JP5311469B2 - MIMO receiver, demodulation circuit, and signal processing program - Google Patents

Description

  The present invention relates to a MIMO receiver, a demodulation circuit, and a signal processing program.

  In recent years, in the field of wireless communication, in order to further improve the communication speed, a MIMO (Multi Input Multi Output) technique using multiple antennas has been developed and its use has started. The MIMO technology is a technology in which, for example, a plurality of data is spatially multiplexed by simultaneously transmitting from a plurality of antennas, and the receiving side extracts the plurality of original data from the received data. Since the amount of data transmitted at the same time can be increased, the communication speed is improved. Also, the modulation method is greatly related to the communication speed. For example, 2 bits in QPSK, 4 bits in 16QAM, and 6 bits in 64QAM, the number of information bits that can be transmitted simultaneously can be increased. However, noise resistance decreases as the number of information bits transmitted simultaneously increases. Therefore, it is called AMC (Adaptive Modulation and Coding) that selects an optimum method according to the quality of the propagation path environment in addition to the coding rate and coding method. Technology is being developed and used. Also in mobile communication, standardization in a new communication method as LTE (Long Term Evolution) is currently in progress in 3GPP (Third Generation Partnership Project), and this MIMO technology and AMC technology are adopted.

  One method for receiving data transmitted by the MIMO technique and extracting the original data is MLD (Maximum Likelihood Detection). Here, QR-MLD (QR decomposition MLD) which is a kind of MLD will be described.

FIG. 4 is a diagram for explaining a communication channel model using the MIMO technique. Here, an example in which the number of transmission antennas and the number of reception antennas are both two is shown. Transmission data x ₁ and x ₂ from the transmission antennas 51 and 52 are received by the reception antennas 61 and 62, respectively. The propagation path values indicating the state of the propagation path between the transmission antennas 51 and 52 and the reception antennas 61 and 62 are h ₁₁ , h ₁₂ , h ₂₁ , and h ₂₂ . Noise n ₁ and n ₂ are mixed in the receiving antennas 61 and 62, and received data y ₁ and y ₂ are obtained. The above data is represented by complex numbers. The transmission data x ₁ and x ₂ are obtained by modulating the original data. The reception data y ₁ and y ₂ are expressed by the formula 1.

Here, Y = [y ₁ , y ₂ ] ^T , H = [[h ₁₁ , h ₂₁ ], [h ₁₂ , h ₂₂ ]] ^T , X = [x ₁ , x ₂ ] ^T , N = [n ₁ , n ₂ ] ^T , Equation 1 is expressed as Equation 2.

In normal MLD, equalization processing is performed by using this Y value and the estimated propagation path estimation value H ′ = [[h ′ ₁₁ , h ′ ₂₁ ], [h ′ ₁₂ , h ′ ₂₂ ]] ^T With respect to the equalized data, errors with all replicas assumed as transmission data X are sequentially calculated, and demodulation is performed so that the error is minimized. The error here is the square Euclidean distance between the equalized data and the replica. The replica means data of lattice points according to the modulation method. However, in this method, it is necessary to calculate an error from all replicas, and the amount of calculation becomes enormous. For example, when the modulation method is 64QAM (Quadrature Amplitude Modulation), the number of lattice points is 64, and considering the combination of x ₁ and x ₂ , it is necessary to calculate 64 × 64 = 4096 errors. QR-MLD is a technique for reducing the calculation required for this error calculation.

なお、ＱＲ−ＭＬＤでは、ｘ₁とｘ₂を一度に処理できないため、まず一方について復調し、その後で、ＹとＨ'を倒置して他方の復調を行うことになる。以降はｘ₂の処理について説明する。 In QR-MLD, the estimated propagation path estimation value H ′ is not used as it is, but QR decomposition is performed. Q is a unitary matrix and R is an upper triangular matrix. By using this R during error calculation, the elements to be calculated are reduced. Further, since a part of the R component is a real number, the amount of calculation can be reduced accordingly. The result of QR decomposition of H ′ is shown in Equation 3.

In QR-MLD, x ₁ and x ₂ cannot be processed at a time, so that one is demodulated first, and then Y and H ′ are inverted and the other is demodulated. After that describes the processing of x _2.

In Expression 3, r ₁₁ and r ₂₂ of R are real numbers. Also, since Q is a unitary matrix, the relationship of Equation 4 is established. I is a unit matrix.

When a matrix QR obtained by QR decomposition of the estimated propagation path estimated value H ′ is substituted into the propagation path value H of Expression 1, the expression shown in Expression 5 is obtained.

When both sides of the formula 5 are multiplied by the conjugate inversion matrix QH of the unitary matrix Q and expanded, the formula shown in the formula 6 is obtained.

When expanded into a matrix as Z = RX + N ′, the equation shown in Equation 7 is obtained.

