JPH0786972A - Adaptive equalizer - Google Patents

Abstract

PURPOSE:To suppress or reduce an interference wave as maximizing the reception S/N of a desired wave even when the arrival direction of the desired wave coincides with that of the interference wave in an adaptive equalizer which eliminates wave distortion and an interference disturbance wave due to multipath fading as displaying an adaptive antenna array function. CONSTITUTION:Signals received by antenna elements 101-10N arranged at the half wavelength interval of radio frequency are inputted to an adaptive array filter 20. The adaptive array filter 20 is a delay line filter with tap which performs complex multiplication with a tap coefficient. A decision device 30 decides a modulation symbol from the output signal of the adaptive array filter 20. A subtractor 40 outputs an error signal representing an error between the input and output signals of the decision device 30. The adaptive filter 20 can be operated as correcting the tap coefficient so as to minimize the mean square value of the error signal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an adaptive equalizer, and more particularly to an adaptive equalizer that removes waveform distortion and interference wave caused by multipath fading while exhibiting an adaptive antenna array function.

[0002] When digital transmission is performed in terrestrial microwave communication, land mobile communication, local radio, satellite mobile communication, etc., waveform distortion due to multipath fading and various interference jamming waves become problems. Therefore, in such a digital transmission system, it is important to effectively remove the waveform distortion due to multipath fading and the interference wave.

[0003]

2. Description of the Related Art Waveform distortion due to multipath (intersymbol interference)
e), the adaptive filter (Adapt)
iveFilter) or maximum likelihood sequence estimation (MLSE)
Various adaptive equalization methods based on the above have been used conventionally.

On the other hand, an adaptive antenna array (Adaptive Array Ante) is used for the interference wave.
(nnas) is used to remove the interference wave by antenna pattern forming. In addition, recently, studies have been made to use the adaptive antenna array not only for removing interference and interference waves but also for removing multipath distortion (for example, RT.
Compton, "Adaptive Antenna"
s-Concept and Performance
e ", Prentice Hall, Inc., 1988).

FIG. 11 is a block diagram showing an example of a conventional adaptive equalizer in which the above adaptive antenna array is used not only for removing interference and interference waves but also for removing multipath distortion. This adaptive equalizer is a Bernard Widro.
row) proposed LMS algorithm (Least-
It is also called an LMS adaptive array because it uses the Mean-Square Algorithm.

In the figure, N antenna elements 51 are arranged at half-wavelength (λ / 2) intervals of a radio frequency, and received signals x ₁ , x ₂ ,. ． ． , X _N to the corresponding complex multiplier 52, where the tap coefficients c ₁ , c ₂ ,. ． ． , C _N , respectively. Here, the received signals x ₁ , x ₂ ,. ． ． , X _N is actually
Although input to the receiver, the receiver is omitted in FIG. This is because the entire signal system is handled by the equivalent baseband system, and a receiver for converting a radio frequency into an intermediate frequency and a baseband frequency will not be described in this specification and the drawings.

Received signal x input to the complex multiplier 52
₁ , x ₂ ,. ． ． , X _N and tap coefficients c ₁ , c
₂ ,. ． ． , C _N are each defined by the following matrix.

X ^T = [x ₁ , x ₂ ,. ． ． , X _N ] (1) C ^T = [c ₁ , c ₂ ,. ． ． , C _N ] (2) Here, T means transpose. The signals that are complex-multiplied by the complex multiplier 52 are respectively generated by the combiner 5
3 is commonly input and linearly combined. As a result, the output signal y of the combiner 53 is expressed by the following equation.

Y = C ^T · X (3) The output signal y of the combiner 53 is input to the subtractor 55, where it is subtracted from the reference signal from the reference signal generator 54. This reference signal is a known digital signal that is the same as the digital signal generated at the transmitting side, and actually corresponds to the training burst in the transmitted digital signal. An error signal ε between the reference signal from the reference signal generator 54 and the combined signal y from the combiner 53 is taken out from the subtractor 55 and supplied to the tap coefficient correction circuit 56.

Here, the reference signal is a known digital signal that is the same as the digital signal generated on the transmission side, and an error signal ε indicating an error between the reference signal and the training burst of the received combined signal y is multipath distortion or interference during propagation. Corresponds to the error caused by the waves. The root mean square value of the error signal ε is MSE (Minimum-Square-Error).
It is called an evaluation function and is expressed by the following equation.

Ξ = E [ε ^* · ε] (4) Here, E [] in the above equation represents a time average, and * represents a complex conjugate.

There is always one tap coefficient vector solution C _opt that minimizes the MSE evaluation function ξ, and it satisfies the normal equation (Wiener-Hop equation). Here, the normal equation can be obtained with the value obtained by partially differentiating the equation (4) with the tap coefficient as zero, but can be easily obtained from the orthogonal principle. That is, it is given by the correlation between the error signal ε and the complex conjugate of the received signal vector X, and is shown as follows.

