WO2008023683A1

WO2008023683A1 - Signal separating device and signal separating method

Info

Publication number: WO2008023683A1
Application number: PCT/JP2007/066155
Authority: WO
Inventors: Kazunari Hashimoto
Original assignee: Panasonic Corporation
Priority date: 2006-08-22
Filing date: 2007-08-21
Publication date: 2008-02-28

Abstract

Provided is a signal separating device in which a reception process is shared by the MIMO mode and the SIMO mode and the reception process is performed in accordance with multi-mode in a communication system in which a multi-stream and a single stream are mixed. The device includes a channel compensation coefficient calculation unit (140). In the case of the multi-stream, the channel compensation coefficient calculation unit (140) calculates a unitary matrix as the channel compensation matrix by QR-decomposing a channel estimation value and calculates an upper triangular matrix as the triangular matrix. On the other hand, in the case of the single stream, the channel compensation coefficient calculation unit (140) calculates a diagonal matrix having channel estimation values h1, h2 as diagonal elements as the channel compensation matrix and calculates a diagonal matrix having |h1|2, |h2|2 as diagonal elements as the triangular matrix.

Description

Specification

Signal separation apparatus and signal separation method

Technical field

The present invention relates to a signal separation device and a signal separation method, and more particularly to a signal separation device and a signal separation method for performing reception processing in a communication system in which multistreams and single streams are mixed.

Background art

In recent years, in a wireless communication system, in order to further increase the subscriber capacity and increase the transmission speed, spatial multiplexing transmission (MIMO: Multi_I beating) in which communication is performed using a plurality of antennas on the transmitting side and the receiving side. ut Multi-Output) is drawing attention. In MIMO, transmission signals (multi-streams) are transmitted from multiple transmission antennas in the same band, and signals received by multiple reception antennas are separated to extract information.

[0003] As a signal separation processing method for MIMO signals (multi-stream), the maximum likelihood half-time IJ method (QR—MLD (Maximum Likelihood Detection with QRdecomposition)) using QR decomposition has been used (Non-patent Document 1). ). Figure 1 shows an example of the main configuration of a receiver that separates MIMO signals using QR-MLD. Receiving apparatus 10 shown in FIG. 1 includes a receiving antenna 11 1, 11 2, the receiving unit 12 1, 12 2, the channel estimator 13, QR decomposition unit 14, and orthogonal (Q ^H multiplication) unit 15, A maximum likelihood determination (MLD) unit 16, an LLR (Log Likelihood Ratio) calculation unit 17, and a decoding unit 18 are provided. Hereinafter, a description will be given of a MIMO signal separation processing operation by the receiving apparatus 10 configured as described above. In Fig. 1, the receiving antenna N

R

= 2 shows the configuration example, but the number of receiving antennas is not limited to receiving antenna N = 2.

R

A receiving unit may be provided depending on the situation.

[0004] Signals transmitted from N transmitting antennas x, signals received by N receiving antennas

T R

The relationship y = Hx + n exists between the number y. H is the channel estimation value, n is the receiver noise component, and the transmission signal χ, reception signal y, channel estimation value H, and receiver noise component n are expressed by equations (1) to (4). The received signal y is expressed as in equation (5) from y = Hx + n.

[Number 1]

(2)

"= (" Γ ·· "^ ϊ (4)

(Five)

[0005] The QR decomposition unit 14 performs QR decomposition on the channel estimation value H as shown in Equation (6). In QR decomposition, as shown in Eq. (6), channel estimation value H is decomposed so that matrix Q after QR decomposition is a unitary matrix 歹 lj and row 歹 IJR is an upper triangular matrix.

[0006] The unitary matrix is a complex matrix (^^ ni. — ¹ ) where the inverse matrix is obtained by complex conjugate transpose.

Q ^H Q = I (I is the unit line 歹 IJ). The diagonal elements of the upper triangular row 歹 IJR are real numbers.

[Equation 6]

[0007] Here, substituting equation (6) into equation (5) yields equation (7) _c

[Equation 7]

[0008] orthogonalization (Q ^H multiplication unit) 15, Keru multiplying the complex conjugate transpose matrix Q ^H of the matrix Q on both sides of the equation (7) (Equation (8)).

[Equation 8]

At this time, if the noise component of the second term on the right side is ignored, Equation (9) is obtained.

[Equation 9]

[0010] Incidentally, are orthogonalized and express applying a Q ^H on the left side of the received signal y, hereinafter, will be representative of the received signal after orthogonalization and z (Equation (10)).

[Equation 10]

∑ = Q ^h y = Rx

in… q _N ^R \ N _T (10)

0

^R N _X N _T J In the above formula (10), the elements of z are numbered sequentially from the bottom for convenience.

[0011] The maximum likelihood determination unit 16 performs demodulation by maximum likelihood determination by the QR-MLD method, using the reception signal z orthogonalized as shown in Equation (10).

[0012] In the following, the maximum likelihood judgment is made using the case where the number of transmitting antennas N = 2 and the number of receiving antennas N = 2 as an example.

T R

The maximum likelihood determination process in the fixed unit 16 will be described. Number of transmitting antennas N = 2, receiving antenna

T

When the number of antennas N = 2, the transmission signal x, the reception signal y, and the channel estimation value H are expressed by Equation (11) to

R

It is expressed as equation (13). (1 1)

An y (1 2)

h (1 3)

[0013] Channel estimation value H is QR-decomposed by QR decomposition unit 14 as shown in equation (14) _c

[Equation 14]

[0014] Further, the received signal z after being orthogonalized by the orthogonalizing unit 15 is expressed as in Expression (15).

[Equation 15]

The maximum likelihood determination unit 16 performs maximum likelihood determination by the QR-MLD method from the received signal z that has been orthogonalized using Equation (15). QR — In the maximum likelihood determination by the MLD method, the received signal y is multiplied by the complex conjugate transpose matrix Q ^H and the orthogonal received signal z is multiplied by the transmit replica and the upper triangular matrix R is multiplied by the symbol replica candidate. The distance between signal points is compared with a point, and a transmission signal is determined from a combination of transmission symbols having the smallest distance between the signal points. The distance between signal points is usually the square Euclidean distance, but the Manhattan distance may be used to reduce the amount of computation.

