JP2000077926A - Controller for array antenna - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the delay of convergence caused by low rating of signal processing due to band division by changing the overlap of window of FET in moving on the time base through a frequency band dividing filter means. SOLUTION: From the start of operation to the reception of an instruction from an input/output control means 9, input/output signal control means 3-1 to 3-n succeed samples buffered at every sample timing to FET calculating means 4-1 to 4-n of following steps. After the instruction is received from the input/output control means 9, on the other hand, the input/output signal control means 3-1 to 3-n dispatch sample values to the FET calculating means 4-1 to 4-n only at the timing of completely exchanging the input samples. As a result, the frequency of dispatching values to the FET calculating means is reduced into 1/band division number. Thus, the windows of FET calculating means 4-1 to 4-n buffer signals so as to overlap the samples on the time base.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an array antenna control device for adaptively controlling a weight coefficient of an element antenna used for various wireless communications.

[0002]

2. Description of the Related Art An adaptive array antenna has a strong directivity in a certain direction or a signal coming from a certain direction by synthesizing a signal received by each element antenna by multiplying the signal by an appropriate weighting factor. It is possible to set the sensitivity to 0 only for the signal to be received and not to receive the signal. Such an antenna is used to perform optimal beamforming and realize spatial filtering by using an appropriate weighting factor according to a change in a radio wave environment, and to avoid interference with other wireless communication systems. .

[0003] With the recent development of digital signal processing technology and the resulting miniaturization of devices, the application of adaptive array antennas using digital signal processing to mobile communications has attracted attention. In a multipath fading environment that poses a problem in a mobile communication system, countless reflected waves, that is, delayed signals of desired signals arrive. By applying the adaptive array antenna to such an environment, ideally, the desired signal and the uncorrelated interference signal are spatially canceled, while the delayed signal of the desired signal is captured and in-phase synthesized with the desired signal, It is possible to contribute to the improvement of S / (N + I) of the output signal.

[0004] A delay element (TDL; T
It is known that a configuration for performing signal processing on the time axis using a mapped delay line) and a configuration for performing signal processing on the frequency axis using a perfect reconstruction filter are effective. 1 and 2a, respectively. FIG. 1 is a configuration diagram of an adaptive array antenna that performs time domain signal processing based on TDL. FIG. 2A is a configuration diagram of an adaptive array antenna that performs frequency domain signal processing using FFT as a band division filter.

[0005] As a document that makes the above-mentioned configuration known, RT
Compton, Jr, "The RelationshipBetween Tapped Dela
y-Line FFT Processing in Adaptive Array ", IEEE
Transaction on Antennas and Propagation, vol. 36 N
o. 1, January 1988. According to the document, there is no essential difference in the performance of both configurations.

The adaptive array antenna for performing frequency domain signal processing shown in FIG. 2A employs FFT as a band division filter, but uses a buffer as a window for performing FFT. This buffer takes in the signal so as to shift by one sample with respect to the signal sample on the time axis. FIG. 2B is an explanatory diagram showing this state as the operation of the buffer for the samples on the time axis.

However, a plurality of configurations derived from the basic configuration shown in FIG. 2A are conceivable especially for an adaptive array antenna which performs frequency domain signal processing using FFT for a band division filter, and characteristics of each performance are investigated. ing. Kamiya and Karasawa, "Characteristic Comparison of Convergence Characteristics of Time-Domain Signal Processing and Frequency-Domain Signal Processing in Adaptive Arrays", IEICE Technical Report AP98-19 (1998-06)
There is. The configuration of the adaptive array antenna that performs the frequency domain signal processing described in the document includes a configuration in which signal processing is performed in parallel for each divided frequency band (subband), thereby reducing the weight coefficient update speed. Has been introduced. FIG. 3A is a configuration diagram of such an adaptive array antenna that performs frequency domain signal processing using an FFT as a band division filter, and reduces the weight coefficient update speed. FIG. 3B is an explanatory diagram showing the operation of the buffer for the samples on the time axis based on the configuration of FIG. 3A.

In the configuration of FIG. 3A, the speed of the input signal in each subband is reduced to one order of the FFT by applying the FFT window for performing the band division so that the signals do not overlap. become. The weight coefficient control algorithm operates independently for each subband, and sequentially obtains a weight coefficient that minimizes the square error between the band-divided reference signal and the subband signal.

