JP2007300512A - System and method for radio transmission - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve characteristics of a QRM-MLD method with which a calculation amount in a maximum likelihood judgment method, though it is an ideal detection method, is reduced in an MIMO transmission system. <P>SOLUTION: In a receiver, after rearranging channel matrixes according to receiving quality, an extended channel matrix extended by information on receiver noise power and the number of transmitting antennas is calculated, a virtual receiving signal vector is calculated from a unitary matrix and an upper triangle matrix to be acquired by a matrix decomposing operation of the extended channel matrix, combination of candidates for transmitting signals is arranged on tree structure consisting of hierarchies for the number of transmitting antennas and virtually having the optional number of branchings in each hierarchy, accumulation metrics corresponding to reliability of candidates for receiving signals in each hierarchy are calculated, a plurality of transmitting signals are separated/detected based on the accumulated metrics remained to the last stage. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a wireless transmission system and a wireless transmission method used for next-generation wireless communication (advanced wireless technology) represented by MIMO (Multiple-Input Multiple-Output).

Multiple input multiple output (MIMO) transmission schemes that use multiple antenna elements for both transmission and reception are being studied as a technology for further expanding transmission capacity without increasing the occupied frequency bandwidth in wireless communications. It is active.
These multiple antenna elements used in the MIMO transmission method include wireless transmission such as a method of spatially separating individual antennas, a method of using different combinations of polarizations (for example, vertical polarization and horizontal polarization), etc. Any method may be used as long as the multi-input multi-output condition is satisfied.

FIG. 5 shows an outline of a wireless communication system using the MIMO transmission system, and FIG. 6 shows an outline of a receiver in the MIMO transmission system. As shown in FIG. 5, in the MIMO transmission method, different signals can be transmitted from a plurality ( _Nt ) of transmission antennas at the same time and the same frequency. A plurality of transmission signals transmitted from the plurality of transmission antennas are received by a plurality ( _Nr ) of reception antennas.

Further, in the receiver, as shown in FIG. 6, in order to separate a plurality of transmission signals, in the channel estimation unit, from each reception antenna in a pilot signal section where a known pilot signal is transmitted, from each transmission antenna to each reception antenna Channel estimation is performed by obtaining a channel impulse response value or a channel estimation value. Then, the signal detection unit 22 detects and separates each transmission signal based on the channel estimation result in the channel estimation unit 21.
The multiple signal (multi-stream) transmission method using the MIMO transmission method is mainly classified into two types according to the transmission method performed in the transmitter of FIG. The first method is Space Division Multiplexing (SDM) that does not require channel information on the transmission side, and the second method is an eigenmode that requires channel information (channel information) on the transmission side. Transmission (Eigenmode Transmission; in some literature, it may be called SVD-MIMO, E-SDM, etc.).
Known reception methods for MIMO transmission receivers include the Zero-Forcing (ZF) method and the Minimum Mean Square Error (MMSE) method, which use simple linear signal processing. There is a Maximum Likelihood Detection (MLD) method classified as a method using nonlinear signal processing as a method for obtaining excellent transmission characteristics (in the following, an example in which SDM is used unless otherwise specified) Explained in).

  This method calculates a metric (usually using a square Euclidean distance as the metric) for all possible combinations of multiple transmitted and received signals transmitted from multiple transmit antennas, and provides a minimum metric. The combination is selected. According to this method, a plurality of signals can be separated with high accuracy, but there is a problem that the calculation load required for signal separation increases due to the amount of calculation of the square Euclidean distance.

For example, four transmission signals (s ₁ , s ₂ , s ₃ , s ₄ ) are transmitted from four transmission antennas (N _t = 4), respectively, with a modulation scheme of 16QAM (modulation multilevel number _Mary = 16). Suppose. In this case, since one transmission signal is mapped to one of _Mary signal points, the total number of combinations of transmission signal candidates included in the reception signal is _Mary ^Nt = 16 ⁴ = 65536. (That is, in the MLD method, the amount of calculation in signal separation increases exponentially as the number of modulation multi-levels and the number of transmission streams increase.)

  Therefore, in order to select the most probable transmission signal combination by the MLD method, it is necessary to calculate the square Euclidean distance for all of these combinations, which requires a very large calculation capacity. For this reason, it is not practical to use the MLD method for a wireless terminal (particularly a mobile terminal) because of its extremely large calculation burden, which increases power consumption and prevents downsizing of the wireless terminal.

Therefore, in recent years, a method called QRM-MLD method has been proposed as a signal separation method improved from the MLD method. The QRM-MLD method uses QR decomposition of the MIMO channel matrix and the M algorithm to narrow down the combinations of transmission signal candidates that calculate the square Euclidean distance in the MLD method, while significantly reducing the amount of computation compared to the MLD method. It has been reported that characteristic deterioration can be minimized (see Non-Patent Document 1 for the QRM-MLD method).
H. Kawai, K. Higuchi, N. Maeda, M. Sawahashi, T. Ito, Y. Kakura, A. Ushirokawa, and H. Seki, “Likelihood function for QRM-MLD suitable for soft-decision turbo decoding and its performance for OFCDM MIMO Multiplexing in multipath fading channel, ”IEICE Trans. Commun. vol.E88-B, no.1, pp.47-57, Jan. 2005. A method of further reducing the amount of calculation using a metric with a small calculation load different from the Euclidean distance has been studied (Non-patent Document 2). Kenichi Higuchi, Hiroyuki Kawai, Noriyuki Maeda, Mamoru Sawahashi, “Adaptive Survival Symbol Replica Candidate Selection Method Using Reliability Information in OFCDM MIMO Multiplexing Using QRM-MLD,” IEICE Technical Report (IEICE Technical Report), RCS2004 -69, May 2005. Takeo Ogane, “Basics and Elemental Technologies of MIMO Systems” Takeo Ogane, IEICE Antenna and Propagation Design and Analysis Workshop (29/30) Text, Nov./Dec. 2004. Masanori Seki, Toshiaki Koike, Eiichi Murata, Susumu Yoshida, Junmichi Araki, “FPGA implementation and performance evaluation of real-time processing MIMO-MLD using improved metrics,” IEICE Technical Report, RCS2004-292, Jan. 2005 . Hiroyuki Kawai, Kenichi Higuchi, Mamoru Sawahashi, Takumi Ito, Yoshikazu Kakura, Akihisa Gokawa, Hiroyuki Seki, “Composition of QR Decomposition-Reduced QR Decomposition-MLD Using Pilot Channel Estimation and Ranking in OFCDM MIMO Multiplexing” , RCS2003-312, Mar. 2004. GD Golden, GJ Foschini, RA Valenzuela, and PW Wolniansky, “Detection algorithm and initial laboratory results using V-BLAST space-time communication architecture,” IEE Electronics Letters, vol.35, no.1, pp.14-16, Jan 1999. H. Matsuda, K. Hojyo, T. Ohtsuki, and T. Kaneko, “Signal Detection Scheme Combining MMSE V-BLAST and Variable K-best Algorithms Based on Minimum Branch Metric,” Proc. IEEE VTC2005-Fall, vol.1, pp.19-23, Dallas, Texas, USA, Sept. 2005. JP 2005-328311, “Noise power estimation device, noise power estimation and signal detection device”

  However, since the orthogonalization of the MIMO channel using QR decomposition in the conventional QRM-MLD method described above is realized by the Zero-Forcing (ZF) standard, the MIMO channel matrix becomes smaller as the search range of transmission signal candidates is narrowed. The characteristics deteriorate due to the effects of noise enhancement and judgment error propagation caused by an adverse condition.