[0016] Furthermore, in QR-MLD maximum likelihood determination, z = r with only one variable X in equation (15)

1

After obtaining the square Euclidean distance e ² expressed by equation (16) based on the equation of X (the first step

twenty two 1)), the squared Euclidean distance e expressed by equation (17) based on the equation z = rx + rx

Estimate the final X and X from the minimum of 2 11 1 12 2 2 (second stage).

1 2

Specifically, when the square Euclidean distance is used as the signal point distance, first, as a first stage, for each symbol point of the transmission replica d (a replica of the transmission signal X), an equation (

1 2

16) is used to calculate the square Euclidean distance, and the transmission replica d having the smallest square Euclidean distance expressed by Equation (16) is provisionally determined as the transmission signal X.

1 2

[Equation 16] = ki-ι | ² · · · (丄 6)

[0018] Furthermore, as a second stage, the transmission replica d and the corresponding squared Euclidean distance e

1

² and for each symbol point of transmit replica d (replica of transmit signal X),

1 2 1

To calculate the squared Euclidean distance.

[Equation 17] e ₂ = | z.-R _n d-

+ e,…, 丄 7)

[0019] For example, when the modulation scheme is QPSK, there are four symbol points, so there are as many values as the transmission replica d can take, and a transmission replica d can be taken for each transmission replica d.

1 1 2

There are four values. Therefore, if the number of transmit antennas N = 2, transmit replica d

There are 4 ² = 16 possible combinations of T 1 and d. When the modulation method is 64QAM, there are 64 ² = 4096 possible combinations of transmit replicas d and d.

1 2

Yes

[0020] In the QR-MLD method, the transmission replication power d used in the second-stage squared Euclidean distance is the power used for all symbol candidates. The number of computations of this squared Euclidean distance is reduced.

1

QR-MLD method (QRM—MLD (Maximum Likelihood Detection with QRdecomposition and the M_algorithm method) / Q Transmit replica used at Euclidean distance d

1 The number of candidates is set to S in order of increasing squared Euclidean distance obtained in the first stage (S is d

1) is less than the total number of possible symbol candidates). ing.

In this way, the maximum likelihood determination unit 16 holds the transmission replica candidate (d 1, d 2) and the corresponding square Euclidean distance in ascending order of the square Euclidean distance calculated from the equation (17).

1 2

[0022] Based on the square Euclidean distance calculated by the maximum likelihood determination unit 16, the LLR calculation unit 17 obtains a log likelihood ratio (LLR: Log Likelihood Ratio) for each bit for channel decoding.

[Equation 18]

LLR = _{ma pb} , one e _min _ _{p 0} … (1 8)

Here, e ² represents the minimum value of e ² when the b-th bit of the transmission antenna p is “1”. same

min, p, b, l 2

Similarly, e ² represents the minimum value of e ² when the b-th bit is “0”. QR — Maximum likelihood by MLD method min, p, b, 0 2

In the determination, the log likelihood ratio required for channel decoding is obtained in this way.

[0024] The decoding unit 18 decodes the transmission signal x by Turbo decoding or the like using the log likelihood ratio.

Yes

[0025] Thus, in the maximum likelihood determination for the multi-stream by the QR-MLD method, the channel estimation value is decomposed into the product of the unitary matrix Q and the upper triangular row IJR, and then the square Euclidean distance is calculated. On the other hand, the equation for calculating the square Euclidean distance E directly from Equation (5) without using QR decomposition is as shown in Equation (19). The transmit replicas d and d in equation (19) are

1 2

It is a replica of transmitted signals X and X.

1 2

[Equation 19]

^E =

… ( ^{1 9} )

[0026] Comparing the calculation formula (16), (17) of the square Euclidean distance by QR—MLD method with the calculation formula (19) for calculating the square Euclidean distance by the conventional MLD method without using QR—MLD As can be seen, in the QR-MLD method, since the lower part of the upper triangular matrix R is zero, the amount of computation required to calculate the square Euclidean distance can be reduced compared to the MLD method.

[0027] By the way, MIMO is capable of increasing the transmission speed S, and information is transmitted reliably. In some cases, single stream transmission by SIMO (Single Input Multiple Output) is suitable. In SIMO, signals transmitted from a single antenna are received by multiple antennas, and the received signals are received or processed by selection or combining diversity, so information is reliably transmitted when the SN (Single to Noise) ratio is small. It is effective when you want to. The SIMD reception method with the best reception performance in a low SNR environment is the ML D method. In the following, regarding the maximum likelihood determination for a single stream by the MLf ¾D method, when the number of transmitting antennas N = 1 and the number of receiving antennas N = 2 (hereinafter also referred to as “IX 2 antennas”)

T R

Will be described as an example.

[0028] When the transmission signal is represented by χ and the channel estimation value is represented by H (Equation (20)), the received signal y is represented by Equation (21). In equation (21), receiver noise is ignored as in equation (10).

[Equation 20] h (2 0)

, J

( ( twenty one )

i

[0029] In the maximum likelihood determination by the MLD method, the signal point distance between the actually received received signal y and the symbol replica candidate point obtained by multiplying the transmission replica by the channel estimation value is compared, and this signal is compared. The transmission signal is determined from the combination of transmission symbols with the smallest point-to-point distance. For example, when the square Euclidean distance is used as the distance between signal points and a single stream is received by multiple receiving antennas and the maximum likelihood is determined, the square Euclidean distance expressed by equations (22) and (23) is used. The transmission replica d to be minimized is determined as the transmission signal X.

[Number 22]

( twenty two )

( twenty three ) [0030] That is, in the maximum likelihood determination for a single stream by the MLD method, a square Euclidean distance is calculated independently for each antenna as in equations (22) and (23), and a transmission replica is determined. In other words, in the maximum likelihood determination for a single stream by the MLD method, when the number of transmit antennas is N = 1 and the modulation scheme is QPSK, there are four symbol points, so

T

When there are four possible values of force d and the number of receiving antennas N = 2, transmit replica d

R

Since there are two power antennas, 4 X 2 = 8 Euclidean distances are calculated. When the modulation method is 64QAM, 64 Eq = 128 Euclidean distance calculations are performed.