Summarizing Equation 7 for n ′ ₁ and n ′ ₂ yields Equation 8 below.

In the QR-MLD, the square errors of the noise components n ′ ₁ and n ′ ₂ are obtained, and x ₁ and x ₂ that minimize the sum are obtained. First, an error n ′ _{2, j} with each replica is calculated with a replica of the transmission data x ₂ as x _{2, j} . j = 0 to J-1, where J is the number of lattice points determined by the modulation method. The number of replicas required is the same as the number of grid points. For example, in the case of 64QAM, since it is modulated as 64 lattice points on the IQ plane, J = 64. n ′ _{2, j} is expressed by the equation (9).

Let e _{2, j} be the square of n ' _{2, j} . This is shown in Equation 10.

Subsequently, the other error n ′ ₁ is obtained, but before that, the replica candidates x _{1, j} of the transmission data x ₁ are narrowed down. This is a process of determining x1 _{, j} using a condition that minimizes the sum of squared errors. Specifically, x _{1, j} that satisfies n ′ _{1, j} = 0 is obtained for each transmission signal replica x _{2, j} . The conditional expression of x _{1, j} is shown in Equation 11.

That is, in Equation (11), x _{1, j} closest to the right side of the following equation is obtained. As a result, x _{1, j} that minimizes the sum of the square errors is obtained for each x _{2, j} , so that the remaining errors n ′ _{1, j} can be reduced to J. Assuming that the square error of n ′ _{1, j} is e _{1, j} , e _{1, j} is expressed by the equation shown in Equation 12.

このｅ_jから、復調結果を尤度として計算する。 E _{1, j} thus obtained, when the sum of e _{2, j} and e _j, e _j is expressed by the formula shown in Formula 13.

This e _j, calculates the demodulation result as the likelihood.

Figure 5 is a diagram for explaining a method of calculating the likelihood of the sum e _j of the square error at QPSK. The IQ plan view shown on the left and right of the figure is QPSK, so there is one grid point in each quadrant. That is, J is 4, and e _j is up to e ₀ , e ₁ , e ₂ , e ₃ . In Figure 5, the right or left of the grid points indicate corresponding e _j. One grid point includes 2 information bits. Numbers in parentheses (bit # 1, bit # 0) shown above or below each lattice point in the figure indicate each information bit. The likelihood of each information bit is calculated individually. When the likelihood for bit # 0 is LLR ₀ and the likelihood for bit # 1 is LLR ₁ , each likelihood is calculated according to the equation (14).

The likelihood calculation takes the difference of the minimum error corresponding to “1” or “0” for each information bit. For bit # 0, the lattice point where bit # 0 = "0" is on the right side of the IQ plane, and the corresponding errors are e ₀ and e ₂ . Of these, the smallest error is obtained in the second term on the right side of the expression of LLR ₀ shown in Equation 14. The lattice point where bit # 0 = “1” is on the left side of the IQ plane, and the corresponding errors are e ₁ and e ₃ . Of these, the smallest error is obtained in the first term on the right side of the expression of LLR ₀ shown in equation (14). Similarly for bit # 1, the minimum error corresponding to “0” and “1” is obtained. The likelihood is calculated by subtracting the minimum error corresponding to “0” from the minimum error corresponding to “1”.

  In general, in QR-MLD, J square errors are calculated by the equations (10) and (12). The calculation of one square error requires two multiplications, but the cost (computation amount, circuit scale, etc.) required for the multiplication is very large. In addition, J is the number of lattice points corresponding to the modulation method. In LTE, the modulation method is also selected and used from QPSK, 16QAM, and 64QAM in a timely manner by AMC technology. The number of each lattice point is 4 in QPSK, 16 in 16QAM, and 64 in 64QAM, and increases exponentially. This means that the cost for calculating the equations (10) and (12) increases. Further, if it is attempted to improve the communication speed by using a modulation method (256QAM or the like) that can send more bits, the cost required to calculate the equations (10) and (12) will increase without limit.

As means for solving such a problem, Patent Document 1 describes that an Euclidean distance is approximated by using an adding circuit having a small circuit scale instead of an integrating circuit having a large circuit scale.
JP 2005-217506 A

  An object of the present invention is to provide a MIMO receiving apparatus, a demodulating circuit, and a signal processing program in which the amount of calculation of QR-MLD for MIMO reception is reduced based on a viewpoint different from Patent Document 1.

  The main feature of the present invention is that the approximation of the Euclidean distance is performed in a state before the square calculation is performed instead of the square calculation in the calculation of the Euclidean distance as in the technique described in Patent Document 1. To do. Since the square calculation is not performed in the approximation of the Euclidean distance, the square calculation is performed by calculating the likelihood.

