Adaptive Processor For An Rf Antenna

Abstract

The elements of an RF antenna array are individually connected to mixer circuits to produce a plurality of distinct IF signals. Each signal is fed to a multi-tapped delay line which makes available an increasingly delayed IF signal at subsequent points therealong. The delayed signals are independently converted from an analog to digital form so that they may be accepted by subfrequency sampling input lines of a digital computer. A pilot signal can also communicate with the computer or be generated internally to produce a control signal for varying weights that control the amplitudes of the IF signals transmitted from the tapped delay line. The weighted signals are fed to a summing circuit which sums these signals to form a system output. The system output is employed for achieving null steering of the array by controlling system gain in the "look direction" and rejecting noise.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to the elimination of interference in an RF antenna array. More particularly, the invention is directed to utilization of a digital computer in an adaptive processor that controls the adaptation rate at a subfrequency below RF.

2. The Prior Art

Adaptive null steering methods for application to low frequency signals are well known in the art. Typically, applications include seismic and sonar information. The basic method includes an antenna array with a transversal filter associated with each element of the array. A transversal filter performs amplitude weighting of element information that has been converted to the IF frequency band. By adjusting the filter weights, desired interference rejection can be obtained thereby maximizing the signal to noise ratio of an antenna system. The basic problem with prior art systems for RF application is that processing of signals is accomplished by analog techniques. Although several authors have recognized the potential of using a digital computer for adaptive processing, there has been a failure by those skilled in the art to design a system where processing can be accomplished by subfrequency sampling at frequencies lower than RF, the sampling rate being computer controlled.

SUMMARY OF THE INVENTION

With the presently disclosed invention, the adaptation rate for processing is independent of the signal frequencies and can be at any arbitrarily slow rate, a fact which allows digital implementation for RF signals. Thus, with the present invention, subfrequency sampling can be used to perform adaptive processing of antenna RF signals. Adaptive null steering for RF systems using digital adaptation offers flexibility not herebefore obtainable with an analog implementation.

Accordingly, a primary object of the present invention is to achieve subfrequency sampling of antenna RF systems by an adaptation rate much slower than the RF signal frequencies thus permitting the use of a digital computer in the signal processing loop.

A further object of the invention is to realize software generation of a pilot signal that is employed in the computation of an output control signal, derived from the computer, which sets the weight of filter constraints in a processing loop. This software generation and insertion into an array model allows the system to be trained in the desired look direction while achieving interference rejection. This approach does not disturb the actual system output as would be the case of hardware insertion of a pilot signal. Thus, there is provided a workable solution when the target frequency estimate is in error. Such errors are likely in practice, especially if the target is uncooperative.

A still further object of the invention is to ahcieve proper control of the pilot signal amplitude and/or formation of an initial beam rather than an initial isotropic pattern so that the array gain can be constrained in the desired look direction, after adaptation, to be close to the theoretical maximum.

The above-mentioned objects and advantages of the present invention will be more clearly understood when considered in conjunction with the accompanying drawings, in which:

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of the present system including a digital computer serving as data processor for the system.

FIG. 2 - 5 illustrate array patterns for elements having various filter configurations connected between the elements and a summing circuit.

FIG. 6 is a block diagram of a basic adaptive element configuration.

FIG. 7 is a block diagram of an adaptive element configuration using a tapped delay line structure.

FIG. 8 is a block diagram of an adaptive element configuration having automatic weight adjusting means.

PREFERRED EMBODIMENTS OF THE INVENTION

At the outset, it is appropriate to define what is meant by adaptive signal processing in the context of the present invention. An adaptive signal processor is a system that continually adjusts its own parameters in the course of time to meet a particular performance criterion. The present invention is most concerned with the performance criterion of maximization of signal-to-interference ratio. Inasmuch as interference is normally time-varying, some adaptive mechanism is required to continually update filter weights, employed in the present system, to meet the changing interference pattern. In essence, the present system senses the interference and automatically adjusts these filter weights to their proper value through feedback loops.

