JP5527604B2 - Rounding canceller and rounding cancellation method - Google Patents

Description

  The present invention relates to a sneak canceller and a sneak canceling method suitable for removing or compensating a sneak interference signal from a transmitter to a receiver in a simultaneous transmission / reception system using a relay station.

  Research on a transmission system using a wireless repeater (hereinafter referred to as a relay station) has been actively conducted. In particular, more reliable signal transmission is achieved by synthesizing and decoding the signal forwarded from the relay node by actively utilizing the broadcast nature of the wireless communication path and the signal directly reaching the destination node from the transmission node. Wireless network coding, which improves frequency utilization efficiency by combining and transferring signals from each transmission node in bidirectional communication using relay stations and bidirectional communication using relay stations, is a conventional wireless transmission method. Compared to this, it has attracted much attention because it has the potential to greatly improve its characteristics. Many various methods related to cooperative communication and network coding have been proposed. In any case, orthogonality between a signal received by a relay station and a signal transmitted is assumed. This means that twice as many radio resources are used as compared with direct transmission.

  Even in such a case, it has been shown that the frequency utilization efficiency can be improved by cooperative communication or network coding, but the same radio resource is used in the communication path from the master station (or base station) to the relay station and from the relay station to the terminal. Can be used, a significant improvement in system characteristics is expected.

  A relay station that performs such simultaneous transmission and reception has been studied in a digital terrestrial television broadcasting system using an OFDM (Orthogonal Frequency Division Multiplexing) system based on SFN (Single Frequency Network). When relay stations simultaneously perform transmission and reception using the same frequency, a serious problem is that the relay transmission signal from the transmission antenna wraps around the reception antenna, causing oscillation and deterioration of characteristics. For this reason, a sneak canceller for removing or compensating a sneak interference signal from a transmitter to a receiver in a simultaneous transmission / reception system using a relay station has been studied.

Robert W. Heath, Jr. and Georgios B. Giannakis, "Exploiting Input Cyclostationarity for Blind Channel Identification in OFDM Systems," IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 47, NO. 3, pp-848-856, March 1999 Sonren, Akira Sano, “Identification of OFDM SFN relaying characteristics with wraparound interference”, IEICE Transactions A, Vol. J88-A, no. 9, pp. 1045-1053, 2005

  FIG. 3 is a schematic block diagram illustrating a configuration of a relay station including a digital signal processing unit having a function of canceling wraparound. FIG. 3 schematically shows data transmission in the propagation path. In this figure, a solid line and a broken line represent data transmission in the relay station and in the propagation path, respectively. The same applies to the block diagrams shown below.

  As shown in the figure, the relay station 9 includes a receiver 901, a digital signal processing unit 910, an amplifier 902, and a transmitter 903. The relay station 9 receives an OFDM signal x (n) transmitted from a base station (not shown), and includes an interference signal that wraps around via a feedback path 950 represented by the transfer function C (z) included in the received signal. The signal with the interference signal suppressed is amplified by the amplifier 902 and transmitted from the transmitter 903 to relay the OFDM signal x (n).

The receiver 901 receives the interference signal that the transmission signal transmitted from the transmitter 903 wraps around via the feedback path 950 together with the OFDM signal x (n) transmitted from the base station. That is, the signal received by the receiver 901 is a signal obtained by combining the OFDM signal x (n) and the interference signal.
The digital signal processing unit 910 receives a signal received by the receiver 901, suppresses an interference signal component included in the signal, and outputs the signal to the amplifier 902. The amplifier 902 amplifies the signal input from the digital signal processing unit 910 and outputs the amplified signal to the transmitter 903. The transmitter 903 transmits the signal input from the amplifier 902.

  The digital signal processing unit 910 includes an analog / digital converter (ADC) 911, a subtractor 912, a control unit 913, an adaptive filter 914, and a digital / analog converter (DAC) 915.

The analog / digital converter 911 converts the reception signal input from the receiver 901 into a digital signal. The subtractor 912 subtracts the signal output from the adaptive filter 914 from the signal converted into a digital signal by the analog / digital converter 911. The error signal s (n), which is the subtraction result output from the subtractor 912, is input to the adaptive filter 914 and the control unit 913 that controls the adaptive filter 914, and is also input to the digital / analog converter 915.
The digital / analog converter 915 converts the error signal s (n) input from the subtractor 912 into an analog signal, and outputs the converted signal to the transmitter 903 via the amplifier 902 of the transfer function G (z). . The transmitter 903 transmits the input analog signal as a transmission signal u (n).

  The adaptive filter 914 includes, for example, an FIR (Finite Impulse Response) filter, and each tap coefficient (that is, the weight of each tap) is controlled by the control unit 913.

  Here, the OFDM signal x (n) is a periodic stationary process with an average of 0 (Non-Patent Document 1), and C (z) and G (z) represent the wraparound path and the transfer function of the amplifier, respectively. Here, n is a time index. The transfer function G (z) of the amplifier is known, but the transfer function C (z) of the sneak path is unknown and may be time-varying. Therefore, the controller 913 uses MMSE to estimate the sneak path. Calculate the transfer function. Here, an adaptive filter (W (z)) 914 using an LMS (Least Mean Square) algorithm is used.

  From FIG. 7, the error signal s (n) is

It can be expressed. Here, using the transfer function C (z) of the wraparound path and the transfer function G (z) of the amplifier provided in the relay station, the OFDM signal x (n) transmitted from the base station apparatus and the error An error function Q (z) representing the relationship with the signal s (n) can be expressed by the following equation (2).

  Using equation (2), equation (1) can be written as the following equation (3).

