Frequency Domain Automatic Equalizer Having Logic Circuitry

1. Field of the Invention

The invention pertains to frequency domain automatic equalization for electrical signals used in transmission of information.

2. Description of the Prior Art

Ideally, it is desirable to transmit electrical signals such that no interference occurs between successive symbols. In practice, however, transmission channels are bandlimited and interference is controlled utilizing clocked systems with equalization conventionally performed in the time domain.

Most conventional automatic equalizers operate in a feedback mode so that the effects of changes in the equalizer transfer function are monitored and used to produce further changes in the transfer function to obtain the best output signals. In such systems, the measurements of the output signal are made in the time domain. Typically, the transfer function may be constructed in the time domain by adjusting the tap gains of a tapped delay line during an initial training period prior to actual message transmission. Examples of such systems are shown in U.S. Pat. Nos. 3,375,473 and 3,292,110.

Frequency domain equalization utilizing time domain adjustments are shown, for example, in U.S. Pat. No. 3,614,673, issued to George Su Kang. Kang utilizes frequency domain measurement and calculations to produce the time domain impulse response of a transversal filter. The impulse response of the transversal filter is applied to set the weights of the transversal filter.

Other approaches to frequency domain equalization require transmission of the discrete Fourier transform of the source signal and require the use of complex analogue circuitry in obtaining an approximation to the desired equalization. See, for example, Weinstein and Ebert, "Transmission by Frequency -- Division Multiplexing Using the Discrete Fourier Transform", IEEE Transactions on Communication Technology, Vol. COM-19, No. 5, October, 1971, pp. 628-634.

The principal object the invention is to provide an automatic equalizer using successive approximations and operable completely in the frequency domain to provide both frequency domain measurements of the transmitted signal and frequency domain corrections to equalize channel amplitude and phase distortion.

Another object of the invention is to provide a frequency domain equalizer employing the discrete Fourier transform and a frequency correction feedback circuit which is responsive to the algebraic sign of the spectral components of the received signal and the algebraic sign of an internally generated error signal.

It is a further object of the invention to provide an alternate approach to generating equalization coefficients in a frequency domain equalizer, as disclosed, for example, in applicant's copending application for "Frequency Domain Automatic Equalizer Utilizing the Discrete Fourier Transform", Ser. No. 706,703 filed July 19, 1976, assigned to the assignee hereof.

Yet, another object of the invention is to utilize time-shared circuitry for minimizing circuit complexity and expense.

Another object of the invention is to provide a frequency domain equalizer having independently adjustable spectral coefficients so that adjustment of one coefficient does not necessitate adjustment of other coefficients.

A frequency domain equalizer of the present invention automatically equalizes the spectral components of a received signal transmitted through a distortion affected transmission channel and comprises multiplying means and feedback control means. The multiplying means performs a complex multiplication of a complex input signal X and a complex multiplicand correction signal C, thereby generating a complex product signal Y. The input signal X has a first factor, D, corresponding to a distortion transfer function of the transmission channel, and a second factor, H, corresponding to an undistorted transfer function of the transmission channel. The feedback control means receives the complex product signal Y and generates a value of the correction signal C, which is inversely proportional to the first factor D of the multiplier input signal X. Further, the feedback control means comprises a reference means for generating a complex reference signal proportional to the second factor, H, of the input signal X, a means for comparing the product signal Y with the reference signal, a means for generating error polarity signals having an algebraic sign corresponding to the difference between the compared product signal Y and the reference signal, and a means for adjusting the value of the correction signal C to produce a value for the product signal Y approximately equal to the reference signal. The adjusting means is responsive to the algebraic sign of the error polarity signals and the algebraic sign of real and imaginary components of the input signal X.