Ψ · C = S (5) Here, Ψ is a correlation matrix of the received signal vector X, and S is a correlation vector of X and the reference signal R, which are respectively expressed by the following equations.

Ψ = E [X ^{* T} · X] (6) S = E [X ^* · R] (7) Then, the inverse matrix Ψ ⁻¹ of the correlation matrix Ψ of the received signal vector X is obtained, The ideal tap coefficient solution C _opt that minimizes MSE is obtained from the following equation by the equation 5).

C _opt = Ψ ⁻¹ · S (8) However, in actual high-speed digital transmission, the tap coefficient is controlled by an adaptive algorithm rather than the direct solution of the correlation matrix. This includes LMS, constant envelope (CM
Various algorithms such as A), Applebaum, and Kalman have been studied. Here, the most simple and well-used LMS is dealt with.

The tap coefficients c ₁ , c ₂ ,. ． ． , C _N as the coordinate axes, the N-dimensional MSE evaluation function ξ is created by the root mean square operation, and thus is represented as a downwardly convex ellipsoid of revolution and always has one minimum point ξ _min . Therefore, the tap coefficients c _i ⁿ at time n using one time slot past tap coefficients c _i ^n-1, the recurrence formula of the formula ^{_{c i n = c i n-}} 1 -μx i * · ε (9 ), The tap coefficient asymptotically approaches the minimum point. However, in the above formula, μ is a correction coefficient.

That is, even if the normal equation is not solved, the LMS algorithm operation of the equation (9) is simply repeated to obtain M
An approximate value of the ideal tap coefficient solution C _opt that minimizes SE is obtained. The tap coefficient correction circuit 56 of FIG. 11 constantly updates the above-mentioned tap coefficient, and the updated tap coefficient c
₁ , c ₂ ,. ． ． , C _N to the complex multiplier 52.

The general characteristics of the LMS-controlled adaptive array (LMS array) are summarized as follows.

When there is neither multipath distortion nor interference wave, the LMS array performs beam forming so that the antenna pattern becomes maximum in the arrival direction of the desired wave. Also, LMS
The array acts as a maximum ratio (MRC) combiner for each antenna element branch.

When an interfering wave exists and the level of the interfering wave is higher than that of the desired wave, the LMS array acts as an antenna pattern null in the arrival direction of the interfering wave so as not to receive the interfering wave. At this time, the antenna pattern in the arrival direction of the desired wave is not always the maximum. On the other hand, when the interference wave level is lower than the desired wave, this null is not created, and the antenna pattern is directed in the desired wave arrival direction while allowing reception of the interference wave.

When there is multipath instead of interference, a multipath wave having a large delay dispersion has a low correlation with the main wave, so that the LMS array is also effective against intersymbol interference.
The same operation as for the interference wave is performed. That is, the LMS array tries to direct the null of the antenna pattern to the echo wave due to the multipath.

When the amplitude phase of the strong interference wave varies with time, the LMS array adaptively varies the null. At this time, the reception pattern for the desired wave also changes with time, and unnecessary modulation is applied to the desired wave.

The above is a general property including advantages and disadvantages, but as described above, since the antenna pattern can be electrically controlled rather than mechanically, its utility value is high. For example, when satellite communication is performed from a mobile body, antenna tracking is always required, and an adaptive array is indispensable. In addition, since the interference from all directions becomes a problem in the premises radio and the car radio, an adaptive array that removes the interference by pattern nulling is used.

Further, as mentioned above, it has been argued that the adaptive array can be used as a kind of adaptive equalizer since echo waves due to multipath can also be treated as interference interference (for example, Clark, "MMSE diversity".
Combining fore wideband digital
Cellular Radio ", IEEE Global Telecommunications Conference 1990, No.
404.5.1). This document evaluates the effect of equivalent adaptive equalization by creating a null antenna pattern in the direction of arrival of an echo wave by an adaptive array and receiving only the main wave.

[0025]

However, the above-mentioned conventional adaptive equalizer (adaptive array) is effective in tracking the arrival direction of the desired wave or performing interference cancellation, but the arrival directions of the desired wave and the interference wave are effective. If there is a match, there is a fatal drawback. That is, if the null is directed to the desired wave, the desired wave cannot be received, and if the null is not directed, interference becomes a problem. When the desired wave and the interference wave match in the arrival direction, the solution of the conventional adaptive array normal equation becomes a solution in which the interference pattern is received at the same time as the antenna pattern toward the desired wave, and the tap coefficient amplitude becomes an extremely small value. , The desired wave can no longer be received at the normal level, and the interference cannot be eliminated.

FIG. 12 shows the result of calculation simulation of a conventional adaptive equalizer. The figure (A) shows the impulse response when the propagation model is a two-wave model by the forward wave away from T, and the figure (B) shows the output response. At this time, the incident angles of the main wave and the forward wave are both 60 °, the signal power to noise power ratio (SN) is 20 dB, the interference wave frequency Ω is 0 (center frequency), the notch depth is 10 dB, and the antenna is two elements. And In FIG. 12 (A), the impulse response is distorted by the two-wave model, but in FIG. 12 (B) as well, it does not become zero at the T intervals and does not satisfy the Nyquist distortion-free condition.