Here, W (z) represents a transfer function of the adaptive filter 914 in a steady state, z ⁻¹ represents a delay operator, and the error function Q (z) is stable. That is, it is assumed that all zeros of 1- (C (z) G (z) -W (z)) exist within the unit circle.
A commonly used LMS algorithm for adaptive filter 914 is:

It can be written as However,

_Wi (n) represents the i-th tap coefficient of the adaptive filter 914, μ represents a positive step size, and the superscript “*” represents a complex conjugate. N represents the number of taps that the adaptive filter 914 has. Further, a bold character such as “w” on the left side of Expression (4) represents a vector. In the text, a character representing a vector is written in the form of “character (vector)”. In equation (4), the error signal s (n) is used as the input signal of the adaptive filter 914. From the averaging method, the stopping point of the equation (4) is given by the following equation (6).

  Here, E [•] described in Equation (6) represents an expected value. However, since s (n) = 0, the condition for i = 0 in Expression (6) is excluded. That is, in the expressions (5-1) and (5-3),

And Further, since the OFDM signal x (n) is a periodic stationary process and its statistics are time-varying, the average of the equation (6) is set average and time average. Thus, x (n) can be treated as a stationary process having a spectral density P (e ^jω ) equal to the cyclic spectrum having a cycle frequency of 0 (Non-patent Document 1). From Expression (3), Expression (6) can be expressed by the following Expression (8).

Here, when z = e− ^jω , Expression (8) can be expressed as the following Expression (9).

  However,

It is. Here, the symbol with “˜” represents an estimated value. That is, ^~ C (z), ^~ G (z) (where the superscript ~ in the text represents a symbol attached just above a character) is an estimated value of the transfer function of the wraparound path Represents the estimated value of the transfer function of the amplifier.
For equation (9) to hold for all i ≧ 1, the integrand below z ⁱ⁻¹ in equation (9) must be expanded with a positive power of z, ie, 9) can be written as the following equation (11).

Here, the operation [X (z)] ₊ included in the equation (11) is

Against

And the causal part, that is, the extraction of the negative term of z and the constant term.
The spectral factorization of the spectral density P (z) can be written as the following equation (14) using the spectral factor R (z).

Here, the spectrum factor R (z) is a minimum phase system, and its constant term is 1. Since 1 / Q (z) is a stable polynomial from Equation (2), 1 / (R (z ⁻¹ ) Q (z ⁻¹ )) can be expanded with a non-negative power of z. Therefore, Formula (11) can be written as the following Formula (15) by multiplying Formula (2) on both sides of Formula (11).

In order to obtain the spectrum factor R (z) included in Equation (15), the OFDM signal x (n) is an OFDM signal having a data length M, a CP (Cyclic Prefix) length L _CP , and a block length T = M + L _CP. The spectral density P (z) is obtained.
Here, assuming that the _kth sample x _k (n) = x (nT + k) within one block length, the OFDM signal xk (n) can be written as the following equation (16).

Since the input symbols s _i (n) have an average of 0 and a variance σ ² , they are not correlated with each other.

It is. Here, δ (r) means δ (r) = 0, 1 for r ≠ 0 and r = 0, respectively (Non-patent Document 1). The cyclic correlation function p (0; τ) at the cycle frequency 0 is an average from 0 to T−1 for each k in the equation (17),

So

And also

Thus, the spectral factor is

It becomes. Here, α is a positive real number. However, γ in the equation (14) is given by the following equation (22).

  here,

Since T / L _CP > 2 in general, 0 <α <1. Thus, the solution of equation (15) can be easily obtained. Moreover, since ^~ R (z) ^~ Q (z) is causal, its value is a constant β ^* from the expression (15). Therefore, the stopping point W ₀ (z) is expressed by Meet.

Further, from the expressions (5-3), (7), and (21), when Z ⁻¹ is included in G (z), that is,

Then β = 1,

It becomes.

That is, the bias term αz- ^M occurs at the stop point. A method for removing the bias term of Equation (26) will be described.

  In the equation (4), a signal represented by the following equation (27) is used instead of the error signal s (n).

  That means

However, this is a whitening operation when the error signal s (n) is equal to the OFDM signal x (n). Thus, Expression (11) can be expressed as the following Expression (29).

  At this time, the stopping point of the equation (29) is given by the following equation (30) using the equation (14).

^The causality portion of z ~ Q (z) is 0, that ^is, because it must be z ~ Q (z) = 1 ,

It is. Accordingly, the stop point of the adaptive filter 104 is expressed by the following equation (32).

  Comparing equation (32) with equation (26) shows that the bias term has been eliminated.

FIG. 4 is a schematic block diagram illustrating a configuration of the relay station 92 including a digital signal processing unit having a wraparound cancellation function considering multipath and bias. FIG. 4 schematically shows data transmission in the propagation path, as in FIG.
The configuration of the relay station 92 illustrated in FIG. 4 is different from the configuration of the relay station 90 illustrated in FIG. 3 in that a whitening filter 926 is provided between the subtractor 912 and the control unit 923. The difference is that a delay device 927 is provided between the digital filter 912 and the digital-analog converter 915 and that the output signal of the delay device 927 is input to the adaptive filter 914.
In the configuration shown in the figure, the same reference numerals are used for the components corresponding to the configuration shown in FIG. The delay unit 927 gives a predetermined time delay between the input and output signals, and outputs the signal input from the subtracter 912 to the digital / analog converter 915 and the adaptive filter 914. The whitening filter 926 is a filter whose transfer function is the inverse of the spectral factor R (z), and outputs a signal s ′ (n). The control unit 923 controls the tap coefficient of the adaptive filter 914 based on the output of the whitening filter 926.
Here, the mean square error of the adaptive filter 914 can be expressed by the following equation (33).

When the tap coefficient vector of the adaptive filter 104 corresponding to the stop point W ₀ (z) is w ₀ (vector), the equation (33) becomes the following equation (34).