[0031] Then, the addition value e ² of the square Euclidean distance for two antennas (Equation (24)) is set as the final square Euclidean distance, and the transmission replica candidates are selected in order of decreasing square Euclidean distance.

[Number 24]

[0032] In other words, the MLD method differs from the QR-MLD method in that the final squared Euclidean distance is obtained by adding the squared Euclidean distance e ² (equation (24)) of the two antennas. On the other hand, in the QR—MLD method, when the number of transmit antennas is N = 2, the square Euclidean distance is calculated.

T

The output is in two stages corresponding to the same number of transmission signals, and the result of the squared Euclidean distance in the first stage is used to calculate the squared Euclidean distance in the second stage.

[0033] In the MLD method, the smaller one is selected from the squared Euclidean distances of the antennas rather than simply calculating the squared Euclidean distances by adding the squared Euclidean distances of the antennas as shown in Equation (24). (E ² = min (e ² ) / 2)

You may make it calculate using methods, such as hanging.

[0034] The square Euclidean distance calculated in this way is held in correspondence with the transmission replica candidates in ascending order of their power in the same manner as in the case of maximum likelihood determination by the QR-MLD method. The distance is used to calculate the log-likelihood ratio for each bit for channel decoding. In SIMO, the number of transmit antennas N = 1, so the log likelihood

T

The ratio LLR is calculated using Equation (25) similar to Equation (18). [Equation 25]

LLR = e _{→ l} ² −e _{→ 0} ² (2 5) where e ² represents the minimum value of e ² when the b-th bit is '1'. Similarly, e ² is a bit

min.b.l min, b, 0

Represents the minimum value of e ² when the force S is '0'.

[0035] As described above, the square Euclidean distance in the QR—MLD method is expressed by the following equations (16) and (

17).

[0036] In Equation (16), z is a received signal after orthogonalization, and d is a transmission replica. Also

1 1

, R are diagonal elements of the R matrix obtained by QR decomposition of the channel estimate H, and are real numbers. In addition,

twenty two

Since the transmission replica is one of the signal points of the transmission symbol point, multiplying the transmission replica by the real number r in Equation (16) changes only the amplitude with respect to the transmission replica.

twenty two

Is equivalent to

[0037] As can be seen from equation (16), in order to calculate the square Euclidean distance in the first stage, real number and complex number multiplication, that is, real number multiplication is performed twice, and then complex subtraction is performed once. After obtaining the value of (z −rd), the absolute value is obtained.

1 22 1

On the other hand, the square Euclidean distance in the case of the 1 × 2 antenna is calculated using Equation (26) to Equation (28) by the MLD method.

[Equation 26]

-h _x d (2 6)

[Equation 27] e ₂ ² = Les ₂ -h ₂ d \ ² … (2 7)

[Equation 28] e ¹ = e _x +…, 2 o) where y and y are received signals, d is a transmitted replica, h and h are channel estimates and complex numbers

1 2 1 2

It is. In Equation (26) and Equation (27), the transmission replica d is multiplied by complex numbers h and h.

1 2

This is equivalent to changing the amplitude and phase with respect to the transmission replica. [0039] At first glance, Eqs. (26) and (27) appear to be equal to the Euclidean distance calculation formula (16) of the first stage of the QR-MLD method.

Non-Patent Literature 1: Complexity-reduced Maximum Likelihood Detection Based on Replica candidate Selection with Decomp osition Using Pilot-Assisted Channel Estimation and Ranking for MIMOMultiplexing Using OFCDM), IEICE Technical Report, RCS2003_312 (2004-3)

Disclosure of the invention

Problems to be solved by the invention

However, the channel estimation values h 1 and h hung on the transmission replica in the expressions (26) and (27) are complex numbers. Therefore, the square Euclidean distance is calculated from these equations.

1 2

To do this, complex multiplication (4 real multiplications and 2 real additions) and 1 complex subtraction must be performed to determine the y -hd value, then the absolute value must be calculated and squared. I must. Tsuma

1 1 1

Therefore, when calculating the square Euclidean distance for one time, the MLD method requires two real multiplications and two real additions more than the QR-MLD method. Therefore, in this state, a MIMO mode receiver that separates a spatially multiplexed signal from a plurality of transmit antennas using the same frequency and time using a plurality of receive antennas, and a single antenna. Force to receive transmitted signals with multiple antennas and select the received signals or receive them by combining diversity. There is a problem that the receiver cannot be downsized.

[0041] An object of the present invention is to share reception processing in the MIMO mode and SIMO mode in a communication system in which multistreams and single streams are mixed, and perform reception processing corresponding to the multimode. A signal separation device and a signal separation method are provided.

Means for solving the problem

[0042] In order to solve this problem, the present invention is based on a force in which a received signal is either a multistream or a single stream, a mode identifying means for identifying, and a channel estimation value. The channel compensation coefficient for calculating the channel compensation matrix and the triangular matrix is switched by switching the method for calculating the channel compensation matrix and the triangular matrix between the case where the received signal is a multi-stream and the case of a single stream. A calculating means; an orthogonalizing means for multiplying the received signal by a complex conjugate transpose matrix of the channel compensation matrix; a multiplication result of the received signal and the complex conjugate transposed matrix; the triangular matrix and a transmission replica; Maximum likelihood determining means for calculating a distance between signal points of a multiplication result with a signal, and likelihood calculating means for calculating the likelihood of the transmission replica signal using the distance between the signal points. Take the configuration.

[0043] According to this configuration, the complex conjugate transposed matrix multiplied by the received signal and the triangular matrix multiplied by the transmission candidate signal can be switched between the multi-stream case and the single stream case. In the case of single stream, it becomes possible to calculate the distance between signal points using the same method as the method for calculating the distance between signal points for multi-stream. The SIMO mode reception processing can be shared to reduce the size of the signal separation device that can handle both modes.