The configuration of FIG. 3a requires the calculation of an inverse FFT to recover the signal in the time domain. In this configuration,
Since the weight coefficient update speed can be reduced, the calculation load imposed on the computer by the adaptive control algorithm of the weight coefficient can be greatly reduced, and it is expected that the configuration will be advantageous in realizing a device. Accordingly, it is possible to adopt a high-performance adaptive control algorithm having a large calculation load.

In an array antenna which performs the above-mentioned frequency band division signal processing, a combination of a band division filter for dividing a received signal on a frequency axis and a band restoration filter for restoring the divided signal to an original spectrum. , But such a filter pair is called a perfect reconstruction filter. The combination of FFT / inverse FFT is one of perfect reconstruction filters. When performing digital signal processing, the FFT /
It is effective to employ a perfect reconstruction filter based on a combination of inverse FFT.

On the other hand, various adaptive control algorithms for array antennas have been conventionally studied. Of these, RL
The S (Recursive Least Squares) method is known as an algorithm having high stability and fast convergence. RLS
The method is, for example, Simon Haykin, "Adaptive Filter Theory",
Prentice-Hall Inc. 1996.
The RLS method is an algorithm that asymptotically obtains a weight coefficient that minimizes the square value of an error between an array antenna output and a reference signal using all past data, and is interpreted as a form of a Kalman filter. You can also.
The RLS method is an algorithm that is excellent in stability and convergence speed, but has a very large calculation load.
When implementing a device using a DSP (Digital Signal Processor), the calculation load on the processor becomes a problem.

For this reason, various methods for reducing the amount of calculation of the RLS method have been conventionally proposed, and high-speed RLS methods and the like have been proposed, but these calculation amount reduction methods cannot be applied to array antennas. This is because the high-speed RLS method is configured using the property that the signal inside the tap shifts by one tap at a time, such as an equalizer,
It cannot be applied to a system in which the values inside all taps change simultaneously, such as an array antenna.

FIG. 4A shows an embodiment to which the fast RLS method can be applied, and FIG. 4B shows an embodiment to which the fast RLS method cannot be applied. It can be said that such a high-performance algorithm having a large calculation amount is very compatible with a frequency-division signal processing type adaptive array capable of operating at a low speed.

[0014]

When the RLS method is applied to the frequency band division signal processing type adaptive array shown in FIG. 3A, the operation of the RLS method can be reduced by performing parallel processing for each divided band. Therefore, it can be said that this combination of the configuration and the algorithm of the array antenna is effective in realizing the device. However, on the other hand, the following problems occur in the configuration of the array antenna to which the RLS method is applied.

FIG. 5 is a graph showing a correlation coefficient between an output signal of an adaptive array antenna for performing frequency domain signal processing and a reference signal when S / N = 10 dB. The configuration of the array antenna is obtained by applying the RLS method to the configuration of FIG. 3A as a weight coefficient control algorithm. The array antenna is a one-dimensional array consisting of eight element antennas. Each received signal is divided into 32 bands. Therefore, the required weighting coefficients are 256.

In FIG. 5, the horizontal axis is the number of steps, that is, RLS.
Shows the number of repetitions of the modulus. From FIG. 5, the search for the weighting coefficient is started, and once the correlation coefficient has been improved, 25
It can be seen that there is a significant deterioration before and after step 0. Such a phenomenon is considered to be a phenomenon peculiar to the RLS method, and the cause can be explained as follows.

The RLS method refers to all past inputs and finds an optimum weighting coefficient at that time. In this operation, an equation obtained in each step is sequentially calculated as a simultaneous equation. Is equivalent to Therefore, since the degree of freedom of the simultaneous equations is not satisfied until the number of simultaneous equations becomes equal to or greater than the number of weight coefficients to be obtained, there are countless solutions satisfying the conditions, and the search for the optimum weight coefficient is not started. For this reason, it once converges to an appropriate solution, but as the degree of freedom is satisfied, the condition is not satisfied with that solution, and once the error increases, the optimal weighting factor is increased with the degree of freedom satisfied. The search starts. As a result, since the calculation of the optimal weighting factor is not started until the same number of steps as the number of weighting factors to be controlled, convergence is significantly slowed down, especially in an adaptive array antenna to which band division signal processing is applied. Cause.