As described above, since the conventional adaptive equalizer removes the interference wave by the null control of the antenna pattern,
When the arrival directions of the desired wave and the interference wave coincide with each other, there is a problem that no removal effect can be obtained.

The present invention has been made in view of the above points,
It is an object of the present invention to provide an adaptive equalizer capable of constructing an optimum receiving system in general digital radio by removing waveform distortion and interference wave due to multipath fading while always maintaining the maximum reception pattern for a desired wave.

[0029]

According to the present invention, in order to achieve the above object, received signals respectively received by a plurality of antenna arrays are complex-multiplied with tap coefficients which are updated and controlled from the outside, and then combined and output. An adaptive array filter, a determiner that receives an output signal of the adaptive array filter and determines a modulation symbol, and an error signal generation unit that obtains an error signal indicating an error of the output signal of the determiner with respect to the output signal of the adaptive array filter. The adaptive array filter is configured to perform complex multiplication by modifying the tap coefficient so as to minimize the root mean square value of the error signal.

[0030]

In the present invention, the plurality of antenna arrays arranged at half wavelength intervals of radio frequency are provided in a one-to-one correspondence with each other, and the signals received by the antenna array are subjected to complex multiplication using the tap coefficient. And a plurality of tapped delay line filters for synthesizing, and a synthesizer for synthesizing each output signal of the plurality of tapped delay line filters,
The adaptive array filter is configured by a tap coefficient correction circuit that corrects and updates the tap coefficients of a plurality of tapped delay line filters so that the mean square value of the error signals is input to the minimum, and the adaptive array filter is configured with the tap coefficients. Since the delay line filter performs the complex multiplication using the tap coefficient modified so as to minimize the root mean square value of the error signal, the antenna pattern and the adaptive equalization can be optimized at the same time.

[0031]

Next, an embodiment of the present invention will be described. FIG. 1 shows a block diagram of an embodiment of an adaptive equalizer according to the present invention. In this figure, this embodiment shows N-element antennas 10 ₁ to 10 ₁ arranged at half-wavelength (λ / 2) intervals of radio frequency.
0 _N , the adaptive array filter 11, and a determiner 30 that receives the output signal of the adaptive array filter 11 as an input signal
And a subtractor 40 that subtracts the input signal and the output signal of the determiner 30 and outputs the error signal to the adaptive array filter 20 to update the tap coefficient.

FIG. 2 shows a circuit configuration diagram of an embodiment of the adaptive equalizer according to the present invention. In the figure, the same components as those in FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted. In FIG.
Each antenna element 10 ₁ to 10 _N of the N-element antenna is individually tapped delay line (TDL) filter 21 ₁ to
Commonly connected to the combiner 22 via 21 _N. The output signal of the combiner 22 is input to the determiner 30 and the subtractor 40, respectively. The output error signal ε of the subtractor 40 is input to the tap coefficient correction circuit 23. The adaptive delay line filters 21 _{1 to} 21 _N , the combiner 22 and the tap coefficient correction circuit 23 constitute the adaptive array filter 20.

The tapped delay line filters 21 _{1 to} 21 _N have the same configuration, and the k-th (k = 1 to N) tapped delay line filter 21 _k is connected in cascade to the antenna element 10 _k . L-1 delay elements 211 each having a delay time τ, tap coefficients c _{k1 to} c _kL, and a delay element 2
11 input signals or 11 output signals are input, and a total of L complex multipliers 212 and these L complex multipliers 212 are input.
And a combiner 213 for combining the respective output signals of 1 and 2.

Here, the tapped delay line filter 21 _k
From the input end to the output end of the (L-1) delay elements 211 of the received signals at x total taps, x _k1 , x _k2 ,
, X _kL , the tap coefficients input to the complex multiplier 212 are c _k1 , c _k2 , ..., C _kL, and the modulation symbols on the transmission side are the sequence {..., a _−1. , A ₀ , a
_Supposing that the signals are transmitted in the order of ₊₁ , ..., The N element antenna reception signal vector X and the tap coefficient vector C of the adaptive array filter 20 can be defined by the following matrices.

X ^T = [x _k1 , x _k2 , ..., X _kL ] (10) C ^T = [c _k1 , c _k2 , ..., c _kL ] (11) The complex multiplier 212 is Since the complex multiplication of the received signal vector and the tap coefficient vector is performed and the multiplication result is output to the combiner 22 through the combiner 213, the combiner 22
The output signal y of can be expressed by the following equation.

[0036]

[Equation 1] The output signal y is input to the determiner 30 and used as determination data, and is also input to the subtractor 40 where it is subtracted from the output determination data of the determiner 30 to be an error signal ε. Here, when the symbol rate is low, the above decision data can be approximated to the transmission modulation symbol a _0, and thus the error signal is represented by the following equation.

