JP2009033231A - Mimo reception device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce the operation amount of MIMO demodulation to lighten the load of demodulation processing in a MIMO communication system which performs demodulation by a maximum likelihood estimation method. <P>SOLUTION: In a MIMO reception device 200, a MIMO demodulation unit 280-1 calculates an error distance Δ2 between all candidate points x'1 of a transmission signal x1 and a transmission signal x2 using matrixes Q and R, calculates an error distance between the all candidate points x'1 of the transmission signal x1 and a reference point determined by respective bits (error distance Δ1 of transmission signal x1), and puts together the error distance Δ2 of the transmission signal x2 and the error distance Δ1 of the transmission signal x1 to calculate composite error distances Δ12 by states of first to fourth bits. Further, branchmetrics of patterns (x1, x2)=(0, 0), (0, 1), (1, 0), and (1, 1) that the first to fourth bits of a four-bit signal of the candidate points x'1 of the transmission signal x1 and candidate points x'2 of the transmission signal x2 can have are calculated and substituted in a trellis map to perform viterbi decoding. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a wireless transmission technology for digital signals, and more particularly, wireless digital in a multi-input multiple-output (hereinafter referred to as “MIMO (Multiple Input Multiple Output)”) propagation environment using a plurality of transmission antennas and a plurality of reception antennas. The present invention relates to a demodulation technique used in signal transmission.

  Conventionally, in broadband mobile communication, the frequency band that can be used is limited, and there is a demand for multimedia communication, so that it is required to achieve high quality and high frequency use efficiency equivalent to fixed communication. Yes. The MIMO communication technique is attracting attention as a technique for coping with this.

[Configuration of MIMO communication system]
FIG. 1 is a diagram illustrating a configuration example of a MIMO communication system. This MIMO communication system is an example configured by a transmission apparatus 100 including two transmission antennas 101 and a reception apparatus 200 including four reception antennas 201. Between the transmission antenna 101 and the reception antenna 201, the MIMO communication system is an example. Is formed with a MIMO propagation path. The transmission device 100 is a terminal device that can move freely, for example, and assigns two different data signals to each of the two transmission antennas 101 so that the radio waves are on the same frequency or in a frequency band overlapping state. Thus, an OFDM signal is output from each corresponding transmission antenna 101. Thereby, each OFDM signal is transmitted through four propagation paths. The receiving apparatus 200 is, for example, a base station apparatus, which separates the received four systems of signals based on the transfer function for each propagation path and demodulates the two systems of different data signals transmitted from the transmitting apparatus 100 ( For example, see Patent Document 1).

[Transmitter]
FIG. 2 is a diagram illustrating a configuration example of the transmission apparatus 100 in the MIMO communication system illustrated in FIG. The transmission apparatus 100 includes an encoding unit 110, a mapping unit 120, a frame configuration unit 130, an IFFT (Inverse Fast Fourier Transform) unit 140, a GI (Guard Interval) signal addition unit 150, an orthogonal modulation unit 160, a mixer 170, and a local oscillator. 171 and the transmission antenna 101 are provided. The mapping unit 120 to the subsequent transmission antenna 101 are configured in two systems.

  The encoding unit 110 receives, for example, a video signal captured by the transmission device 100, performs encoding such as energy spreading, error correction encoding, and interleaving, and separates the signal into two different signals. The mapping unit 120 receives the signal encoded by the encoding unit 110 and maps the signal on a carrier modulation constellation arrangement such as QAM (Quadrature Amplitude Modulation). The frame configuration unit 130 receives the signal that has been mapped to carrier modulation by the mapping unit 120, adds this signal as a data signal, an orthogonally encoded pilot signal that becomes a demodulation reference, and the like, and sets it in advance A frame is configured by arranging at a frequency that is set, and is output as an OFDM (Orthogonal Frequency Division Multiplexing) signal.

  The IFFT unit 140 receives the OFDM signal framed by the frame configuration unit 130, performs IFFT (inverse Fourier transform), and converts the frequency axis data into time axis data. The GI signal adding unit 150 receives the OFDM signal converted into time axis data by the IFFT unit 140, and adds the GI signal to the OFDM signal. The quadrature modulation unit 160 receives the OFDM signal to which the GI signal is added by the GI signal addition unit 150, and processes the OFDM signal that has been processed as a combination signal (complex number) of two real numbers and imaginary numbers so far. And quadrature modulation for orthogonalization on the orthogonal signal. The mixer 170 receives the OFDM signal orthogonalized by the orthogonal modulation unit 160 and performs frequency conversion from IF (Intermediate Frequency) to RF (Radio Frequency) in a required frequency band using the signal from the local oscillator 171. . The two systems from the mapping unit 120 to the mixer 170 perform the same processing, and the OFDM signal output from the mixer 170 is transmitted from the transmission antenna 101 as a transmission signal.

  FIG. 3 is a diagram illustrating a configuration example of the encoding unit 110 in the transmission device 100 illustrated in FIG. The encoding unit 110 includes an energy spreading unit 111, an outer encoding unit 112, an outer interleaving unit 113, an inner encoding unit 114, and two systems of inner interleaving units 115. The energy spreader 111 receives a video signal and spreads the energy of the data of the video signal subjected to data frame synchronization. The outer encoding unit 112 receives the data signal that has been energy diffused by the energy spreading unit 111 and encodes the Reed-Solomon code. The outer interleaving unit 113 receives the data signal encoded by the outer encoding unit 112 and performs convolutional interleaving. The inner coding unit 114 receives the data signal interleaved by the outer interleaving unit 113, and performs convolutional coding at a coding rate of 1/2 as shown in FIG.

  Inner interleaving section 115 receives the signal encoded by inner encoding section 114 and performs bit interleaving, frequency interleaving, and time interleaving processing. Here, bit interleaving replaces data to be transmitted in bit units and distributes carrier errors. In the receiving apparatus 200 described later, the received data is demodulated by replacing the received data in bit units in the reverse procedure. Further, frequency interleaving is to disperse data by exchanging data to be transmitted in units of carriers in the frequency axis direction. In the receiving apparatus 200 described later, the received data is demodulated by replacing the received data in units of carriers in the reverse procedure. Also, time interleaving is to replace data to be transmitted in units of carriers in the time direction and to distribute the data. In the receiving apparatus 200 described later, the received data is demodulated by replacing the received data in units of carriers in the reverse procedure. By this frequency interleaving or time interleaving, a burst-like error can be changed to a random error for a signal deteriorated in the propagation path, and the accuracy of error correction based on the convolutional code can be improved. The inner interleaving unit 115 is composed of two systems, each performing the same processing, and the two systems of signals are output to the corresponding two systems of mapping sections 120.

  FIG. 4 is a diagram illustrating a configuration example of the inner encoding unit 114 in the encoding unit 110 illustrated in FIG. 3. The inner coding unit 114 has a function of performing convolutional coding at a coding rate of ½, receives a data signal from the outer interleaving unit 113, and generates an original code generating polynomial (G1 = 171 oct, G2 = 133 oct). In accordance with this, two systems of signals are generated and output to the two systems of interleaving sections 115, respectively. When the coding rate is set to 2/3, 3/4, etc. other than 1/2, it is performed by puncturing which uses some of the outputs of the two systems according to a rule. Since the convolutional encoding in the inner encoding unit 114 is known, detailed description thereof is omitted here.

[Receiver]
FIG. 5 is a diagram illustrating a configuration example of the reception device 200 in the MIMO communication system illustrated in FIG. This receiving apparatus 200 includes a receiving antenna 201, a mixer 210, a local oscillator 211, an orthogonal demodulation unit 220, a symbol synchronization detection unit 230, a GI signal removal unit 240, an FFT unit 250, a frame separation unit 260, a propagation path estimation unit 270, a MIMO A demodulator 280 and a decoder 290 are provided. The reception antenna 201 to the subsequent frame separation unit 260 are configured in four systems.

  The four receiving antennas 201 receive OFDM signals that have been mixed on the same frequency via the propagation paths between the two transmitting antennas 101 as received signals. The mixer 210 performs frequency conversion from RF to IF on the received OFDM signal using the signal from the local oscillator 211. The quadrature demodulator 220 receives the OFDM signal frequency-converted by the mixer 210, performs quadrature demodulation, separates it into an in-phase signal and a quadrature signal, and generates two combined signals (complex numbers) of real and imaginary numbers. The symbol synchronization detector 230 receives the OFDM signal demodulated by the orthogonal demodulator 220, performs guard correlation, and detects the symbol timing that is the head of the symbol of the OFDM signal.

  GI signal removal section 240 removes the GI signal from the OFDM signal according to the symbol timing detected by symbol synchronization detection section 230. The FFT unit 250 receives the OFDM signal from which the GI signal has been removed by the GI signal removal unit 240, performs FFT (Fourier transform), and converts the time axis data into frequency axis data. The frame separation unit 260 receives the OFDM signal converted into the frequency axis data by the FFT unit 250, separates the data signal, the pilot signal, and the like from the OFDM signal (frame separation), and extracts each signal. From the four receiving antennas 201 to the frame separation unit 260, the same processing is performed in each of the four systems.