  That is, according to the first aspect of the present invention, in a MIMO receiver that receives data transmitted from a plurality of transmission antennas using a plurality of reception antennas, the difference between the received data and a replica assumed as transmission data is different. A demodulator that performs demodulation to reduce the error, the demodulator obtains a difference between the received data and each replica, and calculates an error by Euclidean distance approximation for each difference, and this error Based on the result of error calculation by the calculation unit, a total error for each replica is obtained, and a minimum error corresponding to information bits “1” and “0” is selected from the total error, and corresponding to information bit “1”. And a likelihood calculation unit that outputs a demodulation result using a difference between the square of the minimum error and the square of the minimum error corresponding to the information bit “0” as a likelihood. Receiving apparatus is provided.

  According to the second aspect of the present invention, in a MIMO receiver that receives data transmitted from a plurality of transmission antennas using a plurality of reception antennas, the difference between the received data and a replica assumed as transmission data is small. The demodulation unit obtains the difference between the received data and each replica, and when this difference is expressed by a complex number a + jb, the absolute value of the real part and the imaginary part is taken and n = | A | + j | b |, an error calculation unit that calculates a total value by complex addition for each replica based on the value of n, and performs error calculation by Euclidean distance approximation on the total value, and this error calculation unit The minimum error corresponding to information bits “1” and “0” is selected from the errors obtained by the above, and the square of the minimum error corresponding to information bit “1” and the maximum corresponding to information bit “0” are selected. The difference between the square of the error as the likelihood, the MIMO receiving apparatus according to claim is provided to have a likelihood calculation unit for outputting a demodulation result.

  According to another aspect of the present invention, in a demodulation circuit that performs demodulation in which data transmitted from a plurality of transmission antennas is received by a plurality of reception antennas, a difference between a replica assumed as transmission data is reduced, Calculate the difference between the received data and each replica, calculate the error for each difference by Euclidean distance approximation, and calculate the total error for each replica based on the error calculation result by this error calculator. The minimum error corresponding to the information bits “1” and “0” is selected from the total error, and the square of the minimum error corresponding to the information bit “1” and the minimum error corresponding to the information bit “0” are selected. There is provided a demodulation circuit characterized by having a likelihood calculation unit that outputs a demodulation result using a difference from the square as a likelihood.

  According to another aspect of the present invention, in a demodulation circuit that performs demodulation in which data transmitted from a plurality of transmission antennas is received by a plurality of reception antennas, a difference between a replica assumed as transmission data is reduced, When the difference between the received data and each replica is obtained and this difference is expressed by a complex number a + jb, the absolute value of the real part and the imaginary part is taken as n = | a | + j | b |, and based on the value of n An error calculation unit that calculates a total value by complex addition for each replica and calculates an error by Euclidean distance approximation to the total value, and information bits “1” and “1” from the errors obtained by the error calculation unit The minimum error corresponding to “0” is selected, and the demodulation result is output with the difference between the square of the minimum error corresponding to information bit “1” and the square of the minimum error corresponding to information bit “0” as the likelihood. Demodulation circuit is provided which is characterized by having a that likelihood calculation unit.

  According to another aspect of the present invention, a processor that performs signal processing obtains a difference between data transmitted from a plurality of transmission antennas and data received by the plurality of reception antennas and a replica assumed as transmission data. An error calculation procedure for calculating an error by Euclidean distance approximation with respect to the difference, and a total error for each replica is obtained based on an error calculation result by this error calculation procedure. The minimum error corresponding to “0” is selected, and the demodulation result is output using the difference between the square of the minimum error corresponding to information bit “1” and the square of the minimum error corresponding to information bit “0” as a likelihood. A signal processing program is provided that executes a likelihood calculation procedure.

  According to another aspect of the present invention, a processor that performs signal processing obtains a difference between data transmitted from a plurality of transmission antennas and data assumed by the reception antennas and a replica assumed as transmission data. Is expressed as a complex number a + jb, the absolute values of the real part and the imaginary part are taken as n = | a | + j | b |, and a total value is calculated by complex addition for each replica based on the value of n. An error calculation procedure for calculating an error by Euclidean distance approximation with respect to a value, and a minimum error corresponding to information bits “1” and “0” is selected from the errors obtained by this error calculation procedure, and an information bit And a likelihood calculation procedure for outputting a demodulation result using a difference between the square of the minimum error corresponding to “1” and the square of the minimum error corresponding to information bit “0” as a likelihood. No. processing program is provided.

  According to the present invention, by approximating the calculation of the square error in the above formulas 10 and 12, multiplication can be replaced with a combination of comparison and addition, and the amount of calculation can be greatly reduced.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[Overall configuration of communication system]
FIG. 1 is a block diagram of a communication system in which the present invention is implemented. The transmission station 10 is a base station, for example, and includes an encoding unit 11, a modulation unit 12, a D / A conversion unit 13, and a plurality of antennas 14. The receiving station 20 is a user terminal, for example, and includes an A / D conversion unit 21, a demodulation unit 22, a decoding unit 23, and a plurality of antennas 24.

