Probability device

FIELD: radio engineering and computer science, possible use in complexes of automated systems for controlling multichannel radio communication networks and in devices for processing and transferring data of local computing networks.

SUBSTANCE: probability device contains indicator of random series 1, block for forming non-integer indicator values 2, correction block 3, block for forming values of matrix 4, control block 5, threshold devices block 6, block for forming indicator values 7, clock pulses generator 8, DENY element 9, AND elements block 10, memory block 11, decoder 12, time setting block 13, OR element 14, block 15 for increasing trustworthiness. Device makes it possible to model controllable semi-markov circuits with high trustworthiness with consideration of controlling effects, and dynamically changes threshold values of states, set both numerically and qualitatively, and untrustworthily, due to serial comparison of source data, received as binary code; taking of decision is possible about their mathematical nature and also transformation of source data in block 15, given in incorrect manner, to form, useable for parametric modeling procedure realization.

EFFECT: increased trustworthiness of parameters modeling of real functioning process of multichannel radiocommunication networks and local area networks under conditions of untrustworthy (not full) given data, of probability device, capable of highly trustworthy modeling of controllable semi-markov circuits, formed with consideration of both quantity and quality of given data, describing threshold values and probability-time mechanism (elements of transfer probability matrices) of status changes in modeled random processes.

3 cl, 6 dwg

The invention relates to radio engineering and computer science and is intended for use in the complexes of automated control systems networks multi-radio (SMCRS) and in the media processing and data transmission in local area networks (LAN).

The probabilistic machine auth. St. USSR №1045232, G 06 F 15/36, 1983, bull. 36, comprising a generator of clock pulses, items, And and, OR, shift register, memory blocks and set a time.

However, this probabilistic machine has a narrow scope, because unmanaged models semi-Markov chain without regard to external control actions, which does not allow to use it for the analysis of real processes in networks of multi-channel radio and LAN, because these processes are mostly manageable.

Known probabilistic machine (patent RF №2099781, G 06 F 17/00, 1997, bull. 35)containing block elements And the memory unit, the unit time, an element of the BAN, the element OR the sensor of a random sequence, the shaping unit corrective sequence, the error correction block, the block forming the values of the matrix, the set of indicator values, the control unit, the decoder and the clock.

However, this probabilistic machine also has limited use as a model which induces controlled semi-Markov processes with once and for all predetermined threshold (boundary) conditions - not able to dynamically adjust the boundary conditions of the simulated process, which does not allow to use it for dynamic analysis of real networks multi-radio and LAN, because a large number of processes in managed networks multi-radio and LAN can in the dynamics of functioning to modify the thresholds of their condition under the influence of control actions or external factors.

The closest to the technical nature of the claimed device (prototype) is a probabilistic machine (see RF patent №2139569, G 06 F 17/18, 1999, bull. 28)containing a sensor of a random sequence, the block formation non-integer values of the indicators, the error correction block, the block forming the values of the matrix, the control unit, the block of threshold devices, the set of indicator values, the block elements And the memory block, the block set a time, the decoder, the item BAN, element OR generator of clock pulses, the output of which is connected to the direct input element PROHIBITION and a clock input unit of time, m≥3 outputs which are respectively m outputs of the machine and connected to the corresponding m inputs of the OR element, the output of which is connected to the negative input item BAN, the output of which is connected to the clock inputs of the block And the block f is Mirovaya indicators and control unit, the control input of which is a control input of the machine, and a control output connected to the inputs of the decoder and block the formation of values of the matrix m output processing unit values of indicators connected to respective m inputs of the block elements And m outputs of which are connected to the corresponding m inputs of the memory block, the m outputs of which are connected with the corresponding m inputs of the unit time, the control input of which is connected to the output of the decoder, m control inputs of the block formation non-integer values of indicators are connected with the corresponding m output processing unit values of the matrix m interpolation unit forming a non-integer values of indicators connected to the corresponding m outputs of the memory block, m outputs of the block formation non-integer values of indicators connected to respective m auxiliary inputs of the processing unit values of indicators and m information inputs of the error correction block, the start input of which is connected to the sensor output of a random sequence, the output of the correction block is connected to the input of the threshold device, the m outputs of which are connected to the corresponding m information input processing unit values of indicators, m-1 threshold outputs a control unit connected to line the existing m-1 threshold inputs of the error correction block and the block of threshold units.

The advantage of the prototype is its ability to simulate controlled semi-Markov chain with dynamic threshold adjustment (boundaries) of States of random processes in managed SMCRS and LAN.

However, the prototype has the disadvantage of relatively low reliability modeling parameters of the real process of functioning networks multi-radio and LAN in terms of unreliability (failure, incompleteness) of the original data, i.e. relatively low reliability modeling States SMCRS and LAN unreliable (insufficient, incomplete) set parameters. This is because the simulation is based on the quantitative input of source data values (elements of the matrix of transition probabilities, thresholds (border States), which makes inefficient use of the prototype for dynamic multi-criteria analysis of real networks multi-radio communication and LAN. This probabilistic machine allows you to simulate controlled semi-Markov chain with dynamic threshold adjustment (borders) state only those random processes, the original data set quantitatively, while the initial data for the simulation of a large number of processes actually take place in controlled Smcrs LAN can be set only qualitatively (at a qualitative level - of inaccurate, insufficient, incomplete), based on the opinions of a large number of experts.

The aim of the invention is the creation of a probabilistic automaton, providing increased reliability modeling parameters of the real process of functioning networks multichannel radio communications and local area networks in terms of unreliability (failure, incompleteness) of the original data, probabilistic machine capable of high accuracy to simulate controlled semi-Markov chain generated by taking into account both quantitatively and qualitatively (false, incomplete) set of initial data, describing threshold (boundary) and probabilistic-temporal mechanism (elements of the matrix of transition probabilities) change the state of the simulated random processes.