  The propagation path estimation unit 270 receives the pilot signals separated by the four frame separation units 260, multiplies the orthogonal code assigned to each transmission system by one bit at a time in symbols, and transmits the transmission antenna 101 and the reception antenna 201. All the propagation path characteristics between are estimated and output as propagation path estimation results (see, for example, Patent Document 2).

  The MIMO demodulator 280 receives the propagation path estimation result estimated by the propagation path estimator 270, inputs the data signals separated by the four systems of the frame separators 260, and separates and demodulates the mixed data signals. Do. Decoding section 290 receives the data signal demodulated by MIMO demodulation section 280 and performs decoding such as error correction and deinterleaving. In this way, the original video signal in the transmission device 100 can be obtained.

  FIG. 6 is a flowchart showing processing of conventional MIMO demodulation section 280 in receiving apparatus 200 shown in FIG. This flowchart shows a typical example of a MIMO demodulator in signal transmission based on a MIMO scheme that performs space division multiplexing. It is assumed that the transmission signal X is formed by m systems and the reception signal is formed by n systems. First, the first process in the conventional MIMO demodulator 280 uses an inverse matrix method. MIMO demodulator 280 receives channel matrix H as a channel estimation result from channel estimator 270, receives data signal Y from frame separator 260, performs the calculation shown in step S611, and performs soft Viterbi decoding. Do. According to this inverse matrix method, the amount of calculation is small and realization by hardware becomes easy. Further, in order to obtain diversity gain, it is necessary to increase the number of reception antennas rather than the number of transmission antennas. In addition, noise may be emphasized.

The second processing in the conventional MIMO demodulator 280 uses a maximum likelihood estimation method. MIMO demodulation section 280 receives propagation path matrix H from propagation path estimation section 270 and receives data signal Y from frame separation section 260, and in step S621, all modulation candidate points S of transmission signals x _{1 to} x _m are received. As for likelihood calculation, error distances Δ1 to Δn are calculated, and in step S622, the demodulated signal Xs and the minimum error distance Δs are calculated and soft Viterbi decoding is performed to determine the maximum likelihood. Here, all the modulation candidate points are 16 candidate points for the transmission signal x _{1 and} 16 candidate points for the transmission signal x _{2 in} the case of the 16QAM modulation scheme. Similarly, there are candidate points for the transmission signal x _m. Since there are 16, there are 16 ^m in total. In the configuration of the two transmitting antennas 101 and the four receiving antennas 201, there are 16 ² = 256 total modulation candidate points. According to this maximum likelihood estimation method, an enormous amount of calculation is required, so that it is difficult to realize by hardware. Here, assuming that the modulation multi-level number M, the number of transmission antennas m, and the number of reception antennas n, n × M ^m error distance calculations are required. On the other hand, high demodulation accuracy can be realized, and diversity gains corresponding to the number of receiving antennas can be obtained.

The third processing in the conventional MIMO demodulator 280 uses a candidate narrowing-down maximum likelihood estimation method using an inverse matrix. The MIMO demodulator 280 receives the channel matrix H from the channel estimator 270, receives the data signal Y from the frame separator 260, performs the calculation shown in step S631, and transmits the transmission signals x _{1 to} x in step S632. extracting four candidate points for each _m, in step S633, the combination _{S all} the four candidate points of each transmitted signal _x 1 ~x _m, calculates the error distance Δ1~Δn as the likelihood calculation, in step S634 Then, the maximum likelihood is determined, the demodulated signal Xs and the minimum error distance Δs are calculated, and soft Viterbi decoding is performed. According to this candidate narrowing-down maximum likelihood estimation method using an inverse matrix, the amount of calculation is larger than that of the inverse matrix method and less than that of the maximum likelihood estimation method. Also, the demodulation accuracy is about halfway between the inverse matrix method and the maximum likelihood estimation method.

  FIG. 7 is a diagram illustrating a configuration example of the conventional MIMO demodulator 280 in the receiving apparatus 200 illustrated in FIG. This MIMO demodulator 280-P performs demodulation using the maximum likelihood estimation method, which is the second processing shown in FIG. 6, and includes all modulation candidate point recording unit 281, replica signal generator 282, likelihood. A calculation unit 283, a likelihood determination unit 284, an inner deinterleave unit 285, and a Viterbi decoding unit 286 are provided.

The replica signal generation unit 282 receives the propagation path estimation result H from the propagation path estimation unit 270, and receives all the modulation candidate points S (the two transmission antennas 101 and the four reception antennas 201 from the total modulation candidate point recording unit 281. In the configuration, there are 256 all modulation candidate points S.), and replica signals h ₁ S to h ₄ S are generated. The likelihood calculation unit 283 receives the replica signals h ₁ S to h ₄ S from the replica signal generation unit 282 and receives the data signals Y (y ₁ , y ₂ , y ₃ , y ₄ ) from the four system frame separation units 260. And the error distances Δ1 to Δ4 between the data signal Y and the replica signals h ₁ S to h ₄ S and the sum of squares thereof ΔS = Δ1 ² + Δ2 ² + Δ3 ² + Δ4 ² are calculated as likelihood calculation. (See step S621 in FIG. 6).

  The likelihood determining unit 284 receives the error distance ΔS (the number of error distances is 256 × 4 = 1024) from the likelihood calculating unit 283, determines the maximum likelihood, and the data signal (demodulated signal) Xs (The number of data is 2) and the corresponding minimum error distance Δs is output (see step S622 in FIG. 6). The internal deinterleave unit 285 receives the data signal (demodulated signal) Xs and the minimum error distance Δs, performs deinterleaving on the data signal Xs with respect to time interleaving, frequency interleaving, and bit interleaving, and performs internal deinterleaving. Data signal (demodulated signal) and minimum error distance are output. The Viterbi decoding unit 286 receives the data signal (demodulated signal) subjected to the internal deinterleaving by the internal deinterleaving unit 285 and the error distance, and performs Viterbi decoding.

  FIG. 20 is a diagram for explaining processing of the conventional Viterbi decoding unit 286 shown in FIG. In FIG. 20, the time-series data signals (00, 10, 01, 10, 01, 10, 01, 01, 01,... / Left-side data input to the Viterbi decoding unit 286 are the first transmission system. The right data corresponds to the second transmission system, that is, corresponds to the output of two systems in the inner encoding unit 114 shown in Fig. 4. However, in Fig. 4, there is a constraint that there are six shift registers. Although the example of length 7 is shown, here, the shift register is assumed to be a case of two constraint lengths 3), and the state of the shift register (00, 01, 10, 10) in the inner encoding unit 114 shown in FIG. 11) shows the trellis map formed. Each path is given a combination of data signal candidates corresponding to the path, and an error distance between the data signal candidate for each path and the data signal input to the Viterbi decoding unit 286. Are calculated respectively. Here, the determination value of the data signal corresponding to the minimum path (the path in which the black circle is added on the arrow, the path in which the sum of the error distances obtained for the path is the minimum) among the paths indicated by the arrows (the arrow indicates a solid line “ 0 ”or a broken line“ 1. ”0, 1, 1, 1, 0, 1, 1, 0, 0) is a time-series data signal output by the Viterbi decoding unit 286. This trellis map is a map that is uniquely determined by the configuration of the inner encoding unit 114. It can be seen that, even if a data signal having an error is input to the Viterbi decoding unit 286, the error is corrected and a correct result is determined by this unique constraint condition.

  When the Viterbi decoding unit 286 inputs a time-series data signal, in the case of hard decision Viterbi decoding, the data signal input to the Viterbi decoding unit 286 is a hard decision result that has already been digitized, and this data signal and the trellis map The Hamming distance between the transmission signal candidates indicated by the connection of each path above is calculated. In the case of soft-decision Viterbi decoding, the data signal input to the Viterbi decoding unit 286 is a demodulation result before digitization including an error, and the transmission signal indicated by the connection between this data signal and each path on the trellis map. Euclidean distance between candidates is calculated. Then, the Viterbi decoding unit 286 calculates a branch metric by adding the Hamming distance or Euclidean distance attached to each path in units of paths. FIG. 20 shows a trellis map in which the branch metrics calculated in this way are entered. The Viterbi decoding unit 286 searches for all combinations of paths existing on the trellis map, identifies a path (minimum path) that minimizes the sum of the Hamming distance or the Euclidean distance, and sets the minimum path as the minimum path. A corresponding data signal (a data signal indicated by the minimum path or a data signal indicated by a change in the state of trellis transition) is output as a data signal corrected by Viterbi decoding. This is a kind of maximum likelihood estimation process. Since the processing of the Viterbi decoding unit 286 is known, detailed description thereof is omitted here.