  In transmitting station 10, a CPU (not shown) in base station 10 inputs data to be transmitted to encoding unit 11 as information bits. The encoding unit 11 performs CRC (Cyclic Redundancy Check) addition and convolutional encoding on the input information bits. The modulation unit 12 modulates the input code data and outputs a plurality of modulation data, which is data after modulation, to the D / A conversion unit 13. The D / A conversion unit 13 converts the modulation data output from the modulation unit 12 from a digital signal to an analog signal. Then, the modulated data converted into the analog signal is transmitted through the plurality of antennas 14.

  The receiving station 20 receives data transmitted from the antenna 14 of the transmitting station 10 via the plurality of antennas 24. However, it should be noted that the data received by the antenna 24 is affected by noise when propagating through space after being output from the antenna 14. Data received by the antenna 24 is input to the A / D converter 21. The A / D converter 21 converts the input data from an analog signal to a digital signal. The A / D converter 21 outputs the converted digital signal to the demodulator 22. The demodulator 22 demodulates the data output from the A / D converter 21. The demodulator 22 outputs the demodulated data obtained by demodulation to the decoder 23. The decoding unit 23 performs error correction decoding on the input demodulated data. Using the decoded data obtained as a result, a processing circuit such as a subsequent CPU performs a predetermined process. Here, the modulation data in the transmission station 10 is referred to as transmission data, and the data output from the A / D conversion unit 21 in the reception station 20 is referred to as reception data.

  The MIMO receiving apparatus according to the present invention is implemented as the receiving station 20 in the communication system shown in FIG. The demodulation circuit according to the present invention is implemented as the demodulation unit 22 in the receiving station 20.

[First Embodiment]
FIG. 2 is a block configuration diagram of the demodulation circuit according to the first embodiment of the present invention, and shows a configuration example of a circuit used as the demodulation unit 22 in the receiving station 20 in the communication system shown in FIG. Here, a configuration example is shown in which a process for demodulating transmission data x ₁ and x ₂ is performed by a common circuit. That is, this demodulation circuit performs a process on one transmission data, and then performs the same process on the other transmission data by inverting the input Y and the propagation path estimation value H ′. The following describes the demodulation processing of the transmission data x ₂ in the example.

  2 includes a propagation path estimation unit 31, a QR decomposition unit 32, an equalization unit 33, a replica generation unit 34, an error calculation unit 35, a quadrant detection unit 36, an error calculation unit 37, and a likelihood calculation unit 38. Have

The propagation path estimation unit 31 performs propagation path estimation using known data included in the reception data Y, and outputs a propagation path estimation value H ′. The QR decomposition unit 32 performs QR decomposition on the propagation path estimation value H ′, and outputs a unitary matrix Q and an upper triangular matrix R in the equation shown in Equation 3. The equalization unit 33 equalizes the reception data Y using the unitary matrix Q as shown in Equation 6 and outputs equalization data Z. The replica generation unit 34 generates a replica X ′ ₂ of the transmission data X.

The error calculator 35 obtains a difference between the received data (each element of the equalized data Z) and each replica (each corresponding element of the replica X ′ ₂ ), and calculates an error by Euclidean distance approximation for each difference. I do. That is, the error calculation unit 35 outputs an error e _{2, j} from the equalized data Z and the replica X ′ ₂ according to the equation of Equation 15 described later using the upper triangular matrix R. Along with the output of the error e _{2, j} , the equalized data Z is also transferred to the next stage.

The quadrant detection unit 36 calculates the replica X ′ ₁ from the equalized data Z and the replica X ′ ₂ by the upper triangular matrix R according to the condition of Equation 11. Further, the equalized data Z and the error e _{2, j} are transferred to the next stage.

The error calculation unit 37 calculates a difference between the received data (each element of the equalized data Z) and each replica (each corresponding element of the replicas X ′ _{1 and} X ′ ₂ ), and performs error calculation by Euclidean distance approximation. . That is, the error calculation unit 37 outputs an error e _{1, j} from the equalized data Z, the replicas X ′ ₁ , X ′ ₂ , and the upper triangular matrix R according to the equation 16 below. At this time, the error e _{2, j} is also passed to the next stage.

Likelihood calculating unit 38 calculates the total error e _j of each replica on the basis of the results of error calculation by the error calculating unit 35, 37 (the error _{e 1, j, e 2,} j), among the total error e _j The minimum error corresponding to information bits “1” and “0” is selected from the above, and the difference between the square of the minimum error corresponding to information bit “1” and the square of the minimum error corresponding to information bit “0” is estimated. As a measure, the demodulation result is output. That is, the likelihood calculating unit 38 outputs a likelihood LLR from the errors e _{1, j} and e _{2, j} according to the equation 18 described later. This likelihood LLR is passed to the decoding unit 23 in the communication system shown in FIG. 1 as demodulated data.