The basic principles of design for an adaptive antenna system are disclosed in an article by Widrow, et al. in the Proceedings of the IEEE, vol. 55, No. 12, Dec., 1967. In the basic system disclosed in this article, a scheme is presented for obtaining desired array processing improvement in real time. The performance criterion used is minimum mean square error. Statistics regarding the information signal are assumed but there is no requirement of prior knowledge or direct measurement of the noise field. The system is automatically adjusted to perform an adaptive function according to a simple iterative algorithm, and the procedure does not directly involve the computation of any correlation coefficients or the inversion of matrices, as was previously required by other systems. The system of Widrow makes use of antenna input signals just once, as they occur, in the adaptation process. Accordingly, no need exists to store past input data; but there is a need to store the processor adjustment values referred hereinafter as weights.

Reference is made to FIGS. 2 - 6 to demonstrate this basic system. Referring particularly to FIG. 2 an example of a linear-array receiving antenna is generally denoted by reference numeral 10. The antenna consists of seven isotropic elements 12. These elements are spaced along a straight line at equal distances of one-half wave length where the wave length is a center frequency of the array. The received signals are summed through equal length transmission lines by an appropriate circuit 14 to produce an array output signal 16. The directivity pattern 18 illustrates the relative sensitivity of response to signals from various directions, and the pattern illustrated in the figure is plotted in a plane over an angular range of 180.degree.. It will be noticed that the pattern is symmetric about the vertical zero angle line 20. The main lobe 22 of the pattern is centered about this vertical line. The largest amplitude side lobe 24 has a maximum sensitivity which is typically 12.5 dB below the maximum main-lobe sensitivity. The pattern would be different if it were plotted for a different frequency.

A similar array configuration is illustrated in FIG. 3. This configuration is generally indicated by reference numeral 26 and as will be noted, the output of each element 12 is delayed by an associated suitable delay 28 which, for example, may be a strip of transmission cable. Each delay is series connected between an element and a common summing circuit 14. As a result, the directivity pattern now has its main lobe 32 at an angle indicated by 30.

The sensitivity of this configuration is maximum at the angle indicated by 30 because signals received from a plane wave source incident at this angle, and delayed by delays 28, are in phase with one another and produce the maximum output signal.

Many different types of configuration for phased arrays exist. Referring to FIG. 4, one such configuration 34 is shown where each of the antenna element outputs is weighted by two weights 36 and 38 that are of equal magnitude.

The weight 38 of each element is preceded by a time delay of a quarter cycle at the center frequency, which is equivalent to a 90.degree. phase shift. The output signal 40 is the sum of all the weighted signals, and inasmuch as all weights are of equal magnitude, the directivity pattern at the center frequency is the same as that shown in FIG. 2 because of symmetry.

With continued reference to FIG. 4, an interferring sinusoidal "noise" at the center frequency is indicated by 42. The angle of incidence of this noise is such that it would be received on one of the side lobes of the directivity pattern with a characteristic sensitivity of only 17 dB less than that of the main lobe at an angle of 0.degree..

The configuration shown in FIG. 5 is generally indicated by reference numeral 44. In this configuration, the weights 36 and 38 associated with each element are of different value. Thus, all the weights shown in configuration 44 have different values. The resultant directivity pattern at the center frequency becomes as shown in the figure. As will be noted, the main lobe 46 is almost unchanged from that shown in FIG. 2 and FIG. 4. However, the side lobe 45 (FIG. 4) which previously intercepted the sinusoidal noise has been shifted so that a null is now positioned in the direction of that noise. Typically, the sensitivity in the noise direction is below the main lobe sensitivity, improving noise rejection by 20 - 30 dB when compared with the sensitivity associated with the configuration of FIG. 4.

To illustrate the existence and calculation of a set of weights that will cause noise rejection as just discussed, consider the illustration of FIG. 6. The signal arriving from the desired direction 48 at 0.degree. will be referred to as the "pilot" signal, while the "noise" is incident to the array along direction 50. Both the pilot signal and the noise signal are assumed, for this example, to be at the same frequency. At a spatial point midway between the antenna array elements 12, the signal and the noise are assumed to be in phase. In the example illustrated, two identical omnidirectional array elements are spaced one-half wave length apart. The signals received by the left element 12 are fed to two variable weights 52 and 54, the weight 54 being preceded by a series connected quarter wave length delay. A similar structure exists between the right element 12 and weights 60, 62. As a result of the configuration shown in FIG. 6, the four weighted signals are summed by a summing circuit 56 to form the array output signal at 58. The determination of a set of weights 52, 54, 60 and 62 which will accept the pilot signal along direction 48 and reject the noise signal along direction 50 can be solved in accordance with the mathematical expressions presented in the aforementioned article by Widrow, et al. Suffice it to say that the solution to this problem presents a set of weights so that the array will have the desired properties, in that it will accept a signal from the desired direction 48 while rejecting noise from direction 50, even if the noise is at the same frequency as the pilot signal due to the fact that the noise comes from a different direction than does the signal.

