RU2530351C2 - Method of context creation for measurement data compression and measurement method - Google Patents

Abstract

FIELD: measuring instrumentation.

SUBSTANCE: invention refers to telemetry and data compression during measurement data translation in monitoring systems, in measurements in inaccessible places, and in measurement data storage, e.g. in aircraft and vessel black boxes. The method involves context creation for measurement data compression where entropy of measurement data typical for particular measurement devices and conditions is adjusted, by measured parameter modulation as well, for actual and/or required measurement accuracy, and compression rate and transmitted/stored data content are adjusted by source data system forming one or more linked data arrays for measurement parameters and one or more data arrays for measurement content.

EFFECT: limitation of data volume during measurements, increased data coding rate and data security ensured.

7 cl, 5 dwg

Description

Technical field

The invention relates to methods for creating context, telemetry and data compression and can find its application in the transmission of measurement data, for example, from spacecraft, in control and monitoring systems, when measuring in hard-to-reach places, and also when storing measurement data, for example, black boxes of aircraft and ships.

Description of the prior art

There are a variety of tasks to control, monitor and conduct remote measurements by means of sensors and / or measuring systems, and the tasks of long-term storage of data of such measurements on storage devices, including devices of increased reliability, are also known. The organization of such measuring systems and devices for data storage in many cases is faced with restrictions related to the volume and / or speed of data transfer between the measuring and receiving devices, or with the limitations of the volume provided for data storage. Such limitations, in turn, lead to the need for preliminary compression of the measurement data before transmission or storage.

Known dictionary-based data compression methods, for example, the Lempel-Ziv family of algorithms (Ziv J. and LempelA. 1977. A universal algorithms for sequential data compression. IEEE Trans. Inf. Theory IT-23, 3, 3 (May), 337 -343 [1]) or probabilistic, for example, Huffman DA (Huffman DA A method for the construction of minimum redundancy codes. In Proceeding of the Institute of Electrical and Radio Engineers. 40 Sept. 1952, pp. 1098-1101 [2]) and arithmetic coding (Langdon GG An introduction to arithmetic coding. IBM J.Res.Dev. 28 Mart 1984, 135-149 [3]), methods for computing the replacement code. However, without taking into account the specificity and / or entropy of the compressible information, these methods assume a relatively low compression effect. Known methods for increasing the efficiency of data compression by methods [1-3] through the use of universal contextual and adaptive data models and / or the use of preliminary, for example, partially sorting, data transformations, however, in comparison with contextual models that compress the information, such methods either give less compression effect, or lead to restrictions associated with additional coding costs. Thus, the problem of fast compression of data with a high coefficient is reduced to the creation of contextual data models that take into account the specifics of the compressible information.

Known methods of data compression designed to compress measuring information of a particular type, taking into account the specifics and offering a high degree of compression for the measurement data corresponding to this specificity, for example, the multi-pass method for compressing vibration data, patent RU No. 2071123 [4]. The disadvantage of this kind of methods for solving real problems of data limitation during measurements is usually their strict specialization in the composition and type of measurements, which, in turn, leads to difficulties when it is necessary to compress a large number of parameters of heterogeneous behavior.

The listed methods of data compression [1-4], including methods for increasing their efficiency, are aimed at solving the same problem of limiting the amount of data (replacing data with a code of shorter length) as the claimed invention, therefore, in a sense, these methods are its analogues. The claimed methods, however, are not, in essence, methods of data compression and do not impose restrictions on the use of any data compression methods in their implementation, but only determine the order of their application, including the order of application of methods to increase their efficiency (forecasting and reducing entropy ) for the task of limiting the amount of data during measurements. This approach provides universality of compression for a large set of measured parameters of various behaviors.

Also significant is the difference between the task of creating a context for data compression and the task of taking measurements, the first of which is aimed mainly at achieving the maximum compression ratio with minimal or no loss of information, reducing the amount of data transferred and / or storing, adapting to formats and / or conditions of the transmission medium. The task of the measurements is aimed at ensuring the sufficiency of the data obtained as a result of the measurements for the implementation of the purposes of their measurement, the choice of the measurement methodology and the composition of the sensors, determining the composition and order of measurements, establishing ranges of possible values of the measured parameters, etc. Based on these differences, we present our technical solution to the problem of limiting the amount of data during measurements as a group of two inventions - a way to create a context for compressing measurement data and a method for taking measurements, each of which also has well-known analogues, while the unity of the inventive concept is ensured using the first second. Most of these analogues, however, have little in common with the claimed methods: methods for creating a context are usually aimed at optimizing the use of a data transmission medium, and methods for carrying out measurements mainly describe specific measurement tasks. Therefore, we give only one, the closest, analogue for the claimed methods of creating a context for compressing measurement data and taking measurements, respectively.