These and other features and advantages of the invention will become apparent when taken in conjunction with the following specification and drawings wherein:

FIG. 1 is a block diagram of the overall theoretical model used in the instant invention;

FIG. 2 is a schematic diagram of an analog circuit for performing the discrete Fourier transform of a sample set;

FIG. 3A illustrates a tree graph for the inverse discrete Fourier transform;

FIG. 3B illustrates a tree graph for the sparse inverse discrete Fourier transform;

FIG. 3C is a schematic diagram of an analog implementation of the tree graph algorithm of FIG. 3B;

FIG. 4 is a tree graph for the complete equalization;

FIG. 5 is a schematic diagram of a complex multiplier used in equalizing the Fourier coefficients of the received signal;

FIGS. 6A and 6B are block schematic diagrams of the invention using complex and simple multipliers in a feedback circuit;

FIG. 7A is a symbolic schematic of the steering logic of FIG. 6A for complex spectral components;

FIG. 7B is a logic schematic representation of the steering logic circuit of FIG. 7A;

FIG. 7C is a symbolic schematic diagram of the steering logic of FIG. 6A for real spectral components;

FIG. 7D is a logic schematic representation of the steering logic of FIG. 7C; and

FIGS. 8A and 8B show the use of the invention in a frequency domain equalization system.

A block diagram of the model of the transmission system is shown in FIG. 1. The system is assumed linear, and it is therefore theoretically immaterial where in the system the distorting elements are located. The transfer function H(w) is a composite of all the ideal elements of the system and is shown in cascade with D(w), which is a composite of all the linear distorting elements of the system. It is assumed that in impulse response h(t) is the ideal symbol and that the information is represented by the magnitude and/or polarity of impulses at the input to H(w) which impulses are spaced in time according to the requirements of h(t) and the detection process. The output of the system before equalization is the Fourier transform of H(w).times. D(w), or the convolution of h(t) and d(t), and is no longer ideal. The equalizer is connected in cascade with the distortion network and functions to eliminate the effects of D(w), i.e., the transfer function of the equalizer is 1/D(w). The equalizer precedes the decision point at the receiver, and the system is capable of determining D(w) and then producing the transfer function 1/D(w) in the transmission path.

FIG. 2 shows an analog discrete Fourier transform (ADFT) circuit which produces a set of electrical signals which represent the real and imaginary coefficients respectively of samples of the Fourier transforms, i.e., frequency spectrum, of the input signal. The input to the ADFT circuit comprises a discrete sample set of, for example, eight samples x.sub.0, x.sub.1, . . . x.sub.7 of the received signal x(t). The sample set may be taken, for example, from terminals of a tapped delay line 5. The discrete i.sup.th sample set [ X].sub.i = x.sub.0 (i), . . . x.sub.7 (i) is transformed by the ADFT circuit into the frequency domain and represented by vectors X.sub.n which are generally complex. Real and imaginary parts of the vector and designated RX.sub.n and IX.sub.n, respectively. Similarly, RH, IH and RD, ID designate respectively the real and imaginary parts of the respective transfer functions H and D.

It is contemplated, however, that the input signal may be digitized in an analog-to-digital converter, and that the sample set could be taken from a digital shift register. In such a case, the digitized shift register output could be converted to analog form (via a D/A converter) and used in the analog circuit shown in FIG. 2. Thus, any equivalent for the tapped delay line or for the digital shift register may be used to provide the sample set to the ADFT.