The aforementioned discussion serves as a proper explanation of the principles involved in the present invention. However, the calculation of weights by this method is impractical. This is due to the fact that the method is only useable when there are a small number of extremely directional noise sources, when the noises are at one frequency, and when the directions of the noises are predetermined. In a practical process, it should not be required to have detailed information regarding the nature and number of the noise sources.

Before discussing more practical methods of adaptive filtering and signal processing to be used in an adaptive array, the configuration shown in FIG. 7 will be discussed. By way of example, two identical system sections 64 and 66 are illustrated, although as will be appreciated, a greater number will usually be used. Referring to section 64, a plurality of delays 68 are series connected with an associated antenna element 70. Individual weights 72 are connected at the junction between adjacent delays and the input of a summing circuit 74. In a similar manner, a weight 76 is connected between the antenna 70 and the summing circuit 74. Likewise, a final weight 78 is connected between the output of the last delay and the summing circuit 74. The various delays 68 may be replaced by a tapped delay line which is discussed hereinafter with particular reference to the present invention. The tapped delay line permits adjustment of gain and phase as desired at a number of frequencies over the band of interest. If the tap spacing is sufficiently close, this network approximates an ideal filter which would allow complete control of the gain and phase at each frequency in the passband.

With continued reference to FIG. 7, the next step in the consideration of an adaptive signal processor is to develop an adaptation procedure which can be used to automatically adjust the multiplying weights 72, 76 and 78 to achieve the desired spacial and frequency filtering. The procedure should produce a given array gain in the specified look direction while simultaneously nulling out interfering noise sources.

FIG. 8 illustrates a basic adaptive element generally denoted by 80 which cooperates with the configuration of FIG. 7. The indicated input terminals to this element are connected to respective junction points at the input of the multiplying weights shown in FIG. 7. Thus, the indicated multiplying weights 82 and 84 in FIG. 8 correspond to two of the multiplying weights shown in FIG. 7. The dots 86 and 88 serve to indicate where the remaining multiplying weights of FIG. 7 would be inserted in the element 80. As indicated in FIG. 8, the element operates upon the input signals with appropriate automatic circuits 90 that accomplish weight adjustment. This adjustment is indicated by control lines 92 and 94. The summing circuit 74 generates an output along line 96. A shunt lead 98 applies a portion of the output from line 96 to a second summing circuit 100. A desired response signal is applied to a second input 102 of the summing circuit 100 to generate an error signal 104 which is in turn fed back to the automatic circuits 90.

Accordingly, the adaptive element shown in FIG. 8 constitutes a closed loop feedback system which operates upon the automatic circuits 90 in a manner adjusting the multiplying weights to produce a given array gain in a specified look direction while simultaneously nulling out interfering noise sources.

In order that adaptation or weight changing take place, a "desired response" signal must be applied to the element 80. A method for obtaining this signal is discussed in the Widrow et al article, supra. Suffice it to say at this point, that the variable weights of the element can automatically be adjusted by an adaptive technique based on the least-mean-square (LMS) algorithm. However, it must be stressed that other algorithms may be employed with different results as to accuracy and computational time requirements.

FIG. 1 shows the adaptive system which forms the present invention. The desired response signal is provided through the use of an artifically injected signal, referred to as a "pilot" signal, which is completely known at the receiver because it can be generated by computer software at the receiver. The pilot signal is constructed to have spectral and directional characteristics similar to those of the incoming signal of interest. These characteristics generally represent estimates of the parameters of the signal of interest. Adaptation with the pilot signal will cause reduced array response (nulling) in the direction of directional noises impinging upon the array.