The known method and system for creating a context for compressing messages in wireless communication, patent RU No. 2273091 [5], found by the authors closest to the claimed method of creating a context for compressing measurement data (prototype). As well as the claimed method, the method and system [5] are aimed at creating a context for data compression and its essential features inherent in the claimed method are synchronization and permanent storage of information specific to specific devices, which is used for encoding (compression) and decoding ( decompression) of transmitted data (messages). However, in accordance with the method and system [5], the composition and type of information used for encoding and decoding does not contain the specificity inherent in the measurement task, which, in turn, does not allow for sufficiently effective transmission of measurement information using the method and system [5] in its purest form. Note that the claimed method of creating a context for compressing measurement information can be used not only to provide wireless communications, for example, it is assumed that it can be used to store compressed archives of measurement information, while characteristic essential features common to the claimed method and method and system [ 5] are saved.

Known methods and systems of data compression, patent RU No. 2371741 [6], found by the authors closest analogue to the claimed method of measurement (prototype). The essential features of the methods and systems [6] inherent in the claimed measurement method are: compressed measurement data (before transmission), use of compression methods for predicting measurements, compression of measurement information with loss and / or loss. However, the methods and systems [6] do not imply the creation of a universal context for joint coding and further, if necessary, transmission of a large set of parameters, which in the general case does not allow compression of a large set of measurement data with proper use when directly using these methods and systems. In addition, in special cases of application of the claimed measurement method, it is possible to jointly create a compression context, that is, forecasting and reducing the entropy of the measurement data of some parameters based on the behavior of others, which is also a distinctive advantage of the claimed measurement method in comparison with methods and systems [6] .

Note that the indicated differences between the claimed methods and the found prototypes relate to the refinement of the context model of the measurement data, and therefore, in accordance with the general considerations set forth above regarding the use of compression methods, they are aimed at increasing the compression efficiency of the measurement data, i.e., increasing compression ratios and / or coding rate.

SUMMARY OF THE INVENTION

The present invention provides a universal solution for the problem of data limitation when measuring a large number of parameters of heterogeneous behavior.

The invention involves the creation of a context for the compression of measurement data, in which information that is specific to devices that provide measurements, that is, generates measurement information or receives from other devices, as well as characteristic conditions, is constantly stored and synchronized between devices that encode and decode. which are measured. More precisely, information is stored between the encoder and the decoder, which is one or more interconnected arrays of data about the measurement parameters and one or more arrays of data about the composition of the measurements. Moreover, in the data arrays of measurement parameters, the indices for the measurement parameters provided on the corresponding device are determined, and the methods and settings for predicting and / or data compression are determined for the measurement parameters. In the data arrays on the composition of measurements, by means of indices from the data arrays on the measurement parameters, information is stored on the composition and coding order of the measurement data and the calculation of the expected values in accordance with the measurement cycles carried out on the corresponding device providing the measurement.

In accordance with one embodiment of the method for creating a context for compressing measurement data, the current and / or starting expected values for the measurement parameters of the devices are additionally stored in the data arrays on the measurement parameters and / or in the associated data arrays, while the current expected values are generated in the process compression and extraction separately (synchronously) on the basis of data that has already been encoded or already extracted from the code, respectively, encode the difference between the real and expected values of the measured parameters, and when decoding the initial value is calculated as the sum of the extracted difference and the expected value, the starting expected values and methods for calculating the current expected values are agreed between the encoder and decoder.

According to one use case of the present invention, the starting expected values, if used, can be used to create a data security key during transmission or storage.

In accordance with the principles of the claimed method of creating a context for compressing measurement data, for the convenience and / or depending on the environment of the transmission of encoded data, when encoding using the above context, the resulting code is divided into blocks that limit the length, time of measurement, or the number of measurement cycles encoded in one such block.

In accordance with the principles of the claimed method for creating a context for compressing measurement data, an approach is recommended in which the measurement data is encoded without loss, while information loss is determined (and thereby controlled) to achieve higher compression ratios at the measurement stage. For these purposes, the invention includes a measurement method in which the measurement data is encoded to limit the volume and / or ensure the safety of the measurement data. In this case, the encoding order of the measurement parameters, as well as the methods for calculating the expected values and compression methods used to encode the measurement parameters, are determined in accordance with the context created by the above-described method of creating a context for compressing the measurement data.