JP 2006-345500 A JP 2005-124125 A

  As described above, the conventional MIMO demodulator 280-P in the MIMO communication system performs decoding by the maximum likelihood estimation method having higher demodulation performance than other demodulation methods, but demodulates the received data signal. The likelihood calculation unit 283 and the likelihood determination unit 284 perform likelihood calculation and maximum likelihood estimation, and the Viterbi decoding unit 286 again performs similar likelihood calculation (each path in FIG. 20) to perform Viterbi decoding. (Calculation of Hamming distance or Euclidean distance) and maximum likelihood estimation. Here, the former calculates the error distance between the actually received data signal and the replica signal generated with reference to all modulation candidate points, and selects the candidate for calculating the minimum error distance. When soft decision Viterbi decoding is performed, an error distance similar to that of the former likelihood calculation is calculated to obtain an Euclidean distance, and a candidate for calculating a minimum Euclidean distance is selected. That is, the likelihood calculation and maximum likelihood estimation in the likelihood calculation unit 283 and the likelihood calculation and maximum likelihood estimation in the Viterbi decoding unit 286 are similar calculations in that an error distance is obtained and a candidate having the minimum value is selected. It can be said that For example, as shown in FIG. 1, in a MIMO communication system including two transmission antennas 101 and four reception antennas 201, in the case of 16QAM modulation scheme, the former likelihood calculation is performed for one transmission. Since there are 16 signal point candidates for the transmission data from the antenna 101, there are 16 × 16 = 256 combination candidate points for the two transmission antennas 101. This combination of candidate points increases exponentially as the number of transmission antennas 101 increases, and the load for obtaining the error distance also increases as the number of transmission antennas 101 increases. Further, the number of reception antennas 201 increases. It increases as the number increases. For this reason, the conventional MIMO demodulator 280-P has a problem that the calculation amount of the MIMO demodulation processing is extremely large and the load is high.

  Accordingly, the present invention has been made to solve the above-described problems, and an object of the present invention is to reduce the amount of operation of MIMO demodulation and reduce the load of demodulation processing in a MIMO communication system that performs demodulation using a maximum likelihood estimation method. The object is to provide a possible MIMO receiver.

  In order to solve the above problems, the invention of claim 1 is directed to a receiving apparatus used in a MIMO communication system configured by a transmitting apparatus having a plurality of transmitting antennas and a receiving apparatus having a plurality of receiving antennas. When the signal transmitted from the first transmission antenna is a first transmission signal, and the signal transmitted from the second transmission antenna of the plurality of transmission antennas is the second transmission signal, QR decomposition is performed on the propagation path characteristics between the transmission antenna and the reception antenna to calculate the matrix Q and the matrix R, and the matrix Q and the matrix R calculated by the QR decomposition unit are used. Using the candidate point generator that generates all candidate points of the first transmission signal, and the matrix Q and the matrix R calculated by the QR decomposition unit, Sending A first candidate point error distance calculation unit that calculates an error distance of the second transmission signal indicating an error distance between the signal and all candidate points of the second transmission signal; and generated by the candidate point generation unit A first transmission indicating an error distance between all candidate points of the first transmission signal and a reference point defined by each bit according to the order of the bit signals represented by all candidate points of the first transmission signal A second candidate point error distance calculation unit for calculating an error distance of the signal, an error distance of the second transmission signal calculated by the first candidate point error distance calculation unit, and a second candidate point error distance calculation The error distance of the first transmission signal calculated by the unit, and an error distance composition unit for calculating a composite error distance for each bit state, and a composite error distance calculated by the error distance composition unit Each bit of the first transmission signal and each bit of the second transmission signal. A Viterbi branch metric calculation unit that calculates a branch metric, and a Viterbi decoding unit that performs Viterbi decoding by substituting the branch metric calculated by the Viterbi branch metric calculation unit into a trellis map. It is characterized by having.

The invention of claim 2 is the receiving apparatus according to claim 1, wherein X is a transmission signal, Y is a reception signal, R11 to R22 are elements of a matrix R, and S is a candidate point. The generation unit generates all candidate points (x′1) of the first transmission signal (x1) when the second transmission signal is set as a candidate point according to Expression (4) and Expression (5) described later. The first candidate point error distance calculation unit calculates all candidate points (y′2 / R) of the second transmission signal (x2) and the second transmission signal according to the expressions (4) and (5) described later. ₂₂ ), the error distance (Δ2) of the second transmission signal, which indicates the error distance from the second transmission signal, is calculated.

Further, the invention of claim 3 is the receiving apparatus according to claim 1 or 2, wherein the second candidate point error distance calculation unit has a modulation scheme of 16QAM.
(A) The error distance between the signal point of the first bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point,
dist_x1_10 = abs (abs (Re (x'1) -2) -1)
dist_x1_11 = abs (abs (Re (x'1) +2) -1)
(B) The error distance between the signal point of the second bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point,
dist_x1_20 = abs (abs (Im (x'1) -2) -1)
dist_x1_21 = abs (abs (Im (x'1) +2) -1)
(C) An error distance between the signal point of the third bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point,
dist_x1_30 = abs (abs (Re (x'1))-3)
dist_x1_31 = abs (abs (Re (x'1))-1)
(D) The error distance between the signal point of the fourth bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point,
dist_x1_40 = abs (abs (Im (x'1))-3)
Dist_x1_41 = abs (abs (Im (x′1)) − 1)

  According to a fourth aspect of the present invention, there is provided a receiving apparatus used in a MIMO communication system including a transmitting apparatus having a plurality of transmitting antennas and a receiving apparatus having a plurality of receiving antennas. When the signal transmitted from the transmission antenna is the first transmission signal and the signal transmitted from the second transmission antenna of the plurality of transmission antennas is the second transmission signal, the transmission antenna and the reception antenna QR decomposition is performed on the channel characteristics between the first transmission signal and the QR decomposition unit for calculating the matrix Q and the matrix R, and the second transmission signal is calculated using the matrix Q and the matrix R calculated by the QR decomposition unit. An error distance of the second transmission signal indicating an error distance between all candidate points of the second transmission signal is calculated, a candidate point having the smallest error distance is selected, and each bit of the candidate point is Different putters The first candidate point error distance calculation unit for calculating each error distance and the matrix Q and the matrix R calculated by the QR decomposition unit. A candidate point selection unit that selects a candidate point of the first transmission signal based on a signal point of the second transmission signal selected by the distance calculation unit, and a first transmission signal selected by the candidate point selection unit Calculating the error distance of the first transmission signal indicating the error distance between the candidate point of the first transmission signal and the reference point defined by each bit according to the order of the bit signals represented by the candidate points of the first transmission signal Two candidate point error distance calculation units, the error distance of the second transmission signal calculated by the first candidate point error distance calculation unit, and the first candidate point error distance calculation unit calculated by the second candidate point error distance calculation unit The error distance of the transmission signal is added to the combined error for each bit state. A combination pattern of bits that can be taken by each bit of the first transmission signal and each bit of the second transmission signal by using an error distance synthesis unit that calculates separation and the synthesis error distance calculated by the error distance synthesis unit A Viterbi branch metric calculating unit that calculates a branch metric, and a Viterbi decoding unit that performs Viterbi decoding by substituting the branch metric calculated by the Viterbi branch metric calculating unit into a trellis map. .

  As described above, according to the present invention, in the MIMO communication system, the likelihood calculation for performing processing by the maximum likelihood estimation method on the received data signal and the likelihood calculation for performing Viterbi decoding are unified. By doing so, the likelihood calculation with a high calculation load is not performed again. As a result, it is possible to realize a MIMO receiving apparatus that can reduce the amount of calculation of MIMO demodulation and reduce the load of demodulation processing as compared with the conventional case.

  The best mode for carrying out the present invention will be described below in detail with reference to the drawings. The present invention is characterized by the demodulation processing of the MIMO demodulator 280 in the receiving apparatus 200 in the example of the MIMO communication system (FIG. 1) configured by the transmitting apparatus 100 shown in FIG. 2 and the receiving apparatus 200 shown in FIG. There is. Also, like the inner encoding unit 114 shown in FIG. 4, the transmission apparatus 100 performs convolutional encoding, and the reception apparatus 200 performs a Viterbi decoding by performing a MIMO demodulation process using a maximum likelihood estimation method. There is application.

[Example 1]
First, Example 1 will be described. FIG. 8 is a diagram illustrating a configuration example (first example) of the MIMO demodulation unit in the reception device according to the embodiment of the present invention. FIG. 9 is a flowchart showing the processing of the MIMO demodulator (Example 1) shown in FIG. This MIMO demodulator 280-1 corresponds to the MIMO demodulator 280 in the receiving apparatus 200 shown in FIG. 5 in the MIMO communication system including the two transmitting antennas 101 and the four receiving antennas 201 shown in FIG. QR decomposition unit 2801, all modulation candidate point recording unit 2802, x1 candidate point generation and x2 error distance calculation unit 2803-1, candidate point x1 'error distance calculation unit 2804, error distance synthesis unit 2805, memory unit 2806, A deinterleaving unit 2807, a Viterbi branch metric calculating unit 2808, and a Viterbi decoding unit 2809 are provided.

