Receiver equalizer apparatus - Patent Review 3992616

Claims

The embodiments of the invention in which an exclusive property or right is claimed are defined as follows:

1. An improvement in a signal filter having an adaptable filter characteristic for filtering input signals, said signal filter being a transversal filter equalizer achieving said adaptable filter characteristic through fixed increment adjustments of values of weighting factors used for multiplying input data samples in said input signal, said fixed increment adjustment either increasing or decreasing a first value of a corresponding weighting factor depending on (i) a sign of an error sample in a discrete time error signal provided by said signal filter, and on (ii) a sign of a pertinent input data sample, taken from said input signal, which is weighted, in providing said error sample, by a second value of said corresponding weighting factor, said improvement in said filter comprising:

a comparator means to provide a determination of whether said input data samples have values which exceed a threshold value, said comparator means being capable of providing a comparator output signal indicating said determination; and

inhibitory means receiving said comparator output signal, said inhibitory means being capable of preventing a said fixed increment adjustment from being made to a value of that said weighting factor corresponding thereto when said pertinent input data sample associated with said fixed increment adjustment fails to exceed said threshold value.

2. The apparatus of claim 1 wherein said input signal is acted upon by a variable gain amplifier before being applied to said signal filter, said variable gain amplifier having gain therethrough controlled by a gain control signal derived from said comparator output signal such that a selected fraction of said fixed increment adjustments are permitted by said inhibitory means to adjust values of said weighting factors.

3. The apparatus of claim 1 wherein an automatic gain control system is used to hold an average value related to said input signal substantially time invariant as said average value occurs when received by said transversal filter equalizer.

4. The apparatus of claim 1 wherein an analog input signal is supplied to an analog to digital converter means to provide a discrete time converter output signal from which said input signal is obtained as a discrete time signal comprising input data samples formed as digital words each comprising magnitude bits and a sign bit.

5. The apparatus of claim 1 wherein said second value of said corresponding weighting factor occurs at a different and earlier time than does said first value.

6. The apparatus of claim 2 wherein said gain control signal is linearly related to a logarithm of said gain.

7. The apparatus of claim 2 wherein an analog input signal is supplied to an analog to digital converter means to provide a discrete time converter output signal from which said input signal is obtained as a discrete time signal comprising input data samples formed as digital words each comprising magnitude bits and a sign bit.

8. The apparatus of claim 2 wherein said second value of said corresponding weighting factor occurs at a different and earlier time than does said first value.

9. The apparatus of claim 4 wherein there is provided a bias signal removal system for adjusting an uncorrected analog input signal toward being symmetrical about a reference signal level wherein said uncorrected analog input signal summed with a correction signal to form said analog input signal and said correction signal is derived from said discrete time error signal.

10. The apparatus of claim 7 wherein there is provided a bias signal removal system for adjusting an uncorrected analog input signal toward being symmetrical about a reference signal level wherein said uncorrected analog input signal summed with a correction signal to form said analog input signal and said correction signal is derived from said discrete time error signal.

11. The apparatus of claim 10 wherein said gain control signal is linearly related to a logarithm of said gain.

12. A discrete time signal filter having an adaptable filter characteristic for filtering discrete time input signals, said filter comprising:

a first tapped delay means through which said discrete input signal passes after being provided at a first tapped delay means input, said first tapped delay means provided with a first tap output at which substantially said discrete input signal appears as a first tap output signal after a selected delay from when provided at said first tapped delay means input;

a first multiplier means to receive said first tap output signal from said first tap output, said first multiplier means capable of multiplying said first tap output signal by a first multiplier signal to provide a first contribution signal, said first mutliplier means subject to being adjusted to alter values of said first multiplier signal;

a summing means which provides as an output signal a summed contribution signal that represents a sum of signals presented to said summing means, said first contribution signal being presented to said summing means;

a first adjusting means capable, when not prevented, of adjusting said first multiplier means to alter values of said first multiplier signal by first fixed increment changes therein, either to increase or to decrease values of said first multiplier signal, said first adjusting means initiating each such first fixed increment change in response to said first tap output signal in conjunction with an error signal derived from said summed contribution signal in conjunction with an estimated output signal which is also derived from said summed contribution signal, with series of said first fixed increment changes used to provide said adaptive filter characteristic;

a first inhibitory means, said first inhibitory means capable of selectively preventing said first adjusting means from adjusting, in manner aforesaid, said first multiplier means in response to a first comparator means output signal; and

a first comparator means which provides said first comparator means output signal depending on whether said first tap output signal exceeds a selected first threshold value.