This goal is achieved by the fact that in the known probabilistic machine that contains the sensor of a random sequence, the block formation non-integer values of the indicators, the error correction block, the block forming the values of the matrix, the control unit, the block of threshold devices, the set of indicator values, the block elements And the memory block, the block set a time, the decoder, the item BAN, element OR generator of clock pulses, yhod of which is connected to the direct input element PROHIBITION and a clock input unit of time, m≥3 outputs which are relevant m outputs of the machine and connected to the corresponding m inputs of the OR element, the output of which is connected to the negative input element PROHIBITION, the output of which is connected to the clock inputs of the block elements And block the formation of indicators and control unit, the control input of which is a control input of the machine m outputs of the processing unit values of indicators connected to respective m inputs of the block elements And m outputs of which are connected to the corresponding m inputs of the memory block, the m outputs of which are connected with the corresponding m interpolation unit forming a non-integer values of indicators and m the unit time, the control input of which is connected to the output of the decoder, m control inputs of the block formation non-integer values of indicators are connected with the corresponding m output processing unit values of the matrix m outputs of the block formation non-integer values of indicators connected to respective m auxiliary inputs of the processing unit values of indicators and m information inputs of the error correction block, the start input of which is connected to the sensor output of a random sequence, the output of the correction block is connected to the input of the threshold device, m output is in which is connected to the corresponding m information input processing unit values of the indicators introduced additional unit increase reliability. The control input unit increase reliability connected to the control output of the control unit, m-1 threshold outputs of which are connected with the corresponding m-1 threshold input unit increase reliability, control output which is connected to the inputs of the decoder and block the formation of values of the matrix m-1 threshold output unit increase reliability connected to the corresponding m-1 threshold inputs of the error correction block and the block of threshold units.

The power increase reliability consists of a controller of the source data and transducer unreliable data, m-1 false threshold input of which is connected to respective m-1 false threshold and outputs the original data, inaccurate control output which is connected to inaccurate control input of the Converter unreliable data. The control input of the controller, the source data is a control unit that controls the output of the controller data source is connected to control the output of the Converter false data, and is the managing unit output, m-1 threshold inputs of the controller of the source data are the corresponding m-1 threshold input block, m-1 threshold outputs of the Converter unreliable data connected with soo the relevant m-1 reliable threshold and outputs the original data and are the corresponding m-1 threshold output unit increase reliability.

The set of values of the matrix consists of the storage element and the counter. The counter input is the input block, the output of the counter is connected to the input storage element, m outputs which are relevant m outputs of the block.

The control unit consists of a control memory element (OSE), the threshold OSA and counter. The first and second outputs of the counter connected to the clock inputs, respectively, of the control and the threshold ONE, the first and second inputs of the counter is connected with the reset outputs, respectively, of the control and the threshold ONE, the third input of the counter is clocked by the input unit, the information input Manager and the threshold ONE United and are managing input unit, the output control OSA is the managing unit output, m-1 outputs the threshold OSE are the corresponding m-1 threshold outputs of the block.

Thanks to the new essential features by introducing a unit increase reliability, provide a consistent comparison (number of bits) of the incoming binary coded source data, decision-making about their mathematical nature and transformation of the original data, in the proposed probabilistic machine allows a preliminary analysis and transformation of source data, the data are unreliable (incomplete) to the form suitable for parametric modeling, which causes an increase in the reliability modeling of the process of changing the state of the network multi-channel radio communications and local area networks in the conditions inherent in the real process of their functioning in the conditions of unreliability (failure, incompleteness) of the original data.

The analysis of the level of technology has allowed to establish that the analogues, characterized by a set of characteristics is identical for all features of the claimed technical solution is available, which indicates compliance of the device to the condition of patentability "novelty".

Search results known solutions in this and related areas of technology in order to identify characteristics that match the distinctive features of the prototype of the features of the declared object, showed that they do not follow explicitly from the prior art. The prior art also revealed no known effect provided the essential features of the claimed invention transformations on the achievement of the technical result. Therefore, the claimed invention meets the condition of patentability "inventive step".

The claimed device is illustrated by drawings, on which:

figure 1 - structural the scheme is managed probabilistic machine;

figure 2 - block diagram of the block to increase reliability;

figure 3 - block diagram of the controller of the source data;

figure 4 - structural diagram of the Converter of false data;

figure 5 is a structural block circuit diagram of the formation of values of a matrix;

figure 6 - block diagram of the control unit.