ここで、伝搬路行列Ｈの要素ｈ_ｉｊは、送信アンテナｊから受信アンテナｉへの伝搬路の周波数応答特性を示す。以下の式によりＱＲ分解を行い、直交行列Ｑ及び上三角行列Ｒを算出する。
Ｈ＝Ｑ・Ｒ （Ｑ：直交行列，Ｒ：上三角行列） ・・・（２） The QR decomposition unit 2801 inputs the propagation path estimation result (estimation result in each propagation path between the transmission antenna 101 and the reception antenna 201) estimated by the propagation path estimation unit 270, performs QR decomposition, the matrix Q and A matrix R is calculated (step S901). A propagation path matrix H which is a propagation path estimation result is shown below.

Here, the element h _ij of the propagation path matrix H indicates the frequency response characteristic of the propagation path from the transmission antenna j to the reception antenna i. QR decomposition is performed by the following equation to calculate an orthogonal matrix Q and an upper triangular matrix R.
H = Q · R (Q: orthogonal matrix, R: upper triangular matrix) (2)

  The x1 candidate point generation and x2 error distance calculation unit 2803-1 receives the matrix Q and the matrix R calculated by the QR decomposition unit 2801, regards the transmission signal x2 as all candidate points S, and transmits all the transmission signals x1. Candidate point x′1 is calculated, and error distance Δ2 between the demodulated signal of transmission signal x2 and all candidate points S is calculated (steps S902 to S904). Here, a signal transmitted from the first transmission antenna # 1 is a transmission signal x1, and a signal transmitted from the second transmission antenna # 2 is a transmission signal x2. Hereinafter, a method of calculating all candidate points x′1 of the transmission signal x1 and the error distance Δ2 of the transmission signal x2 will be described.

Since the matrix Q is an orthogonal matrix, the relationship between the transmission signal X and the reception signal Y is as follows.
Y = HX = (Q · R) X
Q ^H Y = {(Q ^H · Q) · R} X = RX (3)
Here, a received signal received by the receiving antennas # 1 to # 4 is Y = [y1, y2, y3, y4] ^T (T represents transposition), and transmissions are transmitted from the transmitting antennas # 1 and # 2. Let the signal be X = [x1, x2] ^T.

ここで、行列Ｒの要素は複素数である。ｘ１候補点生成及びｘ２誤差距離演算部２８０３−１は、式（４）及び全変調候補点記録部２８０２に記録された全変調候補点Ｓを用いて、送信信号ｘ１の全候補点ｘ’１及び送信信号ｘ２の誤差距離Δ２を、以下の式により算出する。

ここで、全変調候補点Ｓは、例えば１６ＱＡＭの変調方式におけるコンスタレーションで示される全ての点のことをいう。図１０は、電波産業会（Ａｓｓｏｃｉａｔｉｏｎ ｏｆ Ｒａｄｉｏ Ｉｎｄｕｓｔｒｉｅｓ ａｎｄ Ｂｕｓｉｎｅｓｓｅｓ）で定められるテレビ番組素材伝送用の無線素材伝送システムの規格（ＡＲＩＢ ＳＴＤ−Ｂ３３）に準拠した場合の、１６ＱＡＭで表される送信信号の配置を示す図である。このコンスタレーション配置は、図１に示したＭＩＭＯ通信システムにおける１６ＱＡＭの変調方式における送信信号ｘ１，ｘ２のとり得る配置を示している。 Since the matrix R is an upper triangular matrix, the equation (3) can be expressed by the following equation.

Here, the elements of the matrix R are complex numbers. The x1 candidate point generation and x2 error distance calculation unit 2803-1 uses all modulation candidate points S recorded in the equation (4) and all modulation candidate point recording unit 2802, and uses all candidate points x′1 of the transmission signal x1. And an error distance Δ2 of the transmission signal x2 is calculated by the following equation.

Here, all modulation candidate points S refer to all points indicated by a constellation in, for example, a 16QAM modulation system. FIG. 10 shows a transmission signal represented by 16QAM when the wireless material transmission system standard (ARIB STD-B33) for television program material transmission defined by the Association of Radio Industries and Businesses is established. It is a figure which shows arrangement | positioning. This constellation arrangement shows an arrangement that transmission signals x1 and x2 can take in the 16QAM modulation scheme in the MIMO communication system shown in FIG.

In equation (5), all candidate points x′1 of transmission signal x1 are 16 points obtained by substituting all modulation candidate points S for transmission signal x2 in equation (4). Further, the error distance Δ2 of the transmission signal x2 is a distance value of 16 points between x2 (= y′2 / R ₂₂ ) obtained by the equation (4) and all the modulation candidate points S of the transmission signal x2. Become.

  Candidate point x1 ′ error distance calculation unit 2804 receives all candidate points x′1 of transmission signal x1 calculated by x1 candidate point generation and x2 error distance calculation unit 2803-1, and all candidate points of transmission signal x1 An error distance (error distance Δ1 of the transmission signal x1) between x′1 and the reference point defined by each bit is calculated in accordance with the order of the 4-bit signal represented by the candidate point x′1 (step S905). Specifically, the candidate point x1 ′ error distance calculation unit 2804 takes the signal point of each bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the value “0” and “1”. The error distance from the reference point determined for each of the values "" is calculated by the following formula. However, in the following formula, the division by the amplitude Z (= √10) of the received signal defined in ARIB STD-B33 is omitted.

(A) Error distance between the signal point of the first bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point
dist_x1_10 = abs (abs (Re (x'1) -2) -1) (referenced to the first bit “0”) (6)
dist_x1_11 = abs (abs (Re (x'1) +2) -1) (referenced to the first bit "1") (7)
(B) Error distance between the signal point of the second bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point
dist_x1_20 ＝ abs (abs (Im (x'1) -2) -1) (referenced to the second bit “0”) (8)
dist_x1_21 ＝ abs (abs (Im (x'1) +2) -1) (based on the second bit “1”) (9)
(C) The error distance between the signal point of the third bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point
dist_x1_30 = abs (abs (Re (x'1))-3) (referenced to the third bit “0”) (10)
dist_x1_31 = abs (abs (Re (x'1))-1) (referenced to the third bit “1”) (11)
(D) An error distance between the signal point of the fourth bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point
dist_x1_40 = abs (abs (Im (x'1))-3) (4th bit "0" as a reference) ... (12)
dist_x1_41 = abs (abs (Im (x'1))-1) (4th bit "1" as a reference) ... (13)
Here, Re represents a real part, Im represents an imaginary part, and abs represents an absolute value. In each equation, 16 error distances are calculated.

  FIG. 11 is a diagram showing a distribution of 4-bit signals represented by 16QAM in the constellation arrangement. Here, the horizontal axis represents a real number (real) or an in-phase component, and the vertical axis represents an imaginary number (imag) or a quadrature component. As shown in FIG. 11, x1_10 is an area when the first bit is "0", x1_11 is an area when the first bit is "1", x1_20 is an area when the second bit is "0", x1_21 Is the area when the second bit is "1", x1_30 is the area when the third bit is "0", x1_31 is the area when the third bit is "1", and x1_40 is the fourth bit is "0" X1_41 is an area when the fourth bit is “1”.

  Whether the first bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 is “0” or “1” is determined at all candidate points x′1 of the transmission signal x1. It is determined by whether the value of the real part is positive or negative. The error distance of the equation (6) is based on the first bit “0” and its real part is 1 or 3, and thus is calculated with the reference point being (2 + 0j). On the other hand, since the error distance of the equation (7) is based on the first bit “1” and its real part is −1 or −3, the reference point is calculated as (−2 + 0j).

  Whether the second bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 is “0” or “1” depends on all candidate points x ′ of the transmission signal x1. It is determined by whether the value of the imaginary part in 1 is positive or negative. The error distance in equation (8) is based on the second bit “0”, and is calculated with the reference point as (0 + 2j). On the other hand, the error distance in equation (9) is based on the second bit “1”, and is calculated with the reference point being (0-2j).

  For the third bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1, the error distance of the equation (10) is based on the third bit “0”, and the reference point Is calculated as (3 + 0j) or (-3 + 0j). On the other hand, the error distance in Expression (11) is based on the third bit “1”, and is calculated with the reference point being (1 + 0j) or (−1 + 0j).

  For the fourth bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1, the error distance of the equation (10) is based on the fourth bit “0”, and the reference point Is calculated as (0 + 3j) or (0-3j). On the other hand, the error distance in Expression (11) is based on the fourth bit “1”, and is an error distance calculated with the reference point as (0 + 1j) or (0-1j).

The error distance synthesis unit 2805 receives the error distance Δ2 (16 pieces) of the transmission signal x2 calculated by the x1 candidate point generation and x2 error distance calculation unit 2803-1 and calculates by the candidate point x1 ′ error distance calculation unit 2804. The error distance Δ1 (16 × 8 = 128) of the transmitted signal x1 is input, the error distance Δ2 of the transmission signal x2 and the corresponding transmission signal x1 (corresponding to the 16 signal points of the error) The error distance Δ1 is added, and the combined error distance Δ12 of the transmission signal x1 and the transmission signal x2 is calculated for each bit state as follows (step S906).
(A) Synthesis error distance of the first bit Δ12_x1_10 = dist_x1_10 + Δ2 (14)
Δ12_x1_11 = dist_x1_11 + Δ2 (15)
(B) Composite error distance of the second bit Δ12_x1_20 = dist_x1_20 + Δ2 (16)
Δ12_x1_21 = dist_x1_21 + Δ2 (17)
(C) Composite error distance of the third bit Δ12_x1_30 = dist_x1_30 + Δ2 (18)
Δ12_x1_31 = dist_x1_31 + Δ2 (19)
(D) Synthesis error distance of the fourth bit Δ12_x1_40 = dist_x1_40 + Δ2 (20)
Δ12_x1_41 = dist_x1_41 + Δ2 (21)

  The memory unit 2806 temporarily records the data amount for the time required in the internal deinterleaving unit 2807, which will be described later, for the combined error distance Δ12 of the transmission signal x1 and the transmission signal x2 combined by the error distance combining unit 2805. (This step is omitted in the processing flow of FIG. 9).