However, MyuY (vector) is obtained by approximating the error covariance matrix of the tap coefficients in the vicinity of stationary points, by solving the Lyapunov equation, obtaining a Y ^{(vector) = γ 2/2 · I} ( Vector). From this, the mean square error at the stopping point can be expressed by the following equation (35).

In the discussion about the wraparound canceller shown above, it is a major premise that the initial state is stable. The error signal s (n) and the error function Q (z) are expressed by the following equation (36).

  Here, assuming that the initial values of the coefficients are all 0 as the initial state of the adaptive filter 914, the corresponding error function Q (z) is expressed by the following equation (37).

At this time, in order for the error function Q (z) to be stable, | C (e ^jω ) G (e ^jω ) | <1 must be satisfied for all ω, and this condition is not satisfied. Will diverge the error function Q (z).

By using the wraparound canceller described above, it is theoretically possible to realize a relay station using the same frequency.
However, in practical use, since the analog-to-digital converter in reception has a finite number of bits, it is difficult to remove the feedback path signal as the theoretical value. For example, when the reception power by the feedback path is large, the reception level of the desired signal that should be amplified is relatively small with respect to the reception signal corresponding to the feedback path, and is greatly deteriorated due to the quantization error. In such a case, there is a problem that a signal to be relayed is deteriorated due to a quantization error and communication quality is deteriorated.

That is, there is a problem that the communication quality deteriorates due to the quantization error of the relay signal generated according to the received power of the feedback path signal that interferes with the relay signal.
Therefore, in order to reduce the quantization error that occurs in the analog-to-digital converter, it is necessary to reduce the level of the received signal corresponding to this feedback path as much as possible before converting the received analog signal into a digital signal.

  The present invention has been made in view of the above circumstances, and in the signal output from the receiver provided in the relay station, the received power of the interference signal that wraps around from the transmitter of the relay station to the receiver is reduced. It is an object of the present invention to provide a sneak canceller and a sneak canceling method that can be performed.

  In order to solve the above problems, the present invention is provided in a radio relay station that receives a radio signal transmitted from a radio communication station, transmits the radio signal at the same frequency as the received radio signal, and relays the radio signal. A sneak canceller that cancels an interference signal sneaking from a transmitter to a receiver, using a filter coefficient calculated during a period in which the radio signal is not received, by an antenna element provided in the receiver An equalizer that performs an equalization process to reduce the received power of an interference signal that wraps around from a transmitter with respect to the received analog received signal, and the received signal that has been equalized by the equalizer is quantized. The digital received signal in which the interference signal is suppressed by suppressing the interference signal included in the received signal based on a transfer function of a path that goes from the transmitter to the receiver. The is a wraparound canceller characterized in that it comprises a digital signal processing unit for outputting to the transmitter is converted into an analog transmission signal.

  Further, the present invention is the invention described above, wherein the radio relay station is provided with a plurality of the receivers, and is provided with a smaller number of the equalizers than the receivers. Of the received signals received by the receiver, received signals of different combinations are input, the convolution operation using the filter coefficient is performed on the input received signals, and the received power of the interference signal that wraps around from the transmitter is reduced. One of the output signals is output to the digital signal processing unit.

  The present invention further includes a known signal generation unit that outputs a predetermined known signal to the transmitter in the invention described above, wherein the equalizer is a period in which the radio signal is not received. The filter coefficient for reducing the reception power of the interference signal is calculated during a period in which the known signal generation unit is transmitting a known signal.

  Further, the present invention provides the digital signal processing unit according to the above-described invention, wherein the digital signal processing unit converts the received signal equalized by the equalizer into a digital signal, and the digital signal. A weight multiplier that multiplies a first weighting factor corresponding to an antenna element provided in the receiver; a subtractor that subtracts a feedback signal from an output of the weight multiplier and outputs a subtraction result as an error signal; A whitening filter that whitens the error signal; a delay unit that delays the error signal and outputs the error signal to the transmitter as a transmission signal; and filters the transmission signal based on a variable transfer function as the feedback signal. Based on the adaptive filter output to the subtractor and the output of the whitening filter, the transfer function of the adaptive filter is controlled so that the interference signal is minimized. A control unit for the output of the weight multiplication unit, characterized in that it comprises a weight variable control unit for calculating the first weighting factor error is minimized with the known signal.

  In the invention described above, the digital signal processing unit may further include a virtual path weight multiplication unit that multiplies a signal obtained by branching the transmission signal output from the delay unit as a virtual path by a second weighting factor. And an adder that adds the output of the weight multiplication unit and the output of the virtual path weight multiplication unit and outputs the result as an array output signal to the subtractor, and the subtracter outputs the adder The feedback signal is subtracted from the array output signal to be output, and the subtraction result is output as the error signal. The weight variable control unit is further configured to reduce the error between the array output signal and the known signal. A weighting factor of 2 is calculated.

  The present invention also provides a radio relay station that receives a radio signal transmitted from a radio communication station and transmits the radio signal at the same frequency as the received radio signal and relays the radio signal. A sneak cancel method for canceling an interference signal that sneaks into a machine, using a filter coefficient calculated in a period in which the radio signal is not received, and an analog signal received by an antenna element provided in the receiver An equalization step for reducing received power of an interference signal that wraps around from a transmitter with respect to the received signal, and the received signal that has been equalized in the equalization step is quantized and quantized. The interference signal contained in the digital signal is suppressed based on a transfer function of a path from the transmitter to the receiver, and the interference signal is suppressed. Signal is a coupling loop canceling method characterized in that it comprises a digital signal processing step of outputting to the transmitter is converted into an analog transmission signal.

According to the present invention, it is possible to reduce the reception power of an interference signal that wraps around from the transmitter of the relay station to the receiver in the signal received by the receiver provided in the relay station.
Thereby, when the reception power of the interference signal that wraps around from the transmitter to the receiver is large, it is possible to reduce the deterioration of the communication quality due to the quantization error that occurs in the signal to be relayed.