[Explanation of operation by formula]
The operations of the error calculators 35 and 37 and the likelihood calculator 38 will be described using mathematical expressions.

The error calculators 35 and 37 approximate the equations shown in Equations 10 and 12 as follows.

APP () is an approximate function. Since it is not yet squared at this stage, e1 _{, j} and e2 _{, j} are not the squared Euclidean distance but merely the Euclidean distance. Therefore, APP () is a function that approximates the Euclidean distance. When expressed as a complex number z = a + jb, Euclidean distance | z | = √ (a ² + b ² ). The most general approximation method is APP (z) = | z | = MAX (ABS (a), ABS (b)) + {MIN (ABS (a), ABS (b)) / 2}. MAX () means a function that returns a maximum value, and ABS () means an absolute value. The absolute value is usually a two's complement, but may be a one's complement. Further, as an approximation method with higher accuracy than this, an equation shown in Expression 17 may be used.

Then, the likelihood calculating unit 38, the number 15 and number 16 the error e _{1, j} obtained by the equation, by using e _{2, j,} calculates a similarly error e _j and the number 13. It becomes to calculating the likelihood using the error e _j, where the error _{e 1, j, e 2,} j is not square. Therefore, for example, in the case of QPSK, the calculation is performed according to Equation 18 instead of Equation 14.

[Effects of First Embodiment]
According to the embodiment described above, the amount of processing required for multiplication can be greatly reduced by approximating the square sum of the square error calculation shown in Equations 10 and 12 in QR-MLD by comparison or addition. Can do. For example, considering the case where the 64QAM modulation scheme is used in the configuration of two transmitting antennas and two receiving antennas in LTE, conventionally, information bits are bit # 0, bit # 1, bit # 2, bit # 3, bit In the demodulation of each of # 4 and bit # 5, the corresponding sum of squares is calculated at 64 lattice points. For this reason, a total of 128 square sums is required. Since one multiplication with squares is performed twice, the number of multiplications is 256. That is, 1536 multiplications are required in total for 6 bits. On the other hand, in the present embodiment, the multiplication of the sum of squares is reduced to 0, and the number of additional multiplications is performed twice for the calculation of 1-bit likelihood, and only 6 bits to be demodulated are performed. Become. That is, the number of multiplications can be reduced 1536-12 = 1524 times.

[Second Embodiment]
FIG. 3 is a block configuration diagram of a demodulation circuit according to the second embodiment of the present invention, and shows a configuration example of a circuit used as the demodulation unit 22 in the receiving station 20 in the communication system shown in FIG. Also in this embodiment, a configuration example in which a process for demodulating the transmission data x ₁ and x ₂ is performed by a common circuit is shown. After the process for one transmission data is performed, A description will be given assuming that the propagation path estimation value H ′ is inverted and the same processing is performed. Further, the following explanation demodulation processing of the transmission data x ₂ in the example.

  The demodulation circuit shown in FIG. 3 includes a propagation path estimation unit 31, a QR decomposition unit 32, an equalization unit 33, and a replica generation unit 34, as well as the demodulation circuit shown in FIG. Has an error calculation unit 41, a quadrant detection unit 42, an error calculation unit 43, and a likelihood calculation unit 44 that are slightly different in operation.

The propagation path estimation unit 31 performs propagation path estimation using known data included in the reception data Y, and outputs a propagation path estimation value H ′. The QR decomposition unit 32 performs QR decomposition on the propagation path estimation value H ′, and outputs a unitary matrix Q and an upper triangular matrix R in the equation shown in Equation 3. The equalization unit 33 equalizes the reception data Y using the unitary matrix Q as shown in Equation 6 and outputs equalization data Z. The replica generation unit 34 generates a replica X ′ ₂ of the transmission data X.

The error calculation unit 41 obtains a difference between the received data (each element of the equalized data Z) and each replica (each corresponding element of the replica X ′ ₂ ), and expresses this difference as a complex number a + jb, its real part The absolute value of the imaginary part is taken as n = | a | + j | b |, and a total value is calculated by complex addition for each replica based on the value of n, and an error is calculated by Euclidean distance approximation for the total value. I do. That is, the error n ′ _{2, j} is output from the replica X ′ ₂ and the upper triangular matrix R in accordance with the following equation (19). At this time, the equalized data Z is also transferred to the next stage.

The quadrant detection unit 42 calculates the replica X ′ ₁ from the equalized data Z, the replica X ′ ₂ and the upper triangular matrix R according to the condition of Equation 11. Also, the equalized data Z and the error n ′ _{2, j} are transferred to the next stage.