With continued reference to FIG. 1, a system block diagram of the present invention is shown. In essence, the invention is a refinement and improvement over the basic systems aforementioned. The principal objective is to include digital adaptation for RF applications. The actual frequency range for which the present invention is suited does not have restrictive limitations because adaptation occurs at IF. A plurality of identical, spaced antenna elements 106, 108 and 110 are connected to respective RF circuits. For purposes of convenience, only the circuitry connected with antenna element 106 will be discussed. The antenna element 106 feeds an RF amplifier 112. The amplifier drives an input to mixer 114, which can be any conventional first detector. A second input 116 is driven by a local oscillator 118 to produce a signal of intermediate frequency (IF) along lead 120. This signal is amplified by the IF amplifier 122 then is passed through the transverse filter 124 that has its output appearing along lead 126. As indicated in the figure, IF signals at 128 and 130 are associated with antenna 108 and 110, respectively. The IF signals at 128 and 130 drive their respective transverse filters (not shown) and the outputs of these respective filters drive a summing circuit 132 along with the driving input of the IF signal at 126. Although three antenna elements 106, 108 and 110 have been shown, the points 111 represent the location of other elements in the array.

Considering the structure of each transverse filter, reference is made to transverse filter 124 which includes a tapped delay line 134 similar to the interconnected delays 68 shown in FIG. 7 and discussed hereinbefore. The IF signal from the IF amplifier 122 is fed to the tapped delay line at input terminal 136. The outputs from the tapped delay line 134 are separately connected to sample-and-hold circuits 138. However, for convenience, only the first and last output terminals 142 and 144 are shown. To each output of the delay line 134 is connected a conventional amplitude weighting circuit. In the case of delay line output 142, the amplitude weighting circuit is denoted 146 while the delay line output 144 is connected to a similar amplitude weighting circuit indicated by 148. All the outputs from the amplitude weighting circuits are connected in parallel with a conventional summing circuit 150. The output from the summing circuit 150 carries the summed signal.

Up to this point, we have dealt only with analog signals. Therefore, if digital implementation is to take place, an analog to digital conversion must be made. This conversion is implemented by an A/D device of known design.

A digital processor (computer) 154 makes the adaptive processing for RF antenna signals possible. Tap point samples from the A/D converter 152 are made available at input terminals 156, 158 and 160. However, a greater number of tap off sample lines can be present as indicated by the dots 161 adjacent the designated tap points. Tap point samples can be delivered sequentially or in parallel to the digital processor 154. It is to be stressed that the adaptation rate is controlled by the processor speed, not by the RF frequencies.

Other inputs required by the processor are some knowledge of the target signal, such as its center frequency and direction, and possibly samples from a pilot signal generator 162. However, as discussed hereinafter in greater detail, the pilot signal can also be software generated. Communication between the pilot signal generator and the computer 154 is indicated by lines 164 and 166. The computer 154 governs timing control and generates control signals for varying the amplitude weighting circuits 146 and 148 so as to achieve desired interference rejection. As will be appreciated, the system of FIG. 1 is a hybrid system, inasmuch as the information path is analog, but the adaptation is performed digitally. The number of weights per filter must be determined by the particular application. However, interference rejection is achieved by iteratively adjusting the weight according to a gradient search criterion that is built into the computer software. At each iteration, the adaptive processor receives samples of all the tap point signals, then makes the necessary computations according to preselected algorithms, and commands the adjustment of the filter weights 146, 148. The sample and hold circuits 138, at all tap points in the systems, are strobed essentially simultaneously. Thus, at each iteration, the processor "sees" more than a single instantaneous sample of the RF environment. It must be stressed once again that the actual sampling rate is controlled by the processor speed, and not by the incoming RF signal frequencies.

To complete the closed feedback loop of each transverse filter 124, control signals are generated at output lines 168 and 170. For brevity, only the closed loop associated with output line 168 will be discussed. The control signal appearing at line 168 drives a conventional weight control circuit 172 which produces a command signal along output line 174 forming a second input to the amplitude weighting circuit 146. The command signal causes the amplitude weighting circuit to change or adapt. In this manner, a closed feedback loop is created with the digital computer 154 forming a very important part of the loop.

Concentrating on the matter of pilot signal generation at 162, a distinct advantage occurs if the generation is by software. The advantage arises in the case when the estimate of the target center frequency is in error, or the target has no appreciable carrier. This situation requires artifically inserting the pilot signal into the system. Such a manipulation is easily accomplished in the software itself. However, this would be very difficult to accomplish with hardware, and would perturb the array output.