It is a schematic block diagram which shows the structure of the relay station 1 which has the wraparound canceller 3 in this embodiment. It is the schematic which shows the structure of the equalizer 300 in the embodiment. It is a schematic block diagram which shows the structure of the relay station provided with the digital signal processing part which has a function of wraparound cancellation. It is a schematic block diagram which shows the structure of the relay station 92 provided with the digital signal processing part which has a roundabout cancellation function in consideration of the multipath and the bias.

In the present invention, before converting a received signal into a digital signal, the reception power of a transmission signal received via a feedback path included in the received signal is reduced using an analog equalizer. As a result, in the signal output from the equalizer, the difference in received power between the signal to be relayed and the transmitted signal can be reduced, and the received power of the signal to be relayed can be relatively increased. It is possible to reduce the quantization error of the relayed signal that occurs when the conversion is performed. As a result, it is possible to reduce deterioration in transmission quality when a signal transmitted from the base station is received and transmitted at the same frequency as the signal.
Embodiments of the present invention will be specifically described below with reference to the drawings.

  FIG. 1 is a schematic block diagram showing a configuration of a relay station 1 having a wraparound canceller 3 as an embodiment of the present invention. FIG. 1 schematically shows data transmission in the propagation path.

As shown in FIG. 1, the relay station 1 includes M (M ≧ 1) receivers 330-1 to 330 -M, a sneak canceller 3, an amplifier 306, and a transmitter 309. Each of the receivers 330-1 to 330-M has the same configuration. Hereinafter, when any one of the receivers 330-1 to 330 -M is indicated or when all of the receivers 330-1 to 330 -M are indicated, the receiver 330 is referred to.
Further, in the same figure, a route through which a signal transmitted from the transmitter 309 wraps around each receiver 330 is represented as M feedback paths 302-1 to 302-M. In addition, a path through which a signal transmitted from a base station (not shown) propagates to each receiver 330 is represented as M forward paths 301-1 to 301-M.

  The relay station 1 is a device that performs radio relay by receiving a radio signal transmitted from a base station (not shown) and transmitting the received radio signal at the same frequency. That is, the relay station 1 shown in FIG. 1 receives a radio signal transmitted from a base station (not shown) using a receiving array antenna including antenna elements provided in each of the plurality of receivers 330. The radio signal is amplified by the amplifier 306 and transmitted by the transmitter 309 using a radio signal having the same frequency as the received signal.

Each of the receivers 330 receives an OFDM signal x _i (n) (i = 1, 2,...) Transmitted from a base station (not shown) via forward paths 301-1 to 301-M by antenna elements included therein. M) and the interference signal v _i (n) (i = 1, 2,..., M) transmitted by the transmitter 309 that wraps around the feedback paths 302-1 to 302-M.
That is, each of the receivers 330 has an OFDM signal x (n) transmitted from the base station and an interference signal v _i (n) in which the signal u (n) transmitted from the transmitter 309 provided in the own station wraps around. The synthesized signal r _i ′ (n) = x _i (n) + v _i (n) is received.

Here, the received signal x _i (n) is transmitted through the forward paths 301-1 to 101-M, which are propagation paths between the base station and the receiver 330-i (i = 1, 2,..., M). Using the transfer function H _i ′ (z) (i = 1, 2,..., M), x _i (n) = H _i ′ (z) · x (n). The interference signal v _i (n) is a sneak path (feedback path) 302-1 that is a propagation path between the transmitter 309 and the receiver 330-i (i = 1, 2,..., M). Using the transfer function C _i ′ (z) (i = 1, 2,..., M) in 302-M, v _i (n) = C _i ′ (z) · u (n). Note that the feedback paths 302-1 to 302-M may include not only a route outside the relay station 1, but also a route transmitted through the relay station 1 and received by the receiver 330.
Further, the interference signal v _i (n) received via the feedback paths 302-1 to 302-M is closer to the received signal x _i (n) because the transmitter 309 of the transmission source is closer to the base station. And received with very large power (| x _i (n) | << | v _i (n) |).

The wraparound canceller 3 cancels the interference signal v _i (n) that wraps around from the transmitter 309 to each of the receivers 330. The amplifier 306 amplifies the signal in which the interference signal v _i (n) is suppressed from the signal received by each receiver 330 by the wraparound canceller 3 and outputs the amplified signal to the transmitter 309. The transmitter 309 transmits the transmission signal input from the amplifier 306 through the transmission antenna element provided.

The wraparound canceller 3 includes m (m = M−α, α is a natural number of 1 or more) equalizers 300-1 to 300 -m and a digital signal processing unit 30.
The equalizers 300-1 to 300-m have the same configuration. Hereinafter, when any one of the equalizers 300-1 to 300-m is indicated, or when all the equalizers are indicated, the equalizer 300 is referred to. Here, m is the number of received signals input to the digital signal processing unit 30.

Α is the number of surplus receiving antennas necessary to suppress the interference signal v _i (n) received by the equalizer 300 via the feedback paths 302-1 to 302-M. Calculated based on the results.
The digital signal processing unit 30 suppresses an interference signal by performing signal processing by regarding the signal input from the equalizer 300 as an output signal of an equivalent receiving array antenna having m elements. .

