorithm only provides The theoretical limit that a detection performance can achieve is generally not adopted in the actual communication system. The performance of the suboptimal algorithm is very close to the performance of the optimal algorithm, but the complexity is greatly reduced, and the suboptimal algorithm is usually used in the actual communication system.

The present invention is a kind of quasi-maximum a posteriori probability detection method for layered space-time code system, and its advantage compared with the traditional optimal algorithm, that is, the maximum a posteriori probability detection method is: performance loss is very small, but, The detection complexity is greatly reduced. The advantages and effects of comparing with the best sub-optimal algorithm, that is, the sphere decoding algorithm are: the performance of this method is slightly better than that of the sphere decoding algorithm, or is equivalent to the performance of the sphere decoding algorithm; however, the complexity of the detection method Greatly reduced: the complexity of spherical decoding has a cubic relationship with the number of transmitting antennas, and the complexity of the present invention has a linear relationship with the number of transmitting antennas; moreover, the complexity of the present invention is controllable. Since there is no need to set other thresholds or parameters after determining the number of nodes selected by each layer of antennas, the complexity of the breadth-first search is determined, which can avoid the problem that the complexity of the sphere decoding algorithm may be uncontrollable.

The technical innovation of the detection method of the present invention is: using the QR decomposition-interference deletion structure of the layered space-time code system to recursively calculate the partial vector posterior probability from the bottommost transmitting antenna to the topmost transmitting antenna. This recursive calculation can use breadth-first search on the symbol vector tree graph, thus avoiding the backtracking process in the depth-first search of the sphere decoding algorithm. In addition, on each layer of transmitting antennas, the posterior probability of the partial vector is used as the measure, and the node with a large posterior probability of the partial vector is reserved (corresponding to the partial vector) to make it continue to generate child nodes; the node with the small posterior probability of the partial vector is deleted node (corresponding to the partial vector), so that it no longer produces child nodes. In this way, the loss of its performance is small compared with the greedy search algorithm (searching all paths in the symbol vector tree graph), but the complexity is greatly reduced.

Furthermore, when the present invention performs breadth-first search in the symbol vector tree diagram, the number of nodes selected for each layer of transmitting antennas is not fixed, but varies with the decreasing number of transmitting antennas. Since it is not necessary to set other thresholds or parameters after determining the number of nodes selected by each layer of antennas, the complexity of the breadth-first search in the present invention is definite, that is, the computational complexity is definite or controllable, and avoids The complexity may be an uncontrollable problem in the sphere decoding algorithm.

Finally, the detection method of the present invention can simultaneously obtain the estimated value of the transmitted vector (the result of the hard decision) and the posterior probability of the transmitted symbol (soft information); that is, after the search is completed, a vector with the maximum posterior probability of the vector can be obtained As a hard decision result, or a set of vectors with the largest vector posterior probability is obtained, and soft information of the posterior probability of the transmitted symbol is obtained according to the set of vector posterior probabilities.

Description of drawings

Figure 1(A) and (B) are schematic diagrams of the symbol vector tree diagrams of the maximum a posteriori probability detection method and the quasi-maximum a posteriori probability detection method when the number of transmitting antennas is assumed to be 4 and QPSK modulation is adopted.

Fig. 2 is a flow chart of the quasi maximum a posteriori probability detection method for a layered space-time code system according to the present invention.

Fig. 3 is a structural composition diagram of a transmitter and a receiver in a layered space-time code system output by a hard decision using a quasi-maximum a posteriori probability detection method according to the present invention.

Fig. 4 is a structural composition diagram of a transmitter and an iterative receiver in a layered space-time code system output by soft information using a quasi-maximum a posteriori probability detection method according to the present invention.

Fig. 5 is a graph comparing the bit error rate curve of the layered space-time code system simulation output by the hard decision of the quasi maximum a posteriori probability detector and the bit error rate curves of several other methods in the present invention.

Fig. 6 is a graph comparing bit error rate curves of different iterations when the present invention utilizes quasi-maximum a posteriori probability detector to output soft information as the first-stage detector of the receiver for simulation.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings.

For a multiple-input multiple-output MIMO system with n _T transmit antennas and n _R receive antennas, the channel is an n _R ×n _T -dimensional fading channel H, and the system noise is an n _R- dimensional Gaussian white noise vector v, that is, in v Each component is Gaussian white noise with mean 0 and variance 1. n _T- dimensional transmit signal vector $\mathrm{the\; s}={[{\mathrm{the\; <figure-callout\; id="1"\; label="s"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="s"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,\&CenterDot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T$ with n _R- dimensional received signal vector $\mathrm{the\; y}={[{\mathrm{the\; <figure-callout\; id="1"\; label="y"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">y</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="y"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">y</figure-callout>}}_2{,\&CenterDot;\&Center\; Dot;\&CenterDot;,\mathrm{the\; y}}_{{\mathrm{no}}_R}]}^T$ There is a relationship model between sending and receiving signals: $\mathrm{the\; y}=\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\mathrm{Hs}+v;$ In the formula, ρ is the total energy of the transmitted signal on n _T transmitting antennas.

The detection purpose of the output hard decision is to obtain the transmitted signal vector through the received signal vector y and the fading channel information H $\mathrm{the\; s}={[{\mathrm{the\; <figure-callout\; id="1"\; label="s"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="s"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,\&CenterDot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T$ estimated value of $\widehat{\mathrm{the\; s}}={[{\widehat{\mathrm{the\; s}}}_1,{\widehat{\mathrm{the\; s}}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\widehat{\mathrm{the\; s}}}_{{\mathrm{no}}_T}]}^T;$ The purpose of detecting the output soft information is to obtain the transmitted signal vector under the condition of received signal vector y by receiving signal vector y and fading channel information H $\mathrm{the\; s}={[{\mathrm{the\; <figure-callout\; id="1"\; label="s"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="s"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T$ The signed posterior probability values P(s ₁ |y), P(s ₂ |y), ..., P(s _nT |y) of each component of .

The present invention is a quasi-maximum a posteriori probability detection method for a layered space-time code system. First, the relationship between the sending and receiving signals should be modeled as a QR decomposition-interference deletion structure: that is, for the fading channel in the relationship model between the sending and receiving signals The matrix H is decomposed by QR: H=QR; where, Q is a unitary unitary matrix, and R is an upper triangular matrix. Substitute the above formula into the relationship model of sending and receiving signals: y=QRs+v; then, multiply both sides of the above formula by Q ^H to get the relationship model of sending and receiving signals after orthogonal transformation:

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; Q</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}{\mathrm{<span\; class="notranslate">\; Q</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; QRs</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; Q</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; =</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}\mathrm{<span\; class="notranslate">\; Rs.</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&omega;</span>}$

Among them, the vector z=Q ^H y is the signal vector after the orthogonal transformation of the received signal vector y, and the vector $z={[{\mathrm{<figure-callout\; id="1"\; label="z"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">z</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">z</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&CenterDot;,z_{{\mathrm{no}}_T}]}^T$ The dimensionality of is n _T . Due to the orthogonal transformation, the signal vector z is equivalent to the received signal vector y. Since Q is a unitary matrix, Q ^H Q is equal to a unitary matrix. ω=Q ^H v, $\mathrm{\&omega;}=[{{\mathrm{\&omega;}}_1,\mathrm{\&omega;}}_2,\&Center\; Dot;\&Center\; Dot;\&CenterDot;,{\mathrm{\&omega;}}_{{\mathrm{no}}_T}]$ is an _nT- dimensional Gaussian white noise vector.

该式就是QR分解-干扰删除结构。The n _T components of the transmitted signal vector s are s ₁ , s ₂ , ..., s _nT ; the n _T components of the signal vector z are z ₁ , z ₂ , ..., z _nT ; the n _T components of the Gaussian white noise vector ω The components are ω ₁ , ω ₂ ,..., ω _nT ; the element of the upper triangular matrix R is r _ij , and i and j are natural numbers in the interval [1, n _T ]. Write the above orthogonally transformed transceiver signal relationship model from vector form to matrix form:

This formula is the QR decomposition-interference deletion structure.

