JP6727005B2 - MIMO-OFDM signal receiver - Google Patents

Description

The present invention relates to a receiver for receiving a MIMO-OFDM signal, and more particularly, a MIMO-OFDM signal receiver for correctly receiving transmitted data even when the C/N of the received signal greatly differs among a plurality of reception sequences. Regarding

There is OFDM (Orthogonal Frequency Division Multiplexing) as a multi-carrier modulation system adopted for digital broadcasting, wireless LAN, and the like. In OFDM, a section called a guard interval (GI: Guard Interval) or a cyclic prefix (CP: Cyclic Prefix) is provided in order to obtain resistance to multipath. In OFDM, it is known that channel equalization is possible when the delay spread of the transmission path from the transmission source to the reception point is within the GI length.

On the other hand, there is a technique for transmitting information by a MIMO (Multiple Input and Multiple Output) configuration using a plurality of antennas on both the transmitting side and the receiving side, and studies on this technique have been made (Non-Patent Documents 1 and 2). ). In particular, space division multiplexing MIMO that transmits different information from each antenna has the advantage that the transmission capacity can be increased in proportion to the number of transmission antennas.

Information transmission by a MIMO configuration using OFDM as a modulation method is called MIMO-OFDM, and the advantages of both can be combined without contradiction.

Known methods for receiving space division multiplexing MIMO (Non-Patent Document 3) include a zero forcing method (Non-Patent Document 4), an MMSE method, and an MLD method. The zero forcing method has an advantage that the amount of calculation is small as compared with the other two. However, when the C/N of the reception signals is small, noise is added to the signals of the respective reception series when the interference components included in the signals of the reception series of each other are removed. The zero-forcing method has a drawback that reception characteristics are deteriorated by such noise enhancement.

It is known that the MLD method is an optimal receiving method, though the amount of calculation is significantly larger than the other two methods. The MMSE method is an intermediate receiving method in terms of the amount of calculation and the receiving characteristic, and although it is necessary to estimate the C/N of the received signal, unlike the zero forcing method, there is no noise enhancement. is there.

On the other hand, the LDPC code has been attracting much attention in recent years and has been adopted in a wide field as an error correction code approaching the Shannon limit. It is effective to use the soft decision value in the error correction code. When a soft decision value is used for decoding an LDPC code, it is common to use a log-likelihood ratio, but when obtaining this, it is necessary to estimate the variance value of the noise component superimposed on the signal. is there.

As described above, when the received signal is separated and equalized by using the MMSE method, and when the log-likelihood ratio is obtained, it is necessary to estimate the variance value of the noise component superimposed on the signal. , It is necessary to estimate this accurately in order to obtain good reception characteristics.

ここで、Ｗ_μは重み行列、Ｈはチャネル行列、Ｎ_tは送信アンテナ数、γ₀は１素子のみで全電力を送信した場合の平均受信ＳＮ比を示す。また、Ｉ_NtはＮ_t×Ｎ_tの単位行列、上付きのＴは転置、上付きのＨは複素共役転置を示す。 Generally, in a MIMO system, it is assumed that the noise included in the received signal is independent for each reception sequence, but the noise components have the same variance value. For example, the weight matrix in the MMSE method is given by the following equation (Non-Patent Document 5).

Here, W _μ is a weight matrix, H is a channel matrix, N _t is the number of transmission antennas, and γ ₀ is an average reception SN ratio when the total power is transmitted by only one element. In addition, I _Nt represents an N _t ×N _t identity matrix, superscript T represents transposition, and superscript H represents complex conjugate transposition.

Here, the presence/absence of the second term in parentheses in the equation (1) is the difference between the zero forcing method and the MMSE method. The second term in the parentheses in the equation (1) does not exist in the zero forcing method, but exists in the MMSE method.

As described above, the zero forcing method has a drawback that reception characteristics are deteriorated by noise enhancement, and the MMSE method has no noise enhancement. That is, it can be said that the second term in the parentheses in the equation (1) is an element for not causing noise enhancement. The second term in the parentheses of the equation (1), which is an element for not causing this noise enhancement, is the same as the unit matrix multiplied by the scalar amount, so that it is the same for all diagonal elements in the matrix. The amount of scalar will be added.

ここで、Ｍ_Tは送信アンテナ数（前記Ｎ_tに相当）、Ｍ_Rは受信アンテナ数、Ｈはチャネルインパルス行列（前記チャネル行列と同義）、γは受信信号ベクトルを示す。また、ρは送信信号電力、σ_n ²は雑音成分の分散値、Ｉ_MRはＭ_R×Ｍ_Rの単位行列を示す。すなわち、前記式（２）においても、考慮すべき雑音成分は、行列内の全ての対角成分において同じスカラ量である。 According to Non-Patent Document 2 described above, the reception processing of the following equation is performed by the MMSE method. The decoded signal vector is shown by the following equation.

Here, M _T is the number of transmitting antennas (corresponding to N _t ), M _R is the number of receiving antennas, H is the channel impulse matrix (synonymous with the channel matrix), and γ is the received signal vector. Further, ρ is the transmission signal power, σ _n ² is the variance value of the noise component, and I _MR is the unit matrix of M _R ×M _R. That is, also in the above equation (2), the noise component to be considered is the same scalar amount in all diagonal components in the matrix.

Further, the S/N after separation and detection used when calculating the log-likelihood ratio (corresponding to the above-mentioned variance value of the noise component when the signal power is 1) is given by the following equation (Non-Patent Document 1). 6).

は伝送路行列（チャネル行列）の逆行列、σ²は各受信アンテナの雑音成分の分散値を示す。 here,

Is the inverse of the transmission line matrix (channel matrix), and σ ² is the variance of the noise component of each receiving antenna.

Karasawa, “MIMO Propagation Channel Modeling”, IEICE, J86-B,9 (2003) “2004 standard technology collection “MIMO (multi input multi output) related technology””, JPO Ohgane, Nishimura, Ogawa, "Spatial Division Multiplexing Method and Its Basic Characteristics in MIMO Channels", IEICE, J87-B,9, pp.1162-1173 (2004) F.Becker, L.N.Holzman, R.LuckyandE.Port, “Automatic equalization for digital communication”, Proceedings of the IEEE, 53, 1, pp.96-97 (1965) Ohgane, Ogawa, "Intelligible MIMO system technology", Ohmsha (2009) Hira, Ishizu, Murakami, "Reception characteristics of MIMO-OFDM system using SDM and STC", IEICE Technical Report RCS, 103, 128, pp.85-90 (2003).

As shown in the equations (3) and (4), when the S/N after separation and detection used when calculating the log-likelihood ratio is obtained, the variance value of the noise component of the reception antenna is the same for each reception sequence. It is said that there is.

However, the variance value of the noise component may differ greatly depending on the reception sequence, for example, when the transmission side transmits with different output magnitudes for each transmission sequence. In this case, the MMSE method has a problem in that the weight matrix for separation and detection and the log-likelihood ratio cannot be accurately calculated.

Therefore, the present invention has been made in order to solve such a problem, and an object thereof is a MIMO- which can accurately estimate a variance value of a noise component included in a received signal and accurately calculate a log likelihood ratio. It is to provide an OFDM signal receiving device.

In order to solve the above problems, the MIMO-OFDM signal receiving apparatus according to claim 1 FFTs an equivalent baseband signal at a time corresponding to an effective symbol period of a MIMO-OFDM signal for each reception sequence, and frequency-domains the time-domain signal. An FFT unit for converting into a domain signal, a channel estimation unit for estimating a channel matrix from the frequency domain signal converted by the FFT unit, and a weight matrix for calculating a weight matrix from the channel matrix estimated by the channel estimation unit. A calculation unit, a spatial filter unit that multiplies the frequency domain signal converted by the FFT unit and the weight matrix calculated by the weight matrix calculation unit, and performs separation and detection of a reception signal for each reception sequence; A band noise variance calculation unit that calculates a noise variance value of the entire band for each reception sequence based on the received signal for each reception sequence that has been separated and detected by the spatial filter unit; and the band noise variance calculation unit. A noise variance calculation unit that calculates a noise variance value for each reception sequence and for each carrier based on the noise variance value of the entire band for each reception sequence calculated by the above and the weight matrix calculated by the weight matrix calculation unit A received signal for each reception sequence that has been separated and detected by the spatial filter, and for each reception sequence calculated by the noise variance calculation unit and for each carrier based on the noise variance value, A QAM demodulation unit that calculates a log-likelihood ratio for each bit, a multiplexing unit that multiplexes the log-likelihood ratios for each reception sequence and for each bit calculated by the QAM demodulation unit, and the above-mentioned multiplexing unit An error correction decoding unit that performs an error correction decoding process using a log-likelihood ratio to generate and output a bit string, and the noise variance calculation unit is the weight matrix of the weight matrix calculated by the weight matrix calculation unit. A column vector for each reception sequence is extracted from the amplitude calculation unit that calculates the amplitude of each element and generates a matrix having the amplitude of each element of the weight matrix as an element, and the matrix generated by the amplitude calculation unit. A column vector extraction unit, a row vectorization unit that generates a row vector from the noise variance value of the entire band for each reception sequence calculated by the band noise variance calculation unit, and the row vector extraction unit that extracts the row vector A column vector for each reception sequence and the row vector generated by the row vectorization unit are multiplied, and a multiplication unit that outputs a multiplication result for each reception sequence, and a multiplication unit for each reception sequence output by the multiplication unit The multiplication result is The arithmetic averaging unit for averaging over the rear and calculating the multiplication result of the entire band for each reception sequence, and the multiplication result for each reception sequence output by the multiplication unit were calculated by the arithmetic averaging unit. A division unit that divides by the multiplication result of the entire band for each reception sequence and outputs a division result for each reception sequence, a division result for each reception sequence that is output by the division unit, and a row vectorization unit And a multiplying unit that obtains a noise variance value for each reception sequence and for each carrier by multiplying the row vector thus generated.