  Therefore, the present invention solves the above problems, and in the QRM-MLD method that reduces the amount of calculation while suppressing deterioration of signal detection accuracy in the MIMO transmission method, orthogonality of MIMO channels using QR decomposition A wireless transmission system and a wireless transmission method that can improve the characteristics of the conventional QRM-MLD method using the extended QRM-MLD method that applies the QR decomposition of the Minimum Mean Square Error (MMSE) standard described later. The issue is to provide.

  In order to solve the above-described problem, the present invention provides a plurality of signals from a plurality of antennas at the same time at the same carrier frequency between a transmission device having a plurality of transmission antennas and a reception device having a plurality of reception antennas. A wireless transmission system for transmitting, wherein a transmission device transmits a plurality of signals including a known pilot signal at the reception device, and the reception device is a signal including a plurality of signals transmitted via a radio propagation path as components. And signal separation means for estimating and separating the individual transmitted signals.

In the present invention, the signal separation means is
Noise power acquisition means for acquiring the noise power of the receiver for each receiving antenna;
A transmission signal candidate generating means for generating a plurality of transmission signal candidates according to the modulation multi-value number based on the modulation scheme determined between the transmission device and the reception device;
Channel information estimation means for estimating a channel matrix composed of channel information between the transmitting and receiving antennas based on a received signal including a known pilot signal;
The channel matrix estimated by the channel information estimating means, the expanded channel matrix configuring means for configuring the expanded channel matrix expanded based on the noise power information acquired by the noise power acquiring means and the number of transmission antennas of the transmitting device;
A channel matrix decomposition means for decomposing the extended channel matrix into a unitary matrix and an upper triangular matrix based on the extended channel matrix, and obtaining the matrix;
Virtual received signal calculation means for calculating a virtual received signal vector by multiplying a received signal vector configured based on the received signal of each receiving antenna by a complex conjugate transpose of a unitary matrix;
A reception signal candidate generation unit that generates a plurality of reception signal candidates for each element of the virtual reception vector by multiplying the transmission signal candidate from the transmission signal candidate generation unit by an upper triangular matrix element;
A determination / separation unit that determines and separates a transmission signal using a metric for each transmission signal candidate generated based on information on the Euclidean distance for each reception antenna calculated from the virtual reception vector and a plurality of reception signal candidates. With.

  In particular, the determination / separation means includes a plurality of stages corresponding to the number of transmission antennas used for transmitting a plurality of transmission signals from the transmitter, and each stage uses metric information for each transmission signal. By sequentially narrowing down candidates and selecting a plurality of surviving transmission signal candidates, a plurality of transmitted signals are determined and separated.

In the wireless transmission method of the present invention, the transmitting device transmits the plurality of signals including a known pilot signal at the receiving device, and the receiving device transmits the plurality of transmitted signals via a radio propagation path. Receiving a signal as a component (1);
In the transmission device, the receiver noise power acquisition unit acquires the receiver noise power for each reception antenna, and the transmission signal candidate generation unit performs modulation based on the modulation scheme determined between the transmission device and the reception device. Generating a plurality of transmission signal candidates according to the multi-value number (2);
In the transmission device, based on the received signal including the known pilot signal by the channel information estimation means, estimating a channel matrix composed of channel information between the transmitting and receiving antennas (3),
In the transmission apparatus, the channel matrix estimated by the channel information estimation means is changed from the information of the noise power acquired by the noise power acquisition means and the number of antennas of the transmission apparatus to the extended channel matrix expanded by the extended channel matrix configuration means. Configuring step (4);
In the transmission device, based on the extended channel matrix, the channel matrix decomposition means decomposes the extended channel matrix into a unitary matrix and an upper triangular matrix and obtains the step (5);
In the transmission device, a virtual received signal vector is calculated by multiplying the received signal vector configured based on the received signal of each receiving antenna by the complex conjugate transpose of the unitary matrix by the virtual received signal calculating means;
In the transmission device, the reception signal candidate generation unit multiplies the transmission signal candidate generated by the transmission signal candidate generation unit by the element of the upper triangular matrix, thereby obtaining a plurality of reception signal candidates for each element of the virtual reception vector. Step (6) to generate each,
In the transmission apparatus, each transmission generated based on the information of the Euclidean distance for each reception antenna calculated from the virtual reception vector and the plurality of reception signal candidates by the determination separation unit including a plurality of stages corresponding to the number of transmission antennas. Using the metrics for signal candidates, in each stage, narrow down candidates sequentially using the information of the metrics for individual transmission signals, and select a plurality of surviving transmission signal candidates. Determining and estimating and separating the individual transmitted signals (7).

  According to the present invention, in the receiving apparatus, the MIMO channel is orthogonalized according to the MMSE standard, thereby converting the simultaneous estimation problem in the maximum likelihood determination into an estimation problem that is hierarchized by the number of stages corresponding to the number of transmission signals. When the received signals are estimated and separated, the selection of the transmission signal candidates is selected based on the metrics for the transmission signal candidates at each stage of the determination / separation means, and the candidates that have been narrowed and survived are further narrowed down after the next stage. In this way, the effects of noise enhancement and decision error propagation are minimized, the degradation of reception characteristics is minimized, and the search range of transmission signal candidates is narrowed to greatly increase the amount of computation processing. It can be reduced and the processing speed can be improved.

  In the above invention, the determination separation unit sets the initial cumulative metric to 0, and in each stage, the number of stages executed for the previous stage cumulative metric that is the cumulative metric for each combination of transmission signal candidates selected in the previous stage. For the second corresponding transmission signal, calculate the metric corresponding to the transmission signal candidates corresponding to the number of modulation multi-values, add the previous stage cumulative metric to calculate the cumulative metric, and execute the stage from the cumulative metric information. A plurality of transmission signal candidate combinations for transmission signals of minutes are selected, narrowed down, and the transmission signals are determined and separated from information on the cumulative metric for the selected combinations of surviving transmission signal candidates that are continuously selected until the final stage. It is preferable.

  In the above invention, the reception quality estimation means for estimating the reception quality of each transmission signal in the reception apparatus based on the reception signal including the known pilot signal, and the channel according to the reception quality acquired by the reception quality acquisition means Sorting means for rearranging the matrix and calculating a sorted channel matrix, the channel matrix decomposing means, based on the noise power information obtained by the noise power obtaining means and the number of transmission antennas Preferably, the channel matrix after sorting is regarded as the channel matrix and expanded to form an extended channel matrix, and the expanded channel matrix is decomposed into a unitary matrix and an upper triangular matrix, and channel matrix decomposition is performed.