The vector posterior probability of sending signal vector s under the condition of receiving signal vector y is P(s|y); the vector posterior probability of sending signal vector s under the condition of receiving signal vector z is P(s|z ). Since the signal vector z is equivalent to the received signal vector y, P(s|y)=P(s|z).

的条件下，部分发送信号矢量

的部分矢量后验概率为 当

中的发送天线层的序号k依次取自然数n_T，n_T-1，...，1时，就组成不同层发送天线上的部分矢量后验概率序列： $P\left({\tilde{S}}_{n_T}\right|{\tilde{Z}}_{n_T}),P\left({\tilde{S}}_{n_T-1}\right|{\tilde{Z}}_{n_T-1}),...,P\left({\tilde{S}}_1\right|{\tilde{Z}}_1).$ 因为 ${\tilde{S}}_1=s,{\tilde{Z}}_1=z,$ 即最顶层发送天线上的部分矢量就是该矢量，所以有 $P\left({\tilde{S}}_1\right|{\tilde{Z}}_1)=P\left(s\right|z),$ 即最顶层发送天线上的部分矢量后验概率就是该矢量后验概率。for vector $\mathrm{the\; s}={[{\mathrm{the\; <figure-callout\; id="1"\; label="s"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">s</figure-callout>}}_1{,\mathrm{the\; <figure-callout\; id="2"\; label="s"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_k,{\mathrm{the\; s}}_{k+1},\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T,$ Its last n _T -k+1 components form the partial vector of vector s ${\tilde{S}}_k={[{\mathrm{the\; s}}_k{,\mathrm{the\; s}}_{k+1},\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T,$ k is the serial number of the transmitting antenna layer. Similarly, for signal vector $z={[{\mathrm{<figure-callout\; id="1"\; label="z"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">z</figure-callout>}}_1{,\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">z</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,z_k,z_{k+1},\&CenterDot;\&Center\; Dot;\&Center\; Dot;,z_{{\mathrm{no}}_T}]}^T,$ Its last n _T -k+1 components form the partial vector of vector z ${\tilde{Z}}_k={[z_k,z_{k+1},\&Center\; Dot;\&Center\; Dot;\&CenterDot;,z_{{\mathrm{no}}_T}]}^T,$ k is the serial number of the transmitting antenna layer. After receiving the partial vector

Under the condition that the partial transmit signal vector

The partial vector posterior probability of is when

When the serial number k of the transmitting antenna layer in the sequence takes the natural number n _T , n _T -1, ..., 1, the partial vector posterior probability sequence on the transmitting antenna of different layers is formed: $P\left({\tilde{S}}_{{\mathrm{no}}_T}\right|{\tilde{Z}}_{{\mathrm{no}}_T}),P\left({\tilde{S}}_{{\mathrm{no}}_T-1}\right|{\tilde{Z}}_{{\mathrm{no}}_T-1}),...,P\left({\tilde{S}}_1\right|{\tilde{Z}}_1).$ because ${\tilde{S}}_1=\mathrm{the\; s},{\tilde{Z}}_1=z,$ That is, the partial vector on the top transmit antenna is the vector, so we have $P\left({\tilde{S}}_1\right|{\tilde{Z}}_1)=P\left(\mathrm{the\; s}\right|z),$ That is, the partial vector posterior probability on the topmost transmitting antenna is the vector posterior probability.

有L^nT-k种可能的部分矢量，记为

j为第k+1层发送天线上的每个可能的部分矢量的序号，j分别取区间[1，L^nT-k]内的自然数。Let the transmitted symbol set be A={a ₁ , a ₂ , ..., a _L }, where L is the number of symbols in the transmitted symbol set, and a ₁ , a ₂ , ..., a _L are the transmitted symbols. For an _nT- dimensional transmitted signal vector s, each component of it can take any transmitted symbol in the transmitted symbol set A. Thus, there are L ^nT possible vectors for the transmitted signal vector s. A transmit symbol vector tree diagram with a ladder depth of n _T and each partial vector having L branches formed by all the vectors generated from the transmit antennas on the n _T layer to the top transmit antenna. partial vector

There are L ^nT-k possible partial vectors, denoted as

j is the sequence number of each possible partial vector on the k+1th layer transmit antenna, and j is a natural number in the interval [1, L ^nT-k ].

其具有的部分矢量后验概率 j是第k+1层发送天线上的部分矢量的序号。该部分矢量向第k层发送天线产生L个分支，形成L个维数是n_T-k+1的部分矢量。这L个维数是n_T-k+1的部分矢量的后n_T-k维分量都是

它们的第1维分量分别为a₁，a₂，...，a_L。这L个维数是n_T-k+1的部分矢量记作

i是发送符号的序号，分别取区间[1，L]的自然数。L个在信号矢量 条件下的维数是n_T-k+1的部分矢量的后验概率记作 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k),$ i是发送符号的序号，分别取区间[1，L]的自然数。第k+1层发送天线上的部分矢量

的部分矢量后验概率 与第k层发送天线上的部分矢量 的部分矢量后验概率 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k)$ 之间存在如下关系： $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k)\&Proportional;P\left(z_k\right|{\tilde{S}}_{k=1}^j,s_k=a_i)P(s_k-a_i)P\left({\tilde{S}}_{k+1}^j\right|{\tilde{Z}}_{k+1});$ 其中i取区间[1，L]的自然数。这就是相邻两层发送天线上的部分矢量后验概率间的递推关系式。Assume that there is a partial vector of dimension n _T -k on the transmitting antenna of layer k+1

It has a partial vector posterior probability j is the serial number of the partial vector on the k+1th layer transmit antenna. This partial vector generates L branches to the transmit antenna of the kth layer, forming L partial vectors whose dimension is n _T −k+1. The last n _T -k dimension components of the part vector whose L dimensions are n _T -k+1 are all

Their first dimension components are respectively a ₁ , a ₂ ,..., a _L . These L dimensions are part vectors of n _T -k+1 denoted as

i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L] respectively. L in signal vector The posterior probability of a partial vector whose dimension is n _T -k+1 under the condition is denoted as $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L] respectively. Partial vector on transmit antenna of layer k+1

The partial vector posterior probability of and the partial vector on the transmit antenna of layer k The partial vector posterior probability of $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k)$ There is the following relationship between: $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k)\&Proportional;P\left(z_k\right|{\tilde{S}}_{k=1}^j,{\mathrm{the\; s}}_k=a_i)P({\mathrm{the\; s}}_k-a_i)P\left({\tilde{S}}_{k+1}^j\right|{\tilde{Z}}_{k+1});$ Where i is a natural number in the interval [1, L]. This is the recursive relationship between the partial vector posterior probabilities on two adjacent layers of transmitting antennas.

Wherein, P(s _k =a _i ) is the symbol prior probability of the symbol a _i on the transmitting antenna of the kth layer; $P\left(z_k\right|{\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i)$ is known The probability of z _k under the condition of s _k ＝a _i , which satisfies the mean $\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\underset{l=k+1}{\overset{{\mathrm{no}}_T}{\mathrm{\&Sigma;}}}{r}_{k,l}{\mathrm{the\; s}}_l^j+\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}{r}_{k,k}{a}_i,$ Gaussian distribution with variance 1:

$\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; \{</span><span\; class="notranslate">\; -</span>{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \}</span>$

In the above formula, n _T is the number of transmitting antennas; ρ is the total energy of the transmitted signal on n _T transmitting antennas; a _i is the transmitted symbol, and i is the serial number of the transmitted symbol, which are natural numbers in the interval [1, L]; is the j-th partial vector on the k+1th layer transmit antenna; s _l ^j is the symbol on the l-th layer transmit antenna of the j-th partial vector on the k+1 layer transmit antenna; r _k,l is the matrix H The element of row k and column l in the upper triangular matrix R obtained after QR decomposition; r _{k, k} is the element of row k and column k in the upper triangular matrix R obtained after QR decomposition of matrix H.