In the configuration of FIG. 3A, the RLS method can be applied independently for each sub-band, so the number of samples required until the degree of freedom is satisfied is eight, which is equal to the number of elements. But,
Since the weight coefficient update speed is reduced, only one sample is included in 32 sampling times, which is the number of band divisions, so that the degree of freedom is not satisfied up to 8 × 32 = 256 sampling times. In the configuration of the array antenna for performing the frequency domain signal processing shown in FIG. 2, when the convergence of the correlation coefficient is examined, the correlation coefficient is also degraded once in 256 sampling times. In this case, the input of the sample is input at a rate equal to the sampling frequency, but since all the weighting factors are to be obtained by one RLS method, 256 samples are required until the degree of freedom is satisfied, and the result is as follows. As a result, the correlation coefficient deteriorates at the same time as the case of FIG. 3A.

Accordingly, the present invention provides an adaptive array antenna based on band division signal processing, in which R is used as a weighting factor control method.
Provided is a control device for an array antenna that improves a convergence delay caused by a reduction in signal processing rate due to band division when the LS method is used.

[0020]

SUMMARY OF THE INVENTION An array antenna control apparatus according to the present invention comprises an FF for dividing a received signal on a frequency axis.
Band dividing filter means by T, and band restoring filter means for restoring a signal on the time axis after multiplying a signal of each frequency band divided by the means by a weight coefficient adaptively controlled by the RLS method And the frequency band division filter means is configured to change the overlap in the movement of the FFT window on the time axis.

In the configuration of an array antenna which performs band division signal processing based on perfect reconstruction filters of FFT and inverse FFT using the RLS method as a weight coefficient control algorithm,
The weight coefficient update speed can be reduced by applying the FFT. However, since it takes time until the degree of freedom is satisfied by being able to reduce, the time until the degree of freedom is satisfied does not reduce the weight coefficient, and the degree of freedom is satisfied early,
Thereafter, the weight coefficient update speed is reduced.

More specifically, the operation of the buffer is overlapped as shown in FIG. 2b until the degree of freedom is satisfied, and the sample signal is fetched and the sample is supplied to the RLS processor at a high speed to satisfy the degree of freedom. When the degree of freedom is satisfied in this way, the operation of the buffer is switched to the operation of supplying the FFT with windowing so as not to overlap as shown in FIG. 3B, and the operation speed of the RLS processor is set to the order of the FFT (the number of subbands). Reduce by a factor of one.

Further, the array antenna control apparatus according to the present invention comprises: an analog / digital signal conversion means for converting an element antenna output from an analog signal to a digital signal;
A signal orthogonalizing means for separating a signal output from the analog / digital signal converting means into an orthogonal signal, and a receiving signal for each element antenna obtained by the spatial analog / digital signal converting means to a subsequent band division FFT Signal-vectoring means for applying a window for performing variable overlap, and a band division filter for frequency-dividing a signal in which each element antenna output obtained by the signal vectorization means is vectorized into sub-bands in a frequency domain, The FFT
Weighting factor multiplying means for multiplying a predetermined number of beam outputs selected by the calculating means with a weighting factor supplied from a weighting factor calculating means to be described later; and each beam output multiplied by the weighting factor by the weighting factor multiplying means. An adder for summing for each of the same sub-bands, a signal reconstruction filter for processing the output for each of the sub-bands summed by the adder to restore a person and a band, And a weighting factor calculating means for adaptively calculating a weighting factor using the signal in (1) and supplying the weighting factor to the weighting factor multiplier.

[0024]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 6 is a block diagram showing an embodiment of an array antenna control device according to the present invention. FIG. 6 illustrates an array antenna including a plurality of antenna elements 1-1 to 1-n and a control device for controlling the array antenna.

The signals received by the antenna elements 1-1 to 1-n are converted into digital signals at a predetermined sampling frequency by A / D conversion and orthogonalization means 2, and are separated into I / Q orthogonal channels. Are input to the input signal control means 3-1 to 3-n. In the input signal control means 3-1 to 3-n, based on information obtained from the input / output control means 9 described later, in the time until the degree of freedom in the RLS method is satisfied, the FFT in the subsequent
The signals of the calculation means 4-1 to 4-n are buffered so that the samples overlap on the time axis, and the signals are supplied to the FFT calculation means.