  Further, in the above invention, the sorting means uses a desired wave power-to-interference wave power and noise power ratio (received SINR) on the receiving side as an index representing the reception quality of the plurality of transmission signals, and each column of the channel matrix It is preferable to calculate the sorted channel matrix by rearranging in order from the smallest received SINR of each transmission signal.

  As described above, according to the present invention, transmission is performed in comparison with the conventional QRM-MLD method by the extended QRM-MLD method that performs orthogonalization of the MIMO channel using QR decomposition according to the MMSE (Minimum Mean Square Error) standard. The characteristics can be improved.

(Outline of wireless transmission method)
An embodiment of the present invention will be described. First, the outline of the wireless transmission method of the present invention will be described. In this embodiment, a single user (the number of users is 1), flat fading in which the influence of delayed waves can be ignored, the average transmission power from each transmission antenna is all equal to P _s , and the receiver noise power at each reception antenna is P _n A case where all are equal to each other will be described as an example.

Figure 1 shows the system model of a wireless transmission system using MIMO transmission. The number of transmit antennas is N _t , the number of receive antennas is N _r , the channel matrix from MIMO transmitter 1 to MIMO receiver 2 is H, the transmit signal vector is s (t), and the receiver noise vector is n (t) And At this time, the received signal vector x (t) in the equivalent low-frequency system expression is expressed by the following equation.

However, when E {·} is an ensemble average, the following equation is established.
Before describing the extended QRM-MLD method, which is an improved QRM-MLD method in the embodiment of the present invention, an overview of a MIMO transmission method using a conventional spatial filtering technique that is closely related to the present invention will be described. This will be described (for the MIMO transmission method using spatial filtering, see Non-Patent Document 3, for example).
The output signal vector S _ZF (t) detected by the spatial filtering based on the ZF standard (ZF method) is expressed by the following equation using the spatial filter (ZF antenna weight) W _ZF of the ZF standard (non-patent document) 3)
Where the superscript + of the matrix represents the Moore-Penrose general inverse of the matrix.
A general matrix is represented by a product of a unitary matrix and an upper triangular matrix by QR decomposition. That is, when the above-described QR decomposition is performed on the channel matrix H, the unit matrix Q and the upper triangular matrix R are used to express the following equation.
Using the relational expression of Expression (5), W _ZF can be expressed by the following expression using the unitary matrix Q and the upper triangular matrix R described above.
By the way, in the ZF method, when the channel matrix is ill-conditioned, the norm of H ⁺ becomes large. Therefore, there is a problem of noise enhancement that emphasizes noise included in the received signal. Therefore, in order to suppress the influence of this noise enhancement, there is a spatial filtering (MMSE method) based on the MMSE standard. The output signal vector S _MMSE (t) when detected by the MMSE method is expressed by the following equation using a spatial filter (MMSE antenna weight) W _MMSE of the MMSE standard (see Non-Patent Document 3).
When the average gain of each element of the channel matrix is 1, the reciprocal P _s / P _n of P _n / P _s represents the average received SNR (Signal-to-Noise-power-Ratio). Therefore, if the average gain of the propagation path is known instead of the noise power, it is not necessary to directly estimate the noise power, and if the average received SNR is known, the MMSE weight W _MMSE can be calculated.
Here, the dimension-expanded virtual channel matrix H ′ and the dimension-expanded virtual received signal vector x ′ (t) are respectively defined as follows.
Here, I _{N × N} represents a unit matrix of N rows and N columns, and 0 _{K × L} represents a zero matrix of K rows and L columns (a zero vector when either K or L is 1). From the equation (8), the following two equations are obtained.
From the equations (7), (9) and (10), the following equation is obtained.
From Equation (4) and Equation (11), the MMSE method in MIMO transmission represented by the system model of Equation (1) is that the channel matrix is a channel matrix H ′ whose dimension is expanded in Equation (8), and the received signal vector is It is equivalent to detecting using the ZF method in MIMO transmission represented by the virtual system model (virtual system model) that is the received signal vector x ′ (t) whose dimension is expanded in Equation (8). Show.
Further, when QR decomposition is performed on the dimension-expanded channel matrix and the unitary matrix Q ′ and the upper triangular matrix R ′ are used, the following expression is obtained.
Based on this relationship, the MMSE antenna weight W _MMSE can be expressed by the following relational expression using the unitary matrix Q ′ and the upper triangular matrix R ′.
From equation (14), signal processing based on the MMSE norm in the system modeled by equation (1) is equivalent to performing signal processing based on the ZF norm considering the virtual system model of equation (12). I know that there is. Therefore, for the sake of simplification of explanation, QR decomposition that does not extend the channel matrix as shown in Equation (5) is called `` ZF norm QR decomposition '', and QR decomposition that extends the channel matrix as shown in Equation (13). Is referred to as “MMSE norm QR decomposition”.

In the conventional QRM-MLD method, first, the received SINR (Signal-to-Interference-and-Noise-power-Ratio) of each transmission signal is determined from the estimation result of the channel matrix using the received signal including a known pilot signal. The transmission signals are ranked based on the information, and each column of the channel matrix is sorted. By performing QR decomposition on the sorted channel matrix and multiplying the vector of the received signal by the complex conjugate transpose Q ^H of the unitary matrix Q, the simultaneous estimation problem in MLD is hierarchized by the number of stages equal to the number of transmitted signals. Convert to a presumed problem. Then, by applying an M algorithm composed of stages corresponding to the number of transmission antennas and reducing the number of transmission signal candidates based on the metric calculated in each stage, the amount of decoding computation in MLD has been reduced (Non-Patent Document 2). reference).
However, the orthogonalization of the MIMO channel using QR decomposition in the conventional QRM-MLD method uses the above-mentioned “ZF-standard QR decomposition”, so as the number of surviving transmission signal candidates is reduced at each stage of the M algorithm. There was a problem that transmission characteristics deteriorated greatly.

  Therefore, in this embodiment, when performing orthogonalization of MIMO channels using QR decomposition, instead of the “ZF norm QR decomposition” used in the conventional QRM-MLD method, the “MMSE norm QR decomposition” described above is used. Is used. As a result, it is possible to suppress the problem of deterioration in transmission characteristics when the number of surviving transmission signal candidates in each stage of the M algorithm, which is generated in the conventional QRM-MLD method, is reduced.

  Specific embodiments of the present invention will be described below. Fig. 2 shows an overview of the signal detection algorithm in the receiver used in the extended QRM-MLD method.

(Virtual system model used in the extended QRM-MLD method)
First, similarly to the conventional QRM-MLD method, the reception quality estimation unit 232 estimates the reception SINR for each transmission signal from the channel estimation result using a known pilot signal or the like in the channel estimation unit 231 (specificity of the reception SINR). (For example, see Non-Patent Document 5). In the first half of processing in extended channel matrix configuration section 242, each column of channel matrix H is sorted (reordered) based on the average received SINR size (usually in ascending order of average received SINR).
At this time, the following equation holds.
As shown in Equation (15), since each column of the channel matrix is sorted based on the average received SINR magnitude of each transmission signal, each element of the transmission signal vector is similarly set based on the average received SINR magnitude. It is necessary to think by sorting.