FIG. 2 shows a plurality of operational amplifiers 10 having input terminals marked + or - for indicating the additive or subtractive function performed therein. The gain of the amplifiers is indicated by the multiplication factor shown. All amplifiers have unity gain except those having gain 0.707. If more than eight sample points are used to increase the range of the equalizer, the structure of FIG. 2 would be expanded and would include additional non-unity gain amplifiers. The ADFT circuit shown in FIG. 2 receives N samples (N=8) of a real input function x(t). For the real time samples x(t), the frequency components X.sub.n for n>N/2 are the complex conjugates of X.sub.n for n<N/2. Additionally, X.sub.0 and X.sub.N/2 are real. Consequently, X.sub.0 and X.sub.4 have real components only. Non-redundant information is obtained using X.sub.0 and X.sub.4 and the real and imaginary parts of vectors X.sub.1, X.sub.2 and X.sub.3 (alternately, vectors X.sub.5, X.sub.6 and X.sub.7 could be used instead of vectors X.sub.1, X.sub.2 and X.sub.3). The complex vectors X.sub.1, X.sub.2 and X.sub.3 specify six parameters, and the real vectors X.sub.0, X.sub.4 give two more parameters yielding a total of eight parameters consistent with the number of sample points of x(t). ADFT circuitry is described more generally, for example, in copending application entitled "Time Domain Automatic Equalizer with Frequency Domain Control", Ser. No. 691,008, filed June 1, 1976, in U.S. Pat. No. 3,851,162 to Robert Munoz, the whole of both references incorporated herein by reference.

The inverse of the DFT may be performed quite straightforwardly by reversing the DFT of FIG. 2. The tree graph for the IDFT is shown in FIG. 3A where the inputs are the real and imaginary spectral components of the non-redundant vectors X.sub.n. In the sliding window sampling of the instant invention, significant circuit simplicity is provided in using only a single output of the IDFT. The simplest approach is to utilize the inverse transforms which require only real inputs thus eliminating complex multiplication. Accordingly, FIG. 3B shows a "sparse" IDFT for the 4th time sampling, and FIG. 3C shows an analog implementation of FIG. 3B. The output signal at the 4th time sampling is representative of the input signal sample x.sub.3, for an input sample set x.sub.0 . . . x.sub.N-1. A subsequent input sample set is taken later, shifted in time by a fixed amount t.sub.0, where generally 0<t.sub.0.ltoreq. (T/N) and T is the sample set window, to provide a sample set x.sub.0 . . . x.sub.N-1. The 4th output sample is again representative of the 4th input time sampling, namely x.sub.3. The input sample is again taken, shifted by t.sub.0 and the process repeated to provide a sliding window input. There is therefore a one-to-one correspondence between the number of samples from the IDFT and the number of signal sample sets. Thus, the output signal can be continuous if the input is continuous, as, for example, in utilizing an analog delay line, or the output can be sampled if the input is sampled, as, for example, in utilizing an input shift register.

FIG. 4 illustrates a tree graph for the complete equalization process operating in a running mode. The DFT of the input sample set x.sub.0 . . . x.sub.N-1 for N=8 is computed in section 7. The frequency domain equalization is computed in section 9 and the inverse DFT in section 11. As is readily apparent, sections 7 and 11 are identical to FIG. 2 and FIG. 3B, respectively. The frequency domain equalization is achieved by multiplying each spectral coefficient X.sub.n by a correction factor C.sub.n which is simply a component of the transfer function of the equalizer C(w). Thus,

the equalized spectral components, Y.sub.n, are then inverse transformed by the IDFT to provide the time domain representation of the input sample set.

The multiplication in equation (1) is performed component by component and may be considered the equivalent of a vector dot product multiplication. Indeed, within the frequency domain, the equivalent transfer function of two transfer functions in series is the component-by-component product of the two functions, and there are no cross products as in the case of convolutions. The multiplication indicated in section 9 of FIG. 4 and by equation (1) is a complex multiplication for complex components X.sub.1, X.sub.2 and X.sub.3 and a simple multiplication for real components X.sub.0 and X.sub.4. In both cases, however, only the real results of the multiplication, namely RY.sub.n, are in fact generated inasmuch as the sparse inverse transform requires only real inputs. Further, the correction factors C.sub.n may be adjusted independently of one another, and only the coefficients X.sub.n and Y.sub.n need be considered when making the adjustment. Thus, there is no requirement for obtaining the system output as is generally true in time domain equalizers. The equalization process of FIG. 4 takes place entirely in the frequency domain and provides an in-line system for automatically equalizing the incoming signals.