方差为1的高斯分布：At the time of initialization, the serial number k=n _T of the transmitting antenna layer is the bottom layer, and its partial vector ${\tilde{S}}_{{\mathrm{no}}_T}={\mathrm{the\; s}}_{{\mathrm{no}}_T},{\tilde{Z}}_{{\mathrm{no}}_T}=z_{{\mathrm{no}}_T};$ The calculation method of the partial vector posterior probability on the bottom transmitting antenna is $P\left({\tilde{S}}_{{\mathrm{no}}_T}\right|{\tilde{Z}}_{{\mathrm{no}}_T})=P\left({\mathrm{the\; s}}_{{\mathrm{no}}_T}\right|z_{{\mathrm{no}}_T})=P\left(z_{{\mathrm{no}}_T}\right|{\mathrm{the\; s}}_{{\mathrm{no}}_T}=a_i)P({\mathrm{the\; s}}_{{\mathrm{no}}_T}=a_i);$ in, $P({\mathrm{the\; s}}_{{\mathrm{no}}_T}=a_i)$ is the symbol a priori probability of the symbol a _i on the transmitting antenna of the nth _T layer; $P\left(z_{{\mathrm{no}}_T}\right|{\mathrm{the\; s}}_{{\mathrm{no}}_T}=a_i)$ is known ${\mathrm{the\; s}}_{{\mathrm{no}}_T}=a_i$ The probability of z _nT under the condition that the mean is

Gaussian distribution with variance 1:

$\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; \{</span><span\; class="notranslate">\; -</span>{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}<span\; class="notranslate">\; -</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \}</span>$

Among them, n _T is the number of transmitting antennas; ρ is the total energy of the transmitted signal on the n _T transmitting antennas; a _i is the transmitted symbol, i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L]; r _{nT, nT} is the element of row _nT and column _nT in the upper triangular matrix R obtained after matrix H is decomposed by QR.

After calculating the partial vector posterior probability in the case of antenna serial number k=n _T , in the recursive relationship between the partial vector posterior probability on two adjacent layers of transmitting antennas, the serial number k of the transmitting antenna layer is sequentially By taking natural numbers from n _T -1 to 1, the partial vector posterior probabilities on the transmitting antennas of different layers can be obtained recursively, and the vector posterior probability P(s|z) can be obtained after the recursion ends.

In the maximum a posteriori probability detection method, all partial vectors on each layer of transmitting antennas are reserved. In the recursive relationship between the posterior probabilities of partial vectors corresponding to two adjacent layers of transmitting antennas, the number of partial vectors on the k+1th layer of transmitting antennas is L ^nT-k . The L ^nT-k partial vectors correspond to the L ^nT-k nodes on the k+1th layer of the transmit antenna in the transmit symbol vector tree diagram. Each of the L ^nT-k nodes generates L branches to the transmit antenna of the kth layer. In this way, there are L ^nT-k+1 nodes on the transmitting antenna of the kth layer. The L ^nT-k+1 nodes on the transmit antennas on the k-th layer in the transmit symbol vector tree diagram correspond to the L ^nT-k+1 partial vectors on the k-th layer transmit antennas. L ^nT-k+1 partial vector posterior probabilities can be given by a recursive relationship between partial vector posterior probabilities on two adjacent layers of transmitting antennas. In this way, the maximum a posteriori probability detection method needs to search all L ^nT paths in the transmitted symbol vector tree diagram. Such complexity is too high for the communication system to tolerate.

Referring to Fig. 1, it can be seen from Fig. 1(A): the detection method of the maximum a posteriori probability (MAP) adopts a greedy search algorithm in the symbol vector tree diagram, retains all possible paths in the tree structure, and the complexity is the exponential power of the number of transmitting antennas. As shown in the figure: when the number of transmitting antennas is 4 and QPSK modulation (the number of transmitting symbols is 4), there are 4 nodes on the transmitting antennas on the fourth layer, and 4×4=16 nodes on the transmitting antennas on the third layer. There are 16×4=64 nodes on the transmitting antennas on the second layer, and 64×4=256 nodes on the transmitting antennas on the first layer. A total of 4 ⁴ =256 paths need to be searched, and the vector posterior probabilities of all possible 256 paths are calculated: P(s ^j |y), j=1, 2, . . . , 256. Then select the maximum value from the 256 vector posterior probabilities, and the corresponding vector is the estimated value of the transmitted symbol vector (hard decision). If the number of transmitting antennas is 4 and 16QAM modulation, 65536 paths need to be searched. Such complexity is too high for the communication system to tolerate. It can be seen from Figure 1(B) that the quasi-maximum a posteriori probability detection method only searches 16 complete paths, while 16QAM modulation needs to search 256 complete paths, and the detection complexity is greatly reduced. Therefore, the quasi maximum a posteriori probability detection method is much less complex than the maximum a posteriori probability detection method, and its performance loss is small compared with the maximum a posteriori probability detection method (see Fig. 5 and Fig. 6).

For the quasi-maximal a posteriori probability detection method, a breadth-first search is performed in the vector treemap of the transmitted symbols. For each layer of transmitting antennas, instead of selecting all possible partial vectors, only the partial vectors that contribute greatly to the output of the detection method are selected, that is, one or a group of partial vectors with the largest posterior probability of partial vectors is selected; for each layer of transmitting antennas, the selected Part of the vector number is not fixed, it is different with the decreasing number of transmitting antennas. Assuming that the number of vectors selected by the k-th layer transmit antenna is m _k , the sequence number k of the transmit antenna layer is a natural number from n _T to 1 in sequence. For the case of hard decision output, the topmost transmit antenna selects one partial vector, and the number of partial vectors selected by other transmit antennas does not decrease monotonically with the decrease of the number of transmit antennas, that is $m_{{\mathrm{no}}_T}\&le;m_{{\mathrm{no}}_T-1}\&le;\&CenterDot;\&CenterDot;\&Center\; Dot;\&le;{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">m</figure-callout>}}_2,$ m ₁ =1; for the case of outputting soft information, the number of partial vectors selected for each layer of transmit antennas does not decrease monotonically with the decrease of the number of transmit antennas, namely $m_{{\mathrm{no}}_T}\&le;m_{{\mathrm{no}}_T-1}\&le;\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;\&le;{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">m</figure-callout>}}_2\&le;{\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">m</figure-callout>}}_1.$

以及其部分矢量后验概率 j是第k层发送天线上的部分矢量的序号，取区间[1，m_k]的自然数。计算天线的序号k＝n_T情况下的部分矢量后验概率后，在相邻两层发送天线上的部分矢量后验概率间的递推关系式中令发送天线层的序号k依次取从n_T-1到1的自然数，就可以递推地得到不同层发送天线上的部分矢量后验概率，递推结束后得到的m₁个矢量及其矢量后验概率 $P\left({\tilde{S}}_1^j\right|{\tilde{Z}}_1)=P\left(s^j\right|z),$ j是第1层发送天线上的部分矢量的序号，即矢量的序号，为区间[1，m₁]内的自然数。Corresponding to the recursive relationship between the partial vector posterior probabilities on two adjacent layers of transmitting antennas, the number of partial vectors on the k+1th layer transmitting antenna is m _k+1 , and its partial vector posterior probability is j is the sequence number of the partial vector on the k+1th layer transmit antenna, which is a natural number in the interval [1, m _k+1 ]. Each of the m _k+1 partial vectors generates L branches to the transmit antenna of the kth layer. In this way, L·m _k+1 candidate partial vectors are generated on the k-th layer transmit antenna, that is, L·m _k+1 candidate partial vectors are generated. L·m _k+1 partial vector posterior probability $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k)$ It can be given by the recursive relationship between the partial vector posterior probabilities on two adjacent layers of transmitting antennas, j is the serial number of the partial vector on the k+1th layer of transmitting antenna, which is the interval [1, m _k+1 ] The natural number in , i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L]. Select m _k with the largest value from L·m _k+1 alternative partial vector posterior probabilities, denoted as j is the serial number of the partial vectors on the transmitting antenna of the kth layer, which are natural numbers in the interval [1, m _k ]. Retain the partial vectors corresponding to the posterior probabilities of these m _k partial vectors, and delete the other L·m _k+1 -m _k partial vectors. In this way, the m _k partial vectors of the transmit antenna of the kth layer are obtained

and its partial vector posterior probability j is the serial number of the partial vector on the transmitting antenna of the kth layer, which is a natural number in the interval [1, m _k ]. After calculating the partial vector posterior probability in the case of antenna serial number k=n _T , in the recursive relationship between the partial vector posterior probability on two adjacent layers of transmitting antennas, the serial number k of the transmitting antenna layer is sequentially taken from n The natural number from _T -1 to 1 can recursively obtain the partial vector posterior probabilities on the transmitting antennas of different layers, and obtain m ₁ vectors and their vector posterior probabilities after the recursion ends $P\left({\tilde{S}}_1^j\right|{\tilde{Z}}_1)=P\left({\mathrm{the\; s}}^j\right|z),$ j is the serial number of the partial vector on the first layer of transmitting antenna, that is, the serial number of the vector, which is a natural number in the interval [1, m ₁ ].