In this embodiment, in order to perform orthogonalization of the MIMO channel by “QR decomposition of MMSE norm”, the channel matrix after sorting is expanded and the virtual received signal vector generation unit in the second half processing unit of the extended channel matrix configuration unit 242 The received signal vector x (t) is expanded in 25 first-half processing units. Specifically, transmission power P _s and receiver noise power P _n are known on the receiving side, and the sorted channel matrix and received signal vector x (t) in equation (15) are replaced according to the following equations, respectively.
Accordingly, a virtual system model for implementing the present invention is given by the following equation.
Hereinafter, x ′ (t) is referred to as a virtual received signal vector, and n ′ (t) is referred to as a virtual receiver noise vector. However, the virtual receiver noise vector n ′ (t) is expressed by the following equation.

When the average gain of each element of the sorted channel matrix is 1, the reciprocal P _s / P _n of P _n / P _s represents the average received SNR. Therefore, if the average gain of the radio channel (channel) is known instead of the receiver noise power, the receiver noise power P _n is directly estimated, and it is not necessary to use the average transmission power P _s per element. If the SNR is known, the same processing can be performed.

(MIMO channel orthogonalization using QR decomposition based on MMSE standard)
Next, the matrix decomposition unit 243 performs QR decomposition of the MMSE standard by performing QR decomposition on the extended channel matrix.

Here, the virtual received signal vector generation unit 25 multiplies the virtual received signal vector x ′ (t) by the complex conjugate transpose of the unitary matrix, and nulling x ′ (t), the received signal y ′ (t) after nulling Is obtained.
Thereby, the equation (20) is expressed by the following equation.
Since the row in which the components of the upper triangular matrix of Equation (21) are all 0 can be ignored, the following equation should be considered after all.

(Signal point candidate reduction type MLD using M algorithm)
Subsequent processing can be considered in the same manner as the conventional QRM-MLD method. That is, in the signal point candidate reduction type MLD unit 26 based on the M algorithm, transmission is performed according to the metric calculated in each stage by the M algorithm composed of stages corresponding to the number of transmission antennas (N _t ) based on the equation (22). A transmission signal is determined while reducing signal candidates.

As a metric used, it is common to use a square metric with a square Euclidean distance between a received signal and a received signal candidate as a metric, so this method is used here as an example of a metric generation method. Other metric generation methods can also be used. A representative example of a metric other than the square metric is a Manhuntan metric that does not require multiplication (for example, Non-Patent Document 4).

  Here, processing in the signal point candidate reduction type MLD unit 26 will be described. However, since the algorithm related to this processing is the same as that of the conventional QRM-MLD method, only the outline will be described here, and refer to Non-Patent Document 5 for details.

In the first stage of the M algorithm, the received signal candidate generation unit 261 selects a candidate for the _Nt- th element of the transmission signal vector after sorting in Equation (15), and therefore is determined by the modulation scheme determined between transmission and reception. _Mary reception signal candidates corresponding to all transmission signal candidates c (m) (1 ≦ m ≦ M _ary , where _Mary is the modulation multi-level number) are generated. Then, the cumulative metric calculation unit 271, a metric based on the square Euclidean distance for the first N _t th row element in which y _'Nt (t) and the received signal candidate virtual received signal vector calculated respectively, the cumulative metric (initial cumulative The metric is added to 0) and the value of the cumulative metric is updated. Next, in the surviving signal candidate selection unit 281, the cumulative metrics e ₁ (m) (1 ≦ m ≦ M _ary ) obtained for _Mary received signal candidates are compared, and S _{1 increases in} ascending order of e ₁ (m). ₁ (≦ M _ary ) corresponding transmission signal candidates are selected as surviving transmission signal candidates in the first stage, and S ₁ accumulated metrics for each surviving transmission signal candidate are held.

In the second stage, the received signal candidate generation unit 262 performs all combinations of two transmission signal candidates (that is, S for the N _t row element of the transmission vector after sorting in Expression (15) selected in the first stage). _All surviving transmission signal candidates and all combinations of M _ary transmission candidates for the N _t -1 row elements of the transmission signal vector after sorting in Equation (15) in the second stage (total S ₁ M _ary A vector of received signal candidates is generated by the combination)). Then, the cumulative metric calculation section 272, M _ary for the first N _t -1 line element y _'Nt-1 (t) and N _t -1 line elements of the transmit signal vector after sorting virtual reception signal vector Metrics based on square Euclidean distances with received signal candidates for each transmission candidate are calculated and added to the cumulative metric of the previous stage (first stage) to update the value of the cumulative metric. Further, in the surviving signal candidate selection unit 282, a combination of two transmission signal candidates corresponding to S ₂ (≦ S1Mary) with a small cumulative metric obtained is selected as a surviving transmission signal candidate in the second stage, and the corresponding holding the S ₂ pieces of cumulative metrics.

By repeating the same operation after the third stage, the cumulative metric corresponding to the combination of S _{3 Mary} surviving transmission signal candidates is output in the final N _t stage. However, since the vector represented by the combination of surviving transmission signal candidates output here is output in a sorted state based on the reception SINR of each transmission signal estimated by the reception quality estimation unit 232, the transmission signal determination unit Original transmission signal vector in 29 first half processing part
Rearrangement (desorting) to the original order is performed so as to correspond to.

Finally, in the latter half processing unit of the transmission signal determination unit 29, each transmission signal is detected based on the cumulative metric for the rearranged surviving transmission signal candidate combinations. When error correction is not performed or when hard decision decoding is performed in error correction, the combination of surviving transmission signal candidates with the smallest cumulative metric in the final ( _Nt ) stage is selected as the most probable transmission signal combination. That's fine.

  On the other hand, when soft-decision decoding is performed in error correction, in order to take advantage of the characteristic improvement effect of soft-decision decoding, a search is also made for candidates in a relatively wide range other than the surviving transmission signal candidates that minimize the cumulative metric in the final stage. In addition, it is desirable to select the most probable combination of transmission signals (see Non-Patent Document 3).

(Configuration of wireless transmission system)
Next, the configuration of the wireless transmission system according to the present embodiment will be described. By operating this wireless transmission system, the above-described wireless transmission method of the present invention can be implemented.

As shown in FIG. 1, a wireless transmission system according to this embodiment includes a MIMO transmitter 1 having N _t antennas #. 1 to # N _t, MIMO with N _r antennas # 1~ # N _r A MIMO wireless transmission scheme is used in which a signal composed of a plurality of transmission signals is transmitted to and received from the receiver 2 from N _t antennas # 1 to #N _{t at the} same time at the same carrier frequency.

  The MIMO transmitter 1 is installed in, for example, a radio base station, and the MIMO receiver 2 is installed in a mobile station or the like. In this embodiment, unidirectional communication is described as an example, but the present invention also applies to bidirectional communication in which both a radio base station and a mobile station are provided with both a MIMO transmitter 1 and a MIMO receiver 2. Of course it is available.

In the figure, the number of transmission antennas is N _t , the channel matrix from MIMO transmitter 1 to MIMO receiver 2 is H, the transmission signal vector is s (t), and the receiver noise vector of the receiver is n (t). To do. The MIMO transmitter 1 transmits a signal composed of one or a plurality of transmission signals using a plurality of transmission antennas.