In order to determine the desired equalizer transfer function C(w), one may assume that an isolated impulse or test signal of known magnitude and polarity is transmitted. This test signal is transmitted during a training period prior to message transmission. In the following description, test or training signals are sequentially transmitted to set up or initialize the equalizer to provide the components C.sub.n. Instead of transmitting separate test pulses, however, a single input sample set may be held in storage and utilized during the adjustment or setup process. In this case, the adjustment steps can be made as rapidly as is consistent with proper operation of the feedback loop circuitry. It is within the scope of this invention to completely adjust the equalizer within one input pulse interval. Inasmuch as convergence is always assured and is not dependent on step size, the steps may be made as large as desired consistent with the accuracy needed in the final setting. Step size is determined by the number of bits in the correction factor word (counter 28 in FIG. 6).

During the adjustment process, the ideal training pulses having known characteristics are transmitted.

The ideal received signal is h(t), the impulse response of H(w). However, the actual received test signal is f(t), the impulse response of f(w)= H(w).sup.. D(w). It is intended that C(w) should equal 1/D(w) or be the best approximation possible. For the test pulses f(t), one can write the following for each frequency component n:

and since, ##EQU1## then for each component n: ##EQU2## The ADFT performed at the receiver can produce a set of coefficients for each input sample set i representing RF and IF at discrete frequencies. Also, since h(t) is known, the coefficients RH and IH at these same frequencies are known. With this information, a sample version of 1/d(w) can be produced and used to equalize any signal which is subsequently transmitted through D(w). The equalization function 1/D(w) can be written as 1/D(w)= C(w)= rc+ jIC, where ##EQU3##

The instant invention provides these correction factors by a series of successive approximations. An alternate approach to providing the correction components C.sub.n is by means of a more direct calculation technique and is described in the aforementioned copending application.

FIG. 5 is a schematic diagram of a complex multiplication apparatus which may be used in equalizing the Fourier coefficients of the received signal according to equation (1). The n.sup.th spectral component X.sub.n comprising real and imaginary coefficients RX.sub.n and IX.sub.n, respectively, is multiplied by the n.sup.th correction factor C.sub.n, having the real and imaginary coefficients RC.sub.n and IC.sub.n, respectively, as shown in the equation below:

each of the components RC.sub.n and IC.sub.n is derived during the initial training period wherein x(t)=f(t). FIG. 5 shows the required multiplication as being performed by multipliers 14a, 14b, and 16a, 16b, the outputs thereof being provided to operational amplifiers 18 and 20. During the training period, when f(t) is received, RC.sub.n and IC.sub.n are adjusted such that the outputs RY.sub.n and IY.sub.n of the complex multiplier are simply RH.sub.n and IH.sub.n, respectively. When adjustment is complete, the correction factors RC.sub.n and IC.sub.n are those given by equations (3A) and (3B) above. Thus, during the training period, as seen by equation (2),

and

when these values of RF and IF are substituted into equations (3A) and (3B), one obtains: ##EQU4## When these values of RC and IC as well as the values for RX and IX from equation (5) are substituted into equation (4B), one has, omitting the frequency index n, ##EQU5## Likewise, equation 3C reduces to

these values of RY and IY are the desired values and meet the criteria which stops adjustment. In the instant invention Y.sub.n and H.sub.n are compared in a feedback system and C.sub.n is iteratively adjusted so that Y.sub.n approaches H.sub.n with increasing accuracy.

In utilizing the sparse inverse transform, only the real coefficients RY.sub.n are required as given by equation (4B). Consequently, while it is necessary to utilize the complex multipliers 14a, 14b, 16a and 16b of FIG. 5 to obtain RC and IC (for X.sub.1, X.sub.2 and X.sub.3), it is only necessary to utilize multipliers 14a and 14b (and operational amplifier 18) in the running mode of operation (FIG. 4) once the setup process is complete. Similarly, for real components X.sub.0 and X.sub.4, it is only necessary to utilize multiplier 14a of FIG. 5 to achieve the multiplication indicated in FIG. 4. As seen below, the multipliers 16a and 16b of FIG. 5 are time shared for X.sub.1, X.sub.2 and X.sub.3 in the adjustment or setup mode of operation, and only the required multipliers 14a and 14b are duplicated for each complex component for use in the running mode of operation. For real components, multiplier 14a is duplicated for each component as used in the setup and running modes of operation.