In the framework of the claimed measurement method, a variant is possible in which some parameters are additionally modulated before encoding in order to reduce entropy and, as a result, increase the degree of data compression. This use case of the invention provides control of information loss and sufficiency of measurement data for the purpose of conducting measurements at the stage of their measurement based on the required and / or real measurement accuracy for the selected measurement procedure.

As an additional possibility of using the claimed measurement method, a variant is proposed in which a reduction in compression ratio (increase in code size) is used to identify abnormal measurements and / or malfunctions.

The purpose of the claimed methods of creating a context for compressing measurement data and making measurements is to limit the amount of data for transmission and / or storage during measurements. The technical result, in the General case (that is, with previously undefined private use cases and applied data compression methods), it is expected to expand the set of technical tools used to limit the quantity and / or ensure the safety of measurement information. The presented technical result in accordance with the Rules of Procedure requires confirmation of the feasibility of the claimed methods, therefore, when describing the implementation, constructive evidence is given of the possibility of sharing these methods.

An additional technical result of the claimed method of creating a context for compressing measurement data in particular cases of its application is to increase the compression ratio and / or encoding speed of the measurement data. The result is achieved due to the sufficiency of the context creation method for the direct application of probabilistic coding methods; by minimizing the amount of information necessary to transmit data on the composition and procedure for extracting measurement parameters from the code (especially effective with a large number of measured parameters); due to the possibility of cross-prediction of parameters, where by cross-prediction of parameters is understood the prediction of the behavior of at least one parameter based on previous values of one or more parameters, that is, already encrypted or already extracted from the code on the encoder and on the decoder, respectively.

A relevant example of the application of the claimed method of creating a context for compressing measurement data and the method of measurement is to obtain information from a wide range of measurements from the orbit of spacecraft (SC) of non-geostationary orbits, most of the time spent out of sight of ground-based information receiving points.

Brief Description of the Drawings

Figure 1 - A simplified example of a measurement scheme on a spacecraft to demonstrate the implementation of the invention.

Figure 2 - The structure of the frames formed in accordance with the measurement cycles of the equipment for an example of a measuring circuit of figure 1.

Figure 3 - An example of a histogram of the differences between adjacent values of single-byte characters, its piecewise linear approximation and the coefficients of the parameterization of the calculation of the probability intervals for their encoding by the arithmetic method.

Figure 4 - General structure of the data arrays on the measurement parameters and data arrays on the composition of the measurements.

Figure 5 - An example of a table of descriptions of the parameters of figure 4 for the parameters that make up the frame structure of figure 2.

The implementation of the invention

To demonstrate the implementation of the claimed inventions, we consider a simple example of a measurement scheme (see Fig. 1) performed, for example, on a spacecraft. Data compression, storage of source data (ID) for compression and the formation of frames and parameters, as well as the formation of a compression context as data are received, are performed on measuring device 1. In accordance with our example (see Fig. 1), the measuring data is transmitted to device 1 by interrogating the sensors directly related to it — dose sensor 5, or interrogating device 1 of other measuring devices — device 3 of standard telemetry of spacecraft. In this case, the survey can be carried out either directly or through other devices, for example, device 2, which is used to modulate the measurements of device 3 in order to reduce entropy. In addition, measurements of the pressure and temperature of the gas of the propulsion system (DU) are carried out by the measuring device 4 and then transferred to the measuring device 1 in accordance with the selected measurement procedure for the DU, for example, these measurements are taken and transmitted only during the operation of the DU.

A characteristic feature of virtually any measurement is the cyclical nature of their measurements. Even in the case of measurements of constant values, without taking into account the nature of their change, usually, in order to achieve the necessary accuracy, several measurements are carried out. In the general case, the information of practically any measuring equipment can be divided into cycles, in accordance with which measuring frames are formed — sequences of data containing the measurement results, if necessary equipped with additional information about their own composition. Such frames within the measurement cycle can be formed in various ways, and in general, the methods for their formation depend on the nature of the measurements. In our example, we will consider the simplest version of the formation of measuring frames, in which frames are formed in the order in which they should be encoded.