  For these transmission signals, in addition to the user data signal that is information that the user wants to transmit, in order to estimate the channel matrix between the transmission and reception apparatuses, a known pilot signal between the transmission and reception apparatuses is time-division multiplexed with respect to the user data signal or Transmission is performed using a multiplexing method such as code division multiplexing. In addition, when it is necessary to exchange control information between transmission and reception, a control signal for exchanging control information between transmission and reception may be multiplexed with these signals and transmitted. Each transmission antenna is provided with a frequency converter (not shown) that is an up-converter that converts a baseband signal into an RF band signal.

On the other hand, the MIMO receiver 2 includes N _r receiving antennas # 1 to #N _r , and each receiving antenna has a frequency converter (such as a down converter) that converts an RF band signal into a baseband signal ( (Not shown) is provided. In the present embodiment, processing at the baseband level is basically used, but equivalent processing may be performed in the IF band or RF band.

An outline of the transmission signal detection algorithm in the MIMO receiver 2 is shown in FIG. In FIG. 2, for the sake of simplicity, the case where both the number of transmission antennas _Nt and the number of reception antennas is four is illustrated, but both may be values other than four or different values.

As shown in FIG. 2, MIMO receiver 2 uses channel estimation section 231 that estimates channel matrix H using the received signal in the pilot signal section, and estimates the reception quality of each transmission signal using the channel estimation result A reception quality estimation unit 232, a receiver noise power acquisition unit 233 for acquiring receiver noise power, a channel sort unit 241 for rearranging (sorting) the channel matrix using the channel estimation result and the reception quality estimation result, An expanded channel matrix configuration unit 242 that configures an extended channel matrix from the sorted channel matrix, receiver noise power, and the number of transmission antennas N _t , a matrix decomposition unit 243 that decomposes the extended channel matrix into a unitary matrix and an upper triangular matrix, A virtual received signal is generated by generating an extended received signal vector x ′ (t) from the received signal vector x (t) and the number of transmitting antennas N _t and multiplying the complex conjugate transpose of the unitary matrix. From the virtual reception signal vector generation unit 25 that generates the vector y ′ (t), the virtual reception signal vector y ′ (t), the upper triangular matrix, and the reception quality estimation result, while reducing the transmission signal candidates based on the M algorithm A signal point candidate reduction type MLD unit 26 for reducing the amount of calculation of MLD and detecting a transmission signal suboptimally is provided.

The signal point candidate reduction type MLD unit 26 uses a metric for each transmission signal candidate generated based on the information of the Euclidean distance for each reception antenna calculated from the virtual reception signal vector and a combination of a plurality of reception signal candidates. , A module that determines and separates a plurality of transmission signals, and includes a plurality of stages corresponding to the number of transmission signals transmitted from the MIMO transmitter 1 (that is, the number of transmission antennas N _t ). In the final stage, a plurality of transmission signals are determined and separated while selecting a plurality of combinations of surviving transmission signal candidates by sequentially narrowing down candidates for the transmission signals using the metric information.

Specifically, this signal point candidate reduction type MLD unit 26 is a module that reduces transmission signal candidates used in MLD by an M algorithm composed of stages corresponding to the number of transmission antennas, and the virtual received signal vector y ′ (t) For each element, received signal candidate generation sections 261 to 261 generate a plurality of received signal candidates in each stage based on transmission signal candidates generated based on the upper triangular matrix and the modulation scheme information determined between transmission and reception. Based on 26N _t , the received signal candidate at each stage and the virtual received signal vector y ′ (t), the metric representing the reliability of each transmitted signal candidate at each stage is added to the cumulative metric at the previous stage (however, the initial cumulative metric 0)), the cumulative metric calculation units 271 to 27N that calculate the cumulative metric of the corresponding stage corresponding to the reliability of each combination of received signal candidates _t and surviving signal candidate selecting sections 281 to 283 for selecting a plurality of combinations of transmission signal candidates having a small cumulative metric (that is, having high reliability) as surviving transmission signal candidates in each stage except the final stage; And a transmission signal determination unit 29 that finally detects a sub-optimal transmission signal based on the accumulated metric, the combination of corresponding transmission signal candidates, and the reception quality information of each transmission signal.

  The channel estimation unit 231 is a module that estimates a channel matrix configured by channel information between transmitting and receiving antennas based on a received signal including a known pilot signal, and uses the received signal in the pilot signal section to determine the channel matrix. H is estimated, and the estimated channel matrix H is output to the channel sort unit 241 and the reception quality estimation unit 232.

  The reception quality estimation unit 232 is a module that estimates the reception quality of each of the plurality of transmission signals in the reception device based on the reception signal including the known pilot signal, and the channel estimation result input from the channel estimation unit 231 Is used to estimate the reception quality of each transmission signal, and the estimation result is output to the channel sort unit 241 and the transmission signal determination unit 29.

  The receiver noise power acquisition unit 233 is a module that acquires the noise power of the receiver for each reception antenna, and the acquired receiver noise is input to the extended channel matrix configuration unit 242.

  The channel sorting unit 241 uses the channel estimation result by the channel estimation unit 231 and the reception quality estimation result by the reception quality estimation unit 232 to perform rearrangement (sort) for each column of the channel matrix estimated by the channel estimation unit 231. This is a module for calculating a sorted channel matrix, and the sorted channel matrix is input to the extended channel matrix configuration unit 242.

The extended channel matrix configuration unit 242 displays the sorted channel matrix sorted by the channel sort unit 241, information on the receiver noise power P _n acquired by the receiver noise power acquisition unit 233, and P of the transmission power of the transmitter. _s (known on the receiver side, not shown) and a module that forms an extended channel matrix extended based on the number of antennas of the transmission apparatus, and this extended channel matrix is input to the matrix decomposition unit 243 .

When the average gain of each element of the channel matrix (after sorting) is 1, the reciprocal P _s / P _n of the noise power P _n / P _s represents the average received SNR. ) Can generally be obtained from the channel estimation result of the channel estimation unit 231. Therefore, it is not necessary to directly estimate the receiver noise power P _n and use the average transmission power P _s per element, It is also possible to construct an extended channel matrix by estimating the average received SNR.

The matrix decomposition unit 243 is a module that decomposes the extended channel matrix input from the extended channel configuration unit 242 and obtains a unitary matrix and an upper triangular matrix. The decomposed unitary matrix is input to the virtual received signal vector generation unit 25, and the upper triangular matrix is input to the reception signal candidate generation units 261 to 26N _t of each stage of the signal point candidate reduction type MLD unit 26.

The virtual received signal vector generating unit 25 calculates a virtual received signal vector y ′ (t) by multiplying the received signal vector configured based on the received signal of each receiving antenna by the complex conjugate transpose of the unitary matrix. In this virtual reception signal vector generation unit 25, an extended reception signal vector x ′ (t) is generated from the reception signal vector x (t) and the number of transmission antennas N _t , and this is input from the matrix decomposition unit 243 A virtual received signal vector y ′ (t) is generated by multiplying the complex conjugate transpose of the unitary matrix, and is output to the cumulative metric calculation units 271 to 27N _t of each stage of the signal point candidate reduction type MLD unit 26.