FIG. 6A is a block schematic diagram showing the complex multiplier of FIG. 5 in a feedback circuit. Multipliers 14 and 16 together with operational amplifiers 18 and 20 form the complex multiplication of X and C as in FIG. 5. The outputs of operational amplifiers 18 and 20 are now fed to a feedback circuit 22 which comprises difference amplifiers 24a and 24b, sign generators 26a-26d, reversible counters 28a and 28b, and steering logic circuit 30. The circuitry of FIG. 6 must, in general, be repeated for each frequency component n, or, alternatively, as set forth below, circuit elements may be time shared. For simplicity, the frequency index n has been omitted although it is understood that signals RY, RX, RC, IY, etc., all are associated with a particular frequency, i.e., RY is RY.sub.n.

FIG. 6B is a simplified version of FIG. 6A of the case of real coefficients n=0 and n=4. In this case, only multiplier 14a is needed, and this multiplier is used both during the setup and running modes of operation.

The product output signals RY and IY are fed to difference amplifiers 24a and 24b, respectively, where they are compared with reference signals RH and IH. These reference signals may be provided as outputs of potentiometers (not shown), which are calculated and set off-line to correspond to the ideal or undistorted transfer function H of the transmission channel. Error signals e.sub.R and e.sub.I are provided as outputs from difference amplifiers 24a and 24b. A distinctive feature of the invention is the utilization of only the polarity or sign of the error signals together with the sign of RX and IX to determine whether the correction signals RC and IC should be incremented or decremented. Thus, sign generators 26a and 26c provide outputs indicative of the sign of e.sub.R and e.sub.I, respectively, and sign generators 26b and 26d provide outputs indicative of the sign of RX and IX, respectively. These sign generators 26a-26d may, for example, comprise slicing circuits, which slice at zero signal level and produce " true" or "false" (positive and negative) levels. The polarity information of e.sub.R, e.sub.I, RX and IX is fed to steering logic circuit 30, which is shown in detail in FIGS. 7 and 8. The output of the steering logic circuit 30 is a logic signal .delta..sub.RC and .delta..sub.IC for incrementing counters 28a and 28b or a logic signal .delta..sub.RC and .delta..sub.IC for decrementing counters 28a and 28b. The counters 28a and 28b thus provide the correction signal RC and IC to multipliers 14 and 16. The values of the correction signals RC and IC are continually adjusted until the outputs RY= RH and IY= IH, which occurs when RC and IC are given by equations (3A) and (3B). Initial values RC.sub.no and IC.sub.no of the correction signal are fed to reversible counters 28a and 28b to speed convergence. These values may represent some compromise or approximate equalization known from experience or they may simply be chosen to produce no equalization, e.g., RC.sub.no = 1, IC.sub.no = 0. These initial values may be taken from potentimeters (not shown) as is the case with RH.sub.n and IH.sub.n.

The instant invention facilitates the utilization of logic circuitry in the feedback control apparatus 22 by developing the required correction multipliers in an incremental fashion using small, fixed increments. This concept permits the examination only of the polarity of the input signals corresponding to RX and IX and the polarity of error signals e.sub.R and e.sub.I. The polarity logic or steering logic results in still further simplification of logic circuitry.

When the correction values RC and IC are equal to the values shown in equations (3A) and (3B), these values are stored and used as inputs to multipliers 14 and 16 during transmission of the message signal x(t).

The requirements for the steering logic circuit 30 may be seen from the following analysis given for complex components X.sub.n.