Ε = ya− ₀ (13) This error signal ε is input to the tap coefficient correction circuit 23,
A tap coefficient vector solution C that minimizes the root mean square value is obtained from a normal equation (Wiener-Hop equation). The normal equation can be easily obtained from the orthogonal principle. That is, E [ε · x ^* _pu ] = 0 (p = 1,2, ..., N u = 1,2, ..., L) (14) A normal equation with a linear first-order tap coefficient as an unknown can get. In the above equation, E [] means an expected value operation, that is, a time average.

Here, in order to evaluate the effect of the embodiment on the multipath distortion, a propagation model having a delay dispersion characteristic is dealt with first. Let h _i be the discrete value of the symbol interval of the impulse response of the transmission system, and set i = 0 to the current reference timing. The discrete value h _{i in} which i is a negative value is the leading edge (Precursor) of the impulse response, and the discrete value in which i is a positive value is the trailing edge (Postcursor) of the impulse response.
Shall be indicated. In this case, the received signal is a convolution of the transmission symbol sequence {a _i } and the discrete value h _i . At this time, the main wave component becomes h ₀ a ₀ , and the multipath wave component due to the i-th impulse response discrete value becomes h _i a _0-i .

FIG. 3 is an explanatory diagram showing signals received by the N-element array antennas 10 ₁ to 10 _N in such a multipath line. 3, those parts that are the same as those corresponding parts in FIG. 2 are designated by the same reference numerals, and a description thereof will be omitted. In FIG. 3, 101, 102, and 103 are main waves incident on the antenna elements 10 ₁ , 10 ₂ , and 10 _N , respectively, and 104, ₁
Reference numerals 05 and 106 denote antenna elements 10 ₁ , 10 respectively.
It is a multipath wave that is incident on ₂ , 10 _N. Also, 10
Reference numeral 7 indicates a wavefront received by the first antenna element 101 when the main wave 101 is used as a reference.

Here, assuming that the main wave 101 is h ₀ a ₀ and the multipath wave 104 is h _i a _0-i , the antenna element 10 ₁
-10 _N are arranged at half-wavelength (λ / 2) intervals of the radio frequency, so that the main wave 102 received by the antenna element 10 ₂ adjacent to the antenna element 10 ₁ is the main wave 10
It lags behind 1 by the phase of exp (-jφ ₀ ). This is because when the main wave 101 is received in the antenna elements 10 _1,
This is because the main wave 102 is located at the intersection with the wavefront 107 and has not yet reached the antenna element 10 ₂ .

The delay phase angle φ ₀ depends on the incident angle θ ₀ of the main wave on the antenna element and is given by the following equation.

Φ ₀ = π · sin θ ₀ (15) Similarly, the main wave 103 incident on the _Nth antenna element 10 _N is the reception main wave 10 of the _first antenna element 10 _1.
The phase lags by 1 by exp {-j (N-1) φ ₀ }. Similarly, if the angle of incidence of the multipath wave on the antenna element is θ _i , the phase transition in units of the phase angle φ _i φ _i = π · sin θ _i (16) given by 105, 10
6 occurs.

From the above, the received signal x _pu on the u-th tap in the TDL filter 21 _p connected to the p-th antenna element 10 _p can be expressed by the following equation.

[0044]

[Equation 2] However, in the above equation, n _p represents the receiver noise of the reception wave of the p-th antenna element 10 _p . Further, when the incident angle of the multipath wave by the nth impulse response discrete value is θ _n , the phase shift amount φ _n due to this is represented by the following equation.

Φ _n = π · sin θ _n (18) The above equations (10) to (14), (17) and (1)
The following equation is obtained by deriving the normal equation using the equation 8).

[0046]

[Equation 3] In the above equation, i-th row and j-th column of submatrix [psi _pq of L × L (i, j
= 1, 2 ,. ． ． , N) of {ψ (p,
q) _ij }, this is expressed by the following equation.

[0047]

[Equation 4] However, in the above equation, σ ² is noise power, and δ _pq is the Kronecker delta of the following equation.

[0048]

[Equation 5] The above L × L small matrix Ψ _pq is a Hermitian matrix, and the transposed complex conjugates with respect to p and q are equal as described below.

Ψ _pq = Ψ _qp ^{T *} (22) Further, S _k ^* (where k = 1 to N) on the right side of the equation (19) is a complex vector of the correlation vector S _k between the judgment data and the array antenna received signal. Indicates the conjugate vector. Here, S _k is expressed by the following equation.

[0050]

[Equation 6] The above is the general form of the normal equation of the present embodiment using the N-element array antennas 10 ₁ to 10 _N and the TDL filters 21 _{1 to} 21 _N with L taps. By actually solving this, it can be confirmed that this embodiment operates as an adaptive equalizer.
Here, in order to facilitate the physical interpretation, the following is considered.