Managed probabilistic machine shown in figure 1, consists of a sensor of a random sequence 1, block formation non-integer values of indicators 2, error correction block 3, block the formation of values of the matrix 4, a control unit 5, block threshold device 6, the processing unit indicators 7, clock pulses 8, item BAN 9, block 10, block memory 11, a decoder 12, a block of time 13, item, OR 14 unit increase reliability 15. The output of clock 8 is connected to the direct input element PROHIBITION 9 and a clock input unit of time 13, m≥3 outputs which are relevant m outputs of the machine and connected to the corresponding m inputs of the OR element 14, the output of which is connected to the negative input element PROHIBITION 9, the output of which is connected to the clock input of block elements And 10, a clock input 73 forming unit indicators 7 and clock input 54 of the control unit 5, the control input 51 to the th is a control input of the machine, the m outputs of the processing unit indicators 7 are connected with the corresponding m inputs of the block elements And 10, m outputs of which are connected to the corresponding m inputs of the memory block 11, m outputs of which are connected with the corresponding m interpolation inputs 22₁-22_mblock formation non-integer values of indicators 2 and with m inputs of the block set a time 13, the control input of which is connected to the output of the decoder 12, m control inputs 21₁-21_mblock formation non-integer values of indicators 2 connected to respective m outputs 41₁-41_mprocessing unit values of the matrix 4, m outputs 23₁-23_mblock formation non-integer values of indicators 2 connected to respective m auxiliary inputs 72₁-72_mprocessing unit indicators 7 and m information inputs 31₁-31_merror correction block 3, the starting input 33 which is connected to the sensor output random sequence 1. The output 34 of the correction block 3 is connected to the input of the threshold device 6, m outputs of which are connected to the corresponding m information input processing unit indicators 7. Control input 150 unit increase reliability 15 is connected to the control output 52 of the control unit 5, m-1 threshold outputs 53_{1 -53m-1which is connected to the corresponding m-1 threshold inputs 1531-153m-1unit increase reliability 15, control output 151 which is connected to the input of the decoder 12 and the input 40 of the processing unit values of the matrix 4, m-1 threshold outputs 1521-152m-1unit increase reliability 15 connected to the corresponding m-1 threshold inputs 321-32m-1error correction block 3 and m-1 threshold inputs 621-62m-1block threshold device 6.}

The number "m", "m≥3" (inputs, outputs, adders, multipliers, counters, comparison and so on) is determined in accordance with necessary and sufficient granularity of the state space (number of States) of the simulated process and typically ranges from 3 (three) up to 20 (twenty).

The power increase reliability 15 (figure 2) is designed to implement procedures consistent comparison (number of bits) of the incoming binary coded source data, decision-making about their mathematical nature and transformation of the original data set false (incomplete), to a form suitable for parametric modeling.

The power increase reliability 15 consists of a controller of the source data 1501 and Converter unreliable data 1502, m-1 false threshold inputs 152.2 1-152.2_m-1which is connected to the corresponding m-1 false threshold outputs 151.6₁-151.6_m-1controller source data 1501. Inaccurate control output 151.5 controller data source 1501 is connected to inaccurate control input 152.1 Converter unreliable data 1502. Control input 151.1 controller data source 1501 is a control input 150 unit 15, the control output 151.3 controller data source 1501 is connected to control the output 152.3 Converter unreliable data 1502 and is control the output 151 of the block 15, m-1 threshold inputs 151.2₁-151.2_m-1controller source data 1501 are the corresponding m-1 threshold inputs 153₁-153_m-1unit 15, m-1 threshold outputs 152.4₁-152.4_m-1Converter unreliable data 1502 connected to respective m-1 reliable threshold outputs 151.4₁-151.4_m-1controller source data 1501 and are the corresponding m-1 threshold outputs 152₁-152_m-1unit increase reliability 15.

The controller of the source data block 1501 improve the reliability 15 (figure 3) is designed to perform a consistent comparison (number of bits) of the incoming binary coded source data and making decisions about their mathematical nature of the source the data set quantitatively (parametrically) or qualitatively (false, incomplete) using analytically formulated subjective opinions (associations) of experts about specific thresholds (borders) and elements of probabilistic time-space mechanism (elements of the matrix of transition probabilities) change the state of the simulated random processes. The controller data source 1501 is composed of the shifting element 1510 and storage element 1511. Entrance 151-1 shifting element 1510 is a control input 151.1 controller source data 1501, direct 151-2 shifting element 1510 is a control output 151.3 controller source data 1501. Transit output 151-3 shifting element 1510 is inaccurate control the output 151.5 controller source data 1501, m-1 inputs 151-4₁-151-4_m-1storage element 1511 are m-1 threshold inputs 151.2₁-151.2_m-1controller source data 1501, m-1 direct outputs 151-5₁-151-5_m-1storage element 1511 are m-1 reliable threshold outputs 151.4₁-151.4_m-1controller source data 1501, m-1 transit outputs 151-6₁-151-6_m-1storage element 1511 are m-1 false threshold outputs 151.6₁-151.6_m-1controller source data 1501. Shift element 1510 may be technically implemented as a shift register to shift to the left as shown in the literature [Sidora the A.M., Gusev V.V., Lebedev O. Fundamentals of pulse and digital techniques. - SPb.: SUVIUS, 1995. S-160, RES(6)]. The storage element 1511 can be implemented on the basis of the model of the storage register at D-triggers described in the literature [Sidorov A.M., Gusev V.V., Lebedev O. Fundamentals of pulse and digital techniques. - SPb.: SUVIUS, 1995. S-158, RIS].

Converter unreliable data block 1502 improve the reliability 15 (figure 4) is intended for the procedure of transformation of the initial data set false (incomplete), to a form suitable for parametric modeling. Converter false information 1502 is a programmable permanent memory element solver (PZEV) allow one entrance, one permissive access, m-1 inputs and m-1 outputs, where m can take values from 3 to 20. Enable input V PZEV is unreliable managing input 152.1 Converter unreliable data 1502, allowing the S PZEV is control the output 152.3 Converter unreliable data 1502, m-1 inputs and m-1 outputs PSM are respectively m-1 false threshold inputs 152.2₁-152.2_m-1and respectively m-1 threshold outputs 152.4₁-152.4_m-1Converter unreliable data 1502. Converter inaccurate data is 1502 may be implemented on the basis of programmable permanent memory device series 155 (e.g., CPR), as shown in [awl V.L. Popular digital circuits. The Handbook. - M.: Radio and communication, 1987. P.172, RIS].

