SU807320A1 - Probability correlometer - Google Patents

Description

(54) PROBABLE CORRELOMETER

one

The invention relates to computing and is intended to determine the statistical characteristics of random processes and may find application in hydrometeorology, automation, communications, biology, medicine, radio electronics, and other fields of science and technology.

A probabilistic correlometer is known, built on the principle of probabilistic encoding of information and providing a more efficient calculation of the statistical characteristics of random processes. The correlometer contains a block of probability rounding, a number register, a block of match, triggers, a block of cumulative counters, a random number generator, a block of probability multiplication, AND a control block 1.

The disadvantage of the device is low accuracy.

A probabilistic correlometer is also known, which has a wider scope of use and can be used for statistical processing of non-stationary random processes.

The correlometer contains an adder, a comparison block, an AND block,

block of probability multiplication, block of memory, block of control, accumulating register, multi-bit shifting register, random number generator, registers 2,

However, the drawback of the device is its complexity,

The purpose of the invention is to simplify the correlometer, which is expressed in replacing a multi-bit shift register with a one-bit register, which is. It eliminates the number of multi-bit blocks of AND elements and changes in the connections between the elements of the device.

five ; This goal is achieved by the fact that, in a probabilistic correlometer, i containing six blocks of elements and shifting (the register, the output of which is connected to its input, adder,

0 the first input to; the mountain is the input of the correlometer; the second and third inputs of the adder are connected respectively to the outputs of the first and second blocks of the AND elements, the information input of the first) block of the AND elements is connected to the information output of the accumulating register, the information input of the second block of the AND elements is connected to the information way out

0 register whose input is connected to

the output of the third block of elements And, the information input of which is .-: connected to the output of the accumulating register, the input of which is connected to the output of the fourth block of elements of And, whose information input is connected to the output of the adder, a random number generator, the output of which is connected to the first input of the comparison block the input is connected to the first output of the control unit, the second output of which is connected to the control input of the memory unit, the information input of which is connected to the output of the probability multiplication unit, the output of the adder n with the second input of the comparison unit, the output of which is connected to the input of the shift register, made single-bit, and to the information inputs of the fifth and sixth blocks of elements And, the outputs of which are connected respectively to the first and second inputs of the probability multiplication unit, the control inputs of the first, , the third and fourth blocks of elements And are connected to the third output of the control block, the second output of which is connected to the control inputs of the fifth and sixth blocks of elements I.

 The drawing shows a block diagram of a probabilistic correlometer.

The probability correlometer contains an adder 1, blocks 2 and 3 of the elements I, a one-bit shift register 4, accumulating register 5, a control block 6, a block 7 of comparison, a random number generator 8, a block

9 power multiplication block

10 spam, register 11, blocks 12-15 elements AND,

 The device works as follows.

Before starting, all the blocks are reset to their original (zero) position. The values Xj of the random process X (t) are fed to the input of adder 1, where they are centered on the current value of the expectation, stored, in register 11.

The centered value Xj arrives at the first input of the comparison block and the second input of the last receives a random number from the generator of 8 random numbers. From the output of block 7 C1I1, there is a probability-coded Xj, which is fed through the fourth block 13 of the AND elements to the first input of the block 9 of probable multiplication. The second input of the block in each of the N clock cycles (where N is the number of register bits 4, which coincides with the number of ordinates of the calculated correlation function) receives a block of 14 elements AND the probable coded values X f, X., ... X, stored in register 4 and coming from its output

at each of N shifts during one cycle.

From the output of block 9, the product of probabilistic-encoded process values is fed to block 10 of memory in which they are summed over each of the N channels.

In each processing cycle of a random number X, its centered value is added on adder 1 with the contents of accumulating register 5. Thus, by the end of the Kto cycle in register 5, the sum

, -iX

E% / - i --ii

where K is the number of cycles (random numbers of the array X (t)), which determine the current value of the metrical expectation;

m; is the expectation value calculated in previous K cycles; increment (positive

Hj or negative) expectation of K cycles from J-1 to j of the array;

j-l; j-numbers of arrays of X-magnitude K. At the end of the Kth cycle, control block 6 sequentially connects blocks 2, 3, 12, and 13 elements And in such a way that the increment + t; - - obtained in accumulating register 5 through K cycles algebraically adds to the contents of register 11. Before the start of the next K cycles, the cumulative register 5 is zeroed out.

The value of K is usually taken to be a multiple of 2 and equal to N, since in this case it is possible to synchronize the work of the deterministic and stochastic (from the point of view of information) parts of the devices most easily.

With this hardware implementation of the calculation of the correlation function, the centering of the first K ordinates of the random process X (t) is performed at zero values m, however, with large arrays of random numbers it does not matter much.

The simplification of the correlometer as compared with the known device is achieved due to the fact that the coding of the information in a probabilistic form in the proposed device is carried out at a level closer to the input, namely, immediately after centering X.

It was this that made it possible to peal a shift register in the form of N triggers, each of which contains a probabilistic coded value of a centered value.

input signal, and if we assume that the input process is input to the device as R-bit binary numbers (with single-byte X - R 8), the number of triggers in the proposed device is reduced by N (R - 1) units, which will be , for example, pr, and N 64 and R 8 more than 400 triggers. In this case, without prejudice to the accuracy of the calculations, one of the two (contained in the known device) R-distance comparing units is excluded.

At the same time and independently of the change in the structure of the stochastic part in the proposed device, the process of calculating the current expectation is somewhat changed.

The subsequent value of t. is defined as the sum of the previous three differences, which is found as the sum of centered values, divided by K. The new implementation allowed to exclude from the circuit diagram of the (R + Rod2K) -discharge valve block.

It should be noted here that these changes did not impair the device performance (as compared with the known) due to the fact that the deterministic and stochastic parts work almost in parallel, and the period of operation of the stochastic part of the device is approximately N times the period of its determinated part .

Claims (2)

1. USSR author's certificate No. 542184, kp. G About F 15/34, 1975. 2.Assignment of the USSR on the application 2534281 / 18-24, cl. G About F 15/34, 1977.