Two kinds of outputs can be obtained by using the obtained m ₁ vectors and their corresponding vector posterior probabilities: hard decision output and soft information output. When outputting a hard decision, the m ₁ =1 vectors on the topmost antenna are the estimated value of the transmitted signal vector, that is, the result of the hard decision. When outputting soft information, use _m1 vectors and their vector posterior probabilities, and obtain the symbol posterior probability soft information of each transmitted symbol on each layer of transmitting antennas according to the following full probability formula: $P({\mathrm{the\; s}}_k=a_i|\mathrm{the\; y})=P({\mathrm{the\; s}}_k=a_i|z)=\frac{1}{W}\underset{j=1}{\overset{{\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">m</figure-callout>}}_1}{\mathrm{\&Sigma;}}}\mathrm{\&delta;}({\mathrm{the\; s}}_k^j=a_i)P\left({\mathrm{the\; s}}^j\right|z);$

In the formula, k is the serial number of the transmitting antenna layer, which is a natural number in the interval [1, n _T ], i is the serial number of the transmitted symbol, and is a natural number in the interval [1, L]; P(s _k =a _i |y) =P(s _k =a _i |z) is the symbol posterior probability soft information of the transmitted symbol a _i on the kth layer transmit antenna; P(s ^j |z) is m ₁ vector posterior probability, and j is the vector The serial number of , which is a natural number in the interval [1, m ₁ ]; s _k ^j is the symbol on the transmitting antenna of the k-th layer in the j-th vector; $W=\underset{j=1}{\overset{m_1}{\mathrm{\&Sigma;}}}P\left({\mathrm{the\; s}}^j\right|z),$ is the sum of m ₁ vector posterior probabilities; δ(·) is an indicator function:

$\mathrm{<span\; class="notranslate">\; \&delta;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\begin{array}{c}\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>& \mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="0"\; label="e"\; filenames="C20051006790600372.png"\; state="\{\{state\}\}">e</figure-callout></span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>& \mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; e</span>}\end{array}\right.<span\; class="notranslate">\; .</span>\end{array}$

Referring to Fig. 2, the specific operation steps of the quasi-maximum a posteriori probability detection method that the present invention is used for layered space-time code system are introduced:

(1) After the system is started, initialize the system parameters: initialize the prior probability of each transmit symbol on each layer of transmit antennas to equal probability, and set the number m _k of partial vectors selected by each layer of transmit antennas, where k is The serial number of the transmitting antenna layer, which is a natural number with an interval of [1, n _T ], where n _T is the number of layers of the transmitting antenna, that is, the serial number of the bottom antenna;

The quasi maximum a posteriori probability detection method of the present invention does not select all possible partial vectors for each layer of transmitting antennas, but only selects the partial vectors that contribute greatly to the output of the detection method; the number of partial vectors selected for each layer of transmitting antennas is not fixed : It varies with the decrease of the number of transmit antennas; for the case of hard decision output, the top transmit antenna selects one partial vector, and the partial vector numbers selected by other transmit antennas monotonically do not decrease with the decrease of the number of transmit antennas, namely $m_{{\mathrm{no}}_T}\&le;m_{{\mathrm{no}}_T-1}\&le;\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;\&le;{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">m</figure-callout>}}_2,$ m ₁ =1; for the case of outputting soft information, the number of partial vectors selected for each layer of transmit antennas does not decrease monotonically with the decrease of the number of transmit antennas, namely

${\mathrm{<span\; class="notranslate">\; m</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&le;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="m"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">m</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="m"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">m</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; .</span>$

(2) Convert the relationship model between the sending and receiving signals into a QR decomposition-interference deletion structure through an orthogonal transformation, wherein the QR decomposition is an orthogonal-triangulation decomposition;

For a multiple-input multiple-output MIMO system with n _T transmit antennas and n _R receive antennas, its channel is n _R × n _T dimensional fading channel H, and the system noise is that each component has a mean value of 0 and a variance of Be the _nR- dimensional Gaussian white noise vector v of 1; assume that the sending symbol set is A={a ₁ ,..., a _i ,..., a _L }, where L is the number of symbols in the sending symbol set, a _i is a transmission symbol in the transmission symbol set, and i is the serial number of the transmission symbol; then n _T- dimensional transmission signal vector $\mathrm{the\; s}={[{\mathrm{the\; <figure-callout\; id="1"\; label="s"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="s"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">s</figure-callout>}}_2,\&CenterDot;\&CenterDot;\&CenterDot;a_{{\mathrm{no}}_T}]}^T$ with n _R- dimensional received signal vector $\mathrm{the\; y}={[{\mathrm{the\; <figure-callout\; id="1"\; label="y"\; filenames="C20051006790600331.png,C20051006790600372.png"\; state="\{\{state\}\}">y</figure-callout>}}_1,{\mathrm{the\; <figure-callout\; id="2"\; label="y"\; filenames="C20051006790600331.png,C20051006790600341.png"\; state="\{\{state\}\}">y</figure-callout>}}_2,\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;{\mathrm{the\; y}}_{{\mathrm{no}}_R}]}^T$ The relationship model between sending and receiving signals is: $\mathrm{the\; y}=\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\mathrm{Hs}+v,$ In the formula, ρ is the total energy of the transmitted signal on the n _T transmitting antennas;

The relationship model between the receiving and sending signals is transformed into a QR decomposition-interference deletion structure through orthogonal transformation. Correspondingly, the fading channel matrix H is transformed into an upper triangular matrix R; n _R- dimensional received signal vector y is transformed into The n _T -dimensional signal vector z composed of n _T components z ₁ , z ₂ ,..., z _nT ; for the n _T- dimensional sending signal vector $\mathrm{the\; s}={[{\mathrm{the\; s}}_1,{\mathrm{the\; s}}_2,\&Center\; Dot;\&CenterDot;\&CenterDot;{,{\mathrm{the\; s}}_k,{\mathrm{the\; s}}_{k+1},\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T,$ Among them, the last n _T -k+1 components form part of the vector s $\widetilde{{\mathrm{the\; s}}_k}={[{\mathrm{the\; s}}_k,{\mathrm{the\; s}}_{k+1},\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T;$ Similarly, for n _T- dimensional signal vector $z={[z_1,z_2,\&CenterDot;\&Center\; Dot;\&Center\; Dot;{,z_k,z_{k+1},\&Center\; Dot;\&CenterDot;\&CenterDot;,z}_{{\mathrm{no}}_T}]}^T,$ Among them, the last n _T -k+1 components form a part vector of vector z $\widetilde{Z_k}={[z_k,z_{k+1},\&Center\; Dot;\&Center\; Dot;\&CenterDot;,z_{{\mathrm{no}}_T}]}^T,$ In the above two formulas, k is the serial number of the transmitting antenna layer;

以及通过最底层发送天线上的部分矢量后验概率的计算方法得到其部分矢量的后验概率

其中j是第n_T层发送天线上备选的部分矢量的序号，分别取区间[1，L]内的自然数，L是发送天线上备选的部分矢量个数；从第n_T层发送天线上的L个备选的部分矢量后验概率中选择最大的m_nT个数值作为第n_T层发送天线上的一组部分矢量的后验概率，记为 其对应的部分矢量即为第n_T层发送天线上的m_nT个部分矢量，记为 j是第n_T层发送天线上的部分矢量的序号，分别取区间[1，m_nT]内的自然数，并删除其余的L-m_nT个部分矢量；(3) Start breadth-first search: shilling the sequence number k=n _T of the transmit antenna layer, calculate the n _T layer, that is, the L candidate partial vectors on the bottom transmit antenna