The error calculation unit 43 obtains a difference between the received data (each element of the equalized data Z) and each replica (each corresponding element of the replica X ′ ₁ , X ′ ₂ ), and performs the same as the error calculation unit 41. The error calculation is performed by Euclidean distance approximation. That is, the error calculation unit 43 outputs the error n ′ _{1, j} from the equalized data Z, the replicas X ′ ₁ , X ′ ₂ , and the upper triangular matrix R according to the equation 20 below. At this time, the noise components n ′ _{2, j} are also transferred to the next stage.

The likelihood calculating unit 44 selects the minimum error corresponding to the information bits “1” and “0” from the errors n ′ _{1, j} and n ′ _{2, j} obtained by the error calculating units 41 and 43, and The demodulation result is output using the difference between the square of the minimum error corresponding to the information bit “1” and the square of the minimum error corresponding to the information bit “0” as a likelihood. In other words, the likelihood calculating unit 44 outputs the likelihood LLR from the errors n ′ _{1, j} , n ′ _{2, j} according to the following equations 21, 22, and 23. This likelihood LLR is passed to the decoding unit 23 in the communication system shown in FIG. 1 as demodulated data.

[Explanation of operation by formula]
The operations of the error calculation units 41 and 43 and the likelihood calculation unit 44 will be described using mathematical expressions.

In the error calculators 41 and 43, the errors n ′ _{1, j} and n ′ _{2, j} are used as real parts instead of the square errors e _{1, j} and e _{2, j} obtained by the equations shown in the equations 10 and 12. Calculate by dividing into imaginary parts. First, the errors n ′ _{2, j} and n ′ _{1, j} are obtained by the equations (19) and (20). The candidates for the replica x _{1, j} necessary for calculating n ′ _{1, j} are performed as usual using the condition of Equation 11.

The likelihood calculating unit 44 calculates the total value n ′ _j of the errors n ′ _{1, j} and n ′ _{2, j} by the equation (21). That is, it is divided into complex I component and Q component, and absolute values are respectively taken and added.

The squared Euclidean distance of the sum n ′ _j may be used as the square error e _j , but here, the error e ′ _j is calculated by the above approximation as shown in Equation 22.

The likelihood calculating unit 44 further calculates the likelihood using the error e ′ _j . Here, the error e ′ _j is not squared. Therefore, for example, in the case of QPSK, the likelihood is obtained not by the equation 14 but by the equation 23.

[Effects of Second Embodiment]
Also in this embodiment, as in the first embodiment, the square error calculation shown in Equations 10 and 12 in QR-MLD is approximated by comparison or addition, and the processing related to multiplication is performed. The amount can be greatly reduced.

[Other embodiments]
Although the MIMO receiver and demodulator according to the embodiments of the present invention have been described above, the present invention can be variously modified without changing the gist. For example, in the above description, the case where the number of reception antennas is 2 has been described as an example, but even when the number of reception antennas is 3 or more, by increasing the number of stages for performing error calculation by Euclidean distance approximation according to the number of reception antennas, The present invention can be similarly implemented. In addition, a configuration in which the processing for demodulating the transmission data x ₁ and x ₂ is performed by separate circuits may be employed.

  The demodulating circuits shown in FIG. 2 and FIG. 3 can also be configured such that some or all of the functions are realized by software with a signal processor. In particular, the error calculation units 35 and 37 and the likelihood calculation unit 44 are assumed to be programs of a signal processor, and data transmitted from a plurality of transmission antennas is assumed to be received and transmitted by a plurality of reception antennas. Error calculation procedure for calculating the error by Euclidean distance approximation for each difference, and calculating the total error for each replica based on the error calculation result by this error calculation procedure. The minimum error corresponding to information bits “1” and “0” is selected from the errors, and the square of the minimum error corresponding to information bit “1” and the square of the minimum error corresponding to information bit “0” are selected. A likelihood calculation procedure for outputting a demodulation result using the difference as a likelihood can also be executed. Similarly, the error calculation units 41 and 43 and the likelihood calculation unit 44 are used as a program for the signal processor, and the data transmitted from the plurality of transmission antennas to the signal processor and the transmission data received by the plurality of reception antennas. When the difference from the assumed replica is obtained and expressed as a complex number a + jb, the absolute value of the real part and the imaginary part is taken as n = | a | + j | b |, and the replica is determined based on the value of n. An error calculation procedure for calculating a total value by complex addition for each time and calculating an error by Euclidean distance approximation for the total value, and information bits “1” and “1” from the errors obtained by this error calculation procedure The minimum error corresponding to “0” is selected, and the difference between the square of the minimum error corresponding to information bit “1” and the square of the minimum error corresponding to information bit “0” is used as the likelihood. It is also possible to execute the likelihood calculation procedure and outputs the demodulation result.