In FIG. 2, N TDL filters 21 are provided.
Applying the superposition theory to _{1 to} 21 _N , N TD
The L filters 21 _{1 to} 21 _N can be integrated into one equivalent TDL filter. That is, in the N number of TDL filters 21 _{1 to} 21 _N , the received signals on the input-side first taps from the first branch to the Nth branch are x _1L + x _2L +. ． ． + X _NL , and tap coefficients are similarly c _1L + c _2L +. ． ． _Overlap like + c _NL . By repeating these superpositions up to the L-th tap, the N TDL filters 21 _{1 to} 21 _N can be equivalently regarded as one TDL filter.

That is, in FIG. 2, it can be understood that the _N TDL filters 21 _{1 to} 21 _N are equivalent to one linear equalizer and perform adaptive equalization for multipath distortion. What should be noted here is that, unlike the normal linear equalizer, the linear equalizer of this embodiment always directs the maximum antenna pattern in the direction of arrival of the desired wave to maximize the SN ratio. This antenna pattern is formed at the reference taps of the _N TDL filters 21 _{1 to} 21 _N and can be expressed by the following equation.

[0053]

[Equation 7] In addition, (L- in the _N TDL filters 21 _{1 to} 21 _N )
1) When each delay time τ of the delay elements 211 is set to T / 2 and a fractional interval, the TDL filters 21 _{1 to} 21 _N simply receive interference (Precursor) from the next symbol.
Distortion) is not only removed, but the desired wave power delayed and dispersed as a matched filter is converged, and the effect of maximizing the SN ratio is produced. This is called implicit diversity gain. Also, by setting the delay time τ to be a fractional interval, the problem of the aliasing spectrum due to the deviation of the sampling timing phase on the receiving side is solved.
That is, the TDL filters 21 _{1 to} 21 _N also have a timing control function.

FIG. 4 shows the result of calculation simulation by solving the above normal equation. The propagation model is a two-wave model with a forward wave separated by T where there is one interference wave that is time T ahead of the main wave, the main wave and the forward wave both have an incident angle of 60 °, and an SN ratio of 20 dB. , The interference frequency Ω is 0
(Center frequency) and the depth of the notch are 10 dB.
Further, in the embodiment shown in FIG. 2, two antennas are used,
The number of TDL filter taps is 7.

FIG. 4A shows the impulse response at the output of the TDL filter of the above two-wave model, and FIG. 4B shows the antenna pattern. The impulse response distorted by the conventional two-wave model shown in FIG. 12A is equalized as shown in FIG. 4A according to the present embodiment, and satisfies the Nyquist distortion-free condition. At this time, the antenna pattern is as shown in FIG.
As shown in (B), it can be seen that the antenna maximum pattern is directed in the arrival direction of the main wave indicated by the solid arrow.

In the conventional adaptive equalizer, an interference wave (echo wave)
In contrast to the case where the null is created in the arrival direction, in the present embodiment, the removal is not performed by the adaptive array pattern, but the anti-phase cancellation removal is performed by the linear synthesis of the adaptive array filter 20. That is, in FIG. 2, L complex multipliers 212
By synthesizing the output signals of
The echo wave is canceled in reverse phase.

Next, the interference wave interference elimination function of this embodiment will be described. In FIG. 5, the number of antenna elements N of the embodiment of the present invention shown in FIG. 2 is “2”, and the number of TDL filter taps is L.
A circuit configuration diagram in which is set to "3" is shown. In FIG.
The same components as in FIG. FIG at 5, 25 ₁₁ and 25 ₁₂ the delay time corresponding to TDL filter 21 ₁ of the delay element 211 shown in FIG. 2, respectively delay elements of the symbol period T, 25 ₂₁ and 2
5 ₂₂ wherein each of the TDL filter 21 _second delay element 2
A delay element with a delay time corresponding to 11 has a symbol period T.

Further, 26 ₁₁ , 26 ₁₂ and 26 ₁₃ are complex multipliers corresponding to the complex multiplier 212 in the TDL filter 21 ₁ , and 26 ₂₁ , 26 ₂₂ and 26 ₂₃ are TDL respectively.
A complex multiplier corresponding to the complex multiplier 212 in the filter 21 ₂ and a combiner 27 corresponding to the combiner 213 and the combiner 22 in the TDL filters 21 ₁ and 21 ₂ . Furthermore, 2
8 ₁ is two complex multipliers of the first tap, 28 ₂ is _two complex multipliers of the second tap, 28 ₃ is 2 of the third tap
3 complex multipliers are shown.

The desired wave 301 incident on the antenna element 10 ₁ and the desired wave 302 incident on the antenna element 10 ₂ respectively have no multipath distortion, and the desired wave 50
It is assumed that one CW interference wave having an angular frequency Ω has arrived at 1,502 as indicated by 303 and 304, respectively.