Figure 2 presents a variant of the structure of the measuring frames generated during the measurements according to the scheme of figure 1, each cell in the tables of figure 2 represents one byte of information (symbol) of the corresponding frame of the system. The use of a byte representation of data frames in the description is determined by the choice of data compression method - for our example, this is the arithmetic coding method, the most effective with a limited alphabet. As noted above, the claimed method of creating a context for compressing measurement data does not limit us in the use of data encoding methods, however, to demonstrate its advantages, the most convenient method is arithmetic coding. Convenience due to the clarity of a simultaneous demonstration:

the sufficiency of the claimed method for the direct use of probabilistic coding methods, the convenience of controlling the data compression coefficient only by changing the settings contained in the ID, and the efficiency of increasing the compression coefficient by choosing the measurement methodology.

The use of arithmetic coding also implies the calculation of probability intervals of symbol values. In our example, differences between the value of the measurement parameter (symbol) and the value calculated for it along the ring will be encoded in accordance with the number of bits of its representation, the expected value. In this case, the single-byte alphabet of the arithmetic coding method is mainly used. We will define the probabilities of single-byte characters as an integer ranging in size from 0 to 2 ¹⁵ (determined by the possibility of limiting it to 4 bytes when calculating the code of the arithmetic method), which is a real probability normalized to 2 ¹⁵ , varying from 0 to 1. Calculation of probability intervals is performed in accordance with the scheme shown in Figure 3, in which a byte for measuring represented piecewise linear approximation of the normalized histogram on February ¹⁵ differences between adjacent values pairs Metrized coefficients: Pmin = 1 ÷ 128 - determines the minimum value of the probability for the symbol; X1, X2, X3 - determine the boundaries of linear sections on the x axis, while X3 also sets the intersection point of the corresponding linear section with Pmin, and X2 sets the intersection point between the inclined linear sections; K1, K2 - slope coefficients of the corresponding linear sections; Method - determines the presence of a piecewise-linear section on the left and on the right (0 - symmetrical sections, plus 1 - a piecewise-linear section only on the right, minus 1 - only on the left). The intervals of the arithmetic coding method are calculated by summing the values of the reduced distribution function on the left or right without going through 0, the total probability in this case should not exceed the value of 2 ¹⁵ , which is ensured by summing the probabilities when calculating the interval of the zero value separately on the left and on the right. Note that with such a composition of the parameterization coefficients, for the fast summation of probabilities, the partial sum formulas of arithmetic and geometric progressions are applicable. We also note that all values of the presented function for calculating the intervals for integer parameterization coefficients are also integer, which, when matching the rules for rounding the mantissa of the arithmetic method between compression and extraction algorithms, makes it unambiguous to calculate all intermediate values of the mantissa during encoding.

So, we will form an ID system consisting of connected arrays of data on the composition of measurements and arrays of data on measurement parameters, including information on forecasting methods (methods for calculating expected values), compression methods, and settings (parameterization coefficients for calculations) for both. The general structure of these arrays and the relationships between them are presented in the form of tables in Fig. 4. The key table of Fig. 4, which closes all the others, is a table of parameter descriptions, which represents the main data array of measurement parameters, so we will consider it in more detail. In the case of the predominantly single-byte alphabet that we have chosen, it is most convenient to consider the parameters described by this table as single-byte characters corresponding to the cells of the tables of figure 2, the indexing of which, in accordance with the claimed method of creating a context for compressing the measurement data, we will perform first, and also assign for all parameters appropriate methods. The table of Fig. 5 shows the indexing option for the parameters of our example, and also methods for calculating the expected values and compression methods for all symbols are selected and assigned in accordance with the indices defined for them (symbol ID).

The numbers in the column “Compression Method ID” are identifiers of compression methods selected for the symbols (the numbering corresponds to the notation of FIG. 4). As can be seen from the table of Fig. 5, most characters are assigned the compression method ID 1, corresponding to the arithmetic coding of a single-byte character, for which the probability is calculated in accordance with the scheme described above and shown in Fig. 3, and the parameterization coefficients are contained in the corresponding table “Compression ID 1 »Corresponding to the associated data array of measurement parameters. The structure of this array is presented in the form of a table in figure 4. Compression method 2 is used to compress single-bit (signal) parameters and corresponds to the arithmetic coding of one bit of information with the parameterization coefficients assigned in accordance with the table “Compression ID 2” of Fig. 4, where the “Bit No.” field determines the number of the encoded bit of information in the corresponding byte, a Pmin determines the probability that this bit does not match the expected value. In our example, such a parameter is the only parameter "sign of the presence of shadow" (see figure 2, in and 5). Note that for coding single-bit values by the arithmetic method, the accuracy of calculating probability intervals increases, compared with single-byte characters, by increasing the number of mantissa characters when calculating probability by 7, while the code can be combined into one stream when coding by methods 1 and 2. Finally compression method 3 in our example demonstrates the possibility of encoding data into several streams, for example, parameter “16 BA status” with a previously unknown compression method most suitable for it, in tvetstvii the assigned compression method 3, it will be transmitted for encoding into another stream. This encoding method also allows you to see the possibility of joint application of the claimed method of measurements with the prototype, for example, further processing and compression of measurement data encoded by compression method 3 can be carried out in accordance with the method [6].