And the posterior probability of its partial vector is obtained through the calculation method of the posterior probability of the partial vector on the bottom transmitting antenna

Wherein j is the sequence number of the optional partial vector on the nth _T layer transmitting antenna, which is a natural number in the interval [1, L] respectively, and L is the number of optional partial vectors on the transmitting antenna; from the n _T layer transmitting antenna Among the L alternative partial vector posterior probabilities on , select the largest m _nT values as the posterior probability of a group of partial vectors on the nth _T layer transmitting antenna, denoted as The corresponding partial vectors are the m _nT partial vectors on the transmit antenna of the nth _T layer, denoted as j is the sequence number of the partial vectors on the nth _T layer transmitting antenna, respectively take natural numbers in the interval [1, m _nT ], and delete the remaining Lm _nT partial vectors;

(4) Let the serial number k=n _T -1 of the transmitting antenna layer;

向第k层发送天线产生L个分支，形成L个维数为n_T-k+1的部分矢量

该L个维数为n_T-k+1的部分矢量的后n_T-k维分量都是

且它们的第1维分量分别为a₁，a₂，...，a_L，该L个在信号矢量

条件下的部分矢量后验概率为 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k),$ 上述两式中，i是发送符号的序号，分别取区间[1，L]内的自然数；再通过相邻两层发送天线上的部分矢量后验概率间的递推关系式计算第k+1层发送天线上的部分矢量

向第k层发送天线产生的L个备选的部分矢量的后验概率 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k),$ 这样，第k+1层发送天线上的m_k+1个部分矢量向第k层发送天线生成了L·m_k+1个备选的部分矢量，该部分矢量的后验概率为 $P({\tilde{S}}_{k+1}^j,s_k=a_i|{\tilde{Z}}_k),$ i是发送符号的序号，分别取区间[1，L]内的自然数，j是第k+1层发送天线上的部分矢量的序号，分别取区间[1，m_k+1]内的自然数；从第k层发送天线上的L·m_k+1个备选的部分矢量的部分矢量后验概率中选择最大的m_k个数值作为第k层发送天线上的一组部分矢量的后验概率，记为 其对应的部分矢量即为第k层发送天线上的m_k个部分矢量，记为 j是第k层发送天线上的部分矢量的序号，分别取区间[1，m_k]内的自然数；删除其余的L·m_k+1-m_k个部分矢量；(5) There are m _k+1 partial vectors with dimension n _T -k on the transmitting antenna of layer k+1 Its partial vector posterior probability is j is the serial number of the partial vector on the transmitting antenna of the k+1 layer, which is a natural number in the interval [1, m _k+1 ]; there is a partial vector

Generate L branches to the k-th layer transmit antenna to form L partial vectors with dimension n _T -k+1

The last n _T -k dimensional components of the L partial vectors with dimensions n _T -k+1 are all

And their first dimension components are respectively a ₁ , a ₂ ,..., a _L , the L ones in the signal vector

The partial vector posterior probability under the condition is $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ In the above two formulas, i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L] respectively; then the k+1th is calculated by the recursive relationship between the partial vector posterior probabilities on the adjacent two layers of transmitting antennas Partial vectors on layer transmit antennas

The posterior probability of the L candidate partial vectors generated by the transmitting antenna to the kth layer $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ In this way, the m _k+1 partial vectors on the k+1th layer transmit antenna generate L m _k+1 candidate partial vectors to the kth layer transmit antenna, and the posterior probability of this partial vector is $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L], and j is the serial number of the partial vector on the k+1th layer transmitting antenna, and is a natural number in the interval [1, m _k+1 ]; Select the largest m _k values from the partial vector posterior probabilities of L m _k+1 candidate partial vectors on the k-th layer transmitting antenna as the posterior probability of a group of partial vectors on the k-th layer transmitting antenna , denoted as The corresponding partial vectors are the m _k partial vectors on the transmitting antenna of the kth layer, denoted as j is the serial number of the partial vector on the kth layer transmitting antenna, which is a natural number in the interval [1, m _k ]; delete the remaining L·m _k+1 -m _k partial vectors;

(6) After the serial number k of the sending antenna layer is subtracted by 1, it is judged whether k>0 is established, if yes, then return to the execution step (5); otherwise, end the breadth-first search process, and perform subsequent steps;

(7) Get a set of m ₁ vectors from the breadth-first search process ${\tilde{S}}_1^j={\mathrm{the\; s}}^j$ and its corresponding vector posterior probability $P\left({\tilde{S}}_1^j\right|{\tilde{Z}}_1)=P\left({\mathrm{the\; s}}^j\right|z),$ Wherein j is the sequence number of the vector on the top-level transmit antenna, which is a natural number in the interval [1, m ₁ ]; if it is necessary to output a hard decision, then perform the subsequent steps; if it is necessary to output soft information, then jump to the execution step (9 );

(8) The number of vectors m ₁ on the topmost antenna = 1, this vector is the estimated value of the transmitted signal vector, that is, the result of the hard decision, and the result of the hard decision is output; the detection method ends;

(9) The number of vectors on the topmost antenna is m ₁ , using the m ₁ vectors and their vector posterior probabilities, according to the total probability formula, the soft information of the symbol posterior probability of each transmitted symbol on each layer of transmitting antennas is obtained, and output The soft message; detection method ends.

Referring to Fig. 3, introduce the layered space-time code system that utilizes quasi maximum a posteriori probability detection method of the present invention to output hard decision, this system is made up of transmitter and receiver, and wherein transmitter comprises: information source, constellation point mapper, Serial-to-parallel converter and n _T transmitting antennas; the receiver includes: n _R receiving antennas, quasi-maximum a posteriori probability detector, constellation point inverse mapper and sink, where the binary data {b _i }, sent to the constellation point mapper to obtain the transmitted signal {s ₁ ,...,s _nT }, and the transmitted signal {s ₁ ,...s _nT } is sent to the serial-to-parallel converter to convert the serial signal into n _T parallel signals are sent out through n _T transmitting antennas; the characteristic is that the detector in the receiver is a quasi-maximum a posteriori probability detector that adopts a quasi-maximum a posteriori probability detection method and outputs a hard decision; the quasi-maximum After the posterior probability detector receives n _R channels of parallel received signals {y ₁ , ..., y _nR } from the n _R receiving antennas in the receiver, after performing quasi-maximum posterior probability detection, the transmitted Estimated value of the signal That is, the hard decision result, and then send the hard decision result to the constellation point inverse mapper to obtain the estimated value of the binary data Finally sent to the destination.

Referring to Fig. 4, introduce the layered space-time code system that utilizes quasi maximum a posteriori probability detection method of the present invention to output soft information, this system is made up of transmitter and iterative receiver, wherein transmitter comprises: information source, channel coder, Interleaver, constellation point mapper, serial-to-parallel converter and multiple (n _T ) transmit antennas; iterative receiver includes: multiple (n _R ) receive antennas, detector, bit likelihood ratio calculator, symbol probability calculation device, deinterleaver, interleaver, channel decoder, two subtractors and sink; where the binary data {c _i } generated by the source of the transmitter is sent to the channel encoder to obtain encoded binary data { _bi }, is sent to the interleaver for processing, and the encoded randomly interleaved binary data {b _π(i) } is sent to the constellation point mapper, and the obtained transmitted signal {s ₁ ,...,s _nT } is sent to the serial-to-parallel converter to convert the serial signal into n _T parallel signals, and then send it out through n _T transmitting antennas; its characteristic is that the detector in the iterative receiver adopts the quasi maximum Method and quasi-maximum a posteriori probability detector outputting soft information of a posteriori probability of transmitted symbols on each transmitting antenna.