The MIMO-OFDM signal receiving apparatus according to claim 2 is the MIMO-OFDM signal receiving apparatus according to claim 1, wherein the band noise variance calculation unit includes an output stage band noise variance calculation unit and an input stage band noise variance calculation. A symbol regenerating unit that reconstructs the received signal for each received sequence separated and detected by the spatial filter unit, and generates a regenerated signal for each received sequence. And subtracting the reception signal for each reception sequence separated and detected by the spatial filter unit from the reproduction signal for each reception sequence generated by the symbol reproduction unit to generate an error signal for each reception sequence. A subtraction unit, an amplitude calculation unit that calculates the amplitude of the error signal for each reception sequence generated by the subtraction unit, and the amplitude of the error signal for each reception sequence calculated by the amplitude calculation unit for all carriers. And an arithmetic averaging unit for averaging over and calculating an output stage band noise variance value of the entire band for each reception sequence, wherein the input stage band noise variance calculation unit is calculated by the weight matrix calculation unit. An amplitude calculating unit that calculates the amplitude of each element of the weight matrix and generates a matrix having the amplitude of each element of the weight matrix as an element, and an inverse matrix that calculates the inverse matrix of the matrix generated by the amplitude calculating unit. A calculation unit, an arithmetic averaging unit that averages the inverse matrix calculated by the inverse matrix calculation unit over all carriers, and calculates an inverse matrix having an average value of the entire band as an element; and the output stage band noise The output stage band noise variance value of the entire band for each reception sequence calculated by the variance calculation unit is multiplied by the inverse matrix calculated by the arithmetic mean unit, and the input stage of the entire band for each reception sequence is multiplied. An input noise variance calculation unit that obtains a band noise variance value and outputs the input stage band noise variance value of the entire band for each reception sequence as a noise variance value of the entire band for each reception sequence. To do.

The MIMO-OFDM signal receiving apparatus according to claim 3 performs FFT on an equivalent baseband signal at a time corresponding to an effective symbol period of a MIMO-OFDM signal for each reception sequence, and transforms a time domain signal into a frequency domain signal. Calculating a weight matrix based on a zero forcing criterion from a channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit, and the channel matrix estimated by the channel estimation unit. 1 weight matrix calculation unit, the frequency domain signal converted by the FFT unit, and the weight matrix calculated by the first weight matrix calculation unit are multiplied to separate the reception signal for each reception sequence, and It is included in the output signal of the first spatial filter unit based on the first spatial filter unit that performs detection and the received signal for each reception sequence that has been separated and detected by the first spatial filter unit. A noise variance value, an output stage band noise variance calculation unit that calculates the output stage band noise variance value of the entire band for each reception sequence, and the entire band for each reception sequence calculated by the output stage band noise variance calculation unit. Based on the output stage band noise variance value and the weight matrix calculated by the first weight matrix calculation unit, the noise variance value included in the input signal of the first spatial filter unit is set to a band for each reception sequence. An input stage band noise variance calculation unit that calculates the overall input stage band noise variance value, the channel matrix estimated by the channel estimation unit, and each of the reception sequences calculated by the input stage band noise variance calculation unit A second weight matrix calculation unit that calculates a weight matrix based on the MMSE criterion from the input stage band noise variance value of the entire band, the frequency domain signal converted by the FFT unit, and the second weight matrix calculation A second spatial filter unit that multiplies the weight matrix calculated by the unit to separate and detect a received signal for each reception sequence; and for each reception sequence calculated by the input stage band noise variance calculation unit. A noise variance calculation unit that calculates a noise variance value for each reception sequence and for each carrier based on the input stage band noise variance value of the entire band and the weight matrix calculated by the second weight matrix calculation unit; Based on the received signal for each received sequence separated and detected by the second spatial filter, and the noise variance value for each received sequence and for each carrier calculated by the noise variance calculation unit, for each received sequence, Log-likelihood ratio for each bit Using a QAM demodulation section for calculating, a multiplexing section for multiplexing the log-likelihood ratio for each reception sequence and for each bit calculated by the QAM demodulation section, and the log-likelihood ratio multiplexed by the multiplexing section. An error correction decoding unit that performs error correction decoding processing to generate and output a bit string, and the noise variance calculation unit is the amplitude of each element of the weight matrix calculated by the second weight matrix calculation unit. And a column vector extraction unit that extracts a column vector for each reception sequence from the amplitude calculation unit that generates a matrix having the amplitude of each element of the weight matrix as an element, and the matrix generated by the amplitude calculation unit. And a row vectorization unit that generates a row vector from the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit and the column vector extraction unit. A multiplication unit that multiplies the column vector for each reception sequence by the row vector generated by the row vectorization unit and outputs a multiplication result for each reception sequence; and for each reception sequence output by the multiplication unit. The arithmetic mean of the multiplication result of each reception sequence is calculated by averaging the multiplication results of all the carriers, and the multiplication result of each reception sequence output by the multiplication unit is calculated by the arithmetic mean. Division by the multiplication result of the entire band for each reception sequence calculated by the unit, a division unit for outputting the division result for each reception sequence, the division result for each reception sequence output by the division unit, the line And a multiplication unit that multiplies the row vector generated by the vectorization unit to obtain a noise variance value for each reception sequence and each carrier.

Further, the MIMO-OFDM signal receiving apparatus according to claim 4 is the MIMO-OFDM signal receiving apparatus according to claim 3, wherein the second weight matrix calculating unit is configured to calculate the channel matrix of the channel matrix estimated by the channel estimating unit. A complex conjugate transpose unit that calculates a complex conjugate transpose matrix, the channel matrix estimated by the channel estimation unit, and the complex conjugate transpose matrix calculated by the complex conjugate transpose unit are multiplied, and the matrix of the multiplication result is The first multiplication unit for outputting and the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit are set to diagonal components, and zero is set to a non-diagonal component. By setting, the diagonal matrix generated by the diagonal matrix generation unit and the diagonal matrix generated by the diagonal matrix generation unit are added to the matrix of the multiplication result output by the first multiplication unit. An addition unit that adds and outputs a matrix of the addition result, an inverse matrix calculation unit that calculates an inverse matrix of the matrix of the addition result output by the addition unit, and the inverse matrix calculated by the inverse matrix calculation unit And a multiplication unit that multiplies the complex conjugate transpose matrix calculated by the complex conjugate transpose unit to obtain the weight matrix.

Further, the MIMO-OFDM signal receiving apparatus according to claim 5 FFTs an equivalent baseband signal at a time corresponding to an effective symbol period of a MIMO-OFDM signal for each reception sequence, and transforms a time domain signal into a frequency domain signal. Calculating a weight matrix based on a zero forcing criterion from a channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit, and the channel matrix estimated by the channel estimation unit. 1 weight matrix calculation unit, the frequency domain signal converted by the FFT unit, and the weight matrix calculated by the first weight matrix calculation unit are multiplied to separate the reception signal for each reception sequence, and It is included in the output signal of the first spatial filter unit based on the first spatial filter unit that performs detection and the received signal for each reception sequence that has been separated and detected by the first spatial filter unit. A noise variance value, an output stage band noise variance calculation unit that calculates the output stage band noise variance value of the entire band for each reception sequence, and the entire band for each reception sequence calculated by the output stage band noise variance calculation unit. Based on the output stage band noise variance value and the weight matrix calculated by the first weight matrix calculation unit, the noise variance value included in the input signal of the first spatial filter unit is set to a band for each reception sequence. An input stage band noise variance calculation unit that calculates a total input stage band noise variance value, an input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit, and the first Based on the weight matrix calculated by the first weight matrix calculating unit, the noise variance calculating unit calculates the noise variance value for each reception sequence and each carrier, and the first spatial filter performs separation and detection. A QAM demodulation unit that calculates a log-likelihood ratio for each reception sequence and for each bit based on the received signal for each reception sequence and the noise variance value for each reception sequence and each carrier calculated by the noise variance calculation unit An error correction decoding process using a multiplexing unit that multiplexes the log-likelihood ratios for each reception sequence and each bit calculated by the QAM demodulation unit, and the log-likelihood ratios that are multiplexed by the multiplexing unit. , An error correction decoding unit that generates and outputs a bit string, the noise variance calculation unit calculates the amplitude of each element of the weight matrix calculated by the first weight matrix calculation unit, and the weight Amplitude math that creates a matrix whose elements are the amplitudes of each element of the matrix An output unit, a column vector extraction unit that extracts a column vector for each reception sequence from the matrix generated by the amplitude calculation unit, and an entire band for each reception sequence calculated by the input stage band noise variance calculation unit A row vectorization unit that generates a row vector from the input stage band noise variance value, a column vector for each reception sequence extracted by the column vector extraction unit, and the row vector generated by the row vectorization unit And a multiplication unit that outputs a multiplication result for each reception sequence, and a multiplication result for each reception sequence output by the multiplication unit is averaged over all carriers, and a multiplication result for the entire band for each reception sequence. The arithmetic mean unit for calculating and the multiplication result for each reception sequence output by the multiplication unit is divided by the multiplication result for the entire band for each reception sequence calculated by the arithmetic mean unit, and the reception sequence A division unit for outputting a division result for each, a division result for each reception sequence output by the division unit, and the row vector generated by the row vectorization unit are multiplied, and each reception sequence and carrier And a multiplication unit that obtains a noise variance value for each.

The MIMO-OFDM signal receiving apparatus according to claim 6 performs FFT on an equivalent baseband signal at a time corresponding to an effective symbol period of a MIMO-OFDM signal for each reception sequence, and transforms a time domain signal into a frequency domain signal. Calculating a weight matrix based on a zero forcing criterion from a channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit, and the channel matrix estimated by the channel estimation unit. 1 weight matrix calculation unit, the frequency domain signal converted by the FFT unit, and the weight matrix calculated by the first weight matrix calculation unit are multiplied to separate the reception signal for each reception sequence, and It is included in the output signal of the first spatial filter unit based on the first spatial filter unit that performs detection and the received signal for each reception sequence that has been separated and detected by the first spatial filter unit. A noise variance value, an output stage band noise variance calculation unit that calculates the output stage band noise variance value of the entire band for each reception sequence, and the entire band for each reception sequence calculated by the output stage band noise variance calculation unit. A noise variance calculation unit that calculates a noise variance value for each reception sequence and each carrier based on the output stage band noise variance value and the weight matrix calculated by the first weight matrix calculation unit; Based on the received signal for each reception sequence that has been separated and detected by the spatial filter, and for each reception sequence and for each carrier calculated by the noise variance calculation unit, based on the noise variance value, for each reception sequence and for each bit A QAM demodulation unit that calculates a log-likelihood ratio, a multiplexing unit that multiplexes the log-likelihood ratios for each reception sequence and each bit calculated by the QAM demodulation unit, and the log-likelihood that is multiplexed by the multiplexing unit. An error correction decoding unit that performs an error correction decoding process using a ratio to generate and output a bit string, and the noise variance calculation unit calculates the weight matrix of the weight matrix calculated by the first weight matrix calculation unit. A column vector for each reception sequence is extracted from the amplitude calculation unit that calculates the amplitude of each element and generates a matrix having the amplitude of each element of the weight matrix as an element, and the matrix generated by the amplitude calculation unit. A column vector extraction unit, a row vectorization unit that generates a row vector from the output stage band noise variance value of the entire band for each reception sequence calculated by the output stage band noise variance calculation unit, and the column vector extraction unit A column vector for each of the reception sequences extracted by , A multiplication unit that multiplies the row vector generated by the row vectorization unit and outputs a multiplication result for each reception sequence, and a multiplication result for each reception sequence output by the multiplication unit to all carriers. And averaging to calculate the multiplication result of the entire band for each reception sequence, and the multiplication result for each reception sequence output by the multiplication unit, the reception sequence calculated by the arithmetic averaging unit. Each division by the multiplication result of the entire band, a division unit for outputting the division result for each reception sequence, the division result for each reception sequence output by the division unit, and the row vectorization unit generated And a multiplication unit for multiplying the row vector and the noise variance value for each reception sequence and each carrier.