Different combinations of received signals r _i ′ (n) (i = 1, 2,..., M) output from the respective receivers 330-1 to 330 -M are input to the equalizers 300. For example, when M = m + 1, the reception signals r _i ′ (n) output from the receivers 330-1 to 330-m are input to the equalizer 300-1, and the receivers 330-1 to 330- Reception signals r _i ′ (n) output from (m−1) and 330- (m + 1) are input to the equalizer 300-2, and receivers 330-1 to 330- (m−2) and 300− are input. m~330- (m + 1) received signal _{r i} '(n) is the received signal _{r i} different combinations in each equalizer 300 and so on, is input to the equalizer 300-3', each of which outputs (n ).
FIG. 2 is a schematic diagram showing the configuration of the equalizer 300 in the present embodiment.
The equalizer 300 calculates the input received signal r _i ′ (n) by the following method and performs a convolution operation using the set equalization weights h _{k, j} .

In the equalizer 300, the input received signal r ₁ ′ (n) is delayed by Δn ₁ to Δn N ′ through delay lines 351-1-1 to 351-1 _{-N ′} . Received signal _{r 1} to which the delay is given _{'(n + Δn 1),} ..., r 1' (n + Δn 1 + ... + Δn N'-1), r 1 '(n + Δn 1 + ... + Δn N') includes a multiplier The equalization weights h _{1, j} (j = 1, 2,..., N ′) are multiplied by 352-1-0 to 352-1-N ′.
The adders 353-1-1 to 353-1 -N ′ add the multiplication results of the multipliers 352-1-0 to 352-1 -N ′, thereby receiving the received signal r ₁ ′ (n). A convolution operation using equalization weights h _1,0 , h _1,1 ,..., H _{1, N ′} is performed.

Similarly, a convolution operation is performed on each received signal r _i ′ (n) input to the equalizer 300.
For example, the delay of Δn ₁ to Δn _{N ′} is given to the received signal r _M ′ (n) by the delay lines 351 -M- _{1 to} 351 -MN _′ . The received signals r _M ′ (n + Δn ₁ ),..., R _M ′ (n + Δn ₁ +... + Δn _N′−1 ), r _M ′ (n + Δn ₁ +... + Δn _{N ′} ) given the delay are multiplied by the multiplier. The equalization weights h _{M, j} (j = 1, 2,..., N ′) are multiplied by 352-M-0 to 352-MN ′.
Then, the adders 353-M-1 to 353-MN ′ add the multiplication results of the multipliers 352-M-0 to 352-MN ′, thereby receiving the received signal r _M ′ (n). A convolution operation using equalization weights h _{M, 0} , h _{M, 1} ,..., H _{M, N ′} is performed.

The adder 354 is a convolution for each received signal r _i ′ (n) calculated by the adders 353-1 to 353-1 -N ′,..., 353 -M- 1 to 353 -MN ′. The calculation results are added, and the addition result r _i (n) (i = 1,..., M) is output to the digital signal processing unit 30.

Here, the equalization weight h _{k, j} represents the amount of phase rotation or attenuation of the received signal from the kth antenna multiplied by the received signal that has passed through the jth delay line 351-k-j. The signal r (n) output from the equalizer 300 is expressed by the following equation (38).

Here, the digital signal processing unit 30 is provided by using an equalization weight h _{k, j} that reduces the interference signal v _i (n) that has passed through the feedback paths 302-1 to 302-M. The received power corresponding to the feedback path occupying the dynamic range of the analog / digital converter is reduced, and the proportion of the received power x _i (n) in the dynamic range is increased.

As a method for calculating the equalization weight h _{k, j, when} the signal x (n) transmitted from the base station is not transmitted or when the signal x (n) is not received, the digital signal processing unit 30, the known signal S ₀ is output to the transmitter 309 via the amplifier 306 to be transmitted, and is input to the equalizers 300-1 to 330 -m via the receivers 330-1 to 330 -M. .

Then, the equalizer 300 calculates an equalization weight h _{k, j} that minimizes the received signal r (n) input to the digital signal processing unit using a predetermined algorithm. As an algorithm used at this time, for example, the sum of absolute values of coefficients or square values of coefficients is larger than 0 by the MMSE (Minimum Mean-square-error) method or Lagrange's undetermined multiplier method. A method is used in which a constraint condition is given so as to obtain a value and a coefficient for obtaining an output of 0 is obtained.

Alternatively, a coefficient that maximizes the transmitted known signal S ₀ is calculated by a least square error method, a maximum SNR method, or a constrained output power minimization method, and a coefficient having a condition orthogonal to the obtained coefficient is used. You can also. It is also possible to use a constant envelope signal such as QPSK for the known signal S ₀ , obtain a coefficient by CMA (Constant Modulus Algorithm), and select a coefficient orthogonal to this coefficient as the coefficient h _{k, j} .
Or, the coefficient h _k, may be stored a plurality of combinations as candidates as _j, the coefficient h _k during the transmission of the known signal S _0, changing the _j, be selected a combination received power is the lowest Good.

Returning to FIG. 1, the digital signal processing unit 30 will be described.
The digital signal processing unit 30 receives a signal in which the reception power of the interference signal is suppressed from each equalizer 300, converts the input signal into a digital signal using an analog / digital converter, and performs digital signal processing. To cancel the interference signal and output the analog signal converted again using the digital / analog converter to the amplifier 306.

  The digital signal processing unit 30 includes m analog / digital converters 320-1 to 320-m, m weight multipliers 31-1 to 31-m, d virtual paths 310-1 to 310-d, d weight multipliers 32-1 to 32-d, adder 34, weight variable control unit 33, subtractor 303, adaptive filter 304, control unit 305, whitening filter 307, delay units 308 and 312, digital / analog A converter 321 and a known signal generator 322 are included. In the present embodiment, m is an integer of 2 or more, and d is an integer of 1 or more.

  However, unlike this, the wraparound canceller 3 may include M receivers 330-1 to 330-M. The amplifier 306 may be partly or wholly included in the transmitter 309. That is, the transmitter 309 may be configured to include the amplifier 306, and the signal may be input from the DAC 321 of the digital signal processing unit 30.