From equations (3B) and (3C), it can be seen that if RC is changed incrementally by .delta..sub.RC, then the outputs, Y, change according to:

where the subscripts n have been omitted for clarity. If IC is changed incrementally by .delta..sub.IC, then the outputs change according to:

the output changes should be opposite in polarity from the output errors, in order for the process to converge on the correct settings. If one defines the output error e.sub.R and e.sub.I to be given by:

then for changes in RC, one can write:

or

similarly, for changes in IC

or

the pair of equations for SGN.delta..sub.RC must be satisfied simultaneously, as must be equations for SGN.delta..sub.IC. It is simpler to summarize the conditions under which these equations are satisfied with Boolean equations wherein positive variable are said to be true and negative are false. The equations are written with + designating a logical "OR", "." designating a logical "and", and the bar - designating a logical "not". Then,

these equations represent the steering logic required in the feedback control system of FIG. 6A. The logic decision tree of FIG. 7A indicates the decision made for each of the sixteen possible combinations of four "variables". An implementation of the decision tree is shown in FIG. 7B, where a plurality of AND gates 32 and OR gates 34 perform the decision of FIG. 7A responsive to the binary variables representing the signs of RX, IX, e.sub.R and e.sub.I. The circuit of FIG. 7B responding to specific combinations and situations effectively steers a pulse to increment the proper multiplier. If .delta..sub.RC is true (positive) RC is made incrementally more positive (less negative). If .delta..sub.RC is false (negative), RC is made incrementally more negative (less positive). A similar relation exists between .delta..sub.IC and IC.

It is noted that only one of the four outputs can occur at the same time, i.e., either the real multiplier, RC, is increased or decreased, or the imaginary multiplier, IC, is increased or decreased.

FIG. 7C illustrates the logic decision tree for the simpler cases n=0 and n=4 in which the spectral components have only real coefficients. FIG. 7D shows the hardware steering logic corresponding to FIG. 7C.

The sign generators 26a-26d, which produce the sign information, will always have an output either positive or negative. Consequently, the steering logic will always have an output and the multipliers will be incremented continually. The system will thus hunt around the correct values even after they have been properly achieved. The error due to such hunting depends on the size of the increments .delta..sub.RC and .delta..sub.IC.

Hunting may be avoided or minimized by employing error signal slicers, which have a null zone or deadband around zero level. When both error signals fall within the null zone, further incrementing of the multipliers is inhibited. The cessation of incrementation may also be used to provide a convenient indication that the setup process is complete.

When the setup or initialization process is complete using the ideal test pulse, the output real coefficient RY equals the reference real coefficient RH, and the output imaginary coefficient IY equals the reference imaginary coefficient IH. Moreover, the real multiplier RC equals the real part, and the imaginary multiplier IC equals the imaginary part of the inverse of the complex quantity representing the distortion transfer function, D(w), which was a factor of the input along with the complex quantity H(w) representing the reference channel transmission function. Thus, when the multiplier settings are retained in appropriate storage means after the initial training period, the feedback control may be disabled and the equalizer will cancel the distortion factor from any complex input. The circuitry operates to equalize the discrete frequency components transmitted through the same distortion.