13. The apparatus of claim 12 wherein said discrete time input signal comprises sequentially presented discrete time input data samples formed as m-bit digital words, there being one said input sample provided per sample period including a said input data sample provided in a first sample period and wherein said first tapped delay means is a first shift register means which stores therein both said input data sample provided in said first sample period and said data input samples received in (N-1) sample periods immediately preceding said first sample period.

14. The apparatus of claim 12 wherein:

a second tap output signal is provided from a second tap output in said first tapped delay means, said second tap output signal being substantially said discrete time input signal selectively delayed from when provided at said first tapped delay means input;

a second multiplier means to receive said second tap output signal from said second tap output, said second multiplier means capable of multiplying said second tap output signal by a second multiplier signal to provide a second contribution signal, said second multiplier means subject to being adjusted to alter values of said second multiplier signal, said summing means having said second contribution signal presented thereto;

a second adjusting means capable, when not prevented, of adjusting said second multiplier means to alter values of said second multiplier signal by second fixed increment changes therein, either to increase or to decrease values of said second multiplier signal, said second adjusting means initiating each such second fixed increment change in response to said second tap output signal in conjunction with said error signal, with series of said second fixed increment changes used to provide said adaptive filter characteristic;

a second inhibitory means, said second inhibitory means capable of selectively preventing said second adjusting means from adjusting, in manner aforesaid, said second multiplier means in response to a second comparator means output signal; and

a second comparator means which provides said second comparator means output signal depending on whether said second tap output signal exceeds a selected second threshold value.

15. The apparatus of claim 12 wherein:

a second tap output signal is provided from a second tap output in said first tapped delay means, said second tap output signal being substantially said discrete time input signal selectively delayed from when provided at said first tapped delay means input;

a second multiplier means to receive said second tap output signal from said second tap output, said second multiplier means capable of multiplying said second tap output signal by a second multiplier signal to provide a second contribution signal, said second multiplier means subject to being adjusted to alter values of said second multiplier signal, said summing means having said second contribution signal presented thereto;

a second adjusting means capable, when not prevented, of adjusting said second multiplier means to alter values of said second multiplier signal by second fixed increment changes therein, either to increase or to decrease values of said second multiplier signal, said second adjusting means initiating each such second fixed increment change in response to said second tap output signal in conjunction with said error signal, with series of said second fixed increment changes used to provide said adaptive filter characteristic; and

a second inhibitory means, said second inhibitory means capable of selectively preventing said second adjusting means from adjusting, in manner aforesaid, said second multiplier means in response to said first comparator means output signal delayed substantially to same extent as said second tap output signal.

16. The apparatus of claim 12 wherein said discrete time input signal is acted upon by a variable gain amplifier before being applied to said first tapped delay means input, said variable gain amplifier having gain therethrough controlled by a gain control signal derived from said first comparator means signal such that a selected fraction of first fixed increment changes initiated by said first adjusting means are permitted by said first inhibitory means to adjust said first multiplier means.

17. The apparatus of claim 13 wherein:

said first tap output is an output taken from a storage section in said first shift register means and said first tap output signal comprises each imput data sample most recently stored in each sample period in said first shift register means as each said most recently stored data input sample is presented at said first output tap in each sample period, including said first sample period;

said first multiplier means is a digital word multiplier and said first multiplier signal is a discrete time signal comprising sequentially presented discrete time weighting factors formed as m-bit digital words, and said first contribution signal is a discrete time signal comprising sequentially presented multiplication products formed as digital words;

said summing means is a digital accumulator and said summed contribution signal is a discrete time signal comprising sequentially presented discrete time sums formed as m-bit digital words; and

said first adjusting means includes a second shift register means which adjusts said first multiplier means by supplying a varying first multiplier signal thereto.

18. The apparatus of claim 14 wherein there are (N-2) further sets of filter output weighting members, each said set of filter output weighting members including:

a set tap output provided in said first tapped delay means;

a set multiplier means operating in manner of said second multiplier means to receive a set tap output signal from said set tap output and presenting a set contribution signal to said summing means;

a set adjusting means operating in manner of said second adjusting means capable, when not prevented, of adjusting said set multiplier means by set fixed increment changes in response to said set output tap signal in conjunction with said error signal;

a set inhibitory means operating in manner of said second inhibitory means capable of preventing said set adjusting means from adjusting said set multiplier means on response to a set comparator means output signal; and

a set comparator means operating in manner of said second comparator means to provide said set comparator means output signal depending on whether said set tap output signal exceeds a set threshold value selectively set in said set comparator means.