To clarify the meaning of software insertion of a pilot signal, consider the following: Having knowledge of the array geometry and pilot frequency, as well as the direction of arrival, the relative phase of the pilot at each antenna element may be computed as though the pilot were actually a signal incident to the array. Then at each sampling instance, these relative phase angles can be used to compute pilot samples which would appear at each tap point in the system if the pilot signal were part of the environment. These pilot samples are linearly added to the actual tap point signals in the software to accomplish artifical insertion. This insertion constrains the array to have a main beam in the pilot direction and frequency since the pilot is considered the desired signal when the error signal is formed. Any signals present in the environment which differ substantially from the pilot in frequency and/or direction will be "nulled" by the adaptive system. The amount of difference allowable is dependent upon the physical configuration of the array. The computations at each sampling instant can be minimized by using a table look-up to obtain the required pilot samples. Pre-computed address differences for the various tap points are used to obtain the appropriate pilot samples for each point.

The block diagram of FIG. 1 illustrates an A/D converter at each tap point as indicated by 156. This allows all the tap point samples to be delivered to the computer in parallel. One alternative would be use of only one A/D converter to service all the sample and hold circuits and thus sequentially feed the data to the computer.

Interference rejection will occur even after the interference environment has been assumed unknown. Any prior knowledge of the interference could be used by computer 154 to improve rejection.

It should be borne in mind that the same techniques as discussed hereinbefore can also provide null steering for a transmitting array. Both of these are achieved by software insertion of the nulling signal, nulling the transmitted signal noise in frequency and direction.

It should be understood that the invention is not limited to the exact details of construction shown and described herein for obvious modifications will occur to persons skilled in the art.

Claims

What is claimed is:

1. An adaptive processor for an antenna array comprising input means for transmitting antenna signals between the array and the processor, frequency translating means for translating the signals to information in a subfrequency band, adjustable filter means for filtering noise from the information, means for converting unfiltered information to digital data and digital computer means responsive to the data for generating control signals which adjust the filter means to discriminate against noise whereby a least mean square error of the filtered information is realized.

2. The circuitry set forth in claim 1 wherein the array includes a plurality of antenna elements, and further wherein the adjustable filter means comprises tapped delay means connected to an information path associated with each element, tap off points accessible at each delay means, adjustable weighting means controlled by the controlled signals and respectively connected to receive an information signal from each tap off point for adjusting the amplitude of the respective information signal, and combiner means receiving weighted information signals from said weighting means for combining the signals to produce a filtered system output signal.

3. The circuitry defined in claim 2 wherein the means for converting the unfiltered information to digital data comprises an analog to digital converter and means for connecting the converter to sampled input terminals of the computer means.

4. The circuitry of claim 3 wherein the computer means includes means for generating a software pilot signal having a frequency and representing a direction, said pilot signal being combined by the computer means with samples of data received at the input terminals of the computer means to effect artifical insertion of the pilot signal, and wherein said computer means further includes means for constraining the array to have a main beam in the pilot signal direction at a frequency corresponding to the pilot signal frequency.

5. The circuitry of claim 4 wherein the computer means generates the control signals in response to addition to the pilot signal to the data samples, and wherein the circuitry further comprises means for enabling the control signals to adjust the state of the weighting means at an adjustment rate governed by the computer means.

6. The circuitry defined in claim 5 wherein a plurality of the weighting means are associated with each antenna element, the plurality having output terminals connected to a first summing means, and further wherein output terminals from all first summing means associated with respective elements are connected to a second summing means to generate a system output.

7. An antenna array adaptive processor for improving signal-to-noise ratio in an unknown electromagnetic environment comprising analog means for carrying a plurality of antenna element signals along a plurality of preselected paths, filter means interposed along each path for minimizing the mean square error of the array signal as compared with a noiseless signal in a selected look direction, and digital means responsive to the array signal for adjusting the filter means in a manner effecting the minimizing of the mean square error, said filter means comprising means for delaying each antenna element signal a plurality of amounts to produce a like plurality of delayed signals, amplitude adjustment means respectively connected for adjusting the amplitude of each delayed signal in response to said digital means and thereby adjusting said filter means, and means for combining the outputs of said amplitude adjustment means to produce the array signal.