Here, the transfer functions of the forward paths 301-1 to 301 -M are assumed to be H ₁ ′ (z) to H _M ′ (z). Also, transfer functions corresponding to the OFDM signals x (n) included in the signals r ₁ (n) to r _m (n) output from the equalizers 300-1 to 300 -m are expressed as H ₁ (z) to H _{Let m} (z).
Furthermore, the transfer functions of the feedback paths 302-1 to 302-M are defined as C ₁ '(z) to C _M ' (z). The transfer function corresponding to the feedback paths 302-1 to 302-M included in the signals r ₁ (n) to r _m (n) output from the equalizers 300-1 to 300-m is expressed as C ₁ (z ) To C _m (z). The equalizers 300-1 to 300-m operate so as to reduce the reception power of the interference signal v _i (n) that wraps around, but the remaining components are changed to C ₁ (z) to C _m (z). Correspond.

One ADC (analogue-digital converter) 320-1 to 320-m is provided corresponding to each of the equalizers 300-1 to 300-m, and the corresponding equalization is performed. The signal r _i (n) input from the multiplier 300 is quantized (digitized) and output to the weight multipliers 31-1 to 31-m.
The m weight multipliers 31-1 to 31-m receive signals r ₁ (n) to r _m (from the equalizers 300-1 to 300-m via the ADCs 320-1 to 320-m, respectively. relative to n), multiplies the m weighting factors corresponding to each of the m elements in the equivalent receiving array antenna w _a ¹ a to w _a ^m respectively.

The adder 34 outputs the outputs of the plurality of weight multipliers 31-1 to 31-m and outputs from d weight multipliers 32-1 to 32-32d described later (that is, the output of the virtual path weight multiplier). ) Are added and output as an array output signal y (n).
The subtractor 303 subtracts the feedback signal output from the adaptive filter 304 from the array output signal y (n) and outputs the subtraction result as an error signal s (n). The whitening filter 307 whitens the error signal s (n). The delay device 308 gives a delay to the error signal s (n) and outputs it to the transmitter 309.
The adaptive filter 304 filters the output signal from the delay unit 308 to the transmitter based on the variable transfer function W (z), and outputs the filtered signal to the subtractor 303 as a feedback signal. Based on the output s ′ (n) of the whitening filter 307, the control unit 305 adjusts the adaptive filter so that the interference signal that wraps around the M receivers 330 via the feedback paths 302-1 to 302-M is minimized. The transfer function W (z) 304 is controlled.

A DAC (Digital-Analogue Converter) 321 converts a digital signal input from the delay unit 308 or the known signal generator 322 into an analog signal and outputs the analog signal to the amplifier 306. The signal output from the delay unit 308 is input to the DAC 321 and branched to d virtual paths 310-1 to 310-d via the delay unit 312.
The amplifier 306 amplifies the signal output from the DAC 321 and outputs the amplified signal to the transmitter 309.

The delay unit 312 gives a delay to the signal output from the delay unit 308, and outputs the delayed signal to the virtual paths 310-1 to 310-d. d number signals branched in the virtual path 310-1 to 310-d of, in weight multipliers 32-1 to 32-d, d pieces of weighting coefficients ^{_{^ w a m + 1 ~ ^}} w a m + d , respectively is multiplied by Is output to the adder 34 (where ^ in the text represents a symbol attached immediately above the letter following the letter ^).

The weight variable control device 33 receives the array output signal y (n), the array output signal y (n), and a predetermined pilot signal included in the known input signal x (n) (that is, the OFDM signal x (n)). ) m pieces of weighting coefficients corresponding to each of the m elements in the equivalent receiving array antenna so as to minimize an error between ^w _{a 1} ^~w _{a ^m,} and, d pieces of weighting coefficients ^ w a _{m +} ¹ ~ ^ W _a ^{m + d} is calculated, and the output results of the weight multipliers 31-1 to 31-m and 32-1 to 32-d are controlled.
The known signal generator 322 outputs a known signal S ₀ used for calculating the equalization weight of the equalizer 330 to the DAC 321. The timing at which the known signal generator 322 outputs the known signal is when the OFDM signal transmitted from the base station is not received.

Here, a method for controlling C (z) out of G (z) and C (z) constituting the transfer function of the wraparound path will be considered. Specifically, an adaptive array antenna is introduced on the receiving side of the relay station 1, and the wraparound wave is suppressed by controlling its weight well, and the transfer function C (z) of the wraparound communication path including the adaptive array is reduced. Think about what to do.
The adaptive array for OFDM signals can be broadly divided into two types: a pre-FFT (Fast Fourier Transform) type array that performs synthesis in the time domain and a post-FFT type array that performs synthesis in the frequency domain. Since a delay of 1 OFDM symbol or more occurs in the relay station 1 and the subsequent wraparound canceller is a filter operating in the time domain, a pre-FFT type array is adopted here.

  In the time domain array, since all delayed waves are generally handled as interference signals, the synthesis method is to capture only the maximum power path, but the relay station is installed within the line-of-sight from the parent station (base station). Therefore, it is considered that the deterioration of characteristics due to the loss of the delayed wave can be reduced. It is considered desirable to capture only the path with the maximum power because the effective delay spread between the master station and the relay station 1 can be reduced and the delay time required in the wraparound canceller can be shortened.

  On the other hand, although it is conceivable to introduce an array to the transmission antenna provided in the transmitter 309 of the relay station 1, it is difficult to control the weight in the transmission beam forming, and the relay station 1 as the target is Since the purpose is to expand the coverage, the transmission antenna is assumed to be one element because it is desired to transmit signals in all directions.

In the block diagram of the system of the present embodiment shown in FIG. 1, BS and RS represent the base station (parent station) and the relay station 1, respectively, and the delay unit 308 means a delay operator z- ^a . Since the transmission antenna of the base station has one element, the sneak path and the path from the base station to the relay station are both 1 × m SIMO (Single Input Multiple Output). Here, m is the number of antennas.