It is a block block diagram of the communication system with which this invention is implemented. FIG. 2 is a block configuration diagram of a demodulation circuit according to the first embodiment of the present invention, showing a configuration example of a circuit used as a demodulation unit in a receiving station in the communication system shown in FIG. 1. It is a block block diagram of the demodulation circuit which concerns on the 2nd Embodiment of this invention, and shows the structural example of the circuit used as a demodulation part in the receiving station in the communication system shown in FIG. It is a figure which shows the communication channel model using a MIMO technique. It is a diagram for explaining a method of calculating the likelihood of the sum e _j of the square error at QPSK.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Transmission station 11 Encoding part 12 Modulation part 13 D / A conversion part 14, 24 Antenna 20 Reception station 21 A / D conversion part 22 Demodulation part 23 Decoding part 31 Propagation path estimation part 32 QR decomposition part 33 Equalization part 34 Replica generator 35, 37, 71, 73 Error calculator 36 Quadrant detector 38 Likelihood calculator 51, 52 Transmit antenna 61, 62 Receive antenna

Claims (8)

MIMO (Multi) receiving data transmitted from a plurality of transmission antennas by a plurality of reception antennas
(Input Multi Output) In the receiving device,
A demodulating unit that performs demodulation for reducing the difference between the received data and a replica assumed as transmission data,
This demodulator
An error calculation unit that calculates a difference between the received data and each replica and calculates an error for each difference by Euclidean distance approximation;
Based on the error calculation result by the error calculation unit, a total error for each replica is obtained, and a minimum error corresponding to information bits “1” and “0” is selected from the total error, and information bit “1” is selected. the difference between the square of the minimum error corresponding to the square and the information bit "0" of the minimum error corresponding as the likelihood, possess a likelihood calculation unit for outputting a demodulation result in,
As the Euclidean distance approximation, the error calculation unit calculates | a | in the case of | a | / 2 ≧ | b | from the absolute value | a | of b and the absolute value | b | of b for the complex number z = a + jb. | + | B | / 4, | a | / 4 + | b | is obtained when | b | / 2 ≧ | a |, and 3 (| a | + | b |) / 4 is obtained otherwise. A MIMO receiver characterized by that. MIMO (Multi) receiving data transmitted from a plurality of transmission antennas by a plurality of reception antennas
(Input Multi Output) In the receiving device,
A demodulating unit that performs demodulation for reducing the difference between the received data and a replica assumed as transmission data,
This demodulator
An error calculation unit that calculates a difference between the received data and each replica and calculates an error for each difference by Euclidean distance approximation;
Based on the error calculation result by the error calculation unit, a total error for each replica is obtained, and a minimum error corresponding to information bits “1” and “0” is selected from the total error, and information bit “1” is selected. the difference between the square of the minimum error corresponding to the square and the information bit "0" of the minimum error corresponding as the likelihood, possess a likelihood calculation unit for outputting a demodulation result in,
As the Euclidean distance approximation, the error calculation unit calculates the maximum value of | a | and | b | from the absolute value | a | and the absolute value | b | of a for a complex number z = a + jb, and | A MIMO receiver characterized by adding a half of the minimum values of a | and | b | . In a MIMO receiver that receives data transmitted from a plurality of transmitting antennas by a plurality of receiving antennas,
A demodulating unit that performs demodulation for reducing the difference between the received data and a replica assumed as transmission data,
This demodulator
When the difference between the received data and each replica is obtained and this difference is expressed as a complex number a + jb, the absolute value of the real part and the imaginary part is taken as n = | a | + j | b |, based on the value of n An error calculation unit that calculates a total value by complex addition for each replica, and calculates an error by Euclidean distance approximation for the total value;
The minimum error corresponding to information bits “1” and “0” is selected from the errors obtained by the error calculation unit, and the square of the minimum error corresponding to information bit “1” and the information bit “0” are selected. as the likelihood of the difference between the square of the minimum error corresponding, it possesses a likelihood calculator for outputting a demodulation result,
As the Euclidean distance approximation, the error calculation unit calculates | a | in the case of | a | / 2 ≧ | b | from the absolute value | a | of b and the absolute value | b | of b for the complex number z = a + jb. | + | B | / 4, | a | / 4 + | b | is obtained when | b | / 2 ≧ | a |, and 3 (| a | + | b |) / 4 is obtained otherwise. A MIMO receiver characterized by that. In a MIMO receiver that receives data transmitted from a plurality of transmitting antennas by a plurality of receiving antennas,
A demodulating unit that performs demodulation for reducing the difference between the received data and a replica assumed as transmission data,
This demodulator
When the difference between the received data and each replica is obtained and this difference is expressed as a complex number a + jb, the absolute value of the real part and the imaginary part is taken as n = | a | + j | b |, based on the value of n An error calculation unit that calculates a total value by complex addition for each replica, and calculates an error by Euclidean distance approximation for the total value;