The numbers in the column of the table "ID of the method of calculating the expected values" of Fig. 5 are the identifiers of the corresponding methods, their numbering coincides with the numbering of the tables "ID of the methods of calculating the expected values" shown in Fig. 4. The method of calculating the expected values of 0 is used in cases where the calculation of the expected values is not performed, or is performed by other methods, for example, described below. The method of calculating the expected values of 1 is used for single-byte parameters, the expected values of which correspond to their previous values that have already been encoded, or are offset from them by a constant value defined in the table "ID of the method of calculating the expected values of 1" in figure 4, the parameterization parameter "offset" . The method of calculating expected values 2 demonstrates the ability to calculate expected values for parameters larger than one byte with the alphabet limitations imposed by us using the source data system of the claimed method of creating a context for compressing measurement data. In our example, this method is used to calculate the expected value of the two least significant bytes (characters) of the measurement time, taking into account their known periodicity for each type of frame. The table corresponding to this method, “ID of calculation of expected values 2”, contains the following parameterization coefficients: “Character ID 1” - defines the index of the first character (for example, high byte); "Character ID 2" - defines the index of the second character (for example, low byte); “Offset” - determines the value of the constant offset for the reference 2-byte value formed from the 2 listed characters. In accordance with the method of calculating the expected values of 2 from the table of descriptions of the parameters (see Fig. 4), the expected values corresponding to the character IDs 1 and 2 are extracted, then an offset is added to the received reference value and the new expected values are returned to the table of descriptions of the parameters by the same ID characters from where they were taken. And finally, the method of calculating expected values of 3 is used for characters whose expected values are assumed to be constant equal to the setting "value" of the table "ID of the method of calculating expected values of 3".

The structure of the parameter description table in FIG. 4 is completed by two columns of the starting expected value (most likely or used as a security key) and the current (calculated for the corresponding symbol) expected value. Note that these columns can also be, if necessary, for example, when the types of measured values are different, separated from the general structure of the parameter description table into a table connected by the symbol ID. Also note that, for example, when using only one compression method or only one context calculation method for all parameters, the corresponding related tables can be combined with the parameter description table.

At the end of the description of the claimed method of creating a context for compressing measurement data, we show how, based on the source data system shown in FIG. 4, measurement encoding can be carried out. At the beginning of the coding of the data block, in the parameter description table, all expected parameter values are recorded in accordance with the starting values assigned to them. Upon receipt or as the frame is formed based on the structure of the frame composition description table in Fig. 4, in accordance with the assigned index “Symbol ID”, a link to the data compression method is extracted from the parameter description table, including data on parameterization coefficients from the corresponding linked table of the compression method ID if any. Based on these IDs, the difference between the value of the corresponding symbol in the frame and the value expected for it from the parameter description table is encoded. Further, in accordance with the method for calculating the expected values established for the same symbol in the parameter description table, the expected values are recalculated in the “expected value” column of the parameter description table, after which the transition to the coding of the next symbol is performed, and so on, in a loop to the end of the frame. In the same way, an arbitrary number of frames from the selected measurement circuit can be encoded. An exit from a coding cycle of a data block can be made under the condition of limiting the length of the code during storage or for transmission in one cycle, or by the time of measurements or by some other condition, providing the convenience of further use of the encoded information. The encoding of the next data block begins with the alignment of the starting and expected values and the initialization of the encoding method settings.