Received signal candidate generator 261～26N _t, at each stage of the M algorithm, to candidates of the transmission signal generated based on information on the modulation method determined between the transmitter, multiplying the elements of the upper triangular matrix Thus, a plurality of received signal candidates are generated for each element of the virtual received signal vector. Note that, in the present embodiment, the reception signal candidate generation units 261 to 26N _t have a function of generating transmission signal candidates based on the modulation scheme determined between the MIMO transmitter 1 and the MIMO receiver 2, For each element of the virtual received signal vector y ′ (t) input from the virtual received signal vector generating unit 25, each stage based on the upper triangular matrix input from the matrix decomposing unit 243 and the transmission signal candidates A plurality of received signal candidates are respectively generated in. The received signal candidates of each stage is input to the cumulative metric calculation unit 271～27N _t arranged in each stage.

In this embodiment, the cumulative metric calculators 271 to 27N _t set the initial cumulative metric to 0, and in each stage, the previous stage cumulative which is a cumulative metric for each combination of the selected surviving transmission signal candidates selected and narrowed in the previous stage. A module that calculates a metric corresponding to transmission signal candidates corresponding to the number of modulation multi-values for the transmission signal corresponding to the number of executed stages for the metric, and adds the respective metrics to the previous stage accumulated metric. It is.

The survival signal candidate selection units 281 to 283 are modules that select and narrow down a plurality of combinations of transmission signal candidates for the transmission signals for the stages executed by the cumulative metric calculation units 271 to 273 in each stage. Note that this survival signal candidate selection unit is arranged with the number of stages N _t −1 up to the stage before the final stage.

The transmission signal determination unit 29 obtains a combination of surviving transmission signal candidates that are continuously selected and narrowed down to the final stage from the cumulative metric calculation unit 27N _t of the final stage, and information on the cumulative metric for the final combination To determine and separate the transmission signal.

  FIG. 3 schematically shows a transmission signal determination method according to the present embodiment. As shown in the figure, a trellis (tree structure) having an arbitrary number of paths (branches) is virtually generated at each stage, consisting of stages (hierarchies) corresponding to the number of transmission signals (ie, the number of transmission antennas) A combination of transmission signal candidates is arranged on the trellis, and a search is performed while narrowing down transmission signal candidates for each stage according to the trellis.

  Then, the surviving signal candidate selection units 281 to 283 use the transmission signal candidates arranged in each stage according to the trellis, and according to the metric for each branch, an arbitrary number of paths (for example, (2 paths with small cumulative metrics, etc.) are selected and narrowed down, and the cumulative metrics are calculated by sequentially adding the surviving path metrics narrowed down from the top to the bottom of the trellis, and hard decision decoding is performed in error correction. When performing, a combination of transmission signal candidates having a minimum accumulated metric within a selected branch range is determined as a transmission signal.

Here, the maximum number of combinations of transmission signal candidates that are narrowed down and selected in each stage is the modulation multi-level number ^{(the number of stages)} . That is, when the number of transmission antennas N _t is 4, the number of transmission symbol candidates S _m narrowed down in the mth (<N _t ) stage of the trellis is expressed as (S ₁ , S ₂ , S ₃ ), for example, When the value M _ary is 16, when searching for all transmission symbol candidates (full search) without narrowing down, (S ₁ , S ₂ , S ₃ ) = (16, 16 ² , 16 ³ ) When the transmission symbol candidates are narrowed down to only one, (S ₁ , S ₂ , S ₃ ) = (1, 1, 1). In order to ensure detection accuracy is usually the S _m has to be set to the somewhat greater than one. Also, in FIG. 3, the number of branches at each stage is reduced, and in order to simplify the illustration, the number of transmission antennas N _t is 4, the number of modulation multilevels 4 (M _ary = 4), and the number of transmission symbol candidates is two. It should be noted that only the case of narrowing down (S ₁ , S ₂ , S ₃ ) = ( ₂ , ₂ , ₂ ) is illustrated.

(Wireless transmission method)
By operating the wireless transmission system having the above configuration, the wireless transmission method of the present invention can be implemented.

  First, when a signal is received by the receiving antenna, the channel estimation unit 231 estimates the channel matrix between the transmission antenna and the receiving antenna using the received signal in the pilot signal section and the like and inputs the channel matrix to the channel sorting unit 241. . Reception quality estimation section 232 estimates the reception SINR of each transmission signal based on the channel estimation result in channel estimation section 231 and inputs the result to channel sorting section 241 and transmission signal determination section 29 (using the channel estimation result). For example, see Non-Patent Document 2 for the received SINR estimation method). Further, the receiver noise power acquisition unit acquires the noise power of the receiver to which each receiving antenna is connected, and inputs the noise power to the extended channel matrix configuration unit 242. (Ref. 1).

  Channel sorting section 241 rearranges the columns of channel matrix H estimated based on the received SINR so as to increase the signal detection accuracy, generates a sorted channel matrix, and inputs it to extended channel matrix construction section 242.

The extended channel matrix configuration unit 242 regards the sorted channel matrix as a channel matrix, configures the extended channel matrix of Equation (17) from the receiver noise power P _n , the transmission power P _s , and the number of transmission antennas N _t , The data is input to the decomposition unit 243. When the average gain of each element of the sorted channel matrix is 1, P _s / P _n that is the transmission power to noise power ratio per element represents the average received SNR, and thus the radio propagation path ( Since the average gain of the channel) can generally be obtained by channel estimation of the channel estimation unit 231, the receiver noise power P _n is directly estimated, and it is not necessary to use the average transmission power P _s per element, and the average reception It is also possible to construct an extended channel matrix by estimating the SNR.

  The matrix decomposition unit 243 performs QR decomposition on the extended channel matrix to calculate a unitary matrix and an upper triangular matrix that satisfy Equation (19), and inputs each element of the unitary matrix to the virtual received signal vector generation unit 25. The elements of the upper triangular matrix are input to the signal point candidate reduction type MLD unit 26.

The virtual received signal vector generation unit 25 calculates an extended received signal vector x ′ (t) from the received signal vector and the number of transmitting antennas N _t , and the unitary matrix complex for the extended received signal vector x ′ (t). Multiply conjugate transpose to calculate the virtual received signal vector y '(t), and calculate the elements from the first row to the _Nt (number of transmit antennas) row of the calculated virtual received signal vector y' (t) Output to the signal point candidate reduction type MLD unit 26 in the subsequent stage.

The signal point candidate reduction type MLD unit 26 reduces the amount of calculation of MLD by the M algorithm including stages for the number of transmission antennas N _t .

First, in the received signal candidate generators 261 to 26N _t arranged in each stage of the signal point candidate reduction type MLD unit 26, the above-mentioned transmission signal candidates generated based on the modulation scheme information between transmission and reception are compared with the above. By multiplying the elements of the triangular matrix, a plurality of reception signal candidates are generated for each element of the virtual reception vector y ′ (t). Specifically, the reception signal candidate generation units 261 to 26N _t generate transmission signal candidates based on the modulation scheme determined between the MIMO transmitter 1 and the MIMO receiver 2, and generate virtual reception signal vectors. For the virtual received signal vector y ′ (t) input from the unit 25, a plurality of received signal candidates at each stage are generated based on the upper triangular matrix input from the matrix decomposition unit 243 and the transmission signal candidates To do. The received signal candidates of each stage is input to the cumulative metric calculation unit 271～27N _t arranged in each stage.