The invention's effect

According to the present invention, in a communication system in which multistreams and single streams are mixed, signals that are spatially multiplexed using the same frequency and time by a plurality of transmitting antennas are received by a plurality of receiving antennas. It is possible to share the reception processing between the MIMO mode that is received by using the SIMO mode and the SIMO mode that receives signals transmitted from a single antenna using multiple antennas, and can perform reception processing corresponding to the multimode.

Brief Description of Drawings

[0045] [FIG. 1] A block diagram showing a main configuration of a conventional receiving apparatus when receiving a MIMO signal using the QR—MLD method.

FIG. 2 is a block diagram showing a main configuration of the receiving apparatus according to the embodiment of the present invention.

[Fig.3] Block diagram showing the main configuration of the QR decomposition unit when receiving and processing MIMO signals using the QR-MLD method

[Fig. 4] The main configuration of the QR decomposition unit when receiving and processing SIMO signals using the MLD method. [Fig. 5] A block diagram showing the main configuration of the channel compensation coefficient calculation unit according to the above embodiment. Lock figure

[Fig. 6] Block diagram showing the main configuration of the orthogonalization unit when receiving and processing MIMO signals using the QR-MLD method.

[Fig.7] Block diagram showing the main configuration of the orthogonalization unit when receiving and processing SIMO signals using the MLD method

[Fig.8] Block diagram showing the main configuration of the maximum likelihood decision unit when receiving and processing MIMO signals using the QR-MLD method

[Fig. 9] The configuration of the main part of the maximum likelihood determination unit when receiving and processing SIMO signals using the MLD method

FIG. 10 is a block diagram showing a main configuration of a maximum likelihood determination unit according to the embodiment.

FIG. 11 is a diagram for explaining the heel mode, SIMO mode, and SISO mode. BEST MODE FOR CARRYING OUT THE INVENTION

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

[0047] (Embodiment)

First, a method for sharing a multi-stream signal demultiplexer using the QR-MLD method as a single-stream signal demultiplexer using the MLD method will be described with reference to equations.

[0048] (Sharing method)

As described above, in the MLD method, the power to calculate the square Euclidean distance using Equations (26) to (28). In this embodiment, the channel estimation values are respectively applied to the y and h d terms in Equation (26).

1 1

After taking the conjugate of, the method of calculating the squared Euclidean distance is taken (Equation (29)).

[Equation 29]

[0049] As in equation (29), the conjugate of the channel estimation value is applied to the y term and h d term in equation (26), respectively.

1 1

After multiplication, the square Euclidean distance is calculated, and the coefficient multiplied to the transmission replica is made real. As a result, the square Euclidean distance can be calculated by performing real multiplication and complex multiplication (two real multiplications) and complex subtraction once. In other words, the square Euclidean distance can be calculated with the same number of operations as the QR—MLD method.

[0050] It should be noted that the number of operations required for calculating the square Euclidean distance is the same as the number of operations required for calculating h.

1 1

The number of calculations is not counted. This is because, in the QR-MLD method, the received signal y is multiplied by the complex conjugate transpose matrix Q ^H to the number of operations required to calculate the square Euclidean distance by the QR-M LD method shown above, and the received signal z The number of operations required to find

This is to meet the conditions.

[0051] QR-received signal z after orthogonalization by MLD method, _c calculated by the equation (32)

[Equation 32] ίζΛ + y ₂ (3 2) 1 /

8 2) + in y ₂ ) Using! / In the expression (32), and using the expression (33) as the complex conjugate transpose matrix Q ^H , the expression (34) is obtained:

[Equation 33]

[0052] Comparing equation (34) and equation (15), equation (34) is the same as equation (15). I understand. In other words, in Equation (32), by using Equation (33) as the complex conjugate transpose matrix Q ^H , when the square Euclidean distance by MLD method is performed using orthogonalization processing similar to QR-MLD method, Necessary h and h can be calculated.

1 1 2 2

[0053] For the received signal _z thus orthogonalized using Equation (34), 1

In order to calculate the square Euclidean distance in the case of the X2 antenna, it is necessary to calculate the square Euclidean distance for each receiving antenna as shown in equations (26) and (27). On the other hand, in the QR-MLD method, the square Euclidean distances in the first and second stages are calculated using Eqs. (16) and (17). As described above, Equation (16) for calculating the square Euclidean distance in the first stage by the QR-MLD method is the same as Equation (26) by the MLD method.

[0054] On the other hand, equation (17) is a force S that is a different calculation formula from equation (26), and rd and e output from the first stage to the second stage in equation (17). ^If you replace ² with zero,

12 1 1

Expression (35), which is a calculation expression similar to Expression (26), can be obtained.

[Equation 35]

[0055] That is, by using a diagonal matrix with r equal to zero as the upper triangular matrix used in the calculation of the square Euclidean distance by the QR-MLD method, each stage is independent.

12

It is now possible to calculate the square Euclidean distance, and to calculate the square Euclidean distance by the MLD method for SIMO using the circuit resource of the QR-MLD method.

[0056] The following is an aside. When calculating the square Euclidean distance for SIMO using the multi-stream QR-MLD circuit resources, SIMO does not have a transmitting antenna 2; There is a case where the input of the corresponding part is set to zero and the QR—MLD method circuit is moved! /, Or not! /.

That is, using channel estimation value H (Equation (37)) in which each component of channel estimation value H is zero as shown in Equation (36), received signal y is obtained from Equation (38). It is something to be calculated 0 \

(3 6)

0

However, for the channel estimation value H shown in Equation (37), since each component of H is zero, r and r of the upper triangular matrix R become zero, and Q cannot be obtained. QR can be decomposed

12 22 2

I can't. In other words, if the input corresponding to transmit antenna 2 is simply zero, the maximum likelihood determination cannot be performed using the MLD method using circuit resources of the QR-MLD method, and the circuit can be shared. I understand!