  Further, the input to the subtractor 303 is expressed by the following equation (39). The superscript “T” indicates transposition.

Further, the transfer function of the wraparound path is C _p (z), (p = 1, 2,..., M), and the transfer function from the base station to the relay station is H _p (z), (p = 1, 2, .., M), the transfer functions C _p (z) and H _p (z) can be expressed by the following equations (40-1) and (40-2), respectively.

  Here, K and L are the channel length of the wraparound path and the channel length of the path from the base station to the relay station. And using these impulse response estimates,

In other words, the input r (n) (vector) to the array antenna is expressed by the following equation (42).

Here, the number of 0 included in the formula (41-2) is d. Note that u (n) included in the equation (42) is a signal transmitted from the relay station, and n (n) (vector) is observation noise.
The noise term included in the equation (42) can be expressed by the following equation (43).

In addition, the noise term represented by the equation (43) is a vector composed of a numerical value indicating m noises and d zeros.
Further, from the equation (42), the array output signal y (n) is expressed by the following equation (44).

Here, w _a (vector) included in the equation (44) is a weight of the transparent receiving array antenna represented by the following equation (45).

  The superscript “H” included in the formula (44) means Hermitian conjugate. The subscript j represents the delay time of the maximum power path, and is represented by the following equation (46).

Here, argmax _i ‖h _i (vector) ‖ ₂ represents i that maximizes ‖h _i (vector) ‖ ₂ . Also, ‖ · は₂ represents the Frobenius norm. If the communication path from the base station to the relay station is not frequency selective, the channel length L from the base station to the relay station becomes L = 1, so j = 0.

Next, in the weight variable control device 33, a method for calculating ^m weighting factors w _a ^{1 to} w _a ^m or ^ w _a ^{m + 1} to ^ w _a ^{m + d} corresponding to each element of the equivalent receiving array antenna. explain. In the following, two types of methods, the MMSE (Minimum Mean-Square-Error) standard and the MVDR (Minimum Variance Distortionless Response) standard, will be described.

(Norm 1) MMSE norm

In the weight control based on the MMSE, a weight that minimizes the evaluation function J (w _a ) expressed by the following equation (47) is an optimal weight.

  here,

Then (I (vector) is a unit matrix), and the evaluation function J (w _a ) is expressed by the following equation (49).

Here, when both sides of the equation (49) are partially differentiated by w _a ^H , the following equation (50) is obtained.

  More

Is obtained, the MMSE standard optimum array weight expressed by the following equation (52) is obtained.

(Norm 2) MVDR norm

In the MMSE-based weight control method, when the number of antenna elements is small, the gain for a desired signal (maximum power path from the base station) may be small. This is because when the interference signal power of the wraparound is very large, it may be desirable in terms of the square error to suppress the interference signal even if the power of the desired signal is reduced.
However, since the relay station of this embodiment includes a sneak canceller after the adaptive array, the role of the adaptive array is to receive the desired signal more reliably than to completely suppress the sneak wave. It is considered important.

Therefore, an array weight w _a (vector) that minimizes the evaluation function J (w _a ) expressed by the equation (40) under the constraint condition expressed by the following equation (53) is derived.

  Such an adaptive array is called an MVDR beamformer. When the Lagrange's undetermined multiplier method is used, the evaluation function can be written as the following equation (54).

  Here, λ is a Lagrange multiplier. The evaluation function is

Therefore, the MVDR optimum weight ^to w _a ^opt (vector) can be written as the following equation (56).

  Here, the equation (53) indicating the constraint condition is

From the equation (56),

The following relational expression holds. Thus, the following formula (59) can be obtained.

By substituting the Lagrange multiplier formula ^(56), obtaining a ~ _w ^{a opt} (vector).

As described above, the variable weight control apparatus 33 receives the array output signal y (n) and the known input signal x (n) as inputs, and the equivalent receiving array antenna so as to minimize the error between them. when controlling the plurality of weighting factors w _a ¹ ~w _a ^m corresponding to each element, or using the least square error method to control a plurality of weighting factors w _a ¹ ~w _a ^m, the minimum variance distortion-free response Can be used. At that time, the weight variable controller 33, as shown in equation (51) or equation (56), a plurality of weighting factors _w ^a 1 _~w ^{a m,} the impulse response of the echo path from the transmitter to the receiver Control can be performed based on the estimated value (formula (40-1)) and the estimated value of the impulse response from the base station to the receiver (formula (40-2)).
Further, the weight variable control device 33, similar to the plurality of weighting factors _w ^a 1 _~w ^{a m,} to control the weighting coefficient ^{_{^ w a m + 1 ~ ^}} w a m + d.

As described above, in the present embodiment, the configuration is such that the whitening filter is monitored and then controlled. Further, the output of the amplifier is branched as a virtual path, multiplied by a weighting factor, and added to the signal before amplification. By combining such a configuration with the equalizer 300, the received power of the signal through the feedback path in the ADCs 320-1 to 320-m is reduced, and the array output signal y (n) and the known input signal x ( A plurality of weighting factors corresponding to each element of the equivalent receiving array antenna are controlled so that an error from n) is minimized.
As a result, even when a transmitter having a large amplification factor is used, it is possible to stably cancel an interference signal that travels from the transmitter to the receiver.

  In the present embodiment, the configuration in which the digital signal processing unit 30 includes the virtual paths 310-1 to 310-d has been described. However, the configuration is not limited thereto, and for example, the digital signal processing unit may not include the virtual paths 310-1 to 310-d.