Furthermore, the chip according to claim 7 is, in a chip mounted in a MIMO-OFDM signal receiving apparatus, FFTs an equivalent baseband signal at a time corresponding to an effective symbol period of a MIMO-OFDM signal for each reception sequence to obtain a time domain signal. To a frequency domain signal, a channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit, and a weight matrix is calculated from the channel matrix estimated by the channel estimation unit. A spatial filter that multiplies a weight matrix calculating unit, the frequency domain signal converted by the FFT unit, and the weight matrix calculated by the weight matrix calculating unit to separate and detect a received signal for each reception sequence. A band noise variance calculation unit that calculates a noise variance value of the entire band for each reception sequence based on the received signal for each reception sequence that has been separated and detected by the spatial filter unit, and the band noise variance Noise variance for calculating a noise variance value for each reception sequence and for each carrier based on the noise variance value of the entire band for each reception sequence calculated by the calculation unit and the weight matrix calculated by the weight matrix calculation unit Based on the received signal for each reception sequence separated and detected by the spatial filter and the noise variance value for each reception sequence and each carrier calculated by the noise variance calculation unit, the reception sequence QAM demodulation section for calculating the log-likelihood ratio for each bit and for each bit, a multiplexing section for multiplexing the log-likelihood ratio for each reception sequence and for each bit calculated by the QAM demodulation section, and An error correction decoding unit that performs an error correction decoding process using the log-likelihood ratio, generates a bit string, and outputs the bit string, wherein the noise variance calculation unit calculates the weights calculated by the weight matrix calculation unit. A column vector for each reception sequence is calculated from the amplitude calculation unit that calculates the amplitude of each element of the matrix and generates a matrix having the amplitude of each element of the weight matrix as an element, and the matrix generated by the amplitude calculation unit. A column vector extraction unit for extracting, a row vectorization unit for generating a row vector from the noise variance value of the entire band for each reception sequence calculated by the band noise variance calculation unit, and a column vector extraction unit for extraction. A multiplication unit that multiplies the column vector for each reception sequence and the row vector generated by the row vectorization unit and outputs a multiplication result for each reception sequence; and the reception sequence output by the multiplication unit. For each The arithmetic averaging unit for averaging the multiplication results over all carriers to calculate the multiplication result for the entire band for each reception sequence, and the arithmetic averaging unit for the multiplication result for each reception sequence output by the multiplication unit. By dividing by the multiplication result of the entire band for each reception sequence calculated by, the division unit for outputting the division result for each reception sequence, the division result for each reception sequence output by the division unit, the row vector And a multiplication unit that multiplies the row vector generated by the conversion unit to obtain a noise variance value for each reception sequence and each carrier.

As described above, according to the present invention, it is possible to accurately estimate the variance value of the noise component included in the received signal and accurately calculate the log-likelihood ratio.

1 is a block diagram showing a configuration example of a MIMO-OFDM signal receiving apparatus according to a first embodiment. FIG. It is a block diagram which shows the structural example of an output stage band noise variance calculation part. It is a block diagram which shows the structural example of an input stage band noise dispersion calculation part. It is a block diagram which shows the structural example of a 2nd weight matrix calculation part. It is a block diagram which shows the structural example of a noise variance calculation part. FIG. 9 is a block diagram showing a configuration example of a MIMO-OFDM signal receiving device according to a second embodiment. FIG. 9 is a block diagram showing a configuration example of a MIMO-OFDM signal receiving device according to a third embodiment.

Hereinafter, modes for carrying out the present invention will be described in detail with reference to the drawings. For simplicity of description, the number of transmitting antennas and the number of receiving antennas are both 2 in the following MIMO system. The same applies to Examples 2 and 3 described later. The present invention does not limit the number of transmitting antennas and the number of receiving antennas to two.

[Example 1]
First, the first embodiment will be described. FIG. 1 is a block diagram illustrating a configuration example of the MIMO-OFDM signal receiving apparatus according to the first embodiment. The MIMO-OFDM signal receiving apparatus 1 includes FFT (Fast Fourier Transform) units 10-1 and 10-2, a channel estimation unit 11, and weight matrix calculation units 12-1 and 12-2 corresponding to the number of reception sequences. , Spatial filter units 13-1, 13-2, output stage band noise variance calculation unit 14, input stage band noise variance calculation unit 15, noise variance calculation unit 16, QAM demodulation units 17-1, 17-2, multiplexing unit 18 And an error correction decoding unit 19.

The frequency conversion unit for the number of reception sequences (not shown) frequency-converts the MIMO-OFDM signal received by the MIMO-OFDM signal receiving apparatus 1 and converts it into an IF signal. The IF signals output from the frequency conversion units corresponding to the number of reception sequences are input to A/D conversion units (not shown). The A/D conversion unit for the number of reception sequences performs A/D conversion on the IF signal input from the frequency conversion unit and outputs a digital IF signal. The digital IF signals output from the A/D conversion units corresponding to the number of reception sequences are input to the orthogonal demodulation units (not shown). The orthogonal demodulation unit for the number of reception sequences orthogonally demodulates the digital IF signal input from the A/D conversion unit and outputs an equivalent baseband signal. The equivalent baseband signals output by the orthogonal demodulation unit for the number of received sequences are input to the FFT units 10-1 and 10-2 for the number of received sequences, respectively. The same applies to Examples 2 and 3 described later.

The FFT units 10-1 and 10-2 corresponding to the number of reception sequences perform FFT on the equivalent baseband signal of the time equivalent to the effective symbol period of the OFDM signal among the equivalent baseband signals input from the orthogonal demodulation unit, and Convert the domain signal to a frequency domain signal. The frequency domain signals output from the FFT units 10-1 and 10-2 corresponding to the number of received sequences are divided into three and input to the spatial filter units 13-1 and 13-2 and the channel estimation unit 11, respectively.

ここで、ｋはキャリヤ番号、上付きのＴは転置を示す。 The frequency domain signal X _k output by the FFT units 10-1 and 10-2 is shown by the following equation.

Here, k represents a carrier number, and superscript T represents transposition.

The channel estimation unit 11 extracts a predetermined signal from the frequency domain signals input from the FFT units 10-1 and 10-2 and estimates a channel matrix (channel matrix) for each carrier. The channel matrix output from the channel estimation unit 11 is divided into two, one is input to the weight matrix calculation unit 12-1 and the other is input to the weight matrix calculation unit 12-2.

In order to estimate the channel matrix, a signal known as a pilot signal or a training signal on the receiving side is used. These signals may be transmitted from the transmitting side so that the transmission sequences are orthogonal to each other, or may be transmitted alternately from only one transmitting antenna among a plurality of transmitting antennas. The method of channel estimation is a known technique, and therefore its explanation is omitted. The channel matrix H _k output from the channel estimation unit 11 is shown by the following equation.

The weight matrix calculation unit 12-1 uses the channel matrix input from the channel estimation unit 11 to calculate the weight matrix based on the zero forcing criterion. The weight matrix output from the weight matrix calculation unit 12-1 is divided into two, one is input to the spatial filter unit 13-1 and the other is input to the input stage band noise variance calculation unit 15.

ｚｆは、ゼロフォーシング規範に基づいていることを示している。 The weight matrix W _k ^zf output from the weight matrix calculation unit 12-1 is shown by the following equation.

zf indicates that it is based on the zero forcing standard.

The spatial filter unit 13-1 receives the frequency domain signals from the FFT units 10-1 and 10-2, receives the weight matrix from the weight matrix calculation unit 12-1, and multiplies the weight matrix by the frequency domain signal. , Separates and detects received signals for the number of received sequences. The received signals (separated and detected signals) for the number of received sequences output from the spatial filter unit 13-1 are input to the output stage band noise variance calculation unit 14.

The received signal Y _k ^zf of the carrier number k corresponding to the number of received sequences output from the spatial filter unit 13-1 is shown by the following equation.

Here, the received signals (separated and detected signals) for the number of received sequences output by the spatial filter unit 13-1 are signals separated and detected using a weighting matrix based on the zero-forcing criterion, and thus spatial Noise is added by the processing of the filter unit 13-1. That is, the reception signals (separation and detection signals) for the number of reception sequences output by the spatial filter unit 13-1 are noise-enhanced signals.

(Output stage band noise variance calculation unit 14)
The output stage band noise variance calculation unit 14 receives the received signals (separated and detected signals) as many as the number of received sequences from the spatial filter unit 13-1, and calculates the variance value of the noise component contained in each of the received signals as a band for each received sequence. It is calculated as the total output stage band noise variance value. Here, the output stage in the output stage band noise variance value means the output side of the spatial filter unit 13-1, and the output stage band noise variance value is the received signal output by the spatial filter unit 13-1 (spatial filter unit 13-1). 13-1 shows the variance value of the noise component included in the output signal).

FIG. 2 is a block diagram showing a configuration example of the output stage band noise variance calculation unit 14. The output stage band noise variance calculation unit 14 includes symbol reproduction units 20-1 and 20-2 corresponding to the number of received sequences, subtraction units 21-1 and 21-2, amplitude calculation units 22-1 and 22-2, and addition. The averaging units 23-1 and 23-2 are provided.