The operation steps of the working method of the iterative receiver are as follows:

(1) After the system is started, initialize the system parameters: initialize the prior probability of each transmitted symbol on each layer of transmitting antennas to equal probability, and send the initialized prior probability P(s) of the transmitted symbol into the quasi-maximum posterior probability Detector; initialize the bit likelihood ratio {λ ₂ [b _π(i) ]} output by the local interleaver to be 0, and set the bit likelihood ratio {λ ₂ [b _π(i )] of the initialized local interleaver output ₎ ]} into the first subtractor; then set the number of partial vectors m _k selected by each layer of transmit antenna, where k is the serial number of the transmit antenna layer, which is a natural number with an interval of [1, n _T ], n _T is The number of layers of the transmitting antenna, that is, the serial number of the bottom antenna;

(2) Convert the relationship model between the sending and receiving signals into a QR decomposition-interference deletion structure (QR Decomposition and Cancellation Structure) through orthogonal transformation:

(3) The detector processes the received information obtained from n _R receiving antennas according to the process of breadth-first search and soft information calculation in the quasi-maximum a posteriori probability detection method, and obtains the soft information p(s| y); then send the soft information p(s|y) of the posterior probability of the transmitted symbol output by the detector to the bit likelihood ratio calculator to obtain the bit likelihood ratio information {Λ ₁ [b _π(i) ]}, Then send the bit likelihood ratio information {Λ ₁ [b _π(i) ]} into the first subtractor;

(4) In the first subtractor, the bit likelihood ratio information {Λ ₁ [b _π(i) ]} is subtracted from the bit likelihood ratio information {λ ₂ [b _π(i) ]} output by the local interleaver to obtain The extrinsic information {λ ₁ [b _π(i) ]}, the extrinsic information {λ ₁ [b _π(i) ]} is sent to the deinterleaver for deinterleaving, and the deinterleaved extrinsic information {λ ₁ [b _i ]};

(5) Divide the deinterleaved external information {λ ₁ [ _bi ]} into two paths: one path is sent to the channel decoder for decoding, and the other path is sent to the second subtractor as prior information;

(6) The deinterleaved external information {λ ₁ [ _bi ]} is sent to the channel decoder for decoding, and the log likelihood ratio {Λ ₂ [ _bi ]} of the symbol soft information is output, and then the symbol soft information The log likelihood ratio {Λ ₂ [b _i ]} of the information is sent to the second subtractor;

(7) In the second subtractor, the log-likelihood ratio {Λ ₂ [b _i ]} of the symbol soft information output by the channel decoder is subtracted from the prior information {λ ₁ [b _i ]} to obtain the extrinsic information { λ ₂ [b _i ]}; after the external information {λ ₂ [b _i ]} is sent to the local interleaver for interleaving, it outputs a new bit likelihood ratio, that is, the new {λ ₂ [b _π(i) ]};

(8) The new bit likelihood ratio {λ ₂ [b _π(i) ]} is sent to the symbol probability calculator to obtain new prior information P(s); the new prior information P(s) is Send it to the quasi-maximum a posteriori probability detector to update the prior information; at the same time, in the process of sequentially executing the above steps (3) to (8) once, complete an iterative operation;

(9) cyclically execute the above steps (3)-(8), that is, after multiple iterations, stop decoding;

(10) After the iteration ends, the hard-decision binary data {Λ ₁ [a ₁ ]} output by the channel decoder is sent to the sink, and the iterative receiver ends its detection and decoding operations.

The number of loop iterative operations performed in the operation steps of the iterative receiver is at least 1, or multiple iterations of 2 or more times; the number of iterations depends on the delay and hardware complexity of detection and decoding. Require.

The present invention is a quasi-maximum a posteriori probability detection method and system for a layered space-time code system. The detection method utilizes the QR decomposition-interference deletion structure of the layered space-time code system, from the bottom transmitting antenna to the bottom The top-level transmitting antenna recursively calculates the posterior probability of some vectors, so that the posterior probability of partial vectors can be calculated, and then breadth-first search is performed in the symbol vector tree diagram, and only some vectors with high posterior probability of partial vectors are kept, and other parts are deleted Vector, since some redundant paths (that is, paths with small posterior probability of vectors or paths with small contribution to detector output) are deleted in the vector tree diagram, the complexity of detection is greatly reduced.

The quasi-maximum posterior probability detection method of the present invention is a suboptimal detection method, and its complexity is greatly reduced compared with the optimal detection method. It has been explained in detail above. Now it is mainly introduced that the performance loss of the quasi maximum a posteriori probability detection method of the present invention is smaller than that of the maximum a posteriori probability detection method.

Referring to Fig. 5, introduce the detector embodiment that adopts quasi maximum a posteriori probability detection method of the present invention to compare with existing several detection methods (comprising maximum a posteriori probability and suboptimal detection method-ball decoding detection method) compared error Schematic diagram of code rate curve comparison. The conditions for the simulation implementation are: the number of transmitting antennas is 4, the number of receiving antennas is 8, flat fading channel, and QPSK modulation. As can be seen from the figure, the bit error performance curve (represented by the solid line) of the quasi maximum a posteriori probability detector is better than the bit error performance curve (represented by the dotted line) of the maximum a posteriori probability detector of the optimal algorithm (indicated by the dotted line) lowest position) with little loss and slightly better than the error performance curve of a ball-decoded detector (i.e., the best suboptimal algorithm known so far) (shown by the dotted line - at a position close to the upper position of the solid line), or Its bit error performance performance is equivalent; its bit error performance is far better than the bit error performance curve of the detector of the QR decomposition-interference deletion algorithm (indicated by a thick dotted line - at the top position). However, the computational complexity of the present invention is greatly reduced: the complexity of spherical decoding is cubically related to the number of transmitting antennas, and the complexity of the quasi-maximum a posteriori probability detection method is linearly related to the number of transmitting antennas; moreover, the quasi-maximum a posteriori The complexity of the probabilistic detection method is manageable. Since there is no need to set other thresholds or parameters after determining the number of nodes selected by each layer of antennas, the complexity of the breadth-first search is determined, which avoids the problem that the complexity of the sphere decoding algorithm may not be controllable.

Referring to FIG. 6 , it introduces the comparison of BER curves of different iterations when the present invention uses the quasi-maximum a posteriori probability detector to output soft information as the first-stage detector of the receiver for simulation. The conditions for the simulation implementation are: the number of transmitting antennas is 4, the number of receiving antennas is 8, flat fading channel, QPSK modulation, and the channel coder is Turbo-R1/2. In the figure, the bit error performance curve of the first iteration is represented by a solid line, the bit error performance curve of the second iteration is a dotted line, and the bit error performance curve of the third iteration is a dashed line. It can be seen that the performance of the iterative receiver of the layered space-time code system improves after multiple iterations: on the order of BER of 10 ^-3 , the bit error performance of three iterations is about 0.25dB higher than that of one iteration.

Claims (9)