Cumulative metric calculation units 271 to 27N _t calculate metrics corresponding to transmission signal candidates corresponding to the number of modulation multi-levels for the transmission signal corresponding to the number of executed stages with respect to the previous stage cumulative metric. The cumulative metric is calculated by adding each to.

The surviving signal candidate selection units 281 to 283 of each stage select combinations of transmission signal candidates for transmission signals corresponding to the number of stages executed by the cumulative metric calculation units 271 to 273, and perform narrowing down sequentially. Then, the transmission signal determination unit 29 obtains the combinations of surviving transmission signal candidates that are continuously selected and narrowed down to the final stage from the cumulative metric calculation unit 27N _t of the final stage, and the cumulative metric for the final combination. From this information, the transmission signal is determined and separated.

(Action / Effect)
Next, a description will be given of a simulation of the effects of the wireless transmission system and the wireless transmission method according to the present embodiment described above. Table 1 shows the main simulation parameters.

In this simulation, the case of a single user is evaluated. The number of transmit antennas N _t 4, securing the number of reception antennas N _r 4. In addition, since the channel model is simple, a quasi-static flat Rayleigh channel is assumed, and an IID channel is assumed in which the spatial correlation of fading is uncorrelated in both transmission and reception. As a premise, the transmission power allocated to each transmission antenna is all the same, and the modulation scheme is fixed at 16QAM × 4 streams regardless of the average received SNR.

  Therefore, in this simulation, the transmission transmission rate of the entire transmission device is always equal regardless of the average reception SNR. Also, assuming that channel estimation and timing detection are ideally performed on the receiver side, and that the receiver noise power can be ideally estimated, the extended QRM-MLD method of the present invention is compared with the conventional QRM-MLD method. And evaluate.

  4 (a) and 4 (b) show average bit error rate characteristics of the conventional QRM-MLD method and the radio transmission method (extended QRM-MLD method) according to the present embodiment without error correction coding, respectively.

  Here, for comparison, an ideal MLD (hereinafter referred to as the Full MLD method) that performs a full search of transmission signal candidates and does not reduce the amount of calculation, and a signal with a high average received SINR in order from the ZF and MMSE norms. The V-BLAST (ZF) method (see Non-Patent Document 6) and the V-BLAST (MMSE) method (for example, see Non-Patent Document 7), which sequentially and repeatedly cancels interference from a stream with a high received SINR by spatial filtering, were used. The characteristics of each case are also shown together (V-BLAST: Vertical Bell Laboratories Layered Space-Time).

In the following, it is assumed that the number of surviving transmission signal candidates S _m selected in the mth (<N _t ) stage of the M algorithm in the conventional QRM-MLD method and the extended QRM-MLD method is selected at each stage. (S ₁ , S ₂ , S ₃ ). For example, when searching for all transmission signal candidates (full search), (S ₁ , S ₂ , S ₃ ) = (16, 16 ² , 16 ³ ), and in this case, the same characteristics as the Full MLD method are obtained. Become.

As shown in FIGS. 4 (a) and (b), comparing the results of each simulation, the number of surviving transmission signal candidates selected in each stage is narrowed down to one (S ₁ , S ₂ , S ₃ ) = (1,1,1), the conventional QRM-MLD method has almost the same characteristics as the V-BLAST (ZF) method, and the extended QRM-MLD method is slightly degraded compared to the V-BLAST (MMSE) method. However, compared to the conventional QRM-MLD method, it can be seen that the average received SNR with an average bit error rate of 10 ⁻³ is improved by about 2 dB.

Furthermore, when the transmission signal candidates are narrowed down to four at each stage (S ₁ , S ₂ , S ₃ ) = (4, 4, 4), when compared with the average received SNR where the average bit error rate is 10 ⁻³ In the conventional QRM-MLD method, the degradation is about 4 dB compared to the Full MLD method, but in the extended QRM-MLD method, the degradation is about 1 dB compared to the Full MLD method.

In addition, when the number of transmission signal candidates is narrowed down to 8 at each stage of the M algorithm (S ₁ , S ₂ , S ₃ ) = (8, 8, 8), average reception with an average bit error rate of 10 ⁻³ When comparing the SNR, the conventional QRM-MLD method degrades by about 2 dB compared to the full MLD method, but the extended QRM-MLD method has almost no degradation compared to the full MLD method, and the transmission characteristics are larger than the conventional QRM-MLD method. It turns out that it is improved. From this, the extended QRM-MLD method can approach the characteristics of the ideal Full MLD method with a search for transmission signal candidates in a smaller range than the conventional QRM-MLD method, confirming the effectiveness of this method. it can. In this simulation, the multiple ( _Nt ) signals transmitted from the transmitter were all evaluated by single user MIMO, which was transmitted to the same user. However, these multiple signals were sent to multiple different users. On the other hand, multi-user MIMO that is assigned and transmitted can be detected in the same manner by extracting a transmission signal for the user on the receiving side using the extended QRM-MLD method.

It is a block diagram which shows the system model of the radio | wireless transmission system using the MIMO transmission system which concerns on embodiment. It is explanatory drawing which shows the outline | summary of the transmission signal detection algorithm in the radio | wireless transmission method which concerns on embodiment. It is explanatory drawing which shows the outline | summary of the narrowing-down method of the transmission signal candidate which concerns on embodiment. It is a graph which shows the result of the simulation regarding the wireless transmission system and method which concern on embodiment. It is a block diagram which shows the outline | summary of a structure of the radio | wireless transmission system using a MIMO transmission system. It is a block diagram which shows typically the outline | summary of the receiver structure of a MIMO transmission system.

Explanation of symbols

1 ... MIMO transmitter
2 ... MIMO receiver
22 ... Signal detector
25. Virtual received signal vector generator
26… Signal point candidate reduction type MLD section
27N _t ... Cumulative metric calculator
29 ... Transmission signal judgment unit
231 ... Channel estimation section
232 ... Receive quality estimation unit
233 ... Receiver noise power acquisition unit
241… Channel sorting section
242 ... Extended channel matrix component
243 ... Matrix decomposition part
261-26N _t ... Received signal candidate generator
271 ～ 27N _t … Cumulative metric calculation part
281 to 283 ... Survival signal candidate selection section

Claims (8)