[0059] (Configuration)

FIG. 2 shows a main configuration of the receiving apparatus according to the embodiment of the present invention. A receiving apparatus 100 shown in FIG. 2 includes receiving antennas 110-1 and 110-2, receiving sections 120-1 and 120-2, a decoding section 180, and a signal separating apparatus 190. Signal separating unit 190, channel estimation section 130, and a mode identification unit 135, channel compensation coefficient calculating section 140, orthogonal (Q ^H multiplication) unit 150, a maximum likelihood-format tough 160, LLR calculation unit 170 .

[0060] Hereinafter, a configuration of main parts of the signal separation device 190 in the case where multi-stream reception using the QR-MLD method and single-stream reception using the MLD method are shared will be described with reference to the drawings.

[0061] In FIG. 2, the main configuration in the case of the number of receiving antennas N = 2 (I X 2 antennas)

R

Although an example is shown, the number of receiving antennas N is not limited to 2, and reception is performed according to the number of receiving antennas N.

R R

A communication unit may be provided.

The receiving antennas 110-1 and 110-2 are multi-streams transmitted from a communication partner (not shown). Stream or single stream is received, and the receivers 120-1 and 120-2 are output.

[0063] Receiving sections 120-1 and 120-2 perform reception and demodulation processing on the multistream or single stream received via receiving antennas 110-1 and 110-2, and receive signals obtained Output y to channel estimation unit 130 and orthogonalization unit 150

[0064] Channel estimation section 130 performs channel estimation from received signal y, and outputs a channel estimation result to channel compensation coefficient calculation section 140.

[0065] Mode identification section 135 identifies whether the transmitted signal is multistream or single stream from a control signal notified from a communication partner (not shown), and the identification result is used as a channel compensation coefficient. The calculation unit 140 and the maximum likelihood determination unit 160 output.

[0066] (Shared channel compensator)

In describing the main configuration of the channel compensation coefficient calculation unit 140 according to the present embodiment, first, QR decomposition processing calculation is performed as channel compensation coefficient calculation processing when receiving a MIMO signal using the QR-MLD method. The QR decomposition unit 240 is explained. Figure 3 shows the main configuration of the QR decomposition unit 240. The QR decomposition unit 240 shown in Fig. 3 includes input terminals 240-1, 240-2 squared norems 241-1, 241-2, inverse square roots 242-1, 242-2, and real multiplication 243-1, 243-2, 244-1, 244-2, inner product 245, complex multiplication 246, complex subtraction 247, and output unit 248.

[0067] The QR decomposition unit 240 decomposes the channel estimation value H into a unitary matrix Q and an upper triangular matrix R using the QR-MLD method (Formula (14)).

Here, when each component of channel estimation value H and unitary matrix Q is placed as shown in equation (39), each component of upper triangular matrix R is calculated from equations (40) to (45). Figure 3 also shows the input / output relationship. The subscripts i and q used in equation (40) represent the real part and the imaginary part, respectively.

[Equation 39]

_}

[Equation 40] r _n = I = i = l ( _n ), ² + () + i), ² + (h ₂₁ ) _q (40)

(42)

r22 = 44)

[0069] The inner product 245 calculates an inner product operation h'q of h = (h, h) q = (q, q) (formula (46)) ₍ abab

[Equation 46]

q = ^H q = h _a q _a + h _b q _b (46)

[0070] If the vector is one-dimensional, the inner product operation h'q is expressed by equation (47) _c

[Equation 47]

The output unit 248 outputs the complex conjugate transpose matrix Q ^H of the unitary matrix Q decomposed based on the channel estimation value H as described above to the orthogonalization unit 150 and outputs the upper triangular matrix R to the maximum likelihood determination unit. Output to 160. [0072] FIG. 4 shows a configuration example of a main part of the QR decomposition unit 340 for calculating a channel compensation coefficient when receiving and processing a SIMO signal using the MLD method. In FIG. 4, parts that are the same as those in FIG. The QR decomposition section 340 shown in FIG. 4 includes an input terminal 240-2, an inner product 245, and an output section 249. Figure 4 also shows the input / output relationship.

[0073] The inner product 245 is h * h =

1 1 I h and h * h =

1 2 2 I h

To calculate the ² I 2.

[0074] The output unit 249 outputs a complex conjugate transpose of a matrix having channel estimation values h and h as diagonal elements.

1 2

Is output to the orthogonalization unit 150 and I h and I h I ² are diagonal elements.

1 2

The sequence is output to maximum likelihood determination section 160.

FIG. 5 shows a configuration example of a main part of channel compensation coefficient calculation section 140 according to the embodiment of the present invention. In FIG. 5, parts that are the same as those in FIG. As shown in FIG. 5, the channel compensation coefficient calculation unit 140 employs a configuration in which a switch 1 41 -1-141 -5 and an output unit 142 are added to the QR decomposition unit 240 shown in FIG.

[0076] Channel compensation coefficient calculation section 140 switches a signal to be output to the arithmetic unit in the subsequent stage of each switch according to mode selection information S output from mode identification section 135, and calculates a channel compensation matrix from the channel estimation value. And calculate the triangular matrix. Here, the mode selection information is information for identifying whether it is multistream or single stream, and is notified from a communication partner (not shown).

[0077] When the mode selection information indicates multi-stream, H = (h, h) is input to the input terminal 240-2, while the mode selection information is a single stream.

2 12 22

In order to indicate that this is the case, H = (h, h) is input to the input terminal 240-2.

1 2

When the mode selection information indicates multi-stream, switch 141-1 outputs Q to inner product 245 in the subsequent stage. On the other hand, the mode selection information is a single stream.

1

Switch 141-1 outputs H = (h, h) to the inner product 245 in the latter stage.

1 2

[0079] That is, when the mode selection information indicates multi-stream, inner product 245 calculates r, and when mode selection information is single stream, inner product 245 is I h

Calculate 12 1 and I h I ² .

[0080] When switch 141-2 indicates that the mode selection information is multi-stream, Q is output to the output unit 142.If the mode selection information is single stream, h is output.

1 1 Outputs to power unit 142.

[0081] The switch 141-3 outputs r to the output unit 142 when the mode selection information indicates multi-stream, and when the mode selection information is single-stream,

11 I h

2 I

² is output to the output unit 142.