The circuitry shown in FIGS. 8A and 8B, when joined as indicated by the identically referenced leads, represents the implementation of the flow diagram of FIG. 4 and incorporates therein the circuits of FIG. 3C, 6A and 7. Specifically, the input sample data is taken from a delay line 40 and sample sets i, i+1, . . . are taken to provide the sliding window. During message reception, the DFT, frequency adjustment and sparse inverse DFT are performed for each sample set i, i+1 . . . Operational amplifiers 41 are similar to those shown in FIG. 2, and the output signals corresponding to the discrete frequency components of the transform X are provided as inputs to the equalizer proper. A single time-shared circuit 42 is used by all complex frequency components and comprises multiplier 44a, 44b, operational amplifier 46 and feedback control apparatus 22'. Multipliers 44a and 44b together with operational amplifier 46 correspond respectively to multipliers 16a, 16b and operational amplifier 20 of FIG. 6A. Other circuit elements corresponding to FIGS. 6A and 6B are appropriately labeled, but now the frequency index n is provided as a suffix. Using the time-shared circuit 42, it is only necessary to provide separately, for each component circuit elements corresponding to multipliers 14a, 14b, operational amplifiers 18 and reversible counters 28a and 28b. It is noted that the feedback circuit 22' is the same as feedback circuit 22 of FIG. 6 except that reversible counters 28a and 28b are now omitted. These counters are now duplicated separately for each component as shown in FIG. 8. Consequently, the outputs of the steering logic 30 of feedback circuit 22' are fed to appropriate up/down counters 28a-n and 28b-n via a multiplexer M1. The inputs RX.sub.n, IX.sub.n, RH.sub.n, IH.sub.n, RY.sub.n, RC.sub.n, IC.sub.n to time-shared circuit 42 are provided from another multiplexer M2. For real components n=0 and n=4, a separate feedback circuit 22" is provided. Thus, feedback circuit 22' corresponds to the logic circuit shown in FIGS. 6A and 7B (without counters 28), and feedback circuit 22" corresponds to the logic circuit shown in FIGS. 6B and 7D (without counter 28). For real components, the time-shared circuit 42 is bypassed, and the feedback circuit 22" is interconnected directly between multiplexers M1 and M2. The interconnecting lines are labeled with the subscript "0,4" to indicate that n sequentially takes on its two values 0 and 4. Thus, the line labeled "RY.sub.0,4 " stands for an interconnecting line carrying signal coefficient value RY.sub.0 or RY.sub.4.

Multiplexers M1 and M2 are controlled by an initializing circuit means comprising setup switch 48, peak detector 50, counter 52, interconnection means 54 and clock means 56. Switch 48 is closed during the training or setup period and, upon detection of the incoming signal f(t) in peak detector 50, counter 52 is actuated to provide a delay so that the equalization will be effective when the peak of the training pulse is near the center of delay line 40. Counter 52 triggers the interconnection means 54 and clock means 56 to actuate multiplexers M1 and M2 and provide their interconnection to the other indicated circuit elements. Interconnection means 54 may comprise switching devices, such as transistors or other solid state devices, together with connecting lines so as to provide the necessary wire connections.

Thus, in the running mode, the incoming message signal x(t) is sampled to produce the samples x.sub.k where k=0 . . . N-1. The samples x.sub.k are transformed by a DFT to provide spectral components X.sub.n for n=0 . . . N-1. The spectral components are equalized to produce equalized spectral components Y.sub.n = C.sub.n.sup. . X.sub.n, n=0 . . . N-1 and the inverse DFT is taken for Y.sub.n to produce a single time domain output sample Y.sub.k corresponding to the original input sample x.sub.k. Employing non-redundant frequency components for the discrete Fourier transforms, where N is an even interger and x(t) and f(t) are real, simplifies the equalization in that circuitry need only be provided for N/2 spectral components. Thus, the DFT circuitry provided components X.sub.n for n=0, 1 . . . N/2. Similarly, the components C.sub.n and Y.sub.n need only be provided for n=0, 1 . . . N/2. Sliding window sampling of the input signal x(t) wherein samples are taken every T/N.sup.th interval along the delay line 5 (FIG. 2), allows utilizing a single output from the sparse IDFT corresponding to each sample set for each window. Consequently, N output signals are provided at the output of the IDFT for each N input sample set. The sliding window technique is appropriate as opposed to a complete replacement of x.sub.0 . . . n.sub.N.sub.-1 inasmuch as multiplication in the frequency domain corresponds to a convolution in the time domain, and for the DFT, convolutions are circular rather than linear. See, for example, Digital Signal Processing by Oppenheim and Schafer, Prentice-Hall, 1975.