The set of values of the matrix 4 (figure 5) is designed to check incoming values of the elements of the matrix of transition probabilities, storage, reading management information and consists of a storage element 401 and the counter 402. The input of the counter 402 is input 40 unit 4, the output of the counter 402 is connected to the input of storage element 401, m outputs which are relevant m outputs 41₁-41_munit 4. The memory element 401 may be implemented as a storage device, described in [Bystrov Y.A., Welikson AM, Wagman E, and other electronics: Reference book / Ed. by Bystrov Y.A - SPb.: Energoatomizdat, 1996. S-292, RIS]. The counter 402 can be implemented on the basis of commercially available counter, as shown in [H. Sobotka, Old Ya Microprocessor systems. - M.: Energoizdat, 1981. Pp.96-100].

The control unit 5 (6) is designed to generate parametric or incorrect (incomplete) control code sequence, as well as parametric or false (partial) sequence of threshold values of the state and consists of the control ASA 501, the threshold ASA 502 and counter 503. The first 5.1 and 5.3 second outputs of the counter 503 is otklucheny respectively to the clock inputs 5.6 managing OSA 501 and 5.9 threshold of OSA 502. The first 5.2 and 5.4 second inputs of the counter 503 are connected respectively with the reset outputs 5.7 managing OSA 501 and 5.10 threshold of OSA 502, the third entrance 5.0 counter 503 is a clock input 54 of the block 5. Informational inputs 5.5 managing OSA 501 and 5.8 threshold of OSA 502 are combined and managing input 51 unit 5, the output 5.12 managing OSA 501 is a control output 52 of block 5, m-1 outputs 5.11₁-5.11_m-1the threshold of OSA 502 are the corresponding m-1 threshold outputs 53₁-53_m-1the control unit 5. Managing OSA 501 can be technically implemented as a standard memory device, which is described in detail in [H. Sobotka, Old Ya Microprocessor systems. - M.: Energoizdat, 1981. S-96]. A special case of the technical implementation of the counter 503 is described in [H. Sobotka, Old Ya Microprocessor systems. - M.: Energoizdat, 1981. Pp.96-100]. Technical implementation the threshold ASA 502 is possible by analogy with the commercially available dynamic random access memory device described in [Sidorov A.M., Gusev V.V., Lebedev O. Fundamentals of pulse and digital techniques. - SPb.: SUVIUS, 1995. P.194-196, RIS].

The sensor random sequence 1, part of the General structural scheme is designed to generate random values of the auxiliary sequence with the normal density distribution of the population and can be implemented in the form of a commercially available sensor, described in [Maltsev L.A., Tranberg AM, Yampolsky B.C. Basics of digital technology. - M.: Radio and communication, 1980. P.42, 25].

Block formation non-integer values of indicators 2, part of the General structural scheme is designed to generate the elements of the vector noninteger values of state indicators. Its block diagram and principle of operation are known and described in the prototype (see RF patent №2139569, figure 2).

The correction unit 3, part of the General structural scheme, designed for dynamic correction of mathematical expectation (MO) and the variance of a random sequence in accordance with the conditions specified in the adopted mathematical model. Its block diagram and principle of operation are known and described in the prototype (see RF patent №2139569, figure 3-7).

The block threshold device 6 included in the overall structural scheme, designed to obtain preliminary values of the elements of the vector of indicators of the state of the process. Its block diagram and principle of operation are known and described in the prototype (see RF patent №2139569, Fig.9).

The set of indicators 7, part of the General structural scheme, designed to obtain the final values of the elements of the vector of indicators of the state of the process. Its block diagram and principle of operation are known and described in the prototype (see RF patent №2139569, Phi is .10).

Clock 8, part of the General structural scheme, designed to determine the moments of the output of a probabilistic automaton from the previous state and is implemented in the form of well-known sawtooth generator described in [Bystrov Y.A., Welikson AM, Wagman E, and other electronics: Reference book / Ed. by Bystrov Y.A - SPb.: Energoatomizdat, 1996. S-151, ri (a)].

Item BAN 9 and the element OR 14 included in the overall structural scheme, designed to support the procedures for determining the moments of the output of a probabilistic automaton from the previous state and can be built on the basis of mass-produced items BAN and OR, are described in detail in [Maltsev L.A., Tranberg AM, Yampolsky B.C. Basics of digital technology. - M.: Radio and communication, 1980. P.20-24, 9, 11].

Block elements And 10, part of the General structural scheme is intended to provide a record of the obtained values of the elements of the vector of indicators of the state in a memory block. Elements included in the block elements And 10, are known and described in [Sidorov A.M., Gusev V.V., Lebedev O. Fundamentals of pulse and digital techniques. - SPb.: SUVIUS, 1995. S-14, fig.1.2].

The memory unit 11 included in the overall structural scheme is designed to store the values of the elements of the vector of indicators of the state until the expiration of the period of the change of state and is implemented as a standard storage register in accordance with the description, presented in [Sidorov A.M., Gusev V.V., Lebedev O. Fundamentals of pulse and digital techniques. - SPb.: SUVIUS, 1995. S-158, RIS].

The decoder 12, a part of the overall structural scheme is designed to convert the control code sequence in a code corresponding to a time interval of a period of transition and built in the form of a commercially available decoder described in the book [Bogdanovich M.I., Grell I.N., Prokhorenko, VA and other Digital integrated circuits: a Handbook. - Minsk: Belarus, 1991. S-436, RIS].

Unit task time 13 included in the overall structural scheme is designed to generate zero combination that defines the expiry of the period of change of state and is implemented as a digital timer, similar to that described in [Frolkina V.T., Popov L.N. Pulse and digital devices. - M.: Radio and communication, 1992. S-267, RIS].