The received signals (separated and detected signals) for the number of received sequences input from the spatial filter unit 13-1 are each divided into two, one to the symbol reproducing units 20-1 and 20-2, and the other to the subtracting unit 21-1. , 21-2, respectively.

The symbol reproduction units 20-1 and 20-2 receive the received signals (separation and detection signals) of the corresponding reception sequence from the spatial filter unit 13-1 and reproduce the symbols to generate reproduced signals. The reproduction signals for the number of reception sequences output from the symbol reproduction units 20-1 and 20-2 are input to the subtraction units 21-1 and 21-2.

ここで、decはシンボル再生の関数を示し、引数に最も近い送信シンボルを返す。また、ｌは受信系列を示す。 The reproduced signals Z _k,l ^{zf corresponding} to the number of reception sequences output from the symbol reproducing units 20-1 and 20-2 are shown by the following equation.

Here, dec indicates a symbol reproduction function, and returns the transmitted symbol closest to the argument. Further, l indicates a reception sequence.

The subtraction units 21-1 and 21-2 receive the reproduction signals of the corresponding reception sequences from the symbol reproduction units 20-1 and 20-2, and receive the reception signals of the corresponding reception sequences (separation and separation from the spatial filter unit 13-1). The detection signal) is input. Then, the subtraction units 21-1 and 21-2 subtract the reception signal (separation and detection signal) from the reproduction signal to obtain an error signal. The error signals for the number of reception sequences output from the subtraction units 21-1 and 21-2 are input to the amplitude calculation units 22-1 and 22-2.

The error signals e _k,l for the number of reception sequences output by the subtraction units 21-1 and 21-2 are shown by the following equation.

The amplitude calculation units 22-1 and 22-2 receive the corresponding error signals from the subtraction units 21-1 and 21-2, and calculate the amplitudes thereof. The amplitudes of the error signals corresponding to the number of reception sequences output from the amplitude calculation units 22-1 and 22-2 are input to the arithmetic mean units 23-1 and 23-2.

The amplitude σ _k,l ² of the error signals for the number of reception sequences output from the amplitude calculation units 22-1 and 22-2 is shown by the following equation.

The arithmetic mean units 23-1 and 23-2 receive the amplitudes of the error signals of the corresponding reception sequences from the amplitude calculation units 22-1 and 22-2, average the amplitudes of the error signals over all carriers, Calculate the noise variance of the entire band. The noise variance value of the entire band corresponding to the number of received sequences output from the arithmetic mean units 23-1 and 23-2 is the input stage band noise variance calculation unit 15 as the output stage band noise variance value of the entire band for each received sequence. Is input to.

The noise variance value σ _l ² of the entire band for each reception sequence output from the arithmetic mean units 23-1 and 23-2 is shown by the following equation.

As a result, in the output-stage band noise variance calculation unit 14, the received signal for the number of received sequences input from the spatial filter unit 13-1 is calculated by the variance value of the noise component included in each received signal for each received sequence. The output stage band noise variance value of the entire band is calculated. Here, as described above, the reception signals for the number of reception sequences output by the spatial filter unit 13-1 are noise-emphasized signals.

Therefore, the output stage band noise variance value of the entire band for each reception sequence calculated by the output stage band noise variance calculator 14 is a value affected by noise enhancement and can be said to include an error. Therefore, the input-stage band noise variance calculation unit 15, which will be described later, calculates a noise variance value excluding the influence of noise enhancement.

(Input stage noise variance calculator 15)
Returning to FIG. 1, the input stage band noise variance calculation unit 15 receives the weight matrix from the weight matrix calculation unit 12-1, and the output stage band noise variance calculation unit 14 outputs the output stage band noise of the entire band for each reception sequence. The variance value is entered. Then, the input stage band noise variance calculation unit 15 eliminates the influence of noise enhancement from the output stage band noise variance value of the entire band for each reception sequence, and obtains the input stage band noise variance value of the entire band for each reception sequence. Here, the input stage in the input stage band noise variance value means the input side of the spatial filter unit 13-1, and the input stage band noise variance value is the received signal (spatial filter unit 13) input by the spatial filter unit 13-1. −1 input signal: frequency domain signal) indicates the variance value of the noise component included in the signal.

FIG. 3 is a block diagram showing a configuration example of the input stage band noise variance calculation unit 15. The input stage band noise variance calculator 15 includes an amplitude calculator 30, an inverse matrix calculator 31, an arithmetic mean 32, and an input noise variance calculator 33.

The weight matrix input from the weight matrix calculation unit 12-1 is input to the amplitude calculation unit 30. The amplitude calculation unit 30 receives the weight matrix from the weight matrix calculation unit 12-1, calculates the amplitude for each element of the weight matrix, and generates a matrix having the amplitude of each element of the weight matrix as an element. The matrix composed of the amplitudes of the elements of the weight matrix output from the amplitude calculation unit 30 is input to the inverse matrix calculation unit 31.

The matrix V _k consisting of the amplitudes of the elements of the weight matrix output from the amplitude calculator 30 is shown by the following equation.

The inverse matrix calculation unit 31 receives the matrix of the amplitudes of the elements of the weight matrix from the amplitude calculation unit 30, and calculates the inverse matrix of this matrix. The inverse matrix formed by the amplitudes of the elements of the weight matrix output from the inverse matrix calculation unit 31 is input to the arithmetic mean unit 32.

The inverse matrix U _k output from the inverse matrix calculator 31 is shown by the following equation.

The arithmetic mean unit 32 receives the inverse matrix from the inverse matrix calculation unit 31, averages the inverse matrix over all carriers, and calculates an inverse matrix having an arithmetic average value of the entire band as an element. The inverse matrix having the arithmetic mean value of the entire band output from the arithmetic mean unit 32 is input to the input noise variance calculation unit 33.

An inverse matrix having the arithmetic mean value of the entire band output by the arithmetic mean unit 32 as an element is shown by the following equation.

The input noise variance calculation unit 33 receives the inverse matrix having the arithmetic average value of the entire band as an element from the arithmetic mean unit 32, and the output stage band noise variance calculation unit 14 outputs the output stage band of the entire band for each reception sequence. The noise variance value is input. Then, the input noise variance calculation unit 33 multiplies the inverse matrix having the arithmetic mean value of the entire band as an element by a matrix having the output stage band noise variance value of the entire band of each reception sequence as an element, and The variance value of a certain noise component is obtained as the input stage band noise variance value of the entire band for each reception sequence. The input stage band noise variance value of the entire band for each reception sequence output from the input noise variance calculator 33 is divided into two, one is input to the weight matrix calculator 12-2 and the other is input to the noise variance calculator 16. It

The input stage band noise variance value of the entire band for each reception sequence output from the input noise variance calculator 33 is shown by the following equation.

The processing of the input noise variance calculation unit 33 is expressed by the following equation.

As a result, in the input stage band noise variance calculation unit 15, the inverse matrix of the arithmetic mean based on the elements of the weight matrix that causes noise enhancement is multiplied by the output stage band noise variance value affected by noise enhancement. , The input stage noise variance value is obtained. Therefore, the input stage band noise variance value excluding the effect of noise enhancement generated in the spatial filter unit 13-1 is calculated.

(Weight matrix calculating unit 12-2)
Returning to FIG. 1, the weight matrix calculation unit 12-2 receives the channel matrix from the channel estimation unit 11 and the input stage band noise variance value of the entire band for each reception sequence from the input stage band noise variance calculation unit 15. To be done. Then, the weight matrix calculating unit 12-2 calculates the weight matrix based on the MMSE norm using the channel matrix and the input stage band noise variance value of the entire band for each reception sequence.

FIG. 4 is a block diagram showing a configuration example of the weight matrix calculation unit 12-2. The weight matrix calculating unit 12-2 includes a complex conjugate transposing unit 40, a multiplying unit 41, a diagonal matrix generating unit 42, an adding unit 43, an inverse matrix calculating unit 44, and a multiplying unit 45.

The channel matrix input from the channel estimation unit 11 is divided into two, one is input to the complex conjugate transpose unit 40 and the other is input to the multiplication unit 41. The complex conjugate transpose unit 40 receives the channel matrix from the channel estimation unit 11 and calculates the complex conjugate transpose matrix of the channel matrix. The complex conjugate transpose matrix of the channel matrix output from the complex conjugate transpose unit 40 is divided into two, one is input to the multiplication unit 41 and the other is input to the multiplication unit 45.

The multiplication unit 41 receives the channel matrix from the channel estimation unit 11 and the complex conjugate transposed matrix of the channel matrix from the complex conjugate transpose unit 40, and multiplies the complex conjugate transposed matrix of the channel matrix by the channel matrix. The matrix of the multiplication result output from the multiplication unit 41 is input to the addition unit 43.

The input stage band noise variance value of the entire band for each reception sequence, which is input from the input stage band noise variance calculation unit 15, is input to the diagonal matrix generation unit 42. The diagonal matrix generation unit 42 sets the input stage band noise variance value of the entire band for each reception sequence input from the input stage band noise variance calculation unit 15 as a diagonal component and zero as a non-diagonal component. As a result, a diagonal matrix of the input stage band noise variance value is generated. The diagonal matrix of the input stage band noise variance value output from the diagonal matrix generation unit 42 is input to the addition unit 43.

The addition unit 43 receives the matrix of the multiplication result from the multiplication unit 41, and the diagonal matrix of the input stage band noise variance value from the diagonal matrix generation unit 42. Then, the adding unit 43 adds the diagonal matrix of the input stage band noise variance value to the matrix of the multiplication result. The matrix of the addition result output from the adder 43 is input to the inverse matrix calculator 44.

The inverse matrix calculation unit 44 receives the matrix of the addition result from the addition unit 43 and calculates the inverse matrix of the matrix of the addition result. The inverse matrix output from the inverse matrix calculator 44 is input to the multiplier 45.

The multiplication unit 45 receives the inverse matrix from the inverse matrix calculation unit 44, receives the complex conjugate transposed matrix of the channel matrix from the complex conjugate transpose unit 40, multiplies the inverse matrix by the complex conjugate transposed matrix of the channel matrix, and outputs the multiplication result. Find the weight matrix of. The weight matrix output from the multiplication unit 45 is divided into two, one is input to the spatial filter unit 13-2 and the other is input to the noise variance calculation unit 16.