19. The apparatus of claim 14 wherein an automatic gain control system is used to hold an average value of said discrete time input signal substantially time invariant as it occurs at said first tapped delay means inputs.

20. The apparatus of claim 14 wherein said error signal and said first and second tap output signals are in digital form, said first and second adjusting means being EXCLUSIVE OR logic gates which operate only on sign bits included in said error signal and in said first and second tap output signals.

21. The apparatus of claim 14 wherein said discrete time input signal is acted upon by a variable gain amplifier before being applied to said first tapped delay means input, said variable gain amplifier having gain therethrough controlled by a gain control signal derived from one of said first and second comparator means signals such that a selected fraction correspondingly of one of said first or second fixed increment changes initiated correspondingly by one of said first and second adjusting means are permitted correspondingly by one of said first and second inhibitory means to correspondingly adjust one of said first and second multiplier means.

22. The apparatus of claim 14 wherein:

said first tap output is an output taken from a storage section in said first shift register means and said first tap output signal comprises each input data sample most recently stored in each sample period in said first shift register means as each said most recently stored data sample is presented at said first output tap in each sample period, including said first sample period;

said second tap output is also said first tap output, and said second tap output signal comprises each input data sample stored in said first shift register means in each sample period immediately preceding storage of said input data sample most recently stored as said preceding input data sample is presented at said first output tap in each sample period, including said first sample period;

said first multiplier means is a digital word multiplier and said first multiplier signal is a discrete time signal comprising sequentially presented discrete time weighting factors formed as m-bit digital words, and said second contribution signal is a distrete time signal comprising sequentially presented multiplication products formed as digital words;

said second multiplier means is also said first multiplier means and said second multiplier signal is a discrete time signal comprising sequentially presented discrete time weighting factors formed as m-bit digital words, and said second contribution signal is a discrete time signal comprising sequentially presented multiplication products formed as digital words;

said summing means is a digital accumulator and said summed contribution signal is a discrete time signal comprising sequentially presented discrete time sums formed as m-bit digital words;

said first adjusting means includes a second shift register means which adjusts said first multiplier means by supplying a varying first multiplier signal thereto; and

said second adjusting means is also said second shift register means which adjusts said second multiplier means by supplying a varying second multiplier signal thereto.

23. The apparatus of claim 15 wherein said second inhibitory means receives substantially said first comparator means output signal a tap in a second tapped delay means through which said first comparator means output signal is passed from a second tapped delay means input.

24. The apparatus of claim 15 wherein there are (N-2) further sets of filter output weighting members, each said set of filter output weighting members including:

a set tap output provided in said first tapped delay means;

a set multiplier means operating in manner of said second multiplier means to receive a set tap output signal from said set tap output and presenting a set contribution signal to said summing means;

a set adjusting means operating in manner of said second adjusting means capable, when not prevented, of adjusting said set multiplier means by set fixed increment changes in response to said set output tap signal in conjunction with said error signal;

a set inhibitory means operating in manner of said second inhibitory means capable of preventing said set adjusting means for adjusting said set multiplier means in response to said first comparator means signal delayed substantially to some extent as said tap output signal.

25. The apparatus of claim 15 wherein said discrete time input signal is acted upon by a variable gain amplifier before being applied to said first tapped delay means input, said variable gain amplifier having gain therethrough controlled by gain control signal derived from said first comparator means signal such that a selected fraction of first fixed increment changes initiated by said first adjusting means are permitted by said first inhibitory means to adjust said first multiplier means.

26. The apparatus of claim 15 wherein an automatic gain control system is used to hold an average value of said discrete time input signal substantially time invariant as it occurs at said first tapped delay means input.

27. The apparatus of claim 15 wherein said error signal and said first and second tap output signals are in digital form, said first and second adjusting means being EXCLUSIVE OR logic gates which operate only on sign bits included in said error signal and in said first and second tap output signals.