Probabilistic machine operates as follows. It is known that for analytical probabilistic-temporal description of the system of indicators of quality of functioning SMCRS and LAN is applied to the apparatus controlled Markov chains in the form of stochastic differential equations [2-4]. However, the analysis of [1, 5-8] allows you to create mathematically correct algorithm of bringing false (inadequate, incomplete) set of initial data - e the elements of the matrices of transition probabilities, threshold (boundary) conditions, to the nearest reliable set. Thus, in the simulation of managed semi-Markov chains generated by taking into account both quantitatively and qualitatively (unreliable, insufficient, incomplete) set of initial data, the number of network features is modeled based on a parametrically specified source data, traditional methods, and modeling unreliable (insufficient, incomplete) set of network parameters, by successive transformations using neural network computational methods and algorithms, means that their relative parametric modeling, i.e. transition from false (inadequate, incomplete) set of tasks to parametric modeling.

For solving the problem of combining incomplete, inaccurate, and sometimes conflicting expert opinions about the values of the initial data for modeling elements of the matrix of transition probabilities and thresholds (boundaries) of the States of the simulated process, use one of the standard computational algorithms theory of neural networks - neural network extraprise computational algorithm, or the so-called extrapola neural network (Ann), which is a variety of well-known computational models of associative memory is [5, 6].

Computational neural network algorithm (extrapola neural network) this class consists of two layers of solvers (neurons) - the input layer S_aand the output layer S_b. The input layer S_aconsists of m_Ineurons with a set of forward and backward linkages with m_othe neurons of the output layer S_band the number of input and output images of m is (m=m_I=m_o) and corresponds to the previously entered number "m". In Enns uses the so-called cognitive map, is completely defined by matrix of relations of the form:

Cognitive map characterizes the causal relationship of the individual parameters of the original data and is formulated by experts, the principle of formation of cognitive maps is described in detail in works [5, 6].

While one of the branches of computational neural network algorithm is designed to convert unreliable (incomplete) expert opinions about the values of the elements of the matrix of transition probabilities, and the remaining m-1 branches of the algorithm responsible for converting unreliable (incomplete) expert opinion on the thresholds (the boundaries of) States of the simulated process. Each element of w_ijmatrix (6) defines the relationship of the i-th element (single parameter) false set of initial data j-th e the amenta, the positive relationships are coded 1, negative 1, and the absence of linkages between elements is coded 0.

At the entrance of computational neural network algorithm (EDF) receives an input imagecharacterizing the variety known as parametrically (reliably), and unreliable set of elements (individual parameters) initial data for modeling SMCRS and LAN. Define the q elements (individual parameters)that make up the subset Ω_qfrom a variety of source dataat this point in time specified quantitatively (parametrically), and what elements (individual parameters) from a variety of source data required (preferred) to perform the task, accurate modeling is uncertain (untrue, inaccurate). In order to obtain accurate simulation results it is necessary to reconstruct unreliable specified elements of the source data.

The functioning of computational neural network algorithm is as follows:

1. Activates the input layer S_anetwork input image. In other words, evaluators (neurons) in the input layer is given in the initial state

2. Is bootstrap is facilitale (neurons) output layer in accordance with the expression:

3. Is the cast of neurons in the input layer to the state of the neurons of the output layer:

4. The computation of the new state evaluators (neurons) output layer for allaccording to the formula:

5. Is the repetition of steps 3 to 4 until Enns reach a stable state:

The output of computational neural network algorithm (EDF) have output vectorreceiving a number of state values, which are determined on the basis of expressions (5) and (accurately) describe integrated expert opinion about the values of the elements of the matrix of transition probabilities and thresholds (the boundaries of) States of the simulated process.

Reviewed computational neural network algorithm allows to specify, to reconstruct, to eliminate uncertainty (inaccuracy, incompleteness) in the formation of the source data to simulate the process of changing States SMCRS and LAN. Clarification, reconstruction unreliable (incomplete) set of options allow you to increase the objectivity of the input data, and ultimately, to increase the reliability of the simulation of the processes occurring in d is selected SMCRS and LAN given false, incomplete, inaccurate) information about the state of the environment of radio wave propagation, interference conditions, schedule and network behavior network management system and the influence of other destabilizing factors.

With this in mind, modeling claimed in probabilistic machine. With sensor output of a random sequence 1 the values of the auxiliary random sequence x' with a normal density distribution in binary code, go to the start input 33 of the correction unit 3. In block formation non-integer values of indicators 2, on the basis of the received binary code with interpolation inputs 22₁-22_melements of a vector of integer values of indicators of the state of the process being modeled in the previous step and received in binary code with the control inputs 21₁-21_melements of the matrix of probabilities of transition from one state to another, is the calculation of the elements of the vector are non-integer values of the status indicators in accordance with the expression:

which is a modification of the well-known equation of Kolmogorov-Chapman [4].

Herethe vector of non-integer values of indicators of the state of the simulated process with mathematically the sense step-by-step and, ultimately, the final probability of finding a random process in a particular state; ϕ^T(k+1,k,u) is the matrix of probabilities of transition from one state to another;the vector of integer values of indicators of the state of the process being modeled in the previous step. With outputs 23₁-23_munit 2 vector elements noninteger values of indicatorsin binary code serves for informational inputs 31₁-31_merror correction block 3 and auxiliary inputs 72₁-72_mprocessing unit indicators 7.