ここで、diagは引数を対角成分に持つ行列を示す。ｍｍｓｅは、ＭＭＳＥ規範に基づいていることを示している。

The process of the weight matrix calculation unit 12-2 is expressed by the following equation.

Here, diag indicates a matrix having arguments as diagonal elements. mmse indicates that it is based on the MMSE standard.

Comparing the conventional weight matrix based on the MMSE norm shown in the equation (1) with the weight matrix of the first embodiment based on the MMSE norm shown in the equation (19), the second term in the parentheses is They are the same in that they are elements for not causing noise enhancement (elements for suppressing). However, it can be seen that there is a clear difference between the second item in the parentheses. The second term in the parentheses of the equation (1) is an identity matrix multiplied by the scalar amount, while the second term in the parentheses of the equation (19) is a pair having different scalar amounts in diagonal components. Both are different in that they are angular matrices.

Returning to FIG. 1, the spatial filter unit 13-2 receives the frequency domain signal from the FFT units 10-1 and 10-2, the weight matrix from the weight matrix calculation unit 12-2, and the frequency domain signal in the weight matrix. By multiplying the signals, the received signals for the number of received sequences are separated and detected. The received signals for the number of reception sequences output from the spatial filter unit 13-2 are input to the QAM demodulation units 17-1 and 17-2.

The received signal Y _k ^mmse of the carrier number k corresponding to the number of received sequences output from the spatial filter unit 13-2 is shown by the following equation.

Here, the received signals for the number of received sequences output by the spatial filter unit 13-2 are signals separated and detected by using a weight matrix based on the MMSE standard, and the weight matrix is noise enhancement. Is a matrix that has different scalar quantities in the diagonal components as elements for suppressing the, and reflects the input stage noise variance. Therefore, unlike the spatial filter unit 13-1, the spatial filter unit 13-2 balances the insufficiency regarding separation between sequences and the noise enhancement accompanying the separation, and as a result, in the signal after separation and detection. Noise and interference components between systems are minimized.

(Noise variance calculator 16)
The noise variance calculation unit 16 receives the weight matrix from the weight matrix calculation unit 12-2, and the input stage band noise variance calculation unit 15 receives the input stage band noise variance value of the entire band for each reception sequence. Then, the noise variance calculation unit 16 calculates the variance value of the noise component as the output stage band noise variance value for each reception sequence and each carrier.

なお、簡潔のため、前記式（２２）（２３）において、重み行列Ｗの上付きのmmseは省略してある。 The output stage band noise variance value for each received sequence and for each carrier output from the noise variance calculation unit 16 is shown by the following equation. This equation represents the processing of the noise variance calculation unit 16.

In addition, for simplification, the superscript mmse of the weight matrix W is omitted in the formulas (22) and (23).

Here, the output stage in this output stage band noise variance value means the output side of the spatial filter unit 13-2, and the output stage band noise variance value is the noise included in the received signal output by the spatial filter unit 13-2. The dispersion value of the component is shown.

FIG. 5 is a block diagram showing a configuration example of the noise variance calculation unit 16. The noise variance calculation unit 16 includes an amplitude calculation unit 50, column vector extraction units 51-1 and 51-2, a row vectorization unit 52, multiplication units 53-1 and 53-2, and arithmetic mean units 54-1 and 54. -2, division units 55-1 and 55-2 and multiplication units 56-1 and 56-2.

The weight matrix input from the weight matrix calculation unit 12-2 is input to the amplitude calculation unit 50. The amplitude calculation unit 50 calculates the amplitude for each element of the weight matrix input from the weight matrix calculation unit 12-2, and generates a matrix having the amplitude of each element of the weight matrix as an element. The matrix composed of the amplitudes of the elements of the weight matrix output from the amplitude calculation unit 50 is input to the column vector extraction units 51-1 and 51-2.

The column vector extraction units 51-1 and 51-2 are supplied with the matrix consisting of the amplitudes of the elements of the weight matrix from the amplitude calculation unit 50, extract the column vector for each reception sequence from the matrix, and calculate the amplitude of the elements of the weight matrix. Generate a column vector for each received sequence. The column vector for each reception sequence, which is composed of the amplitudes of the elements of the weight matrix output from the column vector extraction units 51-1 and 51-2, is input to the multiplication units 53-1 and 53-2.

The input stage band noise variance value of the entire band for each reception sequence, which is input from the input stage band noise variance calculation unit 15, is input to the row vectorization unit 52. The row vectorization unit 52 generates a row vector with the input stage band noise variance value of the entire band for each reception sequence input from the input stage band noise variance calculation unit 15 as an element. The row vector of the input stage band noise variance value output from the row vectorization unit 52 is divided into two, one is input to the multiplication units 53-1 and 53-2, and the other is input to the multiplication units 56-1 and 56-2. ..

To the multiplication units 53-1 and 53-2, the column vector for each reception sequence formed of the amplitudes of the elements of the weight matrix is input from the column vector extraction units 51-1 and 51-2, and the row vectorization unit 52 inputs the input stage band. A row vector of noise variance values is input. Then, the multiplication units 53-1 and 53-2 multiply the column vector and the row vector. The multiplication results for each reception sequence output from the multiplication units 53-1 and 53-2 are divided into two, one to the arithmetic mean units 54-1 and 54-2, and the other to the division units 55-1 and 55-2. Is entered.

The arithmetic averaging units 54-1 and 54-2 receive the multiplication results of each reception sequence from the multiplication units 53-1 and 53-2, average the multiplication results of each reception sequence over all carriers, and receive the reception sequences. The multiplication result of the entire band for each is calculated. The multiplication results of the entire band for each reception sequence output from the arithmetic mean units 54-1 and 54-2 are input to the division units 55-1 and 55-2.

The division units 55-1 and 55-2 receive the multiplication results for each reception sequence from the multiplication units 53-1 and 53-2, and the arithmetic mean units 54-1 and 54-2 calculate the entire band for each reception sequence. The multiplication result is input, and the multiplication result is divided by the multiplication result of the entire band. The division results for each reception sequence output from the division units 55-1 and 55-2 are input to the multiplication units 56-1 and 56-2.

In the multiplication units 56-1 and 56-2, the division result for each reception sequence is input from the division units 55-1 and 55-2, and the row vector of the input stage band noise variance value is input from the row vectorization unit 52, The division result and the row vector are multiplied to obtain an output stage band noise variance value for each reception sequence and each carrier. The output stage band noise variance value for each reception sequence and for each carrier output from the multiplication units 56-1 and 56-2 is input to the QAM demodulation units 17-1 and 17-2.

As a result, the input stage band noise variance value input by the noise variance calculation unit 16 is excluded from the influence of noise enhancement. Therefore, the noise variance calculation unit 16 eliminates the influence of noise enhancement for each reception sequence and carrier. The output stage band noise variance value is calculated for each.

Returning to FIG. 1, the QAM demodulation units 17-1 and 17-2 receive the received signals (separation and detection signals) for each reception sequence from the spatial filter unit 13-2 and the noise variance calculation unit 16 for each reception sequence. And the output stage band noise variance value for each carrier is input. Then, the QAM demodulation units 17-1 and 17-2 receive the received signals (separated and detected signals) for each received sequence and the output stage band noise variance value for each received sequence and for each carrier, and for each received sequence and each bit. The log-likelihood ratio of is calculated. The log-likelihood ratios for each reception sequence and for each bit output from the QAM demodulation units 17-1 and 17-2 are input to the multiplexing unit 18.

ここで、Ｌ（ｋ，ｎ）は、キャリヤ番号ｋにおけるｎ番目のビットに対する尤度を示し、Ｓ_n0およびＳ_n1は、ｎ番目のビットがそれぞれ０，１である信号点の集合を示し、σ_k ²は、キャリヤ番号ｋの雑音分散値を示す。また、（ｘ，ｙ）は分離および検出後のキャリヤシンボルを示す。 For example, the log-likelihood ratio for each bit is calculated by the following equation.

Here, L(k,n) indicates the likelihood for the n-th bit in the carrier number k, S _n0 and S _n1 indicate the set of signal points in which the n-th bit is 0 and 1, respectively. σ _k ² indicates a noise variance value of carrier number k. Further, (x, y) indicates the carrier symbol after separation and detection.

The multiplexing unit 18 receives the log-likelihood ratios for each reception sequence and each bit from the QAM demodulation units 17-1 and 17-2, and multiplexes the log-likelihood ratios for the number of reception sequences. The multiplexed log-likelihood ratio output from the multiplexing unit 18 is input to the error correction decoding unit 19.

The error correction decoding unit 19 receives the log-likelihood ratio after multiplexing from the multiplexing unit 18, performs error correction processing using the log-likelihood ratio after multiplexing, generates a bit string, and outputs it to the outside.

As described above, according to the MIMO-OFDM signal receiving apparatus 1 of the first embodiment, the weight matrix calculating unit 12-1 calculates the weight matrix based on the zero-forcing criterion using the channel matrix. Then, the spatial filter unit 13-1 multiplies the frequency domain signal, which is the received signal, by the weighting matrix based on the zero-forcing criterion, to separate and detect the received signals for the number of received sequences. As a result, the received signal separated and detected by the spatial filter unit 13-1 becomes a noise-emphasized signal.

The output stage band noise variance calculation unit 14 averages over all carriers for each received sequence based on the received signals of the number of received sequences separated and detected by the spatial filter unit 13-1, and the noise variance of the entire band. The value is calculated as the output stage band noise variance value for each reception sequence. As a result, the output stage band noise variance value for each reception sequence affected by noise enhancement is obtained.

The input-stage band noise variance calculation unit 15 calculates the inverse matrix of the entire band, which consists of the amplitudes of the elements of the weight matrix based on the zero-forcing criterion. Then, the input stage band noise variance calculation unit 15 adds the output stage band noise of the entire band for each reception sequence calculated by the output stage band noise variance calculation unit 14 to the inverse matrix of the entire band formed by the amplitudes of the elements of the weight matrix. The matrix having the variance value as the element is multiplied, and the variance value of the noise component as the multiplication result is calculated as the input stage band noise variance value for each reception sequence. As a result, the input stage band noise variance value for each reception sequence, which eliminates the influence of noise enhancement, can be obtained.

