SU1653132A1 - Digital interpolating filter - Google Patents

Description

(21) 4472360/09

(22) 07.12.88

(46) 05.30.91. Bul V 20 (72) N.N. Kozlov, Yu.A. Korneev, D.Yu.Krasnykh, R.A.M. lk and V.V. Tomashevsky

(53) 621.396.6 (088.8)

(56) Goldenberg L.M. and others. Digital signal processing. Directory. M.: Radio and Communications, 1985, p.53, fig.2.6.

(54) DIGITAL INTERPOLATING FILTER

(57) The invention relates to experimental information processing means. The purpose of the invention is to enhance the functionality by providing control of the filter memory interval. The digital interpolating filter contains accumulative adder 1, divider 2, control unit 12 consisting of a counter 13 and a digital comparator 14, and a quadratic

an interpolator 16, consisting of registers 3-6, weight blocks 7-10 and adder 11. Through the use of controlled weight blocks 7-10, in which weights are controlled depending on the position of the current point in time relative to the interpolation nodes, Interpolator 16 is used to perform the symmetric quadratic interpolation (SRI) of the values of the current average of four counts. The use of SRS allows one to form, with a small total filtering error, an estimate of the average value of the processes with a dynamically varying in time transient average value. In this case, the filter memory interval is determined by the grouping interval of the samples nr. A change in the value of the binary code pgr supplied to the control input of the device provides for a change in the memory interval. 5 il.

from 8

(L

oh ate

00

. 00

Yu

The invention relates to the processing of experimental information and can be used in real-time processing of random processes, which are an additive mixture of a random low-frequency component and discrete white (high-frequency) noise.

The purpose of the invention is to enhance the functionality by providing control of the filter memory interval.

| FIG. 1 shows an electric circuit diagram of a digital interpolating filter} in FIG. 2 and 3 are time diagrams for his work; in fig. 4 shows an embodiment of a weight unit; Fig. 5 shows the plots of functions implemented by functional converters.

The digital interpolating filter contains accumulative adder 1, first divider 2, registers 3-6 from first to fourth, weight blocks 7-10 from first to fourth, adder 11, control unit 12, counter 13, digital comparator 14, second divider 15 and quadratic interpolator 16.

The weight block contains a functional converter 7-1 and a multiplier 7-2.

The digital interpolation filter works as follows.

The process | CL in the parallel binary code is processed to the input of the digital interpolating filter (Fig. 3a). The clock input of the interpolating filter is supplied to the clock input of the TI (Fig. 3b), which determines the step t of the time sampling of the input process. The binary code corresponding to the length of the grouping interval pg is fed to the control input of the filter, which is connected to the second inputs of the first 2 and second 15 dividers, the digital comparator 14.

The control unit 12, which includes a counter 13 and a digital comparator 14, generates pulses with a period pgr & t (Fig. 2b, c, fig.c), fed to the reset input of accumulating adder 1 and clock inputs 3, second 4, third 5 and fourth 6 registers.

At the output of accumulating adder 1, at the i-th time-sampling step, the sum

v (i)

Umf (, jW

(one)

| AND

R)

The current i-th interpolation step (and discretization) is represented as

1 p

+ v

(2)

rpj y de pgr - grouping interval,

j is the number of the nearest interpolation node located to the left of the current time sampling interval;

v - the number of the current interval

sampling counts from the nearest (left) interpolation node:

(3) where E («) is a whole part of the number.

At the output of the first divider 2. normalized values are formed (Fig. 3 d):

UN (1)

one

(four)

1- y (i)

Pgr

In the sampling intervals corresponding to the interpolation nodes, samples are formed at the output of the first divider 2 (FIG. 3 d):

and.,

(i) 4i

GR

yi

 GR

(five

The ones which are synchronized at the output of the digital comparator 14 (Fig. 3b) are recorded in the first 3, second 4, third 5 .and fourth 6 registers (in this case, successively connected registers .3-6 form the rdvig register).

The clock pulses at the output of the digital comparator 14 (Fig. D) are formed in the sampling intervals corresponding to the interpolation nodes.

The variable code v is formed at the output of the counter 13. At the output of the second divider 15, codes of the normalized values of the variable are formed

vH (i)

y

P

one

gr Gr

-E (-i-) pgr

(6)

Codes for the variable vH (i) are fed to the second (control) control inputs of the first 7, second 8, third 9 and fourth 10 weight blocks, the first inputs of which are supplied with binary sample codes, respectively.

V Уц J УН У „, J-2.

Due to the use of controlled weight blocks, in which weights are controlled depending on the position of the current point in time relative to the interpolation nodes, the interpolator 16 performs symmetric quadratic interpolation of the values of the current average mx (i) over four counts UN, J + 1 Un, J, Un, J-1, Un, J-2, supplied to the first inputs of the weight blocks.