28. The apparatus of claim 15 wherein:

said first tap output is an output taken from a storage section in said first shift register means and said first tap output signal comprises each input data sample most recently stored in each sample period in said first shift register means as each said most recently stored data sample is presented at said first output tap in each sample period, including said first sample period;

said second tap output is also said first tap output, and said second tap output signal comprises each input data sample stored in said first shift register means for each sample period immediately preceding storage of said input data sample most recently stored as said preceding input data sample is presented at said first output tap in each sample period, including said first sample period;

said first multiplier means is a digital word multiplier and said first multiplier signal is a discrete time signal comprising sequentially presented discrete time weighting factors formed as m-bit digital words, and said first contribution signal is a discrete time signal comprising sequentially presented multiplication products formed as digital words;

said second multiplier means is also said first multiplier means and said second multiplier signal is a discrete time signal comprising sequentially presented discrete time weighting factors formed as m-bit digital words, and said second contribution signal is a discrete time signal comprising sequentially presented multiplication products formed as digital words;

said summing means is a digital accumulator and said summed contribution signal is a discrete time signal comprising sequentially presented discrete time sums formed as m-bit digital words;

said first adjusting means includes a second shift register means which adjusts said first multiplier means by supplying a varying first multiplier signal thereto; and

said second adjusting means is also said second shift register means which adjusts said second multiplier means by supplying a varying second multiplier signal thereto.

29. The apparatus of claim 16 wherein said error signal and said first tap output signal is in digital form, said first adjusting means being EXCLUSIVE OR logic gates which operate only on sign bits included in said error signal and in said first tap output signal.

30. The apparatus of claim 27 wherein said estimated signal is provided by a level detector means where said estimated signal can be one of a selected number of selected signal levels, said signal level selected being that one closest to a value taken on by said summed contribution signal at time of said selection.

31. The apparatus of claim 30 wherein said first adjusting means includes an up/down counter, from which an output signal is obtained to adjust said first multiplier means and said second adjusting means includes an up/down counter from which an output signal is obtained to adjust said second multiplier means.

32. A digital signal filter having adaptable filter characteristics for filtering digital input signals comprising sequentially presented discrete time input data samples formed as m-bit digital words, there being one said input data sample provided per sample period including a said input data sample provided in a first sample period, said filter comprising:

a sample storage shift register means, capable of storing N digital words of m-bits, to receive said digital input signal and to store both said input data sample received in said first sample period and said input data samples received in (N-1) sample periods immediately preceding said first sample period, said sample storage shift register means being capable of presenting in an ordered sequence, ordered by time of receipt, each said stored input data sample at an output thereof within said first sample period to form a sample storage shift register output signal, and said sample storage shift register means being capable of recirculating said sample storage shift register output signal in said first sample period back through said sample storage shift register means while discarding in said first sample period that said stored input sample which was earliest received;

a weighting storage shift register means capable of storing N words of (q+r) bits having an up/down counter means in a first storage position of said weighting storage shift register means, said up/down counter means to receive single bits representing weighting changes and being capable of using each said weighting change received to increment therein by increase or decrease, an existing count of said weighting changes previously recieved to form a weighting factor in form of a digital word of (q+r) bits, said weighting storage shift register means being capable of presenting in an ordered sequence each said weighting factor, corresponding through initial relative shift register storage position to each said stored input data sample, at an output thereof in said first sample period to form a weighting storage shift register output signal corresponding to said sample storage shift register output signal, and said weighting storage shift register means being capable of recirculating said weighting storage shift register output signal in said first sample period back through said first storage position and remainder of said weighting storage shift register means;

a digital multiplier means to receive both said storage shift register output signal and said weighting storage shift register output signal, such digital multiplier means being capable of multiplying together each said stored input data sample in said sample storage shift register output signal with m most significant bits of each said corresponding weighting factor in said tap weighting storage shift register output signal in said first sample period to form sequentially, as digital words, N multiplication products in said first sample period, said digital multiplier means being capable of presenting said N multiplication products sequentially as formed at an output thereof in said first sample period to form a digital multiplier output signal;

an accumulator means to receive said digital multiplier output signal, said accumulator means being capable of providing a total in said first sample period of said N multiplication products received in said first sample period in form of an accumulator output digital word of m bits, said accumulator means to present said accumulator output digital word at an output thereof to form a filter output signal;