The modulation symbols of the desired waves 301 and 302 are a _i , and those of the CW interference waves 303 and 304 are √J · ex.
p (jΩt), and each antenna element 10 ₁ , 1
Assuming that the incident angles to 0 ₂ are θ _a and θ _j , at this time, as shown in FIG. 5, there is a phase difference exp (−jφ between the desired waves 301 and 302 incident on the antenna elements 10 ₁ and 10 _{2. a} ) occurs, and similarly, the phase difference exp (-jφ _j ) also occurs between the CW interference waves 303 and 304. Where φ _a = π sin θ
_a , φ _j = π sin θ _j .

The normal equation in the model shown in FIG. 5 is expressed by the following equation.

[0062]

[Equation 8] Here, Ψ ₂₂ = Ψ ₁₁ , Ψ ₂₁ = Ψ ^{T *} _12, which are expressed by the following equations.

[0063]

[Equation 9] In the above equation, η ₀ is the desired wave to interference wave power ratio (D / U)
, Ρ is the reciprocal of the signal-to-noise power ratio (SN ratio), and these are respectively expressed by the following equations.

Η ₀ = J / (a ^* _i · a _i ) (28) ρ = σ ² / (a ^* _i · a _i ) (29) Further, in the normal equation of the equation (25), the judgment data and the antenna are Correlation vectors S ₁ and S ₂ with the received signal and tap coefficient vectors C ₁ and C ₂ are respectively expressed by the following equations.

S ₁ ^T = [1 0 0] (30) S ₂ ^T = [exp (−jφ _a ) 0 0] (31) C ₁ ^T = [c ₁₁ c ₁₂ c ₁₃ ] (32) C ₂ ^T = [C ₂₁ c ₂₂ c ₂₃ ] (33) FIG. 6 shows a calculation simulation result obtained by solving the normal equation. Here, both the incident angle of the desired wave and the CW interference wave are 45 °, the SN ratio is 20 dB, and the CW interference wave frequency Ω is 0.
(Center frequency). FIG. 5A shows an antenna pattern by each tap of the embodiment shown in FIG.
In the figure drawn by solving the normal equation of the equation (5) and by the equation (24), the arrival direction of the desired wave is indicated by a solid arrow, and the arrival direction of the CW interference wave is indicated by a dashed arrow.

As can be seen from FIG. 6 (A), the antenna maximum pattern is oriented in the desired wave arrival direction. As described above, in the conventional adaptive equalizer, the null of the adaptive array pattern is created in the arrival direction of the interference wave. However, in this embodiment, the null is not performed by the adaptive array pattern, but the inverse by the linear synthesis of the TDL filter is performed. Perform phase cancellation removal. That is, in the present embodiment, in FIG. 5, the complex multipliers 26 ₁₁ , 26 ₁₂ , 26 ₁₃ , 2 of the respective taps of the synthetic TDL filter are shown.
The output signals of 6 ₂₁ , 26 ₂₂ and 26 ₂₃ are combined by the combiner 27 to cancel the CW interference in anti-phase.

The antenna pattern shown in FIG. 6A is the gain of the adaptive array filter in the spatial domain. The gain in the frequency domain, that is, the evaluation of the frequency characteristic of the adaptive filter will be described below. It is assumed that a signal exp (jωt) of each frequency ω having a unit amplitude is incident on the array antenna in the arrival direction (θ = 45 °) of the interference wave indicated by the dashed arrow in FIG. At this time, the combined frequency characteristic of the _three TDL filter multipliers 28 _{1 to} 28 ₃ is obtained by the following equation.

H (ω) = | {c ₁₁ + c ₂₁ · exp (-jφ _j )} + {c ₁₂ + c ₂₂ · exp (-jφ _j )} · exp (jωT) + {c ₁₃ + c ₂₃ · exp ( −jφ _j )} · exp (j2ωT) | (34) The frequency characteristic of the above equation is shown in FIG. 6 (B). In the same figure (B), the horizontal axis is -0.5f _s to + 0.5f _s (where f
_s is the frequency of the modulation speed), and the vertical axis has one scale of 10
The amplitude of dB is shown. This frequency characteristic shows a notch characteristic at the center frequency of f = 0. That is, due to this notch characteristic, the CW interference wave at f = 0 is significantly reduced.

The above is the case of a single CW interference wave.
The present embodiment also has a similar effect of removing a plurality of CW interference waves. This will be described with reference to FIGS. 7 and 8. FIG. 7 is a circuit configuration diagram for explaining a plurality of interference wave interference removing functions of one embodiment of the present invention, and has the same configuration as FIG. However, the array antenna elements 10 ₁ in FIG. 7, the first interference wave 401 in addition to the desired wave 301 and the second interference wave 4
03 and is incident, another in the array antenna elements 10 ₂ Further, the first interference wave 402 and a second interference wave 404 is incident on the other of the desired wave 302.