  The inner deinterleaving unit 2807 reads the combined error distance Δ12 recorded in the memory unit 2806, and performs the time interleaving, frequency interleaving, and bit interleaving added to the transmission signal by the inner interleaving unit 115 of the encoding unit 110 in the transmission device 100. Then, deinterleaving is performed, and the synthesis error distances are rearranged according to the deinterleaving. Then, the signal is returned to the signal before the inner interleaving (this step is omitted in the processing flow of FIG. 9).

  The Viterbi branch metric calculation unit 2808 receives the combined error distance (16 × 8 = 128) for each bit state, which is internally deinterleaved by the internal deinterleaving unit 2807, and each bit is a candidate point of the transmission signal x1 Regarding four patterns (x1, x2) = (0, 0), (0, 1), (1, 0), (1, 1) that can be taken by the candidate point x′2 of x′1 and the transmission signal x2, A corresponding synthesis error distance group is selected, a synthesis error distance having a minimum value is selected from the group, and is output as a branch metric for each pattern (step S907). Hereinafter, a method for generating a branch metric will be described.

  FIG. 12 is a diagram illustrating all candidate points and corresponding numbers of transmission signals represented by 16QAM. 12, numbers corresponding to the signal points of the transmission signal shown in FIG. 10 are shown.

  FIG. 13 is a diagram for explaining a method for selecting candidate points in a 4-bit signal represented by 16QAM. As an example, a method for selecting the synthesis error distance of the first bit will be described. First, a case where the pattern that can be taken by the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2 is (x1, x2) = (0, 0) will be described. The Viterbi branch metric calculation unit 2808 selects a synthesis error distance group (16 synthesis error distances) when the first bit shown in the equation (14) at the candidate point x′1 of the transmission signal x1 is “0”. The Viterbi branch metric calculation unit 2808 is most reliable when the first bit of the candidate point x′2 of the transmission signal x2 is “0” from the group of synthesis error distances (16 synthesis error distances). Eight possible synthetic error distances are selected. The eight synthesis error distances are distances in the range where the real number of the candidate point x′2 of the transmission signal x2 is positive. In FIG. 13, the signal points from No. 1 to No. 8 on the coordinates of the first bit. Is the synthesis error distance corresponding to. Then, the Viterbi branch metric calculation unit 2808 selects a minimum one of the eight combined error distances, and uses the combined error distance as a candidate point x′1 of the transmission signal x1 and a candidate of the transmission signal x2. The branch metric is when the first bit of the point x′2 is (0, 0).

  Further, the case where the pattern (x1, x2) = (0, 1) that can be taken by the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2 will be described. The Viterbi branch metric calculation unit 2808 selects a synthesis error distance group (16 synthesis error distances) when the first bit shown in the equation (14) at the candidate point x′1 of the transmission signal x1 is “0”. The Viterbi branch metric calculation unit 2808 is most reliable when the first bit of the candidate point x′2 of the transmission signal x2 is “1” from the group of synthesis error distances (16 synthesis error distances). Eight possible synthetic error distances are selected. The eight synthesis error distances are distances in the range where the real number of the candidate point x′2 of the transmission signal x2 is negative. In FIG. 13, the signal points from No. 9 to No. 16 on the first bit coordinates. Is the synthesis error distance corresponding to. Then, the Viterbi branch metric calculation unit 2808 selects a minimum one of the eight combined error distances, and uses the combined error distance as a candidate point x′1 of the transmission signal x1 and a candidate of the transmission signal x2. A branch metric when the first bit of the point x′2 is (0, 1) is used.

  Further, a case where the pattern (x1, x2) = (1, 0) that can be taken by the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2 will be described. The Viterbi branch metric calculation unit 2808 selects a synthesis error distance group (16 synthesis error distances) when the first bit shown in the equation (15) at the candidate point x′1 of the transmission signal x1 is “1”. The Viterbi branch metric calculation unit 2808 is most reliable when the first bit of the candidate point x′2 of the transmission signal x2 is “0” from the group of synthesis error distances (16 synthesis error distances). Eight possible synthetic error distances are selected. The eight synthesis error distances are distances in the range where the real number of the candidate point x′2 of the transmission signal x2 is positive. In FIG. 13, the signal points from No. 1 to No. 8 on the coordinates of the first bit. Is the synthesis error distance corresponding to. Then, the Viterbi branch metric calculation unit 2808 selects a minimum one of the eight combined error distances, and uses the combined error distance as a candidate point x′1 of the transmission signal x1 and a candidate of the transmission signal x2. The branch metric is set when the first bit of the point x′2 is (1, 0).

  Further, a case where the pattern (x1, x2) = (1, 1) that can be taken by the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2 will be described. The Viterbi branch metric calculation unit 2808 selects a synthesis error distance group (16 synthesis error distances) when the first bit shown in the equation (15) at the candidate point x′1 of the transmission signal x1 is “1”. The Viterbi branch metric calculation unit 2808 is most reliable when the first bit of the candidate point x′2 of the transmission signal x2 is “1” from the group of synthesis error distances (16 synthesis error distances). Eight possible synthetic error distances are selected. The eight synthesis error distances are distances in the range where the real number of the candidate point x′2 of the transmission signal x2 is negative. In FIG. 13, the signal points from No. 9 to No. 16 on the first bit coordinates. Is the synthesis error distance corresponding to. Then, the Viterbi branch metric calculation unit 2808 selects a minimum one of the eight combined error distances, and uses the combined error distance as a candidate point x′1 of the transmission signal x1 and a candidate of the transmission signal x2. It is assumed that the branch metric is when the first bit of the point x′2 is (1, 1).

  Similarly, the Viterbi branch metric calculation unit 2808 can take patterns (x1, x2) = (candidate points x′1 of the transmission signal x1 and candidate points x′2 of the transmission signal x2 in each of the second to fourth bits. For (0,0), (0,1), (1,0), (1,1), select the corresponding composite error distance group, select the composite error distance that takes the minimum value, and select it. Output as branch metric for each pattern. In this way, the first bit, the second bit, the third bit, and the fourth bit of the 4-bit signal of the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2 can be taken, respectively. A branch metric (error distance) is calculated for the pattern (x1, x2) = (0, 0), (0, 1), (1, 0), (1, 1).

  The Viterbi decoding unit 2809 calculates the first and second bits of the 4-bit signal calculated by the Viterbi branch metric calculation unit 2808 of the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2. A branch metric of patterns (x1, x2) = (0, 0), (0, 1), (1, 0), (1, 1) that can be taken by the third bit and the fourth bit is input, and the trellis Viterbi decoding is performed by assigning to the map.

  FIG. 21 is a diagram illustrating the processing of the Viterbi decoding unit 2809. In FIG. 21, (1) to (4) expressed corresponding to the path indicated by the arrow are one of the 4-bit signals of the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2. The branch metrics of the pattern (x1, x2) = (0, 0), (0, 1), (1, 0), (1, 1) that can be taken by each bit are shown. Also, (5) to (8) are patterns (x1, x2) that can be taken by the second bit of the 4-bit signal of the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2. = Indicates branch metrics of (0, 0), (0, 1), (1, 0), (1, 1). Also, (9) to (12) are patterns (x1, x2) that can be taken by the third bit of the 4-bit signal of the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2. = Indicates branch metrics (error distance) of (0, 0), (0, 1), (1, 0), (1, 1). Also, (13) to (16) are patterns (x1, x2) that can be taken by the fourth bit of the 4-bit signal of the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2. = Indicates branch metrics (error distance) of (0, 0), (0, 1), (1, 0), (1, 1). FIG. 21 corresponds to FIG.

  That is, as shown in FIG. 21, the Viterbi decoding unit 2809 substitutes the input branch metric into the trellis map, and the total sum of error distances (path metrics) for each path calculated from the trellis map using the branch metric is the minimum. And a data signal corresponding to the minimum path (a data signal indicated by the minimum path or a data signal indicated by a change in the state of trellis transition) is output as a Viterbi-decoded data signal.