[0082] When the mode selection information indicates multi-stream, the switch 141-4 outputs r to the output unit 142, and when the mode selection information is single-stream,

22 I h

1 I

² is output to the output unit 142.

[0083] The switch 141-5 outputs Q to the output unit 142 when the mode selection information indicates multi-stream, and outputs h when the mode selection information is single-stream.

2 2 Outputs to force section 142.

[0084] In this way, the switch 141— ;! to 141-5 can be switched according to the mode selection information S indicating whether the stream is multistream or single stream. Channel compensation matrix and triangular matrix required for stream and single stream reception processing can be acquired.

That is, in the case of multi-stream, channel compensation coefficient calculation section 140 calculates a unitary matrix as a channel compensation matrix by QR decomposition of the channel estimation value, and calculates an upper triangular matrix as a triangular matrix.

[0086] On the other hand, in the case of single stream, channel compensation coefficient calculation section 140 calculates a diagonal matrix having channel estimation values h and h as diagonal elements as a channel compensation matrix, and I h I ² and I h

1 2 1 2

A diagonal matrix having I ² as a diagonal element is calculated as a triangular matrix.

[0087] The output unit 142 in the case of multi-stream, and outputs the complex conjugate transposed matrix Q ^H of Yunitari matrix Q to the orthogonalization section 150, while outputting the upper triangular ascending歹IJR to maximum likelihood determination unit 160 In the case of single stream, the diagonal with channel estimation values h and h as diagonal elements

1 2

The complex conjugate transpose matrix of the matrix is output to the orthogonalization unit 150 and I h is

1, I h

2 Outputs a diagonal matrix as diagonal elements to maximum likelihood determination section 160.

[0088] Specifically, in the case of multi-stream, the received signal y = y + jy, y = y + jy

1 li lq 2 2i Corresponding to the inputs y, y, y., Y to the orthogonalization unit 150, q, q, q , Q Imaginary part z of force z is q, q, q, q Real part z of force z is q,

12¾ 12i 22¾ 22i 2 2i Hi q, q The imaginary part z of the force z is input q, q q, q respectively.

21i 21¾ 2 2¾ l id 1

[0089] On the other hand, in the case of over-time, the input y to the orthogonalization section 150 of the received signal y = y + jy, corresponding to y, the the real part z of z h, h force S, _z

li Id 1 li li Id 1

H and h are input to the imaginary part z of. Also, the orthogonalization part of the received signal y = y + jy 15 lq lq li 2 2i 2q

Corresponding to inputs y and y to 0, h to the real part z of z, h to the imaginary part z of the h force z,

2i 2q 2 2i 2i 2q 2 2q 2q h is input.

2i

[0090] (Sharing of orthogonalization unit)

First, orthogonalization when receiving and processing MIMO signals using the QR-MLD method is explained. In the QR-MLD method, the complex conjugate transpose matrix Q ^H is multiplied as a weight to the received signal y as shown in Equation (48).

[Number 48]

) + · / (_ ^

) + Zo (_ _2?

-(4 8)

FIG. 6 shows a main configuration of orthogonalizing section 250 that realizes equation (48). The orthogonalization unit 250 includes a multiplier 251— ;! to 251—4, a delay unit 252— ;! to 252—4, 254— ;! to 254-4, and a calorie calculator 253— ;! to 253—. 4 and.

[0092] On the other hand, in orthogonalization when receiving and processing SIMO signals using the MLD method, the complex conjugate transpose Η * of the channel estimation value Η is multiplied as a weight on the received signal y as shown in Equation (49). It is done.

[Number 49]

FIG. 7 shows the main configuration of the orthogonalizing unit 350 that realizes the equation (49). The orthogonalization unit 350 shown in FIG. 7 employs the same configuration as the orthogonalization unit 250 using the QR-MLD method shown in FIG. 251— ;! to 251-4, delay units 252— ;! to 252—4, 254—;! To 254-4, and Calorie calculator 2 53— ;! to 253-4. Therefore, as shown in Equation (49), the orthogonalization unit can be shared by multiplying the received signal y by the complex conjugate transpose H * of the channel estimation value H as a weight. As can be seen from Eqs. (48) and (49), in the case of the MLD method, orthogonalization is performed with half the amount of computation compared to the QR-MLD method.

[0094] (Share the maximum likelihood determination unit)

As described above, in the maximum likelihood determination of the QR-MLD method, the maximum likelihood determination is performed in the two stages of the first stage and the second stage. The square Euclidean distance calculation for each stage is calculated from Equation (16) and Equation (17), and the square Euclidean distance e ² at each transmit replica d of the first stage.

1 1 is used to calculate the squared Euclidean distance in the second stage.

FIG. 8 shows a main configuration of maximum likelihood determination section 260 when receiving and processing a MIMO signal using the QR-MLD method. The maximum likelihood determination unit 260 shown in FIG. 8 includes transmission replica generation units 261-1 and 261-2, £ separation calculation 262 262-1 and 262-2, and surviving replica selection 咅 263-1 and 263 -2. It has.

On the other hand, when receiving and processing a SIMO signal using the MLD method, for example, the square Euclidean distance in the case of a 1 × 2 antenna is calculated from the equations (29) and (30). As described above, in the MLD method, the concept of the first stage and the second stage is not calculated, and the square Euclidean distance is calculated individually for each receiving antenna, and finally, the square Euclidean distance for each receiving antenna is added to obtain the final. The square Euclidean distance is calculated.

FIG. 9 shows a main configuration of maximum likelihood determination section 360 using the MLD method. The maximum likelihood determination unit 360 shown in FIG. 9 replaces the maximum likelihood determination unit 260 shown in FIG. 8 by replacing the transmission replica generation unit 261-2 with the transmission replica generation unit 261-1, and replacing the distance calculation unit 262-2 with the distance. Instead of the calculation unit 262-1, the surviving replica selection unit 263-2 is replaced with the surviving replica selection unit 263-1, and a symbol selection unit 361 is added. In other words, the maximum likelihood determination unit 260 shown in FIG. 8 is configured to have two first stages.