The minimum error corresponding to information bits “1” and “0” is selected from the errors obtained by the error calculation unit, and the square of the minimum error corresponding to information bit “1” and the information bit “0” are selected. as the likelihood of the difference between the square of the minimum error corresponding, it possesses a likelihood calculator for outputting a demodulation result,
As the Euclidean distance approximation, the error calculation unit calculates the maximum value of | a | and | b | from the absolute value | a | and the absolute value | b | of a for a complex number z = a + jb, and | A MIMO receiver characterized by adding a half of the minimum values of a | and | b | . In a demodulation circuit that performs demodulation in which data transmitted from a plurality of transmission antennas is received by a plurality of reception antennas, and the difference between the replica assumed as transmission data is reduced,
An error calculation unit that calculates a difference between the received data and each replica and calculates an error for each difference by Euclidean distance approximation;
Based on the error calculation result by the error calculation unit, a total error for each replica is obtained, and a minimum error corresponding to information bits “1” and “0” is selected from the total error, and information bit “1” is selected. the difference between the square of the minimum error corresponding to the square and the information bit "0" of the minimum error corresponding as the likelihood, possess a likelihood calculation unit for outputting a demodulation result in,
As the Euclidean distance approximation, the error calculation unit calculates | a | in the case of | a | / 2 ≧ | b | from the absolute value | a | of b and the absolute value | b | of b for the complex number z = a + jb. | + | B | / 4, | a | / 4 + | b | is obtained when | b | / 2 ≧ | a |, and 3 (| a | + | b |) / 4 is obtained otherwise. A demodulation circuit characterized by that. In a demodulation circuit that performs demodulation in which data transmitted from a plurality of transmission antennas is received by a plurality of reception antennas, and the difference between the replica assumed as transmission data is reduced,
When the difference between the received data and each replica is obtained and this difference is expressed as a complex number a + jb, the absolute value of the real part and the imaginary part is taken as n = | a | + j | b |, based on the value of n An error calculation unit that calculates a total value by complex addition for each replica, and calculates an error by Euclidean distance approximation for the total value;
The minimum error corresponding to information bits “1” and “0” is selected from the errors obtained by the error calculation unit, and the square of the minimum error corresponding to information bit “1” and the information bit “0” are selected. as the likelihood of the difference between the square of the minimum error corresponding, it possesses a likelihood calculator for outputting a demodulation result,
As the Euclidean distance approximation, the error calculation unit calculates | a | in the case of | a | / 2 ≧ | b | from the absolute value | a | of b and the absolute value | b | of b for the complex number z = a + jb. | + | B | / 4, | a | / 4 + | b | is obtained when | b | / 2 ≧ | a |, and 3 (| a | + | b |) / 4 is obtained otherwise. A demodulation circuit characterized by that. In the processor that performs signal processing,
An error calculation procedure for obtaining a difference between data transmitted from a plurality of transmission antennas by data received by a plurality of reception antennas and a replica assumed as transmission data, and calculating an error for each difference by Euclidean distance approximation; ,
Based on the error calculation result by this error calculation procedure, a total error for each replica is obtained, and a minimum error corresponding to information bits “1” and “0” is selected from the total error, and information bit “1” is selected. A likelihood calculation procedure for outputting a demodulation result using the difference between the square of the minimum error corresponding to and the square of the minimum error corresponding to information bit “0” as a likelihood ,
As the Euclidean distance approximation, for the complex number z = a + jb, the error calculation procedure described above is based on the absolute value | a | and the absolute value | b | of b, and | a | if | a | / 2 ≧ | b | | + | B | / 4, | a | / 4 + | b | is obtained when | b | / 2 ≧ | a |, and 3 (| a | + | b |) / 4 is obtained otherwise. A signal processing program. In the processor that performs signal processing,
When the difference between the data transmitted from a plurality of transmitting antennas and the data received by the plurality of receiving antennas and the replica assumed as the transmission data is obtained and this difference is expressed by a complex number a + jb, the absolute values of the real part and the imaginary part An error calculation procedure for calculating n = | a | + j | b |, calculating a total value by complex addition for each replica based on the value of n, and calculating an error by Euclidean distance approximation for the total value; ,
The minimum error corresponding to information bits “1” and “0” is selected from the errors obtained by this error calculation procedure, and the square of the minimum error corresponding to information bit “1” and the information bit “0” are selected. A likelihood calculation procedure for outputting a demodulation result with the difference from the square of the corresponding minimum error as a likelihood , and
As the Euclidean distance approximation, for the complex number z = a + jb, the error calculation procedure described above is based on the absolute value | a | and the absolute value | b | of b, and | a | if | a | / 2 ≧ | b | | + | B | / 4, | a | / 4 + | b | is obtained when | b | / 2 ≧ | a |, and 3 (| a | + | b |) / 4 is obtained otherwise. A signal processing program.

Images