During the training period, the correction values RC and IC are produced by feedback adjustment for each channel, N, to satisfy equations (3A) and (3B). Thus, reversible counter 28a-0 having an initial value RC.sub.00 (RC.sub.n0 where n=0) is successively incremented or decremented to approach, within the slicer deadband, the equalization value of equations (3A) and (3B). For complex components, n=2, for example, the reversible counters 28a-3 and 28b-3 having initial values RC.sub.30 and IC.sub.30 are each adjusted so that RC.sub.3 and IC.sub.3 satisfy equations (3A) and (3B). After all channels n have the equalization values of RC.sub.n and IC.sub.n stored in their respective counters or storage means, setup switch 48 is opened, and the equalizer is ready to receive message signals x(t). Each spectral coefficient X.sub.n of the message signal is now multiplied by the equalization values of RC.sub.n and IC.sub.n (RC.sub.n only for n=0,4) stored in reversible counters 28a-n and 28b-n. The coefficients Y.sub.n are thus frequency equalized and given by Y.sub.n = C.sub.n.sup. . X.sub.n as desired. The sparse IDFT of the set Y.sub.n produces the desired output time sample Y.sub.k.

It is noted that in a facsimile system, the sync pulses used to achieve line synchronization of the scanning and printing mechanism can provide an ideal set of known pulses for the purpose of setting up the automatic equalizer. Furthermore, if the sync pulses are continued throughout the transmission of facsimile information, the automatic equalizer settings can be regularly updated. The system can thus be adaptive in the sense that the equalizer can be made to track changes in channel characteristics which occur during the transmission of a document.

In utilizing an analog tapped delay line to provide the input sets x.sub.k, the equalizer bandwidth is determined by the tap spacing, .tau. = T/N and is BW= 1/2.tau. = N/2T. In such systems, filtering may be used to limit the bandwidth of the incoming signal to BW to avoid aliasing. Images do not occur in an analog delay line since samples are continuously available and the sampling rate may be thought of as infinite. If the input sample set is taken from stages of a shift register, for example, the sampling rate must be at least the Nyquist rate to avoid aliasing. It is important to note that the input sampling rate may not necessarily be the same as the sampling rate seen by the DFT since one could connect, for example, every other stage of the input shift register to the DFT input circuitry. The input sampling rate determines the rate at which the output samples appear and the image points of the output signal spectrum. The transform sampling rate determines the equalizer transfer function which is continuous in the analog delay line case since the transform sampling frequency, N/T, is twice the bandwidth BW= N/2T. The equalization transfer function may also be made continuous with shift registers or sample and holding circuits at the input if the transform sample interval is taken at T/N seconds using a total of N inputs. If the number of samples taken during time T is N, then the equalization will exactly cancel the distortion at N/2 positive frequencies, plus dc, evenly spaced by 1/T, and the impulse response of the equalized system will be exactly correct at N equally spaced points. This type of equalizer is thus ideally suited to digital transmission; however, the equalization function will be a smooth curve between the sample frequencies so that it is also well suited for non-digital transmission such as facsimile and video. Thus, although control of the equalizer occurs at discrete points, the transfer function itself is continuous from dc to BW= N/2T and beyond, as an image, where T/N is the transform sample spacing. The response in between the control frequencies is a result of the continuous "windowing" in the time domain.

If the sample set does not include all the non-zero samples of the unequalized system response, then the equalization between the sample frequencies will not be good enough to eliminate intersymbol interference in the digital sense. If the samples are not close enough, the equalization bandwidth will be too narrow. The equalization function is periodic in the frequency domain with period of 1/.tau.. The sample spacing is easily changed without changing the system complexity. However, if the number of samples is increased, the circuit complexity increases faster than linearly, since the number of nodes in the discrete Fourier transform algorithm used is N Log.sub.2 N.

While the invention has been described with reference to a particular embodiment thereof, it is apparent that modifications and improvements may be made by those of skill in the art without departing from the spirit and scope of the invention.