1. A quasi-maximum a posteriori probability detection method for a layered space-time code system, which is to obtain the estimated value of the transmitted signal vector as a hard decision result according to the received signal vector and the MIMO channel matrix, or obtain the transmitted symbol The soft information of posterior probability; It is characterized in that: described detection method is a kind of quasi-maximum posterior probability detection method; This method utilizes the QR decomposition-interference deletion structure of layered space-time code system, from the bottom transmit antenna to the bottom The top-level transmitting antenna calculates the posterior probability of partial vectors sequentially. When performing breadth-first search in the symbol vector tree diagram, only the largest partial vector or a group of partial vectors in the partial vector posterior probability is retained, and the rest of the partial vectors are deleted; and in the width-first The number of partial vectors of each layer of transmitting antennas selected during the search is different; one or a group of vectors with the largest posterior probability of the vector is obtained after the search; one of the vectors with the largest posterior probability is used as the transmitted signal vector The estimated value of , that is, the result of the hard decision; or the soft information of the posterior probability of the transmitted symbol is obtained according to the set of vector posterior probabilities. 2. The detection method according to claim 1, characterized in that: the method comprises the following steps: (1) After the system is started, initialize the system parameters: initialize the prior probability of each transmit symbol on each layer of transmit antennas to equal probability, and set the number m _k of partial vectors selected by each layer of transmit antennas, where k is The serial number of the transmitting antenna layer, which is a natural number with an interval of [1, n _T ], where n _T is the number of layers of the transmitting antenna, that is, the serial number of the bottom antenna; (2) Convert the relationship model between the sending and receiving signals into a QR decomposition-interference deletion structure through an orthogonal transformation, wherein the QR decomposition is an orthogonal-triangulation decomposition; For a multiple-input multiple-output MIMO system with n _T transmit antennas and n _R receive antennas, its channel is n _R × n _T dimensional fading channel H, and the system noise is that each component has a mean value of 0 and a variance of Be the _nR- dimensional Gaussian white noise vector v of 1; assume that the sending symbol set is A={a ₁ ,..., a _i ,..., a _L }, where L is the number of symbols in the sending symbol set, a _i is a transmission symbol in the transmission symbol set, and i is the serial number of the transmission symbol; then n _T- dimensional transmission signal vector $\mathrm{the\; s}={[{\mathrm{the\; s}}_1,{\mathrm{the\; s}}_2,\&CenterDot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T$ with n _R- dimensional received signal vector $\mathrm{the\; y}={[{\mathrm{the\; y}}_1,{\mathrm{the\; y}}_2,\&CenterDot;\&CenterDot;\&Center\; Dot;,{\mathrm{the\; y}}_{{\mathrm{no}}_R}]}^T$ The relationship model between sending and receiving signals is: $\mathrm{the\; y}=\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\mathrm{Hs}+v,$ In the formula, ρ is the total energy of the transmitted signal on the n _T transmitting antennas; The relationship model between the receiving and sending signals is transformed into a QR decomposition-interference deletion structure through orthogonal transformation. Correspondingly, the fading channel matrix H is transformed into an upper triangular matrix R; n _R- dimensional received signal vector y is transformed into n _T -dimensional signal vector z composed of n _T components z ₁ , z ₂ ,..., z _nT ; For n _T- dimensional transmitted signal vector $\mathrm{the\; s}={[{\mathrm{the\; s}}_1,{\mathrm{the\; s}}_2,\&CenterDot;\&CenterDot;\&Center\; Dot;,{\mathrm{the\; s}}_k,{\mathrm{the\; s}}_{k+1},\&Center\; Dot;\&Center\; Dot;\&CenterDot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T,$ Among them, the last n _T -k+1 components form part of the vector s ${\tilde{S}}_k={[{\mathrm{the\; s}}_k,{\mathrm{the\; s}}_{k+1},\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,{\mathrm{the\; s}}_{{\mathrm{no}}_T}]}^T;$ Similarly, for n _T- dimensional signal vector $z={[z_1,z_2,\&Center\; Dot;\&Center\; Dot;\&CenterDot;,z_k,z_{k+1},\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,z_{{\mathrm{no}}_T}]}^T,$ Among them, the last n _T -k+1 components form a part vector of vector z ${\tilde{Z}}_k={[z_k,z_{k+1},\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;,z_{{\mathrm{no}}_T}]}^T,$ In the above two formulas, k is the serial number of the transmitting antenna layer; (3) Start breadth-first search: shilling the sequence number k=n _T of the transmit antenna layer, calculate the n _T layer, that is, the L candidate partial vectors on the bottom transmit antenna

And the posterior probability of its partial vector is obtained through the calculation method of the posterior probability of the partial vector on the bottom transmitting antenna

Wherein j is the sequence number of the optional partial vector on the nth _T layer transmitting antenna, which is a natural number in the interval [1, L] respectively, and L is the number of optional partial vectors on the transmitting antenna; from the n _T layer transmitting antenna Among the L alternative partial vector posterior probabilities on , select the largest m _nT values as the posterior probability of a group of partial vectors on the nth _T layer transmitting antenna, denoted as

The corresponding partial vectors are the m _nT partial vectors on the transmit antenna of the nth _T layer, denoted as

j is the sequence number of the partial vectors on the nth _T layer transmitting antenna, respectively take natural numbers in the interval [1, m _nT ], and delete the remaining Lm _nT partial vectors; (4) Let the serial number k=n _T -1 of the transmitting antenna layer; (5) There are m _k+1 partial vectors with dimension n _T -k on the transmitting antenna of layer k+1 Its partial vector posterior probability is j is the serial number of the partial vector on the transmitting antenna of the k+1 layer, which is a natural number in the interval [1, m _k+1 ]; there is a partial vector

Generate L branches to the k-th layer transmit antenna to form L partial vectors with dimension n _T -k+1

The last n _T -k dimensional components of the L partial vectors with dimensions n _T -k+1 are all And their first dimension components are respectively a ₁ , a ₂ ,..., a _L , the L ones in the signal vector

The partial vector posterior probability under the condition is $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ In the above two formulas, i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L] respectively; then the k+1th is calculated by the recursive relationship between the partial vector posterior probabilities on the adjacent two layers of transmitting antennas Partial vectors on layer transmit antennas The posterior probability of the L candidate partial vectors generated by the transmitting antenna to the kth layer $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ In this way, the m _k+1 partial vectors on the k+1th layer transmit antenna generate L m _k+1 candidate partial vectors to the kth layer transmit antenna, and the posterior probability of this partial vector is $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ i is the serial number of the transmitted symbol, which is a natural number in the interval [1, L], and j is the serial number of the partial vector on the k+1th layer transmitting antenna, and is a natural number in the interval [1, m _k+1 ]; Select the largest m _k values from the partial vector posterior probabilities of L m _k+1 candidate partial vectors on the k-th layer transmitting antenna as the posterior probability of a group of partial vectors on the k-th layer transmitting antenna , denoted as

The corresponding partial vectors are the m _k partial vectors on the transmitting antenna of the kth layer, denoted as j is the serial number of the partial vector on the kth layer transmitting antenna, which is a natural number in the interval [1, m _k ]; delete the remaining L·m _k+1 -m _k partial vectors; (6) After the serial number k of the sending antenna layer is subtracted by 1, it is judged whether k>0 is established, if yes, then return to the execution step (5); otherwise, end the breadth-first search process, and perform subsequent steps; (7) Get a set of m ₁ vectors from the breadth-first search process ${\tilde{S}}_1^j={\mathrm{the\; s}}^j$ and its corresponding vector posterior probability $P\left({\tilde{S}}_1^j\right|{\tilde{Z}}_1)=P\left({\mathrm{the\; s}}^j\right|z),$ Wherein j is the sequence number of the vector on the top-level transmit antenna, which is a natural number in the interval [1, m ₁ ]; if it is necessary to output a hard decision, then perform the subsequent steps; if it is necessary to output soft information, then jump to the execution step (9 ); (8) The number of vectors m ₁ on the topmost antenna = 1, this vector is the estimated value of the transmitted signal vector, that is, the result of the hard decision, and the result of the hard decision is output; the detection method ends; (9) The number of vectors on the topmost antenna is m ₁ , using the m ₁ vectors and their vector posterior probabilities, according to the total probability formula, the soft information of the symbol posterior probability of each transmitted symbol on each layer of transmitting antennas is obtained, and output The soft message; detection method ends. 3. The detection method according to claim 2, characterized in that: in the step (6), the recursive calculation formula of the partial vector posterior probability on the adjacent two layers of transmitting antennas is that the k+1 layer transmits Partial vectors on the antenna

Generate L candidate partial vectors to the k-th layer transmit antenna

The posterior probabilities of the two partial vectors are and $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k),$ The recursive calculation formula of the posterior probability between them is: $P({\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i|{\tilde{Z}}_k)\&Proportional;P\left(z_k\right|{\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i)P({\mathrm{the\; s}}_k=a_i)P\left({\tilde{S}}_{k+1}^j\right|{\tilde{Z}}_{k+1});$ In the formula, j is the serial number of the partial vector on the transmitting antenna on the k+1th layer, any natural number in the interval [1, m _k+1 ], i is the serial number of the transmitted symbol, and i is in the interval [1, L] A natural number in ; P(s _k =a _i ) is the symbol prior probability of the symbol a _i on the transmitting antenna of the kth layer; $P\left(z_k\right|{\tilde{S}}_{k+1}^j,{\mathrm{the\; s}}_k=a_i)$ is known