a level detector means to receive said filter output signal, said level detector means being capable of determining which allowed transmitted signal level each said accumulator output digital word is most nearly equal to in said first sample period said accumulator means capable of providing said determined allowed transmitted signal level in form of an estimated output digital word of m bits at an output thereof in said first sample period to thereby form an estimated output signal, said level detector means being capable of further determining a sign of that difference which results between said accumulator output digital word and corresonding said estimated output digital word and of providing a current error bit in said first sample period representing said sign as a sign bit error signal;

a modulo-2 adder means to provide at an output thereoof in said first sample period single bits representing said weighting changes to form a weighting change output signal, each said weighting change being provided to said up/down counter as that said weighting factor corresponding thereto is recirculated through said first storage position, each said weighting factor corresponding to each said weighting change also corresponding to a preceding input data sample with which said weighting factor was multiplied in said digital multiplier means in a sample period preceding said first sample period in manner of multiplication occuring in said digital multiplier means in said first sample period as said above, each said weighting change being provided in response to a result of modulo-2 adding (i) a sign bit of each corresponding said preceding input data sample, all such sign bits received by said modulo-2 adder means obtained from said digital input signal to form a first adder input signal, to (ii) a preceding error bit, formed in said level detector means in said sample period preceding said first sample period in manner of forming said current error bit to form a delayed signal bit error signal received by said modulo-2 adder means; and

a comparator means to receive a comparator means input signal obtained from said digital input signal, said comparator means being capable of determining whether magnitude portions of each said preceding input data sample exceeds a threshold value and further being capable of preventing said incrementing of said up/down counter means by each said weighting change corresponding to each said preceding input data sample having a magnitude portion which does not exceed said threshold.

33. The apparatus of claim 32 wherein said comparator means input signal is a magnitude portion of a delayed sample storage shift register output signal formed in said sample storage shift register means in said sample period preceding said first sample period in manner of forming said sample storage shift register output signal in said first sample; and, said first adder input signal is a sign portion of said delayed sample storage shift register output signal.

34. The apparatus of claim 32 wherein said comparator means input signal is a magnitude portion of said digital input signal and said comparator means determines whether magnitude portions of each said preceding input data sample exceeds a threshold value during that sample period in which said preceding input data sample was presented to said sample storage shift register means, a signal representing said determination being supplied to a comparator means delay followed by a comparator means storage shift register which provides a output signal to said weighting storage shift register, said comparator means storage shift register being a recirculating shift register means operating in step with said sample storage shift register means; and, said first adder input signal is an output signal from a sample sign storage shift register, a recirculating shift register operating in step with said sample storage shift register means, said sample sign storage shift register means receiving a sign portion of said digital input signal delayed by a sample sign delay means.

35. The apparatus of claim 32 wherein said modulo-2 adder is an EXCLUSIVE OR logic gate.

36. The apparatus of claim 32 wherein said preceding sample period immediately precedes said first sample.

37. The apparatus of claim 32 wherein said comparator means provides an output signal related to occurrences of said preventing of said incrementing of said up/down counter means.

BACKGROUND OF THE INVENTION

The present invention relates to adaptive equalizers for communication receivers of the transversal filter type and particularly to those utilizing fixed increment tap weight adjustments therein.

As is well known in data communication, the channel between the transmitter and the receiver over which data is transmitted not only distorts, but variably distorts, the data transmitted. To overcome the effects of this distortion on the data transmitted at the receiver, i.e to reduce the intersymbol interference which occurs as a result of the distortion, adaptive equalizers have been used. Such adaptive equalizers, also known as adaptive transversal filters or adaptive nonrecursive digital filters, are well-known and often used for this purpose.

Such a known adaptive equalizer is shown in block diagram form in FIG. 1 and may be considered as a sampled analog, i.e. discrete time system or it may be considered a digitized discrete time system with either an infinite number of bits in each digital word or sufficient bits in each digital word such that any quantization errors are entirely negligible. The operation of the system in FIG. 1 is well-known and is set out below to show the general principles of adaptive equalization.

The adaptive equalizer shown in FIG. 1 operates by measuring the intersymbol interference in the output, cross-correlating this intersymbol interference with each tap output and, on the basis of this cross-correlation result, the adaptive equalizer adjusts itself so that the remaining residual intersymbol interference in the output is uncorrelated with any of the tap output signals occurring at nodes, 10, about each of the delay blocks designated 11 in FIG. 1. The delay blocks 11, and nodes 10 viewed as taps, may be considered together as a tapped delay line. As is shown below, when the output error is uncorrelated with any of the tap inputs, the weights of each of the tap output signals are properly adjusted to an optimum for minimizing the intersymbol interference.