Symmetric quadratic interpol qi is implemented as follows:

a) on three counts y. ,, j-2, un, J-1 UN J the left interpolating polynomial is formed

P, i, Ui, J-2, Vn, J-1, Un. j

 UI, J-2 + Ui. J-1 - Um ,. x (1-v, (i)) + (un, j-2 - 2un, J-1 +

+ UN J) VH (i) (1 + vH (i)

(7)

f) on three counts un, j-1, un, .1, Hz J + 1, a right-interpolating polynomial is formed

Un, J-1, Un.J UN J + 1J

 Woo, j-1 + (un, j-un, j-D-vH (i) +

+ (Un. J-1 + 2un, j + ui, J + D- x vH (i) (4 |) i) -1), (8)

In expressions (7), (8), the variable Vu (i) is determined according to formula (B).

The interpolating polynomial symmetric with respect to the estimation interval (j-1) ... j is formed as the average of the left and right interpolating polynomials:

P (i) JJP, (vM (i) + P2 (v (i))

(9)

1653132 "

l ... Evaluation of the current average (.i; in

each time point i is equal to the value of the interpolating polynomial P (i):

5 mx, (i) Winto) P (i, UN, j-2, un, J-1, Y „, J, UN. J + 1 1 °)

0

five

0

five

0

five

0

five

0

five

The formation of the left P, (i), right P2 (i), and symmetric P (i) interpolating polynomials is illustrated by the plots of FIG. Substituted the expression (7), (8) in (9) and the drive like members, you can write

P (i) A (vH (i)) vH, j-2 + B (vH (i)). YM, j-1 + C (vM (i)) yH, j -f D (vH (i)) x

 Un, J + 1CD

The formation of each addend that is included in the expression (10) is carried out by controlled weight blocks 7-10, the first input of which is fed by the counts un, and the second input is the variable VH (I). Due to this, the values of the coefficients A, B, C, D in formula (11) change depending on the position of the current time point i relative to the interpolation nodes (JO ... J, the interval of arguments between which is the interpolation interval. On transition of the argument i, through the interpolation node, four counts of the meterings one meter participating in the interpolation procedure are shifted one node to the right.

The use of symmetric quadratic interpolation allows one to form, with a small total filtering error, an estimate of the average value of the processes with a dynamically varying in time transient average value. In this case, the memory interval of the interpolating filter is determined by the grouping interval of the samples Pgr,

A change in the smoothing interval (memory intervals) in the proposed device does not entail a change in the amount of hardware costs and is provided by changing the value of the binary code for the GTP supplied to the control input of the device. In this case, the period of the pulses from the output of the digital comparator 14, which correspond to the distance of the interpolation nodes on the time axis (Fig. 3 c, d, e), changes.

The structure of the weight unit 7 directly follows from the expression (11) and is shown in FIG. 4 (weight blocks 8-10 have a similar structure), where 7-1 is a functional converter that implements the function A (vH) (blocks 8-10 implement the functions B (v), C (vrt), D (vH), respectively), 7-2 is a multiplier.

Functions implemented by functional converters 7-1, 8-1, 9-1, 10-1 are as follows:

..;

Ј.

4-3 v n - VH.

five;

2

5vH -VH. ,

VH (v and 1) 5

(12) (13)

(14) (15)

Claims (1)

From the above expressions it is clear that A (vH) D (vH). The graphs of the functions A (vH), B (vH), C (vH), D (vh) are shown in FIG. 5. Invention Formula A digital interpolating filter containing a quadratic interpolator made in the form of an adder, the first, second, third, and fourth weight blocks and the first, second connected in series, five 0 the third and fourth registers, the outputs of which are connected through the respective weight blocks with the first, second, third and fourth inputs of the adder, respectively, as well as the control unit, the first output of which is connected to the clock inputs of the first to fourth registers, characterized in that expanding the functional capabilities due to the control of the filter memory interval, a series accumulating adder and the first divider are introduced into it, the output of which is connected to the input of the first regul and a second divider, the output of which is connected to the second inputs of all weight units, and the control unit is designed as a series-connected counter, the input of which is the clock input of the control unit and connected to the clock input of the accumulating adder, and a digital comparator, the output of which is the first output of the control unit and is connected to the reset inputs of the accumulating adder and counter, the output of which is connected to the first input of the second divider, the second input of which is connected to the second inputs of the first divide l and the digital comparator and is the control input of the digital interpolating filter, the input of which is the progress of the accumulating adder. 0 five TI Ai Fm.1 / 0hd . (fi. () d + ((jd fi V 9-f L-f 9-f f-.f b - f c-S z-f V Olh p / f. . f.Mfi 6Hfi i i. i j. it / ji  WXI4Q B   Hjt a, / (g Zirgwg  ; 1 I I I I I I I I I I I I I I I I I I I I I M | I And And i1     7 j IIIIIIIIIIIIIIIINIIIINIIIII 7; P From 277 $ WWfiWd  -L. # wow I; /.  WXI4Q (g Zirgwg  i1 I 7 / w RVRS () w to x ; rif.54i Phage. five