FIGS. 7A and 7B are explanatory diagrams showing the operation of the input signal control means in the embodiment of FIG. This figure illustrates the operation of the input signal control means by taking the case of four band divisions as an example.

FIG. 7A shows input signal control means 3 from the start of operation to the reception of a command from input / output control means 9.
It is explanatory drawing showing operation | movement of 1-3 -n. That is, the samples buffered at each sample timing are transferred to the FFT calculation means 4-1 to 4-n at the subsequent stage.

FIG. 7B is an explanatory diagram showing the operation of the input signal control means 3-1 to 3-n after receiving an instruction from the input / output control means 9. That is, only at the timing when the input samples are completely replaced, the FFT calculation means 4-1
Pass sample value to ~ 4-n. As a result, the value is FF
The frequency of handing over to the T calculating means 4-1 to 4-n is reduced to 1 / band division number.

The FFT calculating means 4-1 to 4-n, to which the sample values are supplied from the input signal control means 3-1 to 3-n at the timing as described above, execute the FFT operation upon receiving the values. Output. Here, the order of FFT, that is, the number of band divisions is
m. The outputs of the FFT calculation means 4-1 to 4-n are multipliers
The signal is multiplied by a weighting factor supplied from 5-1-1 to 5-mn and supplied from a weighting factor calculation means 10 based on the RLS method described later. It is added for each band multiplication result. The output of each adder is IF
It is supplied to FT (inverse FFT) calculation means 7. The IFFT
The result of the calculation means 7 is input to the output signal control means 8,
Upon receiving a command from the input / output control means 9 described later, two kinds of operations are performed in synchronization with the operations of the input signal control means 3-1 to 3-n. That is, while the operation of the input signal control means is performing the FFT at each sample timing, the output signal control means takes in the output of the first port of the IFFT at each sample timing and makes it an output signal. Therefore, IFFT
Ignore the output of other ports. Next, the operation of the input signal control means determines that the FF
In the state where the sample value is supplied to the T calculation means, the output signal control means adopts the entire output of the IFFT and adopts it as a signal on the time axis as the output signal.

As described above, the input signal control means 3-1 to 3-n
The input and output control means 9 controls the two types of operations of the output signal control means 8 and the output signal control means 8 to operate in synchronism. Basically, the two types of operation are switched by RLS
At the timing when the freedom of law is satisfied,
Switching may be performed at a later timing.

The weight coefficient calculating means (RLS method) 10 receives the outputs of the adders 6-1 to 6-m and the reference signal 11 divided into corresponding subbands, and the square error is calculated. Multiplier for calculating minimum weighting coefficient based on RLS method
Supply to 5-1-1 to 5-mn. The weight coefficient calculating means 10 also
An instruction is received from the input / output calculation means 9, and the RLS method is executed at a speed equal to the operation speed of the FFT calculation means. Also, the band-divided reference signal 11 receives an instruction from the input / output control device 9 and applies a method of applying FFT to perform band division.
The reference signal is band-divided by performing the same operation as the T calculation means 3-1 to 3-n.

Various changes, modifications, and omissions of the technical idea and scope of the present invention can be easily made by those skilled in the art with respect to the embodiment described in detail above. Therefore, the above-described embodiment is merely an example, and does not intend to limit the present invention. The invention is limited only as defined by the appended claims and equivalents thereof.

[0034]

FIGS. 8A to 8D are graphs of correlation coefficients between an output signal calculated from weighting coefficients obtained in each simulation step and a reference signal in the embodiment of the present invention. Samples from 1 to 4 times 32 steps satisfying the degree of freedom of the RLS method are subjected to FF by the operation of FIG.
7B shows the transition of the output signal of the array antenna calculated by using the weighting factor obtained in each step and the correlation coefficient of the reference signal when the operation is switched to the operation shown in FIG.

According to the present invention, as is clear from FIGS. 8A to 8D, the deterioration of the correlation coefficient in 256 steps is not seen as shown in FIG. 5, and the convergence characteristic is improved. .