[0098] Symbol selection section 361 calculates the squared Euclidean distance for each receiving antenna. Specifically, the square Euclidean distance shown in Expression (31) is calculated. However, since reciprocal calculation is required in Equation (31), even if the square Euclidean distance shown in Equation (50) is calculated, good.

[Equation 50] e ' ²

² ₊ | A * | V… (5 0)

FIG. 10 shows a main configuration of maximum likelihood determination section 160 according to the present embodiment. The maximum likelihood determination unit shown in FIG. 10 adopts a configuration in which a symbol selection unit 361 and a switch 161 are added to FIG.

[0100] The switch 161 switches a signal to be output to the LLR calculation unit 170 in accordance with the mode selection information S indicating multistream or single stream. Specifically, when the mode selection information indicates that the mode selection information is multi-stream, the switch 161 outputs the square Euclidean distance calculated using Equation (17) to the LLR calculation unit 170, and the mode selection information is In the case of a single stream, the squared Uterid distance calculated by the symbol selection unit 361 is output to the LLR calculation unit 170.

It should be noted that surviving replica selection section 263-1 also outputs replica signal candidates to symbol selection section 3601.

[0102] That is, when the mode selection information indicates multi-stream, the square Euclidean distance is calculated in two stages of the first stage and the second stage, and when the mode selection information is single stream, reception is performed. The squared Euclidean distance is calculated separately for each antenna.

[0103] (Shared LLR calculator)

The QR—MLD method performs multi-stream reception and the MLD method performs single-stream reception, so the number of bits for calculating the log-likelihood ratio differs for each mode, but the equations used in the LLR calculation logic (18), (25 ) Can be realized with the same configuration, so it is necessary to share the LLR calculation unit 170 with the force S.

[0104] As described above, according to the present embodiment, the channel compensation coefficient calculation unit 140 multiplies the received signal by the channel compensation coefficient calculation unit 140 and the transmission replica signal in both the multi-stream case and the single-stream case. In the case of multi-stream, the channel estimation value is QR-decomposed to obtain the unitary matrix and upper triangular matrix, and the unitary matrix is converted to the channel compensation matrix. And upper triangular matrix is triangular row 歹 IJ, and in the case of a single stream, a diagonal matrix with the channel estimation value as the diagonal element is the channel compensation matrix, and the diagonal matrix is the product of the channel estimate and the complex conjugate of the channel estimation value. Is a triangular matrix.

[0105] Then, the signal point distance between the multiplication result obtained by orthogonalizing the received signal and the complex conjugate transpose matrix of the channel compensation matrix and the multiplication result of the triangular matrix and the transmitted replica signal is The maximum likelihood is determined by calculating for each stage corresponding to the number of transmission signals, and the likelihood of the transmission replica signal is calculated using the distance between signal points. Since the received signal z after orthogonalization, which is obtained by multiplying the received signal y by the complex conjugate, is in a state in which the phase rotation has been restored, the resource for calculating the distance between signal points using the QR-MLD method for the multi-stream is reduced. The distance between signal points can be calculated using the MDL method.

As a result, in a signal separation device used in a reception device including a plurality of reception antennas, a plurality of transmission antennas allow signals that are spatially multiplexed using the same frequency and time to be received by the plurality of reception antennas. It is possible to reduce the size of a receiver that can handle both modes by sharing the reception processing of the MIMO mode for reception and the SIMO mode for receiving signals transmitted from a single antenna by multiple antennas. it can.

[0107] In the above description, the case of switching between the MIMO mode and the SIMO mode has been described. However, reception processing is performed using the receiving apparatus 100 even for a single stream in the SISO (Single Input Single Output) mode. be able to.

Specifically, in the SISO mode, maximum likelihood determination section 160 calculates the square Euclidean distance using only equation (16), and outputs the calculation result to LLR calculation section 170. . In other words, the square Euclidean distance in the first stage according to the QR-MLD method is equal to the square Euclidean distance between the received signal received via a single receiving antenna and the transmitted replica, so the QR for the multistream in MIMO mode — By using only the square Euclidean distance in the first stage using circuit resources of the MLD method, the square Euclidean distance by the MLD method can be calculated for the SISO.

[0109] Since the SISO receives and processes a signal transmitted from a single antenna with a single antenna, power consumption can be reduced compared to MIMO and SIMO. Form of this implementation According to the situation, even when the MIMO, SIMO, and SISO modes are switched according to the usage situation, the circuit resources of the QR—MLD method for multistreams in the MIMO mode can be used for SIMO and SISO. In addition, the square Euclidean distance can be calculated by the MLD method, and it becomes possible to support multimode (Fig. 11).

[0110] One aspect of the signal separation device of the present invention includes mode identifying means for identifying whether a received signal is a multistream or a single stream, a channel compensation matrix and a triangular matrix based on a channel estimation value. Channel compensation coefficient calculating means for calculating the channel compensation matrix and the triangular matrix by switching the method of calculating the channel between the case where the received signal is a multi-stream and the case of a single stream, and the received signal Orthogonalizing means for multiplying the channel compensation matrix by the complex conjugate transpose matrix, a multiplication result of the received signal and the complex conjugate transpose matrix, and a multiplication result of the triangular matrix and the transmission replica signal. A configuration comprising: a maximum likelihood determination unit that calculates a signal point distance; and a likelihood calculation unit that calculates the likelihood of the transmission replica signal using the signal point distance. Take.

[0111] According to this configuration, the complex conjugate transposed matrix multiplied by the received signal and the triangular matrix multiplied by the transmission candidate signal can be switched between the multi-stream case and the single stream case. In the case of single stream, it becomes possible to calculate the distance between signal points using the same method as the method for calculating the distance between signal points for multi-stream. The SIMO mode reception processing can be shared to reduce the size of the signal separation device that can handle both modes.

[0112] According to one aspect of the signal separation device of the present invention, the channel compensation coefficient calculating means performs QR decomposition on the channel estimation value to obtain a unitary matrix and an upper triangular matrix when the received signal is a multi-stream. When the received matrix is the single stream, the diagonal matrix having the channel estimation value as a diagonal element is used as the channel compensation. A matrix is used, and a diagonal matrix having a diagonal element as a product of the channel estimation value and the complex conjugate of the channel estimation value is used as the triangular matrix.

