SU398950A1 - DEVICE FOR DETECTION AND CORRECTION OF ERRORS IN THE SYSTEM OF RESIDUAL CLASSES - Google Patents

Description

one

The invention relates to a computing technique, in particular, "devices for detecting and diffusing error-checking in the system of residual classes, M1 can be used in the construction of reliable digital devices.

A device for detecting and correcting information shredded in a system of residual classes is known. The device contains an input register, adders on test m bases, the outputs of which are connected to the I / O m of the analysis block — to 1 zero of the remainder on the control basis: m and the block “stant corrected errors connected to the error location block” -bki IT is characterized by sequential operation; nullification to detect erroneous figures in the original number and inadequate speed.

The described device is distinguished by the fact that there are two blocks in its “B | Veda”. An example of current number, outputs - which are connected to the accumulators on the test bases, a correcting block, a key input to the input register, and an output mi | Klyuche1N | NYY to the entrances of the su | .m, matorov on the control oopov and m. and the input m of the computing units of the net rank. This increases the speed of the device.

In the drawing, an example of the block diagram of the device for monitoring and correcting single modular units.

mistakes with a bad thing: a computation of the rank in one of the variants of the | | Neni.

For a more complete explanation of the operation of the device, a description is given of the algorithm for detecting and correcting errors using inaccurate rank values.

Alignment algorithm. A device for the formation of a pa is essentially reduced in equipment if, instead of an exact rank and a non posi tion code with n symbols, to form its approximate value For, we take J - i +; - where c is an indefinite value that can take the value "O or" 1.

Let a nonpositional code (HI, 02, ..., au) be given. By a single modular error, or simply by a modular error, is meant an arbitrary distortion of one of the symbols of the nonpositional code. In this case, it is assumed that the distorted symbol belongs to the clause of symbols in the module under consideration. Under the expansion operation, the pharmacization of the deduction modulo the number of l, represented by the opposition code (aa oa, ..., an) on the bases of PI, p2, ..., P., olera expansion, requires exact restoration of the value of the unshared code (o

02, ...., On).

The equipment of the form-exact rank is measurable with the equipment for converting the number represented by the “positional code” into the non-positional one.

The device is used in the process of decoding and expansion operations with a 1-line basis: by increasing the rank value, thereby achieving a significant reduction in the equipment of the decoding device.

The possibility of constructing such a device is justified by the following 1P | Supplement M. Conditional b.Values and:

| L: | P - the smallest, non-negative deduction of an integer — a, x, a modp;

q

whole part from dividing integer j

la X on the NUMBER of p, as a result of one, the remainder is obtained

t, PnQKiiKz is the code range of the AU of a digital computer, where Ki, K2 are test bases, Q Pn -} - l;

-3Pii - V; 3Pn2 - the range of valid values of the number of numbers x.

Let I be the syndrome of perversion | 5ki, formed on the basis of the expansion operation from bases n to KiKz and from bases Q to / Ci / C2 with a non-volume definition of ranks. Then the syndrome t, modular 0SH1gI KI has the structure (/), if the Vi error occurred on the total from the ooHOBaiHHU gPnQ range,

+) l

IMKiKzKil 2, L Pi

(ii) if the error VKi occurred on the control base K,

| SA / (i - / (2lXl / 2,

S - I b (similarly, if the error V / Co occurred on the basis of Kg),

, | Л, / С2,

where (/ (, - 1Д / С, {P QKz} - i К ,.)

As it follows from this P-presentation, ra: zbros syndromes associated with | p1p; MI | irova.viem (Not accurate rank, not similar (determined by the parameter p). Analytical form of the syndromes indicates the algorithm it chooses the selection of test grounds K, Kz, satisfying the requirement of separation of errors.

The Control and Correction Information device co, de f (see drawing) input register 1, designed for temporary storage; n 1I -degrees of x in the corresponding modules separately; a device 2 for the xy-aneni correction constants, which are unilateral long-term memory devices for 1 each module; block 3, designed to store constants; 4 calculations of inaccurate rank of the number when expanding from the bases of tPn to QK.K.2, consisting of summator at 5, 6 h of scheme 7 a1} 1nnza transfer from higher order, connected in parallel-series with each other; a block 5 for calculating an inaccurate number rank when it is cleared from the bases of Q to the summation of the matrices 9, 10 of the high-order transfer analysis circuit 11, connected in series-parallel to each other; adders 12, 13 on the test basis 1М, Intended for the sum of world constructions during the expansion. С Р, on QKiK, coming from the device 2 and block 4; adders 14, 1-5 on reference bases, designed to sum the constants when expanding from Q to Pn / i / C2, coming from device 2 and block 3; a correction coder unit 16, which is the KOHCTaiHT long-term rectifying memory, verifying the number of x; the analysis unit i / 7 to zero of the difference of digits on the test bases when expanding from t, Pn to QKiKz and the digits. on the test base of the expansion scale from Q to, P1 is intended for processing the signal of the presence or absence of an error in the tested number. Code of the checked number l; buses are received at the input register 7. The output signals of the register are fed to (inputs of the storage unit 3 of constants. AddressoF for addressing the block of constants serves a deduction code recorded at the input register on a corresponding basis. base. From the outputs of storage unit 3 of constants The check UN codes for the control UNOs. M come for the inputs of adders J2, 13, 14, 15, and the codes for the constants designed to calculate inaccurate number rant,. enter the inputs c} of the mattors 5, 6 and 9 , 10 blocks 4 and &.

Summaries 5, 6 TL 9, 10 They are n one-digit dry meters.

The sum of the cyiMiMa mat and the transfer produced in each discharge are transferred as One of the terms of the same name and the next one, respectively, of the subsequent adders. The transfer signals produced in the older bits of the adders 5, 6, 9, 10, as well as the signal from the output of Schemes 7, 11 of the analog-transfer, are fed to the inputs of the adders on the test bases of names 12, 13 and 14, 15, as these signals. are components of the inexact welt of the number x. Adders 12, 13 and 14, 15 add up the incoming On x inputs codes for the corresponding control bases as the number expands into the one and the other. From the outputs of the sum of the controllers, but for the control base, the signals. Are supplied to the inputs; block 16 const. Ont correction and block / 7 of the analysis (Zero.

The zero analysis block 17 is a logic circuit of the right-of-code codes of the control logic bases, coming from the outputs of the adders 12 and M, IS and 75.

If these codes are nonaipiHo equal, the output of block 17 produces a signal of the correctness of the tested number. In the opposite case, an error signal is generated. Both of these signals are output to the control circuit of the device. M -control and corrections. At the outputs of block 16 of the correction constants, the code of correction of the tested number x is generated if there is an error in it. The address for addressing the block / 5 correction constants is

SIZE 10 OF 15 QUANTITY but control OCiiOBailHHM,

produced by the outputs of the summators / 2, 13 and 14, 15.

Subject invention

A device for detecting and correcting errors in the system of residual classes, containing input iperHCTp, adders on test bases, the outputs of which are connected to the inputs of the analysis unit and the residual zero. On the control equipments of the block of correction, which is different in that In order to increase the speed, the number of the brush-raig computations has been entered into it, two outputs connected to adders

on the control signals, the storage unit of the constants by the correction, the inputs connected to the input register, and the outputs connected to the inputs of the adders on the test bases and the inputs of the calculation blocks

rank.

Nod norrmtstsii

Signal