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an adaptive array antenna that performs time domain signal processing using TDL.

FIG. 2A is a configuration diagram of an adaptive array antenna that performs frequency domain signal processing using an FFT as a band division filter.

FIG. 2B is an explanatory diagram showing a buffer operation for a sample on a time axis in the configuration diagram of FIG. 2A.

FIG. 3a is a configuration diagram for reducing a weight coefficient update speed in an adaptive array antenna that performs frequency domain signal processing using an FFT as a band division filter.

FIG. 3B is an explanatory diagram showing a buffer operation for a sample on a time axis in the configuration diagram of FIG. 3A.

FIG. 4A is an explanatory diagram of an embodiment to which the high-speed RLS method can be applied, in which a signal is shifted in a tap;

FIG. 4B is an explanatory diagram of an embodiment to which the high-speed RLS method cannot be applied, and is a diagram of a configuration in which a signal changes simultaneously in a tap.

FIG. 5 is a graph showing a correlation coefficient between an output signal of an adaptive array antenna that performs frequency domain signal processing and a reference signal.

FIG. 6 is a configuration diagram illustrating an embodiment of an array antenna control device according to the present invention.

FIG. 7A is an explanatory diagram showing an operation of transferring a value to the FFT calculating means at each sample timing (an example of band division into four) until the degree of freedom of the input signal control means is satisfied in the configuration of FIG. 6; .

FIG. 7B shows an operation in which, after the degree of freedom of the input signal control means is satisfied, a value is passed to the FFT calculation means at a timing several times the sample timing in the configuration of FIG. 6 (an example of dividing into four bands). FIG.

FIG. 8A is a graph showing a correlation coefficient between an output signal and a reference signal calculated from a weight coefficient obtained in each simulation step in the embodiment according to the present invention, wherein a buffer operation is switched by eight samples.

FIG. 8B is a graph showing a correlation coefficient between the output signal and the reference signal calculated from the weighting coefficients obtained in each simulation step in the embodiment according to the present invention, wherein the buffer operation is switched by 16 samples.

FIG. 8c is a graph showing a correlation coefficient between an output signal calculated from weighting coefficients obtained in each simulation step and a reference signal in the embodiment according to the present invention, in which a buffer operation is switched by 32 samples.

FIG. 8D is a graph showing the correlation coefficient between the output signal and the reference signal calculated from the weight coefficient obtained in each simulation step in the embodiment according to the present invention, wherein the buffer operation is switched at 64 samples.

[Explanation of symbols]

 1-m array antenna 2 A / D and signal orthogonalization means 3-n input signal control means 4-n FFT calculation means (m order) 5-mn multiplier 6-m adder 7 inverse FFT calculation means 8 output signal control Means 9 Input / output signal control means 10 Weight coefficient calculation means (RLS method) 11 Band-divided reference signal

Claims (2)

[Claims] 1. An array antenna control apparatus for adaptively controlling a weight coefficient of an element antenna, comprising: a band division filter means by FFT for dividing a received signal on a frequency axis; A band restoring filter unit for multiplying a frequency band signal by a weighting coefficient adaptively controlled by an RLS method and restoring the signal on a time axis, wherein the frequency band division filter unit Is configured to change the overlap in the movement of the FFT window on the time axis. 2. An analog / digital signal converter for converting an element antenna output from an analog signal to a digital signal, a signal orthogonalizer for separating a signal output from the analog / digital signal converter into a quadrature signal, A signal vectorizing means for applying a window for variably overlapping a window for supplying each element antenna reception signal obtained by the spatial analog / digital signal conversion means to a subsequent band division FFT; A band division filter that performs frequency division of the signal in which the element antenna output is vectorized into sub-bands in the frequency domain, and a predetermined number of beam outputs selected by the FFT calculation unit are supplied from a weight coefficient calculation unit described later. Weighting factor multiplying means for multiplying the weighting factor by An adder that sums each beam output multiplied by the weight coefficient for each subband, a signal reconstruction filter that processes the output for each subband summed by the adder to restore a person and a band, Weight coefficient calculating means for adaptively calculating a weight coefficient using a signal at an arbitrary point in the configuration and supplying the weight coefficient to the weight coefficient multiplier. Control device.