The weight matrix calculation unit 12-2 calculates the weight matrix based on the MMSE standard using the channel matrix and the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit 15. .. As a result, a weighting matrix having different scalar amounts in diagonal components as elements for suppressing noise enhancement and reflecting the input stage noise variance value is obtained.

The spatial filter unit 13-2 multiplies a frequency domain signal, which is a received signal, by a weighting matrix based on the MMSE standard to separate and detect received signals for the number of received sequences. As a result, the received signal separated and detected by the spatial filter unit 13-2 becomes a signal in which noise emphasis is difficult.

The noise variance calculation unit 16 multiplies the column vector formed by the amplitudes of the elements of the weight matrix based on the MMSE norm by the row vector of the input stage band noise variance value calculated by the input stage band noise variance calculation unit 15. Then, the noise variance calculation unit 16 obtains a multiplication result of the entire band averaged over all the carriers with respect to this multiplication result, divides the multiplication result of the column vector and the row vector by the multiplication result of the entire band, The noise variance value for each reception sequence and each carrier is obtained by multiplying the division result by the row vector. As a result, the noise variance value for each reception sequence and for each carrier, which eliminates the influence of noise enhancement, can be obtained.

The QAM demodulation units 17-1 and 17-2 include the received signals for the number of reception sequences separated and detected by the spatial filter unit 13-2, and the noise for each reception sequence and each carrier calculated by the noise variance calculation unit 16. The log-likelihood ratio for each reception sequence and each bit is calculated based on the variance value.

Therefore, even when the C/N for each reception sequence is significantly different, the variance value of the noise component included in the reception signal can be accurately estimated. Further, the multiplexed signal can be separated and detected without noise enhancement, and the log-likelihood ratio can be calculated accurately.

[Example 2]
Next, a second embodiment will be described. FIG. 6 is a block diagram illustrating a configuration example of the MIMO-OFDM signal receiving apparatus according to the second embodiment. This MIMO-OFDM signal receiver 2 includes FFT units 10-1 and 10-2 corresponding to the number of reception sequences, a channel estimation unit 11, a weight matrix calculation unit 12-1, a spatial filter unit 13-1, and an output stage band noise dispersion. The calculation unit 14, the input stage band noise variance calculation unit 15, the noise variance calculation unit 16, the QAM demodulation units 17-1 and 17-2, the multiplexing unit 18, and the error correction decoding unit 19 are provided.

Comparing the MIMO-OFDM signal receiver 1 of the first embodiment shown in FIG. 1 with the MIMO-OFDM signal receiver 2 of the second embodiment shown in FIG. 6, the MIMO-OFDM signal receiver 2 calculates the weight matrix. The difference from the MIMO-OFDM signal receiving apparatus 1 is that the unit 12-2 and the spatial filter unit 13-2 are not provided. Further, the MIMO-OFDM signal receiving apparatus 2 receives the weight matrix from the weight matrix calculator 12-2 in that the noise variance calculator 16 inputs the weight matrix from the weight matrix calculator 12-1. Different from the receiving device 1. Further, the MIMO-OFDM signal receiving apparatus 2 receives from the spatial filter unit 13-2 in that the QAM demodulation units 17-1 and 17-2 input the received signals for the number of reception sequences from the spatial filter unit 13-1. This is different from the MIMO-OFDM signal receiving apparatus 1 which inputs the received signals for the number of sequences.

The FFT units 10-1 and 10-2 perform the same processing as in the first embodiment, and output the time domain signal to the spatial filter unit 13-1 and the channel estimation unit 11. The channel estimation unit 11 performs the same processing as in the first embodiment and outputs the channel matrix to the weight matrix calculation unit 12-1. The weight matrix calculation unit 12-1 performs the same processing as that of the first embodiment, and the weight matrix calculated based on the zero forcing criterion is used as the spatial filter unit 13-1, the input stage band noise variance calculation unit 15, and the noise variance calculation unit 16. Output to.

The spatial filter unit 13-1 performs the same processing as that of the first embodiment, and receives the reception signals (separation and detection signals) for the number of reception sequences in the output stage band noise variance calculation unit 14 and the QAM demodulation units 17-1 and 17-2. Output to. The output stage band noise variance calculation unit 14 performs the same processing as that of the first embodiment, and outputs the output stage band noise variance value (output stage band noise variance value affected by noise enhancement) for each reception sequence. It is output to the calculation unit 15. The input stage band noise variance calculation unit 15 performs the same processing as that of the first embodiment, and calculates the input stage band noise variance value (the input stage band noise variance value excluding the effect of noise enhancement) for each reception sequence. Output to.

The noise variance calculation unit 16 receives the weight matrix from the weight matrix calculation unit 12-1 and the input stage band noise variance value from the input stage band noise variance calculation unit 15 for each reception sequence. Then, the noise variance calculation unit 16 performs the same processing as that of the first embodiment, and the band noise variance value (band noise variance value excluding the effect of noise enhancement) for each reception sequence is QAM demodulation units 17-1 and 17-2. Output to.

The QAM demodulation units 17-1 and 17-2 receive the received signals (separated and detected signals) for each reception sequence from the spatial filter unit 13-1 and output the entire bandwidth of the reception sequences for each reception sequence from the noise variance calculation unit 16. The band noise variance value is input. Then, the QAM demodulation units 17-1 and 17-2 perform the same processing as that of the first embodiment, and output the log likelihood ratio for each reception sequence and for each bit to the multiplexing unit 18. The multiplexing unit 18 and the error correction decoding unit 19 are the same as those in the first embodiment.

As described above, according to the MIMO-OFDM signal receiving apparatus 2 of the second embodiment, the weight matrix calculating unit 12-1 calculates the weight matrix based on the zero-forcing criterion, as in the first embodiment. Then, as in the first embodiment, the spatial filter unit 13-1 uses the weighting matrix based on the zero-forcing criterion to separate and detect the received signals for the number of received sequences. As a result, the received signal separated and detected by the spatial filter unit 13-1 becomes a noise-emphasized signal.

The output stage band noise variance calculation unit 14 calculates the output stage band noise variance value for each reception sequence, as in the first embodiment. As a result, the output stage band noise variance value for each reception sequence affected by noise enhancement is obtained. The input stage band noise variance calculation unit 15 calculates the input stage band noise variance value for each reception sequence, as in the first embodiment. As a result, the input stage band noise variance value for each reception sequence, which eliminates the influence of noise enhancement, is obtained.

The noise variance calculation unit 16 uses a weight matrix based on the zero forcing standard instead of the weight matrix according to the first embodiment based on the MMSE standard. The noise variance calculation unit 16 multiplies the column vector composed of the amplitudes of the elements of the weight matrix by the row vector of the input stage band noise variance value calculated by the input stage band noise variance calculation unit 15, and thereafter the embodiment The same process as 1 is performed to calculate the noise variance value for each reception sequence and each carrier. As a result, the noise variance value for each reception sequence and for each carrier can be obtained by eliminating the influence of noise enhancement.

The QAM demodulation units 17-1 and 17-2 are separated and detected by the spatial filter unit 13-1 instead of the reception signals corresponding to the number of reception sequences of the first embodiment separated and detected by the spatial filter unit 13-2. Received signals corresponding to the number of received sequences are used. The QAM demodulators 17-1 and 17-2 receive the received signal for each received sequence, and the received signal for each received sequence and the noise variance value for each carrier calculated by the noise variance calculation unit 16 for each received sequence and each bit. The log-likelihood ratio of is calculated.

Therefore, even when the C/N for each reception sequence is significantly different, the variance value of the noise component included in the reception signal can be accurately estimated. Further, the log-likelihood ratio can be calculated with high accuracy with a configuration simpler than that in the first embodiment.

[Example 3]
Next, a third embodiment will be described. FIG. 7 is a block diagram illustrating a configuration example of the MIMO-OFDM signal receiving apparatus according to the third embodiment. This MIMO-OFDM signal receiving device 3 includes FFT units 10-1 and 10-2 corresponding to the number of reception sequences, a channel estimation unit 11, a weight matrix calculation unit 12-1, a spatial filter unit 13-1, an output stage band noise dispersion. The calculation unit 14, the noise variance calculation unit 16, the QAM demodulation units 17-1 and 17-2, the multiplexing unit 18, and the error correction decoding unit 19 are provided.

Comparing the MIMO-OFDM signal receiving apparatus 2 of the second embodiment shown in FIG. 6 with the MIMO-OFDM signal receiving apparatus 3 of the third embodiment shown in FIG. 7, the MIMO-OFDM signal receiving apparatus 3 has an input stage band. The difference from the MIMO-OFDM signal receiving apparatus 2 is that the noise variance calculating unit 15 is not provided. Further, in the MIMO-OFDM signal receiving apparatus 3, the input stage band noise variance is that the noise variance calculating unit 16 inputs the output stage band noise variance value of the entire band for each reception sequence from the output stage band noise variance calculating unit 14. This is different from the MIMO-OFDM signal receiving apparatus 2 which inputs the input stage band noise variance value of the entire band for each reception sequence from the calculation unit 15.

FFT sections 10-1 and 10-2, channel estimation section 11, weight matrix calculation section 12-1, spatial filter section 13-1, QAM demodulation sections 17-1 and 17-2, multiplexing section 18 and error correction decoding section 19 Is the same as in the first embodiment.

The output stage band noise variance calculation unit 14 performs the same processing as that of the first embodiment, and outputs the output stage band noise variance value (output stage band noise variance value affected by noise enhancement) for each reception sequence. Output to. The noise variance calculation unit 16 receives the weight matrix from the weight matrix calculation unit 12-1 and the input stage band noise variance value for each reception sequence from the output stage band noise variance calculation unit 14. Then, the noise variance calculation unit 16 performs the same processing as that of the first embodiment to calculate the band noise variance value (band noise variance value excluding the influence of noise enhancement) for each reception sequence, and the QAM demodulation unit 17- 1 and 17-2.

As described above, according to the MIMO-OFDM signal receiving apparatus 3 of the third embodiment, the weight matrix calculation unit 12-1 calculates the weight matrix based on the zero-forcing criterion, as in the first and second embodiments. Then, the spatial filter unit 13-1 uses the weighting matrix based on the zero-forcing criterion to separate and detect the received signals for the number of reception sequences, as in the first and second embodiments. As a result, the received signal separated and detected by the spatial filter unit 13-1 becomes a noise-emphasized signal.