[0113] According to this configuration, in the case of multi-stream, signal separation can be performed while reducing the amount of computation using the QR-MLD method, and in the case of single-stream, channel estimation is performed. Since the received signal z after orthogonalization, in which the complex conjugate of the value is multiplied by the received signal y, is in a state in which the phase rotation is returned, the procedure for calculating the signal point distance by QR-MLD method for multi-stream Using the same procedure as above, the distance between signal points can be calculated by the MDL method. Therefore, the reception processing in MIMO mode and SIMO mode is shared, and a signal separation device that can handle both modes is used. Miniaturization can be achieved.

[0114] In one aspect of the signal separation device of the present invention, the maximum likelihood determination means includes a plurality of processing means for calculating the distance between the signal points corresponding to the same number of stages as the number of multi-stream transmission signals. Selection means for calculating the distance between the signal points used by the likelihood calculation means based on the distance between the signal points calculated in each stage; and when the received signal is a multi-stream, The distance between the signal points calculated in the final stage is selected as the distance between the signal points used by the likelihood calculating means, and when the received signal is a single stream, the signal point calculated by the selecting means. And a switching means for selecting the distance.

[0115] According to this configuration, in both the multi-stream and single-stream cases, the same resource is used to calculate the distance between the signal points in the same number of stages as the number of multi-stream transmission signals. In the case of a stream, it becomes possible to select the distance between signal points in the final stage calculated by processing each stage in series and perform the maximum likelihood determination using the QR-MLD method. Based on the distance between multiple signal points calculated by processing in parallel for each stage, the maximum likelihood determination by the MLD method can be performed by selecting the distance between the signal points used for the likelihood.

One aspect of the signal separation device of the present invention employs a configuration further comprising receiving means for receiving the received signal with any one of a plurality of receiving antennas.

[0117] According to this configuration, it is possible to perform signal estimation using the MLD method for a single stream in the SISO mode while using circuit resources that perform signal separation for the multi-stream using the QR-MLD method.

[0118] The embodiments of the present invention have been described above.

Note that each functional block used in the description of each of the above embodiments is typically realized as an LSI that is an integrated circuit. These may be individually integrated into one chip, or part or all of them. It may be integrated into a single chip to include Here, it may be called IC, system LSI, super LSI, or ultra LSI, depending on the difference in power integration as LSI. Also, the method of circuit integration is not limited to LSI, and may be realized by a dedicated circuit or a general-purpose processor. Field programmable gate arrays (FPGAs) that can be programmed after LSI manufacturing and reconfigurable processors that can reconfigure the connection and settings of circuit cells inside the LSI may be used. Furthermore, if integrated circuit technology that replaces LSI emerges as a result of advances in semiconductor technology or other derived technologies, it is naturally also possible to integrate functional blocks using this technology. For example, the possibility of applying technology is possible.

Yes

[0120] The disclosure of the specification, drawings and abstract contained in Japanese Patent Application No. 2006-225935, filed on August 22, 2006, is hereby incorporated by reference.

Industrial applicability

[0121] The signal separation device and signal separation method of the present invention share reception processing in the MIMO mode and the SIMO mode in a communication system in which multistreams and single streams are mixed, and receive signals corresponding to the multimodes. For example, it is useful for a signal separation device and a signal separation method for performing reception processing in a communication system in which multistreams and single streams are mixed.

Claims

The scope of the claims

[1] the power of the received signal being either multistream or single stream, mode identification means,

Based on the channel estimation value! /, The channel compensation matrix and the triangular matrix calculation method are switched between the case where the received signal is multi-stream and the case of single stream, and the channel compensation matrix and the triangle are calculated. Channel compensation coefficient calculating means for calculating a matrix, orthogonalizing means for multiplying the received signal and the transposed matrix of the channel compensation matrix, a multiplication result of the received signal and the transposed matrix, and the triangular matrix Maximum likelihood determination means for calculating a distance between signal points of a multiplication result with a transmission replica signal;

A likelihood calculating means for calculating the likelihood of the transmitted replica signal using the distance between the signal points;

A signal separation device comprising:

[2] The channel compensation coefficient calculation means includes:

When the received signal is multi-stream, the channel estimation value is subjected to QR decomposition to obtain a unitary matrix and an upper triangular matrix, the unitary matrix is used as a front channel compensation matrix, and the above triangular matrix is used as the triangular matrix. age,

When the received signal is a single stream, the channel estimation value is a diagonal element, the diagonal matrix is the channel compensation matrix, and the product of the channel estimation value and the complex conjugate of the channel estimation value is a diagonal element. Let the diagonal matrix be the triangular matrix

The signal separation device according to claim 1.

[3] The maximum likelihood determination means includes:

A plurality of processing means for calculating the distance between the signal points corresponding to the same number of stages as the number of multi-stream transmission signals;

Selection means for calculating the distance between the signal points used by the likelihood calculating means based on the distance between the signal points calculated in each of the stages;

When the received signal is a multi-stream, the signal point distance calculated in the final stage of the stages is selected as the signal point distance used by the likelihood calculating means, and the received signal is a single stream. If calculated by the selection means Switching means for selecting the distance between the signal points

The signal separation device according to claim 1.

[4] Receiving means for receiving the received signal with any one of a plurality of receiving antennas,

Further comprising

The signal separation device according to claim 1.

[5] Based on the force that the received signal is multistream or single stream, and the channel estimate to be identified! /, The method for calculating the channel compensation matrix and the triangular matrix is described below. And switching between the case of single stream and the case of single stream, calculating the channel compensation matrix and the triangular matrix,

Multiplying the received signal by the transposed matrix of the channel compensation matrix, a multiplication result of the received signal and the transposed matrix, and a signal point distance between the multiplication result of the triangular matrix and the transmission replica signal. Calculating step;

Calculating the likelihood of the transmitted replica signal using the distance between the signal points; and a signal separation method.