  The configuration of the MIMO demodulator 280-1 in FIG. 8 has been described above. Here, the conventional MIMO demodulator 280-P shown in FIG. 7 is compared with the MIMO demodulator 280-1 shown in FIG. FIG. 14 is a diagram illustrating a configuration when the MIMO demodulator 280-1 according to the first embodiment is made to correspond to the conventional MIMO demodulator 280-P. In other words, the MIMO demodulator 280-1 in FIG. 14 is different from the expression in FIG. 8 in order to make the configuration correspond to the conventional MIMO demodulator 280-P in FIG. In FIG. 14, this MIMO demodulator 280-1 includes a QR decomposition unit 2811, a total modulation candidate point recording unit 2812, a replica signal generation unit 2813, a likelihood calculation unit 2814, an internal deinterleave unit 2815, a branch metric generation unit 2816, and A Viterbi decoding unit 2817 is provided.

  QR decomposition section 2811 corresponds to QR decomposition section 2801 shown in FIG. 8, and all modulation candidate point recording section 2812 corresponds to all modulation candidate point recording section 2802 shown in FIG. In addition, the replica signal generation unit 2813 and the likelihood calculation unit 2814 include the x1 candidate point generation and x2 error distance calculation unit 2803-1, the candidate point x1 ′ error distance calculation unit 2804, and the error distance synthesis unit 2805 shown in FIG. Correspond. Further, the inner deinterleaving unit 2815 corresponds to the memory unit 2806 and the inner deinterleaving unit 2807 shown in FIG. 8, and the branch metric generation unit 2816 corresponds to the Viterbi branch metric calculation unit 2808 shown in FIG. Reference numeral 2817 corresponds to the Viterbi decoding unit 2809 shown in FIG. The functions of the components shown in FIG. 14 have already been described with reference to FIGS.

  FIG. 15 is a flowchart illustrating processing when applied to a MIMO communication system including m transmission antennas 101 and n reception antennas 201 in the first embodiment. That is, this flowchart is a generalized process in the MIMO communication system including the two transmission antennas 101 and the four reception antennas 201 shown in FIG. Hereinafter, the process of FIG. 15 will be described.

In this MIMO communication system, the transmission apparatus 100 transmits OFDM signal transmission signals X = [x ₁ ,..., X _m ] mapped to 16QAM via m transmission antennas 101 on the same frequency. Then, the reception apparatus 200 receives the received signal Y = [y ₁ ,..., Y _n ] via the n reception antennas 201 via the MIMO propagation path.

The receiving apparatus 200 calculates a propagation path matrix H between the transmission and reception antennas using the pilot carrier, performs QR decomposition on the propagation path matrix H, and calculates a matrix Q and a matrix R (step S1501). Further, by using the matrix Q and the matrix R, the transmission signal _{x m} all candidate points S (multilevel variable content: 16 points) and regarded to transmit to the transmission signal _{x m} signal _{x m-1,} ··· _x 1 All candidate points _{x 'm-1, ···,} x' of the calculated _one, and calculates the error distance Δm between the demodulated signal and all candidate point S of the transmission signal _{x m} (step S1502~ step S1504) . The transmission signal _x m-1, all candidate points x of _{··· x 1 'm-1,} ···, x' _1, in the order of 4 bit signal indicative of the candidate points, the first bit to 4 When each bit of the bit can take values of “0” and “1”, error distances Δm−1 to Δ1 between the reference points defined by the respective bits are calculated (step S1505).

Then, the receiving apparatus 200, the error distance Δm of the transmission signal _{x m,} corresponding to (corresponding to the 16 signal points of the error) transmission signal _x m-1, the · · · _{x 1} error distance Δm- 1 to Δ1 are added together, and a combined error distance dX is calculated for each bit state (step S1506). In addition, with respect to this combined error distance dX, possible patterns (x ₁ ,..., X _m ) = (0,..., 0), (0,..., 1),. , (1,..., 1), a synthesis error distance group is selected, a synthesis error distance having a minimum value is selected from the group, and it is output as a branch metric for each pattern (step S1507).

  The processing in FIG. 15 has been described above, but these correspond to steps S901 to S907 shown in FIG. 9 respectively, so please refer to them for specific explanations and mathematical expressions.

  As described above, according to the MIMO demodulator 280-1 of the first embodiment, the x1 candidate point generation and x2 error distance calculator 2803-1 uses the matrix Q and the matrix R to generate all candidate points x1 of the transmission signal x1. The error distance Δ2 between '1 and the transmission signal x2 is calculated, and the candidate point x1' error distance calculation unit 2804 calculates the error distance between all candidate points x'1 of the transmission signal x1 and the reference point defined by each bit. (Error distance Δ1 of the transmission signal x1) is calculated, and the error distance combining unit 2805 combines the error distance Δ2 of the transmission signal x2 and the error distance Δ1 of the transmission signal x1 for each state of the first to fourth bits. And the Viterbi branch metric calculation unit 2808 calculates the first to fourth bits of the 4-bit signal of the candidate point x′1 of the transmission signal x1 and the candidate point x′2 of the transmission signal x2. Each putter can take (X1, x2) = (0, 0), (0, 1), (1, 0), (1, 1) branch metrics are calculated, and the Viterbi decoding unit 2809 converts the branch metrics into a trellis map. Substituted Viterbi decoding to generate a decoded signal. Whereas the conventional MIMO demodulator 280-P performs the likelihood calculation (error distance calculation) twice, the MIMO demodulator 280-1 performs processing by the maximum likelihood estimation method on the received data signal. The likelihood calculation for performing and the likelihood calculation for performing Viterbi decoding are unified so that only one likelihood calculation is required. In addition, Viterbi decoding is performed without generating a demodulated signal. Therefore, the amount of calculation in MIMO demodulation can be reduced, and the load of demodulation processing can be reduced. In addition, it is possible to realize signal separation performance substantially equivalent to that of the conventional maximum likelihood estimation method, and it is easy to realize hardware.

[Example 2]
Next, Example 2 will be described. In the first embodiment, a branch metric for Viterbi decoding is calculated from all candidate points x′1 (16) of the transmission signal x1 and the error distance Δ2 (16) of the transmission signal x2. On the other hand, in the second embodiment, the calculation amount is further reduced by calculating the error distance for four of the 16 candidate points instead of calculating the error for all 16 candidate points. That is, in the second embodiment, a branch metric for Viterbi decoding is calculated from the four candidate points of the transmission signal x1 and the four error distances of the transmission signal x2 in order to further reduce the demodulation processing load.

  FIG. 16 is a diagram illustrating a configuration example (example 2) of the MIMO demodulation unit in the reception device according to the embodiment of the present invention. FIG. 17 is a flowchart showing the processing of the MIMO demodulator (Example 2) shown in FIG. This MIMO demodulator 280-2 corresponds to the MIMO demodulator 280 in the receiving apparatus 200 shown in FIG. 5 in the MIMO communication system including the two transmitting antennas 101 and the four receiving antennas 201 shown in FIG. QR decomposition unit 2801, all modulation candidate point recording unit 2802, x1 candidate point selection and x2 error distance calculation unit 2803-2, candidate point x1 ′ error distance calculation unit 2804, error distance synthesis unit 2805, memory unit 2806, A deinterleaving unit 2807, a Viterbi branch metric calculating unit 2808, and a Viterbi decoding unit 2809 are provided.

  Comparing the configuration of the MIMO demodulator 280-1 (Example 1) shown in FIG. 8 with the configuration of the MIMO demodulator 280-2 (Example 2) shown in FIG. Although common to the modulation candidate point recording unit 2802, the candidate point x1 ′ error distance calculation unit 2804, the error distance synthesis unit 2805, the memory unit 2806, the inner deinterleave unit 2807, the Viterbi branch metric calculation unit 2808, and the Viterbi decoding unit 2809, it is the MIMO. Demodulation section 280-2 includes x1 candidate point selection and x2 error distance calculation section 2803-2 having functions different from x1 candidate point generation and x2 error distance calculation section 2803-1 of MIMO demodulation section 280-1. It is different in point. 9 is compared with the processing flow of the MIMO demodulator 280-2 (embodiment 2) shown in FIG. 17, the MIMO demodulator 280- 2 is different in that step S1703 is different from step S903 of the MIMO demodulator 280-1. That is, the MIMO demodulator 280-2 operates on 16 candidates that are all candidates, whereas the MIMO demodulator 280-2 is different in that it operates only on four candidate points.

x1 candidate point selection and x2 error distance calculation unit 2803-2 receives the matrices Q and matrix R calculated by the QR decomposition unit 2801, firstly, x2 is obtained by Equation _{(4) (= y'2 / R} 22 ) And all the modulation candidate points S of the transmission signal x2 are calculated, and a candidate point at the smallest error distance among the error distances Δ2 (16 error distances) of the transmission signal x2 is selected. This is used as the demodulation point of the transmission signal x2. Further, the x1 candidate point selection and x2 error distance calculation unit 2803-2 extracts signal points in which each bit of the 4-bit signal of the transmission signal x2 has a different pattern for the demodulation point of the transmission signal x2, and this 4 Using the signal points as candidate points S, the candidate points x′1 (four) of the transmission signal x1 are calculated by the above-described equation (5). As will be described later, for example, when the demodulation point of the transmission signal x2 is [0, 0, 1, 0], [1, 0, 1, 0] [0, 1, 1, 0] [0, 0,0,0] [0,0,1,1] are extracted.