The output stage band noise variance calculation unit 14 calculates the output stage band noise variance value for each reception sequence, as in the first and second embodiments. As a result, the output stage band noise variance value for each reception sequence affected by noise enhancement is obtained.

The noise variance calculation unit 16 uses the weighting matrix based on the zero forcing criterion as in the second embodiment, but unlike the second embodiment, the input stage band noise variance value calculated by the input stage band noise variance calculation unit 15. Instead of, the output stage band noise variance value calculated by the output stage band noise variance calculator 14 is used. The noise variance calculation unit 16 multiplies the column vector composed of the amplitudes of the elements of the weight matrix by the row vector of the output stage band noise variance value, and thereafter performs the same processing as in Examples 1 and 2, A noise variance value is calculated for each series and each carrier. As a result, the noise variance value for each reception sequence and for each carrier, which eliminates the influence of noise enhancement, can be obtained.

Similar to the second embodiment, the QAM demodulation units 17-1 and 17-2 receive as many reception signals as the number of reception sequences separated and detected by the spatial filter unit 13-1 and reception calculated by the noise variance calculation unit 16. A log-likelihood ratio for each received sequence and each bit is calculated based on the noise variance value for each sequence and each carrier.

Therefore, even when the C/N for each reception sequence is significantly different, the variance value of the noise component included in the reception signal can be accurately estimated. Further, the log-likelihood ratio can be calculated with high accuracy with a configuration simpler than that in the second embodiment.

Further, since the MIMO-OFDM signal receiving apparatus 3 of the third embodiment does not include the input stage band noise variance calculation unit 15, it is not necessary to calculate the inverse matrix (the inverse matrix of the input stage band noise variance calculation unit 15). The calculation unit 31 calculates the inverse matrix). Generally, the calculation of the inverse matrix is heavy and the obtained value is unstable. Therefore, according to the MIMO-OFDM signal receiving apparatus 3 of the third embodiment, the calculation load can be suppressed as compared with the first and second embodiments.

The MIMO-OFDM signal receiver 1 of the first embodiment shown in FIG. 1, the MIMO-OFDM signal receiver 2 of the second embodiment shown in FIG. 6, and the MIMO-OFDM signal receiver of the third embodiment shown in FIG. The processing of each component of the device 3 may be realized by an LSI chip which is an integrated circuit mounted on the MIMO-OFDM signal receiving devices 1, 2, and 3. These may be individually made into one chip, or some or all of them may be made into one chip.

Further, instead of the LSI, it may be realized by a chip such as VLSI or ULSI having a different degree of integration. Further, the chip is not limited to an LSI or the like, and a dedicated circuit or a general-purpose processor may be used, or an FPGA (Field Programmable Gate Array) may be used.

1, 2 and 3 MIMO-OFDM signal receiving device 10 FFT unit 11 Channel estimation unit 12 Weight matrix calculation unit 13 Spatial filter unit 14 Output stage band noise variance calculation unit 15 Input stage band noise variance calculation unit 16 Noise variance calculation unit 17 QAM Demodulation unit 18 Multiplexing unit 19 Error correction decoding unit 20 Symbol reproduction unit 21 Subtraction unit 22, 30, 50 Amplitude calculation unit 23, 32, 54 Arithmetic mean unit 31, 44 Inverse matrix calculation unit 33 Input noise variance calculation unit 40 Complex conjugate Transposition unit 41, 45, 53, 56 Multiplication unit 42 Diagonal matrix generation unit 43 Addition unit 51 Column vector extraction unit 52 Row vectorization unit 55 Division unit

Claims (7)