In block 3 on non-integer values of indicators of the state (the state probability), the incoming binary information inputs 31₁-31_mand with the threshold information into binary code on the threshold inputs 32₁-32_m-1is dynamic correction of mathematical expectation (MO) and the variance of a random sequence x' in accordance with the conditions described in the approved model [2-4]. As a result, at the output 34 of the correction block 3 in binary code, in the moments of the output of the machine from the previous state (change of time), have a selective value of a random variable x^*obtained from a modified random sequence x' with MO and variance is iej, corresponding to predetermined conditions of the simulation.

A sample value of a random variable x^*with MO and variance corresponding to predetermined conditions of the simulation comes in binary code to the information input of the threshold device 6. M-1 threshold inputs 62₁-62_m-1unit 6 with m-1 threshold outputs 152₁-152_m-1unit increase reliability 15 receives binary coded threshold values of the state process X_{then 1}-X_{since m-1}. The result is a preliminary indicator value stateissued in binary code outputs 64₁-64_mblock threshold device 6. Thus, the m outputs of the block threshold device 6 has m elements of the vectorpreliminary indicators of the state of the process being modeled, which are later used to generate values of the indicators in accordance with the expression described in [4].

Calculating values of status indicatorsimplements the set of values of indicators 7 with regard to coming in binary code on the auxiliary inputs 72₁-72_mthe parameters of the elements of the vector of non-integer values of indicatorsfrom the block 2. The resulting vector elements indicator is in the States through the outputs of the processing unit indicators 7 served in the binary code to the inputs of block elements And 10.

The control unit 5 is a cube memory in which is recorded the program operation of the device and can be implemented according to the scheme shown in Fig.6. Forming a control code sequence and the sequence of threshold values is performed as follows. With an external power source via the control input 51 of the control unit 5 on the information input 5.5 managing OSA 501 is recorded in binary code quantitative or qualitative (false, incomplete) values of the elements of the matrix of transition probabilities (PT)corresponding to the input control, in the memory cell in the control ASA 501. Through the same control input 51 of the control unit 5 on the information input 5.8 threshold of OSA 502 is recorded in binary code the quantitative thresholds X_por-X_{since m-1}either qualitative (false, incomplete) thresholdsstate of the simulated process, in the memory cell threshold of OSA 502. Samples of the moments of the output of the machine from the previous state come from item BAN 9 through the clock input 54 of the control unit 5 5.0 on the third input of the counter 503, and determines, acting with outputs 5.1 and 5.3 accounts is Chica 503 on the clock inputs of 5.6 managing OSA 501 and 5.9 threshold of OSA 502, respectively, the beginning of the read values of the elements stored in ONE 501 new (quantitative or incorrect (incomplete)) matrix PV in binary code via the control output 52 of the control unit 5 to the control input 150 of the block to enhance the reliability of the 15th and beginning reading in binary code new quantitative (X_por-X_{since m-1}or qualitythreshold values-state of the process being modeled, stored in the memory cells, the threshold of OSA 502. Read thresholds is performed with m-1 outputs 5.11₁-5.11_m-1the threshold of OSA 502 through m-1 threshold outputs 53₁-53_m-1the control unit 5 on the threshold m-1 inputs 153₁-153_m-1unit increase reliability 15. With dropping outputs 5.7 and 5.10 managing OSA 501 and the threshold ASA 502 respectively on the first 5.2 and 5.4 second inputs of the counter 503 in the sampled matrix PV or threshold signal that resets the counter 503 and giving a command to the counter 503 to start a new countdown for the newly introduced quantitative or qualitative (false) control actions and thresholds (X_por-X_{since m-1}or

The power increase reliability 15 may be implemented according to the scheme presented in figure 2. Sequential comparison (on which kolichestvo bits) incoming binary coded source data, decision-making about their mathematical nature and transformation of the original data set false (incomplete), to a form suitable for parametric modeling, as follows. Initially, qualitative and quantitative (false, incomplete) information received from the control unit 5 varies according to the number of digits to display:

for recording binary coded quantitative information (as control, and threshold) takes 5 (five) digits of the binary code, while inaccurate, incomplete (qualitative) information knowingly excessive, carries in addition to the usual number of the characteristic causal cognitive relations formulated by experts, which objectively requires the use of ten (10) digits of the binary code for inaccurate, incomplete management, and unreliable (incomplete) threshold information.

With this in mind built the controller of the source data block 1501 improve the reliability 15, is shown in figure 3. The procedure of sequential comparison (number of bits) of the incoming binary coded source data and making decisions about their mathematical nature in the controller data source 1501 is as follows. Shift element 1510 and the storage element 1511 controller recognize the data 1501 is designed to hold five digits of incoming information, if the number of digits exceeds this number, then, from the point of view of mathematics, this information comes in inaccurate, incomplete form. In this case, both elements (1510 and 1511) controller source data 1501 performs the function of a transit node, sending the information directly to the inputs of the Converter unreliable data block 1502 improve the reliability 15. If the input 151-1 shifting element 1510 receives binary control information in the amount of five bits, the shift element 1510 writes this information and with direct access 151-2 sends it via the control output 151.3 controller data source 1501 and control output 151 of the block 15 to the input 40 of the processing unit values of the matrix 4 and to the input of the decoder 12. When the number of bits of control information more than five, the shifting element 1510 does not record this information, and with the transit exit 151-3 sends itthrough unreliable control output 151.5 controller source data 1501 on incorrect control input 152.1 Converter unreliable data 1502. If each of m-1 inputs 151-4₁- 151-4_m-1storage element 1511 receives binary coded threshold information in the amount of five bits (X_por-X_{since m-1}), the storage element 1511 writes this information and with direct output is s 151-5 ₁- 151-5_m-1directs it through a reliable threshold outputs 151.4₁-151.4_m-1controller source data 1501 and the threshold outputs 152₁-152_m-1block 15 to the threshold inputs 32₁-32_m-1error correction block 3 and threshold inputs 62₁-62_m-1block threshold device 6. When the threshold number of bits of information more than fivethe storage element 1511 does not record this information, and with m-1 transit outputs 151-6₁-151-6_m-1directs it through unreliable threshold outputs 151.6₁-151.6_m-1controller source data 1501 m-1 false threshold inputs 152.2₁-152.2_m-1Converter unreliable data block 1502 improve the reliability 15.