  FIG. 18 is a diagram illustrating candidate points with the smallest error distance among the error distances Δ2 of the transmission signal x2. Here, the candidate point with the smallest error distance is [0, 0, 1, 0]. FIG. 19 is a diagram for explaining a method for extracting candidate points in each bit. In FIG. 19, when the demodulation point of the transmission signal x2 is [0, 0, 1, 0], [1, 0, 1, 0] [0, 1, 1, 0] [0, 0, 0] are used as candidate points. , 0] [0, 0, 1, 1] are extracted.

  For example, when the value of the first bit of x2 is “0”, the x1 candidate point selection and x2 error distance calculation unit 2803-2 supports when the transmission signal x2 is [0, 0, 1, 0]. The candidate point x′1 of the transmission signal x1 to be calculated is calculated by Expression (5). Candidate point x1 ′ error distance calculation unit 2804 obtains error distances from the reference point whose first bit is “0” and “1” with respect to this candidate point x′1 (formula (6) and formula ( 7)). The error distance synthesis unit 2805 selects the error distance Δ2 corresponding to the value [0, 0, 1, 0] of the transmission signal x2, and adds this error distance Δ2 and the error distance Δ1 of the candidate point x′1. To determine the composite error distance.

  When the value of the first bit of the transmission signal x2 is “1”, the x1 candidate point selection and x2 error distance calculation unit 2803-2 sets the candidate point [1, 0, 1, 0] of the transmission signal x2. The candidate point x′1 of the transmission signal x1 corresponding to this time is extracted and calculated by the equation (5). Candidate point x1 ′ error distance calculation unit 2804 obtains error distances from the reference point whose first bit is “0” and “1” with respect to this candidate point x′1 (formula (6) and formula ( 7)). The error distance synthesis unit 2805 selects the error distance Δ2 corresponding to the candidate point [1, 0, 1, 0] of the transmission signal x2, and adds this error distance Δ2 and the error distance Δ1 of the candidate point x′1. To obtain the composite error distance. The Viterbi branch metric calculation unit 2808 calculates (x1, x2) = (0, 0), (0, 1), (1,0), (1, 0) with respect to the combined error distance obtained by the above calculation process. As a branch metric when taking the value of 1), the Viterbi decoding unit 2809 assigns this branch metric to the trellis map.

  Similarly, when the value of the second bit of the transmission signal x2 is “0”, the x1 candidate point selection and x2 error distance calculation unit 2803-2 sets the transmission signal x2 to [0, 0, 1, 0]. The candidate point x′1 of the transmission signal x1 corresponding to the occasion is calculated by the equation (5). The candidate point x1 ′ error distance calculation unit 2804 obtains the error distance from the reference point whose second bit is “0” and “1” with respect to this candidate point x′1 (formulas (8) and (9) )). The error distance synthesis unit 2805 selects the error distance Δ2 corresponding to the candidate point [0, 0, 1, 0] of the transmission signal x2, and adds this error distance Δ2 and the error distance Δ1 of the candidate point x′1. To obtain the composite error distance.

  When the value of the second bit of the transmission signal x2 is “1”, the x1 candidate point selection and x2 error distance calculation unit 2803-2 sets the candidate point [0, 1, 1, 0] of the transmission signal x2. The candidate point x′1 of the transmission signal x1 corresponding to this time is extracted and calculated by the equation (5). The candidate point x1 ′ error distance calculation unit 2804 obtains the error distance from the reference point whose second bit is “0” and “1” with respect to this candidate point x′1 (formulas (8) and (9) )). The error distance combining unit 2805 selects the error distance Δ2 corresponding to the candidate point [0, 1, 1, 0] of the transmission signal x2, and adds the error distance Δ2 and the error distance Δ1 of the candidate point x′1. To determine the composite error distance. The Viterbi branch metric calculation unit 2808 calculates (x1, x2) = (0, 0), (0, 1), (1,0), (1, 0) with respect to the combined error distance obtained by the above calculation process. As a branch metric when taking the value of 1), the Viterbi decoding unit 2809 assigns this branch metric to the trellis map.

  Similarly, when the value of the third bit of the transmission signal x2 is “0”, the x1 candidate point selection and x2 error distance calculation unit 2803-2 converts the transmission signal x2 to [0, 0, 0, 0]. The candidate point x′1 of the corresponding transmission signal x1 is calculated by Equation (5). The candidate point x1 ′ error distance calculation unit 2804 obtains an error distance from the reference point whose third bit is “0” and “1” for the candidate point x′1 (Equation (10) and Equation (11)). )). The error distance synthesis unit 2805 selects the error distance Δ2 corresponding to the candidate point [0, 0, 0, 0] of the transmission signal x2, and adds this error distance Δ2 and the error distance Δ1 of the candidate point x′1. To obtain the composite error distance.

  When the value of the third bit of the transmission signal x2 is “1”, the x1 candidate point selection and x2 error distance calculation unit 2803-2 sets the candidate point [0, 0, 1, 0] of the transmission signal x2. The candidate point x′1 of the transmission signal x1 corresponding to this time is extracted and calculated by the equation (5). The candidate point x1 ′ error distance calculation unit 2804 obtains an error distance from the reference point whose third bit is “0” and “1” for the candidate point x′1 (Equation (10) and Equation (11)). )). The error distance synthesis unit 2805 selects the error distance Δ2 corresponding to the candidate point [0, 0, 1, 0] of the transmission signal x2, and adds the error distance Δ2 and the error distance Δ1 of the candidate point x′1. To determine the composite error distance. The Viterbi branch metric calculation unit 2808 calculates (x1, x2) = (0, 0), (0, 1), (1,0), (1, 0) with respect to the combined error distance obtained by the above calculation process. As a branch metric when taking the value of 1), the Viterbi decoding unit 2809 assigns this branch metric to the trellis map.

  Similarly, when the value of the fourth bit of the transmission signal x2 is “0”, the x1 candidate point selection and x2 error distance calculation unit 2803-2 sets the transmission signal x2 to [0, 0, 1, 0]. The candidate point x′1 of the transmission signal x1 corresponding to the occasion is calculated by the equation (5). The candidate point x1 ′ error distance calculation unit 2804 obtains an error distance from the reference point whose fourth bit is “0” and “1” with respect to the candidate point x′1 (Equation (12) and Equation (13)). )). The error distance synthesis unit 2805 selects the error distance Δ2 corresponding to the candidate point [0, 0, 1, 0] of the transmission signal x2, and adds this error distance Δ2 and the error distance Δ1 of the candidate point x′1. To obtain the composite error distance.

  When the value of the fourth bit of the transmission signal x2 is “1”, the x1 candidate point selection and x2 error distance calculation unit 2803-2 sets the candidate point [0, 0, 1, 1] of the transmission signal x2. The candidate point x′1 of the transmission signal x1 corresponding to this time is extracted and calculated by the equation (5). The candidate point x1 ′ error distance calculation unit 2804 obtains an error distance from the reference point whose fourth bit is “0” and “1” with respect to the candidate point x′1 (Equation (12) and Equation (13)). )). The error distance synthesis unit 2805 selects the error distance Δ2 corresponding to the candidate point [0, 0, 1, 0] of the transmission signal x2, and adds the error distance Δ2 and the error distance Δ1 of the candidate point x′1. To determine the composite error distance. The Viterbi branch metric calculation unit 2808 calculates (x1, x2) = (0, 0), (0, 1), (1,0), (1, 0) with respect to the combined error distance obtained by the above calculation process. As a branch metric when taking the value of 1), the Viterbi decoding unit 2809 assigns this branch metric to the trellis map.

  Thus, according to the MIMO demodulator 280-2 of the second embodiment, the x1 candidate point selection and x2 error distance calculator 2803-2 sets the value of each bit of the transmission signal x2 to “0” and “1”. For each case, the candidate point x′1 of the transmission signal x1 is calculated, the candidate point x1 ′ error distance calculation unit 2804 calculates the error distance for the candidate point x′1, and the error distance synthesis unit 2805 The synthetic error distance was calculated. That is, in the second embodiment, the candidate point x′1 of the transmission signal x1 having a different pattern for each bit is selected for the demodulation point of the transmission signal x2, and the error distance is set only for the selected candidate point x ′. It is calculated, branch metric is calculated, and Viterbi decoding is performed. As a result, compared with the first embodiment, it is possible to further reduce the calculation amount of the MIMO demodulation and further reduce the load of the demodulation process.

  In the first and second embodiments, the likelihood calculation for performing processing by the maximum likelihood estimation method on the received data signal and the likelihood calculation for performing Viterbi decoding are unified, and the likelihood of one time. MIMO demodulation was performed by calculation. However, since QR decomposition is performed to calculate the error point Δ2 between the candidate point x′1 of the transmission signal x1 and the transmission signal x2, when the component of the matrix R used in Expression (5) is 0, the expression ( The denominator of 5) becomes 0 and the value diverges. Further, although depending on the QR decomposition algorithm, sufficient power is supplied to the first reception antenna # 1 and the second reception antenna # 2 among the reception antennas # 1 to # 4 which are the four reception antennas 201. When either or both of the transmission signal x1 and the transmission signal x2 cannot be received, the value of the matrix R itself diverges. In order to deal with these, for example, the matrix Q and the matrix R may be derived by rearranging the reception signals in descending order of the reception signal level or by exchanging the transmission signal x1 and the transmission signal x2.