In the propagation model of FIG. 7, there is no multipath propagation as in FIG. 5, and the CW interference waves 401 and 402 are incident on the antenna elements 10 ₁ and 10 ₂ at the angular frequency Ω ₁ and the angle θ ₁ , respectively. It is assumed that the interference waves 403 and 404 are incident on the antenna elements 10 ₁ and 10 ₂ at an angular frequency Ω ₂ and an angle θ ₂ . The normal equation of this propagation model is given by the equation (25), and the small matrix Ψ in the normal equation has a size of 3 × 3 and is given by the following equation.

[0071]

[Equation 10] However, in the above equation, η ₁ and η ₂ are respectively D / U
It is the reciprocal of and is expressed by the following equation.

Η ₁ = J ₁ / (a _i ^* a _i ) (37) η ₂ = J ₂ / (a _i ^* a _i ) (38) Further, φ _a = π sin θ _a , φ ₁ = π sin θ ₁ , φ ₂
= Πsin θ ₂ . Further, the correlation vector S and the tap coefficient vector C between the determination data and the antenna element received signal in the normal equation given by the equation (25) are also the equations (30) to (33) in the case of the propagation model of FIG. It is given by the following formula, which is the same as the formula.

S ₁ ^T = [1 0 0] (39) S ₂ ^T = [exp (−jφ _a ) 0 0] (40) C ₁ ^T = [c ₁₁ c ₁₂ c ₁₃ ] (41) C ₂ ^T = [C ₂₁ c ₂₂ c ₂₃ ] (42) FIG. 8 shows a simulation result obtained by solving the above normal equation. As shown in FIG. 7A, the desired wave D is incident at the incident angle θ.
_a = 45 °, the first CW interference wave J ₁ arrives at an angular frequency Ω = 0, the incident angle θ ₁ = 45 °, and the second CW interference wave J _{2 further} reaches an angular frequency Ω = + 0. Suppose that the incident angle is 0.3 _fs × 2π, the incident angle θ ₂ = 45 °, and the SN ratio is 6
TDL obtained by solving the normal equation of equation (25) using equations (35) to (42) when set to 0 dB.
The composite pattern of the antenna pattern by each tap coefficient of the filter is as shown in FIG.

According to this embodiment, even if a plurality of interference waves arrive and one of them coincides with the arrival direction of the desired wave, the desired combination pattern of the antenna patterns is desired as in the above-described example. In order to receive the wave D at the maximum, the maximum pattern is formed in the incident angle direction of 45 °. Therefore, at the same time, the individual D / U also receives two-wave CW interference of 0 dB,
These are greatly reduced by the frequency notch filtering function of the synthesis of the two TDL filters of FIG.

The frequency characteristic in the direction of arrival angle θ of the adaptive array filter having the configuration shown in FIG. 7 is obtained by the following equation.

[0076]

[Equation 11] When this is calculated under the above conditions, the frequency characteristic in the direction of arrival angle θ of the adaptive array filter having the configuration shown in FIG. 7 is as shown in FIG. 8 (B). The horizontal axis of the figure shows the frequency, and the vertical axis shows the amplitude of 10 dB on one scale. As is clear from the figure, the notch characteristics are shown at the frequencies 0 and +0.3 f _s , respectively. Therefore, the above two interference waves J ₁ and J ₂ are significantly reduced by the notch characteristics.

By the way, the above description has been made in the case where the arrival directions of the desired wave and the interference wave are the same or substantially the same, but according to the present embodiment, the interference wave comes from a direction different from the desired wave. In that case, the interference wave can be removed or reduced. For example, among the simulation conditions of FIG. 8, when only the arrival direction of the interference wave J ₂ is θ ₂ = 5 °, the synthetic pattern of the antenna pattern by each tap coefficient of the TDL filter of this embodiment is as shown in FIG. become,
The frequency characteristic in the direction of the arrival angle θ is as shown in FIG.

As can be seen from FIG. 9, the synthesized antenna pattern is the desired wave D and the first C having the same arrival direction as the desired wave D.
The maximum pattern is shown for the W interference wave J ₁ , and the desired wave D
And the second CW interference wave J having different arrival directions (θ ₂ = 5 °)
_{For 2} , the smallest pattern is shown.

On the other hand, the adaptive array filter having the configuration shown in FIG.
The arrival angle of Ruta is 45 ° (= θ₁) Frequency for direction
The characteristics are shown in Fig. 10 (A), and the arrival angle is 5 ° (= θ₂)
The frequency characteristic with respect to the direction is as shown in FIG.
It The horizontal axis of the figure is the frequency, and the vertical axis is 1 dB on one scale.
Shows the amplitude of. As is clear from FIG.
The frequency characteristic for the direction of arrival angle of 45 ° is frequency 0.
Since the notch characteristic is shown, the above interference wave J₁Box's
Significantly reduced due to notch characteristics.

The frequency characteristic with respect to the direction of arrival angle of 5 ° is frequency + 0.3f _s as shown in FIG. 10 (B).
Since it has a notch characteristic, the above interference wave J ₂ of frequency + 0.3f _s is removed together with the antenna minimum pattern shown in FIG.