Converter unreliable data block 1502 improve the reliability 15 may be implemented as programmable (from the point of view of the matrix of weights (1) - causal cognitive relations formulated by experts) PSAV that implements computational neural network algorithm (2)-(6) in accordance with the scheme depicted in figure 4. Transformation of the original data set false (incomplete), to a form suitable for parametric modeling, as follows. False (incomplete) upravlyayemaya (the elements of the matrix of transition probabilities in binary code comes with unreliable control output 151.5 controller source data 1501 on incorrect control input 152.1 Converter unreliable data 1502 and enable input V PZEV. False (incomplete) threshold informationin binary code comes from unreliable threshold outputs 151.6₁-151.6_m-1controller source data 1501 on unreliable threshold inputs 152.2₁-152.2_m-1Converter unreliable data 1502 and m-1 inputs (1,...,m-1) reprogrammable PZEV. Reprogrammable PSM in accordance with the programmed elements (w_ij) matrix of weights W (1) is analytically described the causal cognitive relations formulated by experts, provides the procedure for calculating (extrapolation) in accordance with the computational neural network algorithm (2)to(6). At the same time as enable input V and m-1 inputs (1,...,m-1) reprogrammable PSM are a group of equal m inputs (max) calculators (neurons) in the input layer S_aEnns, which in binary code are elements of the vectorhaving a physical meaning uncertain (incomplete) managementand thresholdinformation. The set of backward and forward linkages _Iwith m_oEnns, software-implemented within the programmable PZEV, allows to take into account when computing the values of the elements (w_ij) matrix of weights W (1) and get the result of an arithmetic calculation in accordance with the computational neural network algorithm (2)to(6), the extrapolated values of the vector elementsfully (accurately) describing integrated expert opinion about the values of the elements of the matrix of transition probabilitiesand thresholds (the boundaries of) States of the simulated process (X_por-X_{since m-1}). The filing (and writing to storage) in binary code on any i-th input PZEV (in any i-th cell memory PSM) value false (incomplete) of a given parameter (administrator or threshold), initiates the issuance of a binary code from the i-th output PZEV (output of the i-th neuron of the output layer S_b) programmed in accordance with the computational neural network algorithm (2)to(6), the values are mathematically transformed, relatively reliably specified control or threshold parameter. Thus, m-1 (1,...,m-1) outputs PZEV obtained quantitative (relatively accurate) threshold (boundary) conditions of the simulated process (X_por-X_{since m-1}in dvoi the nome code through the threshold outputs 152.4 ₁-152.4_m-1Converter unreliable data 1502 and the threshold outputs 152₁-152_m-1unit increase reliability 15 arrive at the threshold inputs 32₁-32_m-1error correction block 3 and threshold inputs 62₁-62_m-1block threshold device 6, and allowing S PZEV obtained quantitative (relatively accurate) values of the matrix elements MFin binary code via the control output 152.3 Converter unreliable data 1502 and control output 151 unit increase reliability 15 is fed to the input of the decoder 12 and to the input 40 of the processing unit values of the matrix 4.

The set of values of the matrix 4 may be implemented in accordance with the scheme proposed in figure 5. Check incoming values of the elements of the matrix of transition probabilities, the storing and reading control information is carried out as follows. The values of the elements of the matrix PV in binary code is fed to the input of the counter 402, where it is counting and recording - is the same whether the number of received values of the matrix elements MF with the dimension of the matrix is determined by simulation. From the output of the counter control information is fed to the input storage element 401, used for storing and retrieving the values of elementality RO. When new control actions the storage element 401 frees memory cell, transmitting control information line by line (each line of the matrix PV - through output) through m outputs 41₁-41_mprocessing unit values of the matrix 4 m respectively the control inputs 21₁-21_mblock formation non-integer values of indicators 2.

The values of the elements of the matrix MF are held constant at the outputs of block 4 during the control cycle determined by the clock 8. The threshold values of the States of the simulated process X_por-X_{since m-1}during the control cycle are maintained constant at the threshold outputs 152₁-152_m-1unit increase reliability 15 and serve to implement the calculations carried out in the correction unit 3, and are used to obtain the elements of the vector prior values of indicators of the state of the process performed in block threshold device 6.

The moments of the output of the machine from the previous state are determined by the clock pulses 8, OR 14 element PROHIBITION 9 during the formation of the zero combination at the output of unit time job 13. Using block elements And 10 for recording the obtained values of the elements of the vector of indicators of the state simulated process in the memory unit 11, where they are stored until the expiration of the period k (period state change is determined by the unit time 13 values code generated on the basis of managing impacts. The values of the control code sequence with control output 151 unit increase reliability 15 arrives at the decoder 12 converts them into code corresponding to the time interval of the period of change of state k, is recorded in the reversible counter unit 13 and read a generator of clock pulses 8, until the zero combination at the output of block 13, indicating about the expiration time of the machine in this condition. The management of probabilistic-temporal change mechanism state machine and status thresholds is carried out on the control code and threshold combinations coming from the control threshold and outputs (respectively 151 and 152₁-152_m-1) unit increase reliability 15 at the time of the machine from the previous state.