The probability of z _k under the condition of s _k ＝a _i , which satisfies the mean $\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}\underset{l=k+1}{\overset{{\mathrm{no}}_T}{\mathrm{\&Sigma;}}}{r}_{k,l}{\mathrm{the\; s}}_l^j+\sqrt{\frac{\mathrm{\&rho;}}{{\mathrm{no}}_T}}{r}_{k,k}{a}_i,$ Gaussian distribution with variance 1: $\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>{\stackrel{<span\; class="notranslate">\; ~</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; \{</span>{<span\; class="notranslate">\; -</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; l</span>}}{\mathrm{<span\; class="notranslate">\; the\; s</span>}}_{\mathrm{<span\; class="notranslate">\; l</span>}}^{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span>\sqrt{\frac{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ;</span>$ In the formula, n _T is the number of transmitting antennas; ρ is the total energy of the transmitted signal on n _T transmitting antennas; a _i is the transmitted symbol, i is the serial number of the transmitted symbol, and i is selected as a natural number in the interval [1, L]; is the j-th partial vector on the k+1th layer transmit antenna; s _l ^j is the symbol on the l-th layer transmit antenna of the j-th partial vector on the k+1 layer transmit antenna; r _k,l is the matrix H The element of row k and column l in the upper triangular matrix R obtained after QR decomposition; r _{k, k} is the element of row k and column k in the upper triangular matrix R obtained after QR decomposition of matrix H. 4. The detection method according to claim 1, characterized in that: in the method, when the partial vectors are selected for each layer of transmitting antennas, only the partial vectors that contribute greatly to the output of the detection method are selected, that is, the posterior probability of the partial vectors is selected The largest one or a group of partial vectors; and the number of partial vectors selected for each layer of sending days is not fixed, it is different with the decrease of the number of sending antennas; when the hard decision is output, the topmost sending antenna selects 1 Partial vectors, the number of partial vectors selected by the transmit antennas of other layers will not decrease monotonically with the decrease of the number of transmit antennas, that is $m_{{\mathrm{no}}_T}\&le;m_{{\mathrm{no}}_T-1}\&le;\&Center\; Dot;\&Center\; Dot;\&Center\; Dot;\&le;m_2,m_1=1;$ When outputting soft information, the number of partial vectors selected for each layer of transmit antennas does not decrease monotonically as the number of transmit antennas decreases, that is, ${\mathrm{<span\; class="notranslate">\; m</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&le;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; .</span>$ 5. The detection method according to claim 1, characterized in that the method is used for a detector of a receiver in a layered space-time code system or a super three-generation mobile communication system. 6, a kind of layered space-time code system that detects according to the detection method of claim 1 is made up of transmitter and receiver, and wherein transmitter comprises: signal source, constellation point mapper, serial-to-parallel converter and n _T root Transmitting antenna; the receiver includes: n _R receiving antennas, detectors, constellation point inverse mapper and sink; where the binary data {b _i } generated by the source of the transmitter is sent to the constellation point mapper to obtain the transmitted signal {s ₁ ,...,s _nT }, the transmission signal {s ₁ ,...,s _nT } is sent to the serial-to-parallel converter to convert the serial signal into n _T parallel signals, and then sent through n _T roots The antenna sends out; it is characterized in that: the detector in the receiver is a quasi-maximum a posteriori probability detector that adopts a quasi-maximum a posteriori probability detection method and outputs a hard decision; After the n _R receiving antennas in the machine receive n _R channels of parallel receiving signals {y ₁ ,...,y _nR }, they perform quasi-maximum a posteriori probability detection to obtain the estimated value of the transmitted signal

That is, the hard decision result, and then send the hard decision result to the constellation point inverse mapper to obtain the estimated value of the binary data

Finally sent to the destination. 7. A layered space-time code system for detection according to the detection method of claim 1, consisting of a transmitter and an iterative receiver, wherein the transmitter includes: a source, a channel encoder, an interleaver, a constellation point mapper, Serial-to-parallel converter and n _T transmitting antennas; iterative receiver includes: n _R receiving antennas, detector, bit likelihood ratio calculator, symbol probability calculator, deinterleaver, interleaver, channel decoder, two A subtractor and a sink; where the binary data {c _i } generated by the source of the transmitter is sent to the channel encoder to obtain the encoded binary data {b _{i }, and then sent to the interleaver for processing, and the obtained encoded binary data {b i} } The randomly interleaved binary data {b _π(i) } is sent to the constellation point mapper, and the transmitted signal {s ₁ ,...,s _nT } obtained is sent to the serial-to-parallel converter to convert the serial signal Be n _T road parallel signals, then send out by n _T root transmitting antennas; It is characterized in that: the detector in the iterative receiver adopts quasi-maximum a posteriori probability detection method and outputs the a posteriori probability of sending symbols on each transmitting antenna A quasi-maximum a posterior probability detector for soft information. 8. A working method of an iterative receiver in a layered space-time code system according to claim 7, characterized in that: comprising the following steps: (1) After the system is started, initialize the system parameters: initialize the prior probability of each transmitted symbol on each layer of transmitting antennas to equal probability, and send the initialized prior probability P(s) of the transmitted symbol into the quasi-maximum posterior probability Detector; initialize the bit likelihood ratio {λ ₂ [b _π(i) ]} output by the local interleaver to be 0, and set the bit likelihood ratio {λ ₂ [b _π(i )] of the initialized local interleaver output ₎ ]} into the first subtractor; then set the number of partial vectors m _k selected by each layer of transmit antenna, where k is the serial number of the transmit antenna layer, which is a natural number with an interval of [1, n _T ], n _T is The number of layers of the transmitting antenna, that is, the serial number of the bottom antenna; (2) Convert the relationship model between the sending and receiving signals into a QR decomposition-interference deletion structure through an orthogonal transformation, wherein the QR decomposition is an orthogonal-triangulation decomposition; (3) The detector processes the received signals obtained from n _R receiving antennas according to the process of breadth-first search and calculation of soft information in the quasi-maximum a posteriori probability detection method, and obtains the soft information of the posterior probability of the transmitted symbol p(s| y); then send the soft information p(s|y) of the posterior probability of the transmitted symbol output by the detector to the bit likelihood ratio calculator to obtain the bit likelihood ratio information {Λ ₁ [b _π(i) ]}, Then send the bit likelihood ratio information {Λ ₁ [b _π(i) ]} into the first subtractor; (4) In the first subtractor, the bit likelihood ratio information {Λ ₁ [b _π(i) ]} is subtracted from the bit likelihood ratio information {λ ₂ [b _π(i) ]} output by the local interleaver to obtain The extrinsic information {λ ₁ [b _π(i) ]}, the extrinsic information {λ ₁ [b _π(i) ]} is sent to the deinterleaver for deinterleaving, and the deinterleaved extrinsic information {λ ₁ [b _i ]}; (5) Divide the deinterleaved external information {λ ₁ [ _bi ]} into two paths: one path is sent to the channel decoder for decoding, and the other path is sent to the second subtractor as prior information; (6) The deinterleaved external information {λ ₁ [ _bi ]} is sent to the channel decoder for decoding, and the log likelihood ratio {Λ ₂ [ _bi ]} of the symbol soft information is output, and then the symbol soft information The log likelihood ratio {Λ ₂ [b _i ]} of the information is sent to the second subtractor; (7) In the second subtractor, the log-likelihood ratio {Λ ₂ [b _i ]} of the symbol soft information output by the channel decoder is subtracted from the prior information {λ ₁ [b _i ]} to obtain the extrinsic information { λ ₂ [b _i ]}; after the external information {λ ₂ [b _i ]} is sent to the local interleaver for interleaving, it outputs a new bit likelihood ratio, that is, the new {λ ₂ [b _π(i) ]}; (8) The new bit likelihood ratio {λ ₂ [b _π(i) ]} is sent to the symbol probability calculator to obtain new prior information P(s); the new prior information P(s) is Send it to the quasi-maximum a posteriori probability detector to update the prior information; at the same time, in the process of sequentially executing the above steps (3) to (8) once, complete an iterative operation; (9) cyclically execute the above steps (3)-(8), that is, after multiple iterations, stop decoding; (10) After the iteration ends, the hard-decision binary data {Λ ₁ [a ₁ ]} output by the channel decoder is sent to the sink, and the iterative receiver ends its detection and decoding operations. 9. The working method according to claim 8, characterized in that: the number of loop iterative operations performed in the operation step of the iterative receiver is at least one, and the number of iterations depends on the time delay and hardware complexity of detection and decoding degree of different requirements.

Images