  The present invention has been described with reference to the embodiments. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the technical idea thereof. For example, in the first and second embodiments, the two transmission antennas and the four reception MIMO systems configured by the two transmission antennas 101 and the four reception antennas 201 have been described. However, the present invention is not limited to this. Absent.

It is a figure which shows the structural example of a MIMO communication system. It is a figure which shows the structural example of a transmitter. It is a figure which shows the structural example of an encoding part. It is a figure which shows the structural example of an inner coding part. It is a figure which shows the structural example of a receiver. It is a flowchart which shows the process of the conventional MIMO demodulation part. It is a figure which shows the structural example of the conventional MIMO demodulation part. It is a figure which shows the structural example (Example 1) of the MIMO demodulation part in the receiver by embodiment of this invention. FIG. 3 is a flowchart showing processing of Example 1; It is a figure which shows arrangement | positioning of the transmission signal represented by 16QAM at the time of complying with ARIB STD-B33. It is a figure which shows distribution of 4 bit signal represented by 16QAM. It is a figure which shows all the candidate points and corresponding numbers of a transmission signal represented by 16QAM. It is a figure explaining the selection method of the candidate point in 4 bit signal represented by 16QAM. It is a figure which shows the structural example (The structural example at the time of making it respond | correspond to the conventional structure) of the MIMO demodulation part of Example 1. FIG. It is a flowchart which shows the process (process in the case of the number m of transmitting antennas, and the number of receiving antennas n) of Example 1. FIG. It is a figure which shows the structural example (Example 2) of the MIMO demodulation part in the receiver by embodiment of this invention. FIG. 10 is a flowchart showing processing of Example 2; It is a figure which shows the candidate point of x2 with the smallest error distance. It is a figure explaining the extraction method of the candidate point in each bit. It is a figure explaining the process of the conventional Viterbi decoding part in FIG. It is a figure explaining the process of the Viterbi decoding part in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Transmitting device 101 Transmitting antenna 110 Encoding unit 111 Energy spreading unit 112 Outer encoding unit 113 Outer interleaving unit 114 Inner encoding unit 115 Inner interleaving unit 120 Mapping unit 130 Frame configuration unit 140 IFFT unit 150 GI signal adding unit 160 Orthogonal modulation Unit 170 mixer 171 local oscillator 200 receiving device 201 receiving antenna 210 mixer 211 local oscillator 220 orthogonal demodulation unit 230 symbol synchronization detection unit 240 GI signal removal unit 250 FFT unit 260 frame separation unit 270 propagation path estimation unit 280 MIMO demodulation unit 281 total modulation Candidate point recording unit 282 Replica signal generation unit 283 Likelihood calculation unit 284 Likelihood judgment unit 285 Deinterleaving unit 286 Viterbi decoding unit 290 Decoding unit 2801 QR decomposition unit 2802 All modulation Complement point recording unit 2803-1 x1 candidate point generation and x2 error distance calculation unit 2803-2 x1 candidate point selection and x2 error distance calculation unit 2804 candidate point x1 'error distance calculation unit 2805 error distance synthesis unit 2806 memory unit 2807 internal data Interleave unit 2808 Viterbi branch metric calculation unit 2809 Viterbi decoding unit 2811 QR decomposition unit 2812 All modulation candidate point recording unit 2813 Replica signal generation unit 2814 Likelihood calculation unit 2815 Intra-deinterleaving unit 2816 Branch metric generation unit 2817 Viterbi decoding unit

Claims (4)

In a receiving apparatus used in a MIMO communication system configured by a transmitting apparatus having a plurality of transmitting antennas and a receiving apparatus having a plurality of receiving antennas,
A signal transmitted from the first transmission antenna among the plurality of transmission antennas is defined as a first transmission signal, and a signal transmitted from the second transmission antenna among the plurality of transmission antennas is defined as a second transmission signal. In case,
A QR decomposition unit that performs QR decomposition on a propagation path characteristic between the transmission antenna and the reception antenna and calculates a matrix Q and a matrix R;
A candidate point generation unit that generates all candidate points of the first transmission signal when the second transmission signal is set as a candidate point using the matrix Q and the matrix R calculated by the QR decomposition unit;
An error distance of the second transmission signal indicating an error distance between the second transmission signal and all candidate points of the second transmission signal is calculated using the matrix Q and the matrix R calculated by the QR decomposition unit. A first candidate point error distance calculation unit to be calculated;
An error between all candidate points of the first transmission signal generated by the candidate point generation unit and a reference point determined by each bit according to the order of bit signals represented by all candidate points of the first transmission signal A second candidate point error distance calculation unit for calculating the error distance of the first transmission signal indicating the distance;
Adding the error distance of the second transmission signal calculated by the first candidate point error distance calculation unit and the error distance of the first transmission signal calculated by the second candidate point error distance calculation unit; An error distance synthesis unit for calculating a synthesis error distance for each bit state;
A Viterbi branch metric that calculates a branch metric for a combination pattern of bits that can be taken by each bit of the first transmission signal and each bit of the second transmission signal, using the combined error distance calculated by the error distance combining unit. A calculation unit;
A Viterbi decoding unit that performs Viterbi decoding by substituting the branch metric calculated by the Viterbi branch metric calculating unit into a trellis map;
A receiving apparatus comprising: The receiving device according to claim 1,
When X is a transmission signal, Y is a reception signal, R11 to R22 are elements of a matrix R, and S is a candidate point,
The candidate point generation unit generates all candidate points (x′1) of the first transmission signal (x1) when the second transmission signal is set as a candidate point according to the following formulas 1 and 2.
The first candidate point error distance calculation unit calculates between the second transmission signal (x2) and all candidate points (y′2 / R ₂₂ ) of the second transmission signal according to the following equations 1 and 2. A receiving apparatus that calculates an error distance (Δ2) of a second transmission signal indicating an error distance.
[Formula 1]

[Formula 2]

The receiving apparatus according to claim 1 or 2,
The second candidate point error distance calculation unit, when the modulation method is 16QAM,
(A) The error distance between the signal point of the first bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point,
dist_x1_10 = abs (abs (Re (x'1) -2) -1)
dist_x1_11 = abs (abs (Re (x'1) +2) -1)
(B) The error distance between the signal point of the second bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point,
dist_x1_20 = abs (abs (Im (x'1) -2) -1)
dist_x1_21 = abs (abs (Im (x'1) +2) -1)
(C) An error distance between the signal point of the third bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point,
dist_x1_30 = abs (abs (Re (x'1))-3)
dist_x1_31 = abs (abs (Re (x'1))-1)
(D) The error distance between the signal point of the fourth bit of the 4-bit signal representing all candidate points x′1 of the transmission signal x1 and the reference point,
dist_x1_40 = abs (abs (Im (x'1))-3)
dist_x1_41 = abs (abs (Im (x'1))-1). In a receiving apparatus used in a MIMO communication system configured by a transmitting apparatus having a plurality of transmitting antennas and a receiving apparatus having a plurality of receiving antennas,
A signal transmitted from the first transmission antenna among the plurality of transmission antennas is defined as a first transmission signal, and a signal transmitted from the second transmission antenna among the plurality of transmission antennas is defined as a second transmission signal. In case,
A QR decomposition unit that performs QR decomposition on a propagation path characteristic between the transmission antenna and the reception antenna and calculates a matrix Q and a matrix R;
An error distance of the second transmission signal indicating an error distance between the second transmission signal and all candidate points of the second transmission signal is calculated using the matrix Q and the matrix R calculated by the QR decomposition unit. A first candidate point error distance calculation unit that calculates, selects a candidate point that minimizes the error distance, selects a signal point in which each bit of the candidate point has a different pattern, and calculates each error distance; ,
Based on the signal point of the second transmission signal selected in the first candidate point error distance calculation unit using the matrix Q and the matrix R calculated by the QR decomposition unit, the first transmission signal candidate A candidate point selection unit for selecting points;
An error distance between the candidate point of the first transmission signal selected by the candidate point selection unit and the reference point determined by each bit according to the order of the bit signals represented by the candidate point of the first transmission signal A second candidate point error distance calculator for calculating the error distance of the first transmission signal,
Adding the error distance of the second transmission signal calculated by the first candidate point error distance calculation unit and the error distance of the first transmission signal calculated by the second candidate point error distance calculation unit; An error distance synthesis unit for calculating a synthesis error distance for each bit state;
A Viterbi branch metric that calculates a branch metric for a combination pattern of bits that can be taken by each bit of the first transmission signal and each bit of the second transmission signal, using the combined error distance calculated by the error distance combining unit. A calculation unit;
A Viterbi decoding unit that performs Viterbi decoding by substituting the branch metric calculated by the Viterbi branch metric calculating unit into a trellis map;
A receiving apparatus comprising:

Images