  The digital signal processing unit described above may have a computer system inside. In this case, the process of the digital signal processing unit 30 is stored in a computer-readable recording medium in the form of a program, and the computer reads and executes the program, whereby the digital signal processing unit 30 Processing will be performed. Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

DESCRIPTION OF SYMBOLS 1,90,92 ... Relay station 3 ... Round-off canceller 30, 910, 920 ... Digital signal processing part 31-1, 31-m, 32-1, 32-d ... Weight multiplier 33 ... Weight variable control part 34 ... Addition 300, 300-1, 300-m ... Equalizer 301-1, 301-M ... Forward path 302-1, 302-M, 950 ... Feedback path 303 ... Subtractor 304, 914 ... Adaptive filter 305, 913 923: Control unit 306, 902 ... Amplifier 307, 926 ... Whitening filter 308, 312, 927 ... Delay device 309, 903 ... Transmitter 310-1, 310-d ... Virtual path 320-1, 320-m, 911 ... ADC
321,915 ... DAC
322: Known signal generators 330, 330-1, 330-M, 330-i, 901: Receivers 351-1-1, 351-1-N ′: Delay lines 351-M-1, 351-MN '... delay line 352-1-0, 352-1-1, 352-1-N' ... multiplier 352-M-0, 352-M-1,352-MN '... multiplier 353-1 1,353-1-2,353-1-N '... adder 353-M-1, 353-M-2, 353-MN' ... adder 354 ... adder

Claims (3)

A radio relay station that receives a radio signal transmitted from a radio communication station using a plurality of receivers , transmits the radio signal at the same frequency as the received radio signal, and relays the radio signal. a wraparound canceller for canceling an interference signal around to the receiver,
A known signal generator for outputting a predetermined known signal to the transmitter;
Received by an antenna element provided in each of the receivers using a filter coefficient calculated during a period in which the wireless signal is not received and the transmitter transmits the known signal. a plurality of equalizer against analog received signal, performs equalization processing for reducing the reception power of the interference signal sneaking from the transmitter,
Quantizing the received signals equalized by the plurality of equalizers, suppressing the interference signal included in the quantized received signals based on a transfer function of a path that goes from the transmitter to the receiver, A digital signal processing unit that converts a digital reception signal in which an interference signal is suppressed into an analog transmission signal and outputs the analog transmission signal to the transmitter ; and
The number of equalizers is less than the number of receivers;
Each of the equalizers receives received signals of different combinations from the received signals received by the receiver, performs a convolution operation using the filter coefficients on the received signals, and transmits the transmitter. One output signal in which the reception power of the interference signal that wraps around is reduced is output to the digital signal processing unit,
The digital signal processor is
A plurality of digital / analog converters for converting the reception signals equalized by each of the equalizers into digital signals;
A plurality of weight multipliers for multiplying each of the digital signals by a first weighting factor corresponding to an antenna element provided in each of the receivers;
An adder that adds the outputs of the plurality of weight multipliers and outputs the result as an array output signal;
A subtractor that subtracts a feedback signal from the array output signal and outputs a subtraction result as an error signal;
A whitening filter for whitening the error signal;
A delay unit that delays the error signal and outputs the error signal to the transmitter as a transmission signal;
Filtering the transmission signal based on a variable transfer function, and outputting the feedback signal to the subtractor as the feedback signal;
A control unit that controls a transfer function of the adaptive filter based on an output of the whitening filter so that the interference signal is minimized;
A weight variable control unit for calculating the first weighting coefficient that minimizes an error between the array output signal and the known signal;
Wraparound canceller characterized in that it comprises a. The digital signal processing unit further includes
A virtual path weight multiplication unit that multiplies a signal obtained by branching the transmission signal output from the delay unit as a virtual path by a second weighting factor ;
The adder is
Adds the output of each output before Symbol virtual path weight multiplication unit of the plurality of weight multiplication unit, and outputs it to the subtracter as said array output signal,
The weight variable control unit further includes:
Wraparound canceller according to claim 1, characterized in that calculating the second weighting factor error is minimized and the array output signal and the previous SL known signal. A radio relay station that receives a radio signal transmitted from a radio communication station using a plurality of receivers , transmits the radio signal at the same frequency as the received radio signal, and relays the radio signal. a coupling loop cancellation method wraparound canceller executes cancel the interference signal around to the receiver,
A known signal generation unit that outputs a predetermined known signal to the transmitter;
A plurality of equalizers are provided for each of the receivers using filter coefficients calculated during a period in which the wireless signal is not received and the transmitter is transmitting the known signal. is the analog of the signal received by the antenna elements are, the equalization step for performing an equalization process for reducing the reception power of the interference signal sneaking from the transmitter,
Digital signal processing unit, the plurality of equalizer quantizes the received signal equalization, the transfer function of the path going around the interference signal included in the received signal quantized to the receiver from the transmitter suppressed based on, viewed contains a digital signal processing step of outputting a reception signal of the digital to the interference signal is suppressed in the transmitter into an analog transmission signal,
The number of equalizers is less than the number of receivers;
In the equalization step, each of the equalizers receives a received signal of a different combination from the received signals received by the receiver, and performs a convolution operation using the filter coefficient on the received signal. Output one output signal with reduced received power of the interference signal that wraps around from the transmitter to the digital signal processing unit,
In the digital signal processing step, the digital signal processing unit
Each of the equalizers converts a reception signal equalized to a digital signal,
Each of the digital signals is multiplied by a first weighting factor corresponding to an antenna element provided in each of the receivers,
Each multiplication result is added to form an array output signal,
Subtract the feedback signal from the array output signal, the subtraction result as an error signal,
Give a delay to the error signal, output to the transmitter as a transmission signal,
Filtering the transmission signal based on a variable transfer function as the feedback signal;
Based on the whitened signal of the error signal, the transfer function is controlled so that the interference signal is minimized,
A wraparound canceling method comprising calculating the first weighting coefficient that minimizes an error between the array output signal and the known signal .

Images