X(kT) in FIG. 1 represents a series of data samples each provided in a sample period of time duration T, i.e. a discrete time input signal, which is supplied to the input node of the tapped delay line mentioned above at time t=kT. Again, this tapped delay line comprises the delay blocks 11, also labelled with T's, and the nodes 10 thereabout.

At each node there is provided a tap output signal which is delayed by a number of intervals of duration T equal to the number of delay blocks 11 between that node and the input node at which X(kT) is provided. Each of these tap output signals is supplied to a weighting multiplier, 12, shown as circles also labelled by X, which multiplies the tap output signal by a weighting value, either .omega..sub.1, .omega..sub.2. . . . .omega..sub.N. The delayed tap output signals as so weighted are then summed by a summing means, 13, to provide a discrete time equalizer output signal, Z(kT).

The difference in value between Z(kT) and an estimate of the actual transmitted signal, G(kT), forms an intersymbol interference indication signal or output error, I(kT). The estimate of the actual transmitter signal is determined by a level detector, 14, which has stored in it the allowed amplitudes of the originally transmitted signals and provides as an output, in each sample period, the allowed transmitted signal amplitude level, G(kT), which is most closely approached by Z(kT). The difference between the allowed transmitted level G(kT) and the filtered signal levels experienced at the filter output, Z(kT), is determined by error summer 15. This difference again is taken as a measure of the intersymbol interference and is used to form the signal I(kT).

The intersymbol interference signal I(kT) is then supplied to a set of adjusting multipliers, 16, these again being represented as circles labelled with an X. These adjusting multipliers also receive the corresponding tap output signal which they multiply with I(kT). The multiplication results are integrated by integrators, 17, to average these multiplication results over time, which completes the cross-correlation between I(kT) and X(kT) to provide the weighting values .omega..sub.1, .omega..sub.2. . . , .omega..sub.N. In a discrete time system, the integrators 17 are usually summing devices which provide a running total over time.

The system of FIG. 1 is derived from the following analysis. The adaptive equalizer shown there can be described as a technique to minimize in some sense the intersymbol interference distortion function I(kT). Since events in the adaptive equalization system occur only in sample periods, time is discrete occurring in increments each of duration T to yield an expression for time t=kT with k an integer. The sample periods T are those of the sampled input signal or discrete time input signal, X(kT).

The intersymbol interference distortion signal is the following, as stated above, and as can be seen from FIG. 1:

i(kT) = Z(kT) - G(kT).

the sense in which the intersymbol interference is chosen to be minimized is the means square sense which leads to defining the following function to be minimized: ##EQU1## where K is an arbitrary and large number, the number of I.sup.2 (kT) samples to be included in the average.

From FIG. 1, the output of the equalizer at a time t=kT is as follows: ##EQU2## which represents a convolution of the discrete time input signal sequence X(kT) and the adpative equalizer filter characteristic as represented by the sequence of tap output signal weights, .omega..sub.n.

Substituting the intersymbol interference equation into the mean square definition and thereafter substituting the equalizer output function into the result of the first substitution yields the following equation for the mean square function: ##EQU3##

To find the optimum tap weight values .omega..sub.n, this last function must now be minimized with respect to the tap weight values .omega..sub.n which are the independent variables therein. The well-known mathematical step for minimization is to take the N partial derivatives ##EQU4## and then setting these partial derivatives equal to zero. For any particular tap weight, .omega..sub.j, the following is the partial differentiation result for l.ltoreq.j.ltoreq.N: ##EQU5## and which can be placed in the following form: ##EQU6## where, as before, l.ltoreq.j.ltoreq.N. This last equation is obtained from the preceding one through substitution of the third and then the first of the equations set out above. There are N equations just like the equation set out prior to the last equation set out above and each is set to zero to solve for the .omega..sub.n values to provide the optimum adaptive equalizer. These last equations, when implemented, lead to the system shown in FIG. 1, see multipliers 16 and averaging integrators 17.

The adaptive equalizer shown in FIG. 1 and described in the foregoing paragraphs works very well should ideal components be available therefor at economically attractive prices. However, the multipliers available for analog multiplication have unavoidable offsets in them which cause errors in their outputs leading to a degraded performance. Also, the number of such multipliers required for the sampled analog version of the system shown in FIG. 1 can be quite large, easily exceeding 50.