Finally, the outputs of the block 13 are written in binary code relatively reliable indicator value state managed probabilistic machinein each of the time points (determined by the clock pulses 8), with consideration of the m control actions and dynamically alterable threshold (boundary) conditions, asked both quantitatively and qualitatively (false, incomplete).

Thus, the analysis of the principle of operation of the inventive probabilistic machine shows the obvious fact that along with the saved simulation capabilities of managed semi-Markov chains with dynamic threshold adjustment (boundaries) of the States of the simulated random processes, the machine is able to increase confidence in modeling the management of semi-Markov chains generated by taking into account both quantitatively and qualitatively (false, incomplete) set of initial data, describing threshold (boundary) and probabilistic-temporal mechanism (elements of the matrix of transition probabilities) change the state of the simulated random processes occurring in real networks multichannel radio communications and local area networks.

This probabilistic machine provides increased reliability modeling parameters of the real process of functioning networks multichannel radio communications and local area networks in terms of unreliability (failure, incompleteness) of the original data, due to the volatility (uncertainty, variability) of the conditions of functioning of networks of this class. This, in turn, can increase the degree of adequacy of the model, the level is dostovernosti results of the analysis of the quality and performance of networks and as a consequence, to increase the validity of the decisions taken by the management structure, parameters and operating modes SMCRS and LAN, which significantly extends the functionality of the equipment, where it was stated probabilistic machine will be used.

Sources of information

1. Wasserman F. Neurocomputer technique: Theory and practice. - M.: Mir, 1992. - 240 S.

2. Roosters G.B. fundamentals of theory of the effectiveness of targeted processes. Part I. Methodology, methods, models. - M.: the USSR Ministry of defense, 1989. - 660 S.

3. Terent'ev, V.M. Sanin J.V. Analysis of efficiency of functioning of the automated networks multi-radio communications. - SPb.: YOU, 1992. - 80 S.

4. Terent'ev, V.M., Parashchuk IN Theoretical foundations of network management multi-channel radio. - SPb.: YOU, 1995. - 195 C.

5. Kosko Century Fuzzy cognitive maps // International Journal of Man-Machine Studies. V.24. N.Y., 1986. P.16-22.

6. Shcherbakov M.A. Artificial neural network. - Penza: PSTU, 1996. - 44 S.

7. Trahtengerts E.A. Computerized decision support. - M.: SINTEG, 1998. - 342 C.

8. Gorban A.N., Rossii D.A. Neural network on a personal computer. - Novosibirsk: Nauka. Siberian publishing firm Russian Academy of Sciences, 1996. - 146 C.

1. Probabilistic machine that contains the sensor of a random sequence, the block formation non-integer values of the indicators, the error correction block, the block of formation values of the matrix BC is to control, the block threshold devices, the set of indicator values, the block elements And the memory block, the block set a time, the decoder, the item BAN, element OR generator of clock pulses, the output of which is connected to the direct input element PROHIBITION and a clock input unit of time, m≥3 outputs which are relevant m outputs of the machine and connected to the corresponding m inputs of the OR element, the output of which is connected to the negative input element PROHIBITION, the output of which is connected to the clock inputs of the block elements And block the formation of indicators and control unit, the control input of which is managing input machine m outputs of the processing unit values of indicators connected to respective m inputs of the block elements And m outputs of which are connected to the corresponding m inputs of the memory block, the m outputs of which are connected with the corresponding m interpolation unit forming a non-integer values of indicators and m inputs of the unit time, the control input of which is connected to the output of the decoder, m control inputs of the block formation non-integer values of indicators are connected with the corresponding m output processing unit values of the matrix m outputs of the block formation non-integer values of indicators are connected with the corresponding m auxiliary inputs of the processing unit values of indicators and m the information inputs of the error correction block, the start input of which is connected to the sensor output of a random sequence, the output of the correction block is connected to the input of the threshold device, the m outputs of which are connected to the corresponding m information input processing unit indicator values, wherein the inputs of the block to enhance the reliability of the control input which is connected to the control output of the control unit, m-1 threshold outputs of which are connected with the corresponding m-1 threshold input unit increase reliability, control output which is connected to the inputs of the decoder and block the formation of values of the matrix m-1 threshold output unit increase reliability connected to the corresponding m-1 threshold the inputs of the error correction block and the block of threshold devices, and a unit increase reliability consists of a controller of the source data and transducer unreliable data, m-1 false threshold input of which is connected to respective m-1 false threshold and outputs the original data, inaccurate control output which is connected to inaccurate control input of the Converter unreliable data, the control input of the controller, the source data is a control unit that controls the output of the controller data source connected with control is the missing Converter output unreliable data is control the output unit, m-1 threshold inputs of the controller of the source data are the corresponding m-1 threshold input block, m-1 threshold outputs of the Converter unreliable data connected to respective m-1 reliable threshold and outputs the original data and are the corresponding m-1 threshold output unit increase reliability.

2. The machine according to claim 1, characterized in that the shaping unit values of the matrix consists of the storage element and the counter, and the counter input is the input block, the output of the counter is connected to the input storage element, m outputs, which is corresponding to the m outputs of the block.

3. The machine according to claim 1, characterized in that the control unit consists of a control memory element, the threshold memory element and the counter, and first and second outputs of the counter connected to the clock inputs, respectively, of the control and threshold memory elements, the first and second inputs of the counter is connected with the reset outputs, respectively, of the control and threshold memory elements, the third input of the counter is clocked by the input unit, an information input control and threshold memory elements are combined and managing input unit, the output control is his memory element is control the output unit, m-1 outputs of the threshold memory element are the corresponding m-1 threshold outputs of the block.