A wireless transmission system for transmitting a plurality of signals from the plurality of antennas at the same time at the same carrier frequency between a transmission device having a plurality of transmission antennas and a reception device having a plurality of reception antennas,
The transmitting device transmits the plurality of signals including pilot signals known by the receiving device, the receiving device receives a signal having the transmitted plurality of signals as components via a radio propagation path, Having signal separation means for estimating and separating individual transmitted signals;
The signal separating means includes
Receiver noise power acquisition means for acquiring the noise power of the receiver for each of the receiving antennas;
Transmission signal candidate generation means for generating a plurality of transmission signal candidates according to the modulation multi-value number, based on the modulation scheme determined between the transmission device and the reception device;
Channel information estimation means for estimating a channel matrix composed of channel information between the respective transmitting and receiving antennas based on the received signal including the known pilot signal;
An expanded channel matrix configuration unit configured to configure an expanded channel matrix that is expanded based on the noise power information acquired by the noise power acquisition unit and the number of transmission antennas of the transmission device, the channel matrix estimated by the channel information estimation unit When,
Channel matrix decomposition means for decomposing the extended channel matrix into a unitary matrix and an upper triangular matrix based on the extended channel matrix, and obtaining the matrix;
Virtual received signal calculation means for calculating a virtual received signal vector by multiplying the received signal vector configured based on the received signal of each receiving antenna by the complex conjugate transpose of the unitary matrix;
Reception signal candidate generation means for generating a plurality of reception signal candidates for each element of the virtual reception signal vector by multiplying the transmission signal candidates from the transmission signal candidate generation means by the elements of the upper triangular matrix; ,
Using the metric for each transmission signal candidate generated based on the reliability information of the transmission signal candidate corresponding to each reception signal candidate calculated from the virtual reception vector and the plurality of reception signal candidates, Determination separation means for determining and separating,
The determination / separation means includes a plurality of stages corresponding to the number of transmission antennas, and in each stage, candidates are sequentially narrowed down using the metric information for each transmission signal, and a plurality of surviving transmission signal candidates are selected. And determining and separating the plurality of transmitted signals. The determination separation means includes
The initial cumulative metric is set to 0, and in each stage, the multilevel modulation is performed for the transmission signal corresponding to the number of stages executed for the previous stage cumulative metric that is the cumulative metric for each combination of transmission signal candidates selected in the previous stage. Calculate a metric corresponding to several minutes of transmission signal candidates, and calculate a cumulative metric by adding to each of the previous stage cumulative metric,
From the information of the cumulative metric, select a plurality of combinations of transmission signal candidates for transmission signals for executed stages, and narrow down,
2. The radio transmission system according to claim 1, wherein a transmission signal is determined and separated from information of an accumulated metric for combinations of surviving transmission signal candidates that are continuously selected and narrowed down to a final stage. Reception quality estimation means for estimating the reception quality of each of the plurality of transmission signals in the reception apparatus based on the reception signal including the known pilot signal;
Sort means for rearranging the channel matrix according to the reception quality acquired by the reception quality acquisition means and calculating a sorted channel matrix;
With
The channel matrix decomposition means considers the sorted channel matrix as the channel matrix and expands the extended channel matrix based on the noise power information acquired by the noise power acquisition means and the number of transmission antennas. 3. The radio transmission system according to claim 1, wherein the channel matrix decomposition is configured to decompose and acquire the extended channel matrix into a unitary matrix and an upper triangular matrix. The sorting means uses, as an index indicating the reception quality of the plurality of transmission signals, a reception side desired wave power-to-interference wave power and a noise power ratio (reception SINR), and each column of the channel matrix is assigned to each transmission signal. 4. The wireless transmission system according to claim 3, wherein the sorted channel matrix is calculated by performing rearrangement in ascending order of reception SINR. A wireless transmission method for transmitting a plurality of signals from the plurality of antennas at the same time at the same carrier frequency between a transmission device having a plurality of transmission antennas and a reception device having a plurality of reception antennas,
A step of transmitting the plurality of signals including a known pilot signal by the receiving device from the transmitting device, and receiving a signal having the plurality of transmitted signals as components via a radio propagation path in the receiving device; (1) and
In the transmission apparatus, the receiver noise power estimation means estimates the noise power of the receiver for each reception antenna, and based on the modulation scheme determined between the transmission apparatus and the reception apparatus by the transmission signal candidate generation means. Step (2) for generating a plurality of transmission signal candidates according to the modulation multi-level number,
In the transmitting apparatus, based on a received signal including the known pilot signal by channel information estimating means, estimating a channel matrix composed of channel information between the transmitting and receiving antennas (3);
In the transmission device, the channel matrix estimated by the channel information estimation unit is expanded by the extended channel matrix configuration unit based on the noise power information acquired by the noise power acquisition unit and the number of transmission antennas of the transmission device. (4) constructing an extended channel matrix,
In the transmitter, based on the extended channel matrix, the channel matrix decomposition means decomposes the extended channel matrix into a unitary matrix and an upper triangular matrix, and obtains (5),
In the transmitting apparatus, a virtual received signal vector is calculated by multiplying a received signal vector configured based on a received signal of each receiving antenna by a complex conjugate transpose of the unitary matrix by a virtual received signal calculating means (6) )When,
In the transmission apparatus, a plurality of reception signal candidates are virtually received by multiplying the transmission signal candidates generated by the transmission signal candidate generation means by elements of the upper triangular matrix by the reception signal candidate generation means. Generating each element of the signal vector (7),
In the transmission apparatus, transmission signal candidates corresponding to each reception signal candidate calculated from the virtual reception vector and the plurality of reception signal candidates by the determination separation unit including a plurality of stages corresponding to the number of transmission antennas. A metric for each transmission signal candidate generated based on reliability is used, and in each stage, candidates are sequentially narrowed down using the metric information for each transmission signal, and a plurality of surviving transmission signal candidates are selected. And a step (8) of determining the plurality of transmitted signals and estimating and separating each transmitted signal. In the step (8), the determination separation means is
The initial cumulative metric is set to 0, and in each stage, the multilevel modulation is performed for the transmission signal corresponding to the number of stages executed for the previous stage cumulative metric that is the cumulative metric for each combination of transmission signal candidates selected in the previous stage. Calculate a metric corresponding to several minutes of transmission signal candidates, and calculate a cumulative metric by adding to each of the previous stage cumulative metric,
From the information of the cumulative metric, select a plurality of combinations of transmission signal candidates for transmission signals for executed stages, and narrow down,
6. The radio transmission method according to claim 5, wherein a transmission signal is determined and separated from information of an accumulated metric for combinations of surviving transmission signal candidates that are continuously selected and narrowed down to a final stage. In the step (4), the reception quality estimation unit estimates the reception quality of the plurality of transmission signals in the reception device based on the reception signal including the known pilot signal, and the sorting unit estimates the reception quality. Depending on the reception quality acquired by the acquisition means, the channel matrix is rearranged to calculate a sorted channel matrix,
In the step (5), the channel matrix decomposition means expands the sorted channel matrix as the channel matrix based on the noise power information acquired by the noise power acquisition means and the number of transmission antennas. 7. The radio transmission method according to claim 5, wherein the channel matrix decomposition is performed to configure the extended channel matrix and to decompose and acquire the extended channel matrix into a unitary matrix and an upper triangular matrix. . In the step (4), the sorting means uses a desired signal power-to-interference signal power and noise power ratio (received SINR) on the reception side as an index representing the reception quality of the plurality of transmission signals, and each channel matrix 8. The radio transmission method according to claim 7, wherein the sorted channel matrix is calculated by rearranging the columns in ascending order of reception SINR of each transmission signal.