Further, for a number of reasons, it is usually quite attractive to digitize the implementation shown in FIG. 1 and, to hold costs down in such a version, to use as few bits as possible to represent each digital word occurring in the system. This leads to substantial quantization errors which again seriously degrade the performance of the system shown in FIG. 1.

SUMMARY OF THE INVENTION

An adaptive transversal equalizer, or an adaptive discrete time signal filter, is disclosed which, in adjusting the weightings of the tap output signals therein by only fixed increments, also uses amplitude information contained in the discrete time input signal to either prevent or allow such adjustments to be made in a particular instance. In a further embodiment of the invention, this amplitude information is used also to control the average amplitude of the discrete time input signal.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 shows an adaptive equalizer system diagram known in the prior art,

FIG. 2 shows a system diagram of a first embodiment of the invention,

FIG. 3 shows a system diagram of a further embodiment of the invention,

Fig. 4 shows a system diagram of a digitized version of the system shown in FIG. 3,

FIG. 5 shows a system diagram of a further embodiment of the invention,

FIG. 6 shows a system diagram of a further embodiment of the invention alternative to that shown in FIG. 5,

FIG. 7 shows a more specific system diagram of the embodiment of the system shown in FIG. 6,

FIG. 8 shows a system and circuit diagram alternative for a portion of the system shown in FIG. 7,

FIG. 9 shows a system and circuit diagram alternative for a portion of the system shown in FIG. 7 and 8,

FIG. 10 shows a system and circuit diagram of a further aspect of the invention, and

FIG. 11 shows a system and circuit diagram of a further embodiment of the system shown in FIG. 10.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

To reduce the number of multipliers in FIG. 1, and hence the cost, the adjusting multiplier shown there can be replaced by fixed increment, but algebraic sign variable, circuits such an monostable multivibrators or, in a digitized version, EXCLUSIVE OR logic gates. Continuing to describe a sampled analog embodiment, the use of fixed increment adjustments avoids the need for any magnitude value multiplications but also throws away this magnitude information contained in the signals that were supplied to the adjusting means multipliers. Thus, once a fixed magnitude increment is to be used, only sign information contained in the signals which were supplied to the adjusting multiplier means need be used to determine the polarity of the fixed magnitude increment to be sent on for the purpose of adjusting the weights applied to the tap output signals. Then series of such fixed increments changes will apparently continue to provide the adaptive filter characteristics required of the equalizer.

To alter the system of FIG. 1, to use fixed increment adjusting means rather than adjusting multipliers, requires that the signals that were supplied to the adjusting multipliers carry the information needed to determine the sign of the errors currently present in the weighting factors to be applied to the tap output signals. Of course, this must be the case for the system of FIG. 1 to operate satisfactorily to reduce these errors. That such information is contained in the foregoing equations for the system of FIG. 1 can be shown as follows.

Define a weighting factor which has an error therein as follows:

.omega..sub.n = .omega..sub.no + e.sub.n

where .omega..sub.n is the optimum value of the tap weight and e.sub.n represents the error in the tap weight.

Substituting the immediately preceding definition into the second to last equation set out above yields: ##EQU7## Noting that by the definition of G(kT) that ##EQU8## the preceding equation for the partial differential can be written as follows by substituting this definition for G(kT) therein: ##EQU9## again l.ltoreq.j.ltoreq.N.

Expanding this last equation yields: ##EQU10##

Bringing the e.sub.n terms out of the summation over K and interchanging the order of summation yields: ##EQU11##

The summations with respect to k in the preceding equation can be recognized as autocorrelation functions of the discrete input signal which, over sufficiently large K, can be assumed to approach zero except where n=j for the class of discrete input signals of concern, i.e. an uncorrelated discrete time input signal.

Thus, when taking statistical expectations, the value of the first bracket term is equal to the average signal power per sample (which will be represented by the symbol x.sup.2) and the value of the second bracket term is equal to zero. Hence, the preceding equation reduces to the following for l.ltoreq.j.ltoreq.N: ##EQU12## where X.sup.2, the discrete input signal power, is always a positive quantity.

Thus, the sign of ##EQU13## provides, after averaging over a sufficient number of samples, the sign of the error e.sub.j as is required for controlling the adaptive process.

Thus, if there exists an error in the tap weight values, the system of FIG.