8. The circuitry of claim 7 wherein means are provided along the path for translating the frequency of the signal at array elements to a subfrequency for permitting the digital means to operate effectively at a subfrequency.

9. The circuitry set forth in claim 8 wherein the digital means include means for converting the translated signal to a digital form which is sampled at a subfrequency rate by the digital means, the rate being governed by the digital means.

10. The circuitry as set forth in claim 9 wherein a pilot signal is generated by the digital means, the digital means further including means for linearly adding the pilot signal to the sampled signal to constrain the array to have a main beam in the pilot direction and at the pilot frequency.

11. An adaptive processor for an array of a plurality of element antennas having a first electrical signal at each element antenna, comprising a like plurality of input circuits respectively receiving said first electrical signals and outputting respective second signals accordingly; a like plurality of delay means each receiving a respective one of said second signals as an input, each of said delay means including circuit means having a plurality of delay outputs for producing at each of said delay outputs a differently delayed signal; a plurality of weighting means each receiving a respective one of said delayed signals from a respective one of said delay outputs for producing a weighted signal, each of said weighting means being individually adjustable by a control signal to change the amplitude of said weighted signal compared with the amplitude of its corresponding delayed signal; summing mean for adding together the weighted signals from the weighting means to produce a final output signal; and control means for producing said control signals including digital computer means for independently simulating the effects upon said delayed signals due to said weighting means and due to said summing means and producing affected data accordingly, an input of said control means being connected with said delay outputs of said delay means for receiving input data depending upon said delayed signals.

12. An adaptive processor as defined in claim 11 and wherein each of said circuit means comprises a plurality of delay devices connected together in cascade for accumulating delays, the outputs of each of said delay devices comprising said delay outputs.

13. An adaptive processor as defined in claim 11 and wherein said affected data comprises data simulating said final output data and said control means further comprises means for generating pilot signal data representing the frequency and direction of a desired signal that is expected to be received by the array, and wherein said digital computer means comprises means utilizing said pilot signal data and the simulated output data for computing and producing said control signals.

14. An adaptive processor as defined in claim 13 and wherein said means utilizing said pilot signal data comprises computation means for computing values of said control signals to achieve least mean square error between said pilot signal data and the simulated output data.

15. An adaptive processor as defined in claim 13 and wherein said means for generating pilot signal data comprises a digital computer and software.

16. An adaptive processor as defined in claim 13 and wherein each of said input circuits comprises a frequency down-converter for producing said second signals at a lower frequency than said first signals in response to said first signals.

17. An adaptive processor for an array of a plurality of element antennas having a first electrical signal at each element antenna, comprising a like plurality of input circuits respectively receiving said first electrical signals and outputting respective second signals accordingly; a like plurality of delay means each receiving a respective one of said second signals as an input, each of said delay means including circuit means having a plurality of delay outputs for producing at each of said delay outputs a differently delayed signal; a plurality of weighting means each receiving a respective one of said delayed signals from a respective one of said delay outputs for producing a weighted signal, each of said weighting means being individually adjustable by a control signal to establish an amplitude of said weighted signal relative to amplitude of its corresponding delayed signal; summing means for adding together the weighted signals from the weighting means to produce a final output signal; and control means receiving data at its input depending upon said second signals for producing said control signals including digital means for receiving and storing software data specifying a direction from which a desired signal is to be received by the array, said digital computer means further comprising algorithm computation means for establishing values of said control signals to produce a final output signal favoring signals received from said direction and discriminating against signals received from other directions.

18. An adaptive processor as defined in claim 17 and wherein said digital means further comprises means for receiving and storing software data specifying a frequency of said desired signal to be received by the array.

19. An adaptive processor as defined in claim 17 and wherein said algorithm computation means comprises least mean square computation means for establishing values of said control signals to produce a final output signal whose mean square error for signals received from said direction is minimum.

20. An adaptive processor as defined in claim 17 and wherein each of said input circuits comprises a frequency down-converter for producing said second signals at a lower frequency than said first signals in response to said first signals.

21. An adaptive processor as defined in claim 17 and wherein said control means is connected with said delay outputs of said delay means for receiving input data depending upon said delayed signals.

22. An adaptive processor as defined in claim 17 and wherein said control means further includes digital computation means for independently simulating the effects upon said delayed signals due to said weighting means and due to said summing means and producing data simulating said final output signal accordingly.