As described above, by applying the present invention to a two-element array of the smallest scale, a plurality of CW interference waves arriving at an arbitrary angle can be removed or significantly reduced, and not only interference removal. Since the adaptive array pattern is formed in the direction of arrival of the desired wave and the SN ratio is maximized, it is possible to configure an optimum receiving system for mobile digital communication that requires downsizing of the device. Such an effect is an effect that cannot be obtained by the conventional adaptive array antenna.

Optimum reception including adaptive equalization and interference cancellation is realized by increasing the number of antenna elements and the number of TDL filter taps in FIGS. 5 and 7 and setting the reference tap of the TDL filter at the center. be able to. In other words, according to the present embodiment, the desired wave is first received maximally by the antenna pattern tracking by the reference tap of the TDL filter, and the desired wave reception signal delay-dispersed by the taps before and after the reference tap is subjected to matched filtering to obtain the implicit diversity. • Gain can be obtained, and further, interference removal and multipath distortion removal can be performed by linear combination between TDL filter taps.

The present invention is not limited to the above embodiments, and by increasing the number of TDL filter taps, not only CW interference waves but also interference of a plurality of modulated waves having a band can be removed. is there. These various effects are
In the configuration of FIG. 1, the adaptive filter array 20 can be simultaneously realized by performing MMSE control based on the output of the subtractor 40.

[0084]

As described above, according to the present invention,
An adaptive array and adaptive equalization with one root mean square error (MSE) evaluation function by performing complex multiplication using tap coefficients modified by a tapped delay line filter so as to minimize the root mean square value of the error signal Since the antenna pattern and the adaptive equalization can be optimized at the same time by performing common control of the two, it is possible to obtain a synergistic effect of both and to eliminate multiple interference waves even with the smallest 2-element antenna array. Even if the arrival directions of the wave and the interference wave match, the interference pattern can be removed while always tracking the antenna pattern in the desired wave direction to maximize the SN ratio.

Further, according to the present invention, it is possible to suppress or remove interference and at the same time remove waveform distortion due to multipath fading, and further to obtain an implicit diversity gain for a delay dispersion line. You can As described above, according to the present invention, it is possible to optimally configure a receiving system of a mobile body and a general fixed digital radio in which multipath distortion and interference are problems.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention.

FIG. 2 is a circuit configuration diagram of an embodiment of the present invention.

FIG. 3 is an operation explanatory diagram of an N-element array antenna.

FIG. 4 is a diagram showing a simulation result of an example of the present invention.

FIG. 5 is a circuit configuration diagram for explaining a single interference and interference elimination function according to an exemplary embodiment of the present invention.

FIG. 6 is a diagram showing a simulation result of the embodiment of FIG.

FIG. 7 is a circuit configuration diagram for explaining a plurality of interference and interference removal functions according to an embodiment of the present invention.

FIG. 8 is a diagram showing a simulation result of the embodiment of FIG.

9 is a diagram showing another example of the antenna pattern of the embodiment of FIG.

FIG. 10 is a diagram showing another example of frequency characteristics of the embodiment of FIG.

FIG. 11 is a configuration diagram of a conventional example.

12 is a diagram showing the impulse response and the output response of FIG. 11. FIG.

[Explanation of symbols]

10 ₁ to 10 _N antenna element (N element antenna array) 20 adaptive array filter 21 _{1 to} 21 _N delay line (TDL) filter with taps 22, 27, 213 combiner 23 tap coefficient correction circuit 25 ₁₁ , 25 ₁₂ , 25 ₂₁ , 25 ₂₂ , 211 Delay element 26 _{11 to} 26 ₁₃ , 26 _{21 to} 26 ₂₃ , 212 Complex multiplier 30 Judgmenter 40 Subtractor 101 to 103 Main wave 104 to 106 Multipath wave 301, 302 Desired wave 303, 304, 401 ~ 404 CW interference wave

Claims (2)

[Claims] 1. An adaptive array filter for complexly multiplying received signals respectively received by a plurality of antenna arrays by a tap coefficient which is updated and controlled from the outside, and then outputting the synthesized signal, and an output signal of the adaptive array filter is inputted. And an error signal generating means for obtaining an error signal indicating an error of the output signal of the decision unit from the output signal of the adaptive array filter, the adaptive array filter including the error signal. An adaptive equalizer characterized in that the tap coefficient is modified so as to minimize the root mean square value of the output error signal of the signal generating means and the complex multiplication is performed. 2. The adaptive array filter is provided in a one-to-one correspondence with each of the plurality of antenna arrays arranged at half wavelength intervals of a radio frequency, and the signal received by the antenna array is tapped with the tap coefficient. , A plurality of tapped delay line filters for synthesizing and performing complex multiplication, a synthesizer for synthesizing output signals of the plurality of tapped delay line filters, and a square of the error signal to which the error signal is input. The adaptive equalizer according to claim 1, further comprising a tap coefficient correction circuit that corrects and updates the tap coefficient of each of the plurality of delay line filters with taps so as to minimize the average value.