An FFT unit that performs FFT on the equivalent baseband signal at a time corresponding to the effective symbol period of the MIMO-OFDM signal for each reception sequence and transforms the time domain signal into a frequency domain signal;
A channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit;
A weight matrix calculation unit that calculates a weight matrix from the channel matrix estimated by the channel estimation unit;
A spatial filter unit that multiplies the frequency domain signal converted by the FFT unit and the weight matrix calculated by the weight matrix calculation unit to separate and detect a received signal for each reception sequence;
Based on the received signal for each received sequence that has been separated and detected by the spatial filter unit, a band noise variance calculation unit that calculates a noise variance value for the entire band for each received sequence,
Based on the noise variance value of the entire band for each reception sequence calculated by the band noise variance calculation unit and the weight matrix calculated by the weight matrix calculation unit, a noise variance value for each reception sequence and for each carrier is calculated. A noise variance calculation unit for calculating,
For each reception sequence and for each bit, based on the received signal for each reception sequence that has been separated and detected by the spatial filter, and the noise variance value for each reception sequence and for each carrier calculated by the noise variance calculation unit A QAM demodulation unit that calculates a log likelihood ratio of
A multiplexing unit that multiplexes the log-likelihood ratios for each reception sequence and each bit calculated by the QAM demodulation unit;
An error correction decoding unit that performs an error correction decoding process using the log-likelihood ratio multiplexed by the multiplexing unit, and generates and outputs a bit string,
The noise variance calculation unit,
An amplitude calculation unit that calculates the amplitude of each element of the weight matrix calculated by the weight matrix calculation unit and that generates a matrix having the amplitude of each element of the weight matrix as an element;
From the matrix generated by the amplitude calculation unit, a column vector extraction unit that extracts a column vector for each reception sequence,
A row vectorization unit that generates a row vector from the noise variance value of the entire band of each reception sequence calculated by the band noise variance calculation unit,
A multiplication unit that multiplies the column vector for each reception sequence extracted by the column vector extraction unit and the row vector generated by the row vectorization unit, and outputs a multiplication result for each reception sequence,
An arithmetic averaging unit that averages the multiplication results for each reception sequence output by the multiplication unit over all carriers and calculates a multiplication result for the entire band for each reception sequence;
A division unit that divides the multiplication result for each reception sequence output by the multiplication unit by the multiplication result for the entire band for each reception sequence calculated by the arithmetic averaging unit and outputs the division result for each reception sequence. When,
A multiplication unit that multiplies the division result for each reception sequence output by the division unit and the row vector generated by the row vectorization unit to obtain a noise variance value for each reception sequence and for each carrier, A MIMO-OFDM signal receiving apparatus comprising: The band noise variance calculator comprises an output stage band noise variance calculator and an input stage band noise variance calculator.
The output stage band noise variance calculation unit,
A symbol regenerating unit that regenerates a received signal for each reception sequence that has been separated and detected by the spatial filter unit, and that generates a regenerated signal for each reception sequence,
Subtraction for subtracting the reception signal for each reception sequence separated and detected by the spatial filter unit from the reproduction signal for each reception sequence generated by the symbol reproduction unit to generate an error signal for each reception sequence Department,
An amplitude calculation unit that calculates the amplitude of the error signal for each reception sequence generated by the subtraction unit;
An arithmetic mean unit that averages the amplitudes of the error signals for each reception sequence calculated by the amplitude calculation unit over all carriers and calculates an output stage band noise variance value of the entire band for each reception sequence. ,
The input stage band noise variance calculation unit,
An amplitude calculation unit that calculates the amplitude of each element of the weight matrix calculated by the weight matrix calculation unit and that generates a matrix having the amplitude of each element of the weight matrix as an element;
An inverse matrix calculator that calculates an inverse matrix of the matrix generated by the amplitude calculator,
An arithmetic mean unit for averaging the inverse matrix calculated by the inverse matrix calculating unit over all carriers to calculate an inverse matrix having an average value of the entire band as an element,
By multiplying the output stage band noise variance value of the entire band for each reception sequence calculated by the output stage band noise variance calculation unit and the inverse matrix calculated by the arithmetic mean unit, An input noise variance calculation unit that obtains an input stage band noise variance value for the entire band and outputs the input stage band noise variance value for the entire band for each reception sequence as a noise variance value for the entire band for each reception sequence, The MIMO-OFDM signal receiving apparatus according to claim 1, further comprising: An FFT unit that performs FFT on the equivalent baseband signal at a time corresponding to the effective symbol period of the MIMO-OFDM signal for each reception sequence and transforms the time domain signal into a frequency domain signal;
A channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit;
A first weight matrix calculation unit that calculates a weight matrix from the channel matrix estimated by the channel estimation unit based on a zero forcing criterion;
A first spatial filter that multiplies the frequency domain signal converted by the FFT unit and the weight matrix calculated by the first weight matrix calculation unit to separate and detect a received signal for each reception sequence. Department,
The noise variance value included in the output signal of the first spatial filter unit is based on the received signal for each received sequence separated and detected by the first spatial filter unit, and the noise variance value included in the output signal of the first spatial filter unit for the entire band for each received sequence. An output stage band noise variance calculation unit that calculates the output stage band noise variance value of
The first stage based on the output stage band noise variance value of the entire band for each reception sequence calculated by the output stage band noise variance calculation unit and the weight matrix calculated by the first weight matrix calculation unit. An input stage band noise variance calculation unit that calculates the noise variance value included in the input signal of the spatial filter unit as an input stage band noise variance value of the entire band for each reception sequence,
A weight matrix is calculated based on the MMSE standard from the channel matrix estimated by the channel estimation unit and the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit. A second weight matrix calculation unit that
A second spatial filter that multiplies the frequency domain signal converted by the FFT unit and the weight matrix calculated by the second weight matrix calculation unit to separate and detect a received signal for each reception sequence. Department,
For each reception sequence, based on the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit and the weight matrix calculated by the second weight matrix calculation unit And a noise variance calculation unit that calculates a noise variance value for each carrier,
For each received sequence based on the received signal for each received sequence separated and detected by the second spatial filter, and the noise variance value for each received sequence and for each carrier calculated by the noise variance calculation unit And a QAM demodulation unit that calculates a log-likelihood ratio for each bit,
A multiplexing unit that multiplexes the log-likelihood ratios for each reception sequence and each bit calculated by the QAM demodulation unit;
An error correction decoding unit that performs an error correction decoding process using the log-likelihood ratio multiplexed by the multiplexing unit, and generates and outputs a bit string,
The noise variance calculation unit,
An amplitude calculation unit that calculates the amplitude of each element of the weight matrix calculated by the second weight matrix calculation unit and generates a matrix having the amplitude of each element of the weight matrix as an element;
From the matrix generated by the amplitude calculation unit, a column vector extraction unit that extracts a column vector for each reception sequence,
A row vectorization unit that generates a row vector from the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit,
A multiplication unit that multiplies the column vector for each reception sequence extracted by the column vector extraction unit and the row vector generated by the row vectorization unit, and outputs a multiplication result for each reception sequence,
An arithmetic averaging unit that averages the multiplication results for each reception sequence output by the multiplication unit over all carriers and calculates a multiplication result for the entire band for each reception sequence;
A division unit that divides the multiplication result for each reception sequence output by the multiplication unit by the multiplication result for the entire band for each reception sequence calculated by the arithmetic averaging unit and outputs the division result for each reception sequence. When,
A multiplication unit that multiplies the division result for each reception sequence output by the division unit and the row vector generated by the row vectorization unit to obtain a noise variance value for each reception sequence and for each carrier, A MIMO-OFDM signal receiving apparatus comprising: The second weight matrix calculation unit is
A complex conjugate transpose unit for calculating a complex conjugate transpose matrix of the channel matrix estimated by the channel estimation unit,
A first multiplication unit that multiplies the channel matrix estimated by the channel estimation unit and the complex conjugate transposed matrix calculated by the complex conjugate transpose unit, and outputs a matrix of a multiplication result;
By setting the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit to a diagonal component and setting zero to a non-diagonal component, a diagonal matrix is obtained. A diagonal matrix generator to generate,
An addition unit that adds the diagonal matrix generated by the diagonal matrix generation unit to the matrix of the multiplication result output by the first multiplication unit and outputs the matrix of the addition result;
An inverse matrix calculator that calculates an inverse matrix of the matrix of the addition result output by the adder,
It is characterized by comprising: a multiplication unit that multiplies the inverse matrix calculated by the inverse matrix calculation unit and the complex conjugate transposed matrix calculated by the complex conjugate transposition unit to obtain the weight matrix. Item 4. The MIMO-OFDM signal receiver according to Item 3. An FFT unit that performs FFT on the equivalent baseband signal at a time corresponding to the effective symbol period of the MIMO-OFDM signal for each reception sequence and transforms the time domain signal into a frequency domain signal;
A channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit;
A first weight matrix calculation unit that calculates a weight matrix from the channel matrix estimated by the channel estimation unit based on a zero forcing criterion;
A first spatial filter that multiplies the frequency domain signal converted by the FFT unit and the weight matrix calculated by the first weight matrix calculation unit to separate and detect a received signal for each reception sequence. Department,
The noise variance value included in the output signal of the first spatial filter unit is based on the received signal for each received sequence separated and detected by the first spatial filter unit, and the noise variance value included in the output signal of the first spatial filter unit for the entire band for each received sequence. An output stage band noise variance calculation unit that calculates the output stage band noise variance value of
The first stage based on the output stage band noise variance value of the entire band for each reception sequence calculated by the output stage band noise variance calculation unit and the weight matrix calculated by the first weight matrix calculation unit. An input stage band noise variance calculation unit that calculates the noise variance value included in the input signal of the spatial filter unit as an input stage band noise variance value of the entire band for each reception sequence,
For each reception sequence, based on the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit and the weight matrix calculated by the first weight matrix calculation unit And a noise variance calculation unit that calculates a noise variance value for each carrier,
For each received sequence, based on the received signal for each received sequence separated and detected by the first spatial filter, and the noise variance value for each received sequence and for each carrier calculated by the noise variance calculator And a QAM demodulation unit that calculates a log-likelihood ratio for each bit,
A multiplexing unit that multiplexes the log-likelihood ratios for each reception sequence and each bit calculated by the QAM demodulation unit;
An error correction decoding unit that performs an error correction decoding process using the log-likelihood ratio multiplexed by the multiplexing unit, and generates and outputs a bit string,
The noise variance calculation unit,
An amplitude calculation unit that calculates an amplitude of each element of the weight matrix calculated by the first weight matrix calculation unit and generates a matrix having the amplitude of each element of the weight matrix as an element;
From the matrix generated by the amplitude calculation unit, a column vector extraction unit that extracts a column vector for each reception sequence,
A row vectorization unit that generates a row vector from the input stage band noise variance value of the entire band for each reception sequence calculated by the input stage band noise variance calculation unit,
A multiplication unit that multiplies the column vector for each reception sequence extracted by the column vector extraction unit and the row vector generated by the row vectorization unit, and outputs a multiplication result for each reception sequence,
An arithmetic averaging unit that averages the multiplication results for each reception sequence output by the multiplication unit over all carriers and calculates a multiplication result for the entire band for each reception sequence;
A division unit that divides the multiplication result for each reception sequence output by the multiplication unit by the multiplication result for the entire band for each reception sequence calculated by the arithmetic averaging unit and outputs the division result for each reception sequence. When,
A multiplication unit that multiplies the division result for each reception sequence output by the division unit and the row vector generated by the row vectorization unit to obtain a noise variance value for each reception sequence and for each carrier, A MIMO-OFDM signal receiving apparatus comprising: An FFT unit that performs FFT on the equivalent baseband signal at a time corresponding to the effective symbol period of the MIMO-OFDM signal for each reception sequence and transforms the time domain signal into a frequency domain signal;
A channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit;
A first weight matrix calculation unit that calculates a weight matrix from the channel matrix estimated by the channel estimation unit based on a zero forcing criterion;
A first spatial filter that multiplies the frequency domain signal converted by the FFT unit and the weight matrix calculated by the first weight matrix calculation unit to separate and detect a received signal for each reception sequence. Department,
The noise variance value included in the output signal of the first spatial filter unit is based on the received signal for each received sequence separated and detected by the first spatial filter unit, and the noise variance value included in the output signal of the first spatial filter unit for the entire band for each received sequence. An output stage band noise variance calculation unit that calculates the output stage band noise variance value of
For each reception sequence, based on the output stage band noise variance value of the entire band for each reception sequence calculated by the output stage band noise variance calculation unit and the weight matrix calculated by the first weight matrix calculation unit And a noise variance calculation unit that calculates a noise variance value for each carrier,
For each received sequence, based on the received signal for each received sequence separated and detected by the first spatial filter, and the noise variance value for each received sequence and for each carrier calculated by the noise variance calculator And a QAM demodulation unit that calculates a log-likelihood ratio for each bit,
A multiplexing unit that multiplexes the log-likelihood ratios for each reception sequence and each bit calculated by the QAM demodulation unit;
An error correction decoding unit that performs an error correction decoding process using the log-likelihood ratio multiplexed by the multiplexing unit, and generates and outputs a bit string,
The noise variance calculation unit,
An amplitude calculation unit that calculates an amplitude of each element of the weight matrix calculated by the first weight matrix calculation unit and generates a matrix having the amplitude of each element of the weight matrix as an element;
From the matrix generated by the amplitude calculation unit, a column vector extraction unit that extracts a column vector for each reception sequence,
A row vectorization unit that generates a row vector from the output stage band noise variance value of the entire band for each reception sequence calculated by the output stage band noise variance calculation unit,
A multiplication unit that multiplies the column vector for each reception sequence extracted by the column vector extraction unit and the row vector generated by the row vectorization unit, and outputs a multiplication result for each reception sequence,
An arithmetic averaging unit that averages the multiplication results for each reception sequence output by the multiplication unit over all carriers and calculates a multiplication result for the entire band for each reception sequence;
A division unit that divides the multiplication result for each reception sequence output by the multiplication unit by the multiplication result for the entire band for each reception sequence calculated by the arithmetic averaging unit and outputs the division result for each reception sequence. When,
A multiplication unit that multiplies the division result for each reception sequence output by the division unit and the row vector generated by the row vectorization unit to obtain a noise variance value for each reception sequence and for each carrier, A MIMO-OFDM signal receiving apparatus comprising: In a chip mounted on a MIMO-OFDM signal receiving device,
An FFT unit that performs FFT on the equivalent baseband signal at a time corresponding to the effective symbol period of the MIMO-OFDM signal for each reception sequence and transforms the time domain signal into a frequency domain signal;
A channel estimation unit that estimates a channel matrix from the frequency domain signal transformed by the FFT unit;
A weight matrix calculation unit that calculates a weight matrix from the channel matrix estimated by the channel estimation unit;
A spatial filter unit that multiplies the frequency domain signal converted by the FFT unit and the weight matrix calculated by the weight matrix calculation unit to separate and detect a received signal for each reception sequence;
Based on the received signal for each received sequence that has been separated and detected by the spatial filter unit, a band noise variance calculation unit that calculates a noise variance value for the entire band for each received sequence,
Based on the noise variance value of the entire band for each reception sequence calculated by the band noise variance calculation unit and the weight matrix calculated by the weight matrix calculation unit, a noise variance value for each reception sequence and for each carrier is calculated. A noise variance calculation unit for calculating,
For each reception sequence and for each bit, based on the received signal for each reception sequence that has been separated and detected by the spatial filter, and the noise variance value for each reception sequence and for each carrier calculated by the noise variance calculation unit A QAM demodulation unit that calculates a log likelihood ratio of
A multiplexing unit that multiplexes the log-likelihood ratios for each reception sequence and each bit calculated by the QAM demodulation unit;
An error correction decoding unit that performs an error correction decoding process using the log-likelihood ratio multiplexed by the multiplexing unit, and generates and outputs a bit string,
The noise variance calculation unit,
An amplitude calculation unit that calculates the amplitude of each element of the weight matrix calculated by the weight matrix calculation unit and that generates a matrix having the amplitude of each element of the weight matrix as an element;
From the matrix generated by the amplitude calculation unit, a column vector extraction unit that extracts a column vector for each reception sequence,
A row vectorization unit that generates a row vector from the noise variance value of the entire band of each reception sequence calculated by the band noise variance calculation unit,
A multiplication unit that multiplies the column vector for each reception sequence extracted by the column vector extraction unit and the row vector generated by the row vectorization unit, and outputs a multiplication result for each reception sequence,
An arithmetic averaging unit that averages the multiplication results for each reception sequence output by the multiplication unit over all carriers and calculates a multiplication result for the entire band for each reception sequence;
A division unit that divides the multiplication result for each reception sequence output by the multiplication unit by the multiplication result for the entire band for each reception sequence calculated by the arithmetic averaging unit and outputs the division result for each reception sequence. When,
A multiplication unit that multiplies the division result for each reception sequence output by the division unit and the row vector generated by the row vectorization unit to obtain a noise variance value for each reception sequence and for each carrier, A chip characterized by comprising.

Images