Data decoding is carried out in the same way, while for each frame, the frame type parameter ID common to all frame types is first extracted, based on it, the corresponding frame composition description table is selected. Note that with a header common to all frame types, a coding order is possible without a frame type identifier, in which the frame type is determined by the header, and the composition and order of the header parameters is strictly the same for all frame types and can even be put in a separate frame composition table . However, this encoding order is less universal, since it imposes quite serious restrictions on the frame structure. If it is necessary to determine the type of incoming frame, the most universal seems to be the approach in which the frame type identifier is computed first (note - based on any parameters from the frame composition) and, further, in accordance with it, the frame composition description table is linked to a specific table. In this case, the calculated identifier, in order to be able to determine the corresponding table of the frame composition during decoding, should be placed in the code first with the encoding method common to all frame types, which brings us back to the order corresponding to the first example.

Turning to the description of the claimed method of measurements, we will analyze in more detail the process of compressing the measurement data in our example for each measurement, and, at the same time, we will reveal in more detail the meaning of the assignments made by us in the table of parameter descriptions in Fig. 5:

The behavior of the identifier "ID frame type" entirely depends on the selected measuring circuit (frame accumulation frequency) and can be determined in various ways. With its previously unknown behavior, for example, a limited number of system frames can be used. So, in our example, there are only three frame types, therefore, by assigning identifiers accordingly and setting equal probabilities for them, while considering the remaining probabilities to be minimal, we can already get a compression ratio of about 5 times (1.5 bits per 8 bit dimension) of the original size. A greater compression effect can be achieved if there are more precise assumptions about the behavior of the frame type identifier. For example, if for the frames in figure 2 it is known that frame 1 with dose readings is interrogated 1 time per hour, frame 3 with the data of telemetric measurements of the spacecraft is interrogated 1 time per minute, and frame 2 with readings of pressure and temperature during operation of the remote control is interrogated 1 time per second, then the probability of encountering a zero change in the identifier of the frame type will be significantly higher, therefore, choosing the coefficients of the distribution function to approximate the histogram of differences (see Fig. 3) in accordance with this assumption, and also setting 0 as a constant, we expect value for the method of calculating the expected values of 3 (see figure 4), we can significantly increase the compression ratio of this parameter.

In all frames of our example, there is a measurement time specified by a four-byte parameter, suppose that time is measured in seconds or in processor cycles, starting with some starting value, no matter what our example is. The two high-order bytes of time, assuming small measurement intervals, will rarely change between frames, therefore, it is quite effective for them to calculate the expected values by method 1 with a zero specified offset, at which they are expected to be equal to the previous values, and compression method 1, where, using approximation coefficients the highest probability of zero difference is selected. Note that the two high-order bytes of time, subject to the unity of the time format of all frames, can be specified in the frame composition tables with the same character identifiers, which further reduces the entropy of their behavior and, therefore, increases their compression ratio.

The two least significant time bytes (time bytes 2 and 1 (low)) change much more often, however, their behavior is also quite accurately predictable with the known measurement method. For example, knowing exactly the intervals at which frames arrive, you can use the method of calculating the expected values of 1 and set the offset for the least significant byte in accordance with this interval. Note that if you set the offset only for byte 1, then the behavior of byte 2 will be much more difficult to predict, therefore, in our example (see table of Fig. 5), for one of these bytes in all frames the method for calculating the expected values of 1 with an offset of 0 ( the expected value is equal to the previous one), and for the second - the method of calculating the expected values 2. In this case, the parameterization coefficients of the method for calculating the expected values 2 contain the symbol IDs corresponding to both bytes and the expected values for them are calculated together with scheniem according to the known type of frame for each measurement period. The compression of time bytes 2 and 1 is also carried out in accordance with compression method 1 and involves a high compression ratio, at least for byte 2. The compression ratio of the least significant time byte (byte 1) is highly dependent on the method of measurement, in this case, choosing the optimal ratio between the accuracy of determining the measurement time and the accuracy of the format of its storage, as well as, possibly, the accuracy of measurements of the time itself on the measuring device. To reduce the entropy of this parameter in accordance with the claimed measurement method, modulation can be applied, associated with limiting the accuracy of the time storage format, taking into account the real or required accuracy of the time measurement for the chosen measurement methodology.

The parameters "dose increment" and "dose" (see frame 1 in figure 2 and the corresponding parameters in figure 5) in our example are dependent on each other: the first of them shows the dose (for example, the number of times the ionization chamber is activated), scored during the measurement interval, the second - the total total dose as a result of summing the increments. In accordance with this, the assignments are made in the table of descriptions of the parameters of figure 5: for incrementing the dose, the calculation of which each time starts again, this is a method of calculating the expected values of 3 with a constant expected value, and for the accumulated dose, the readings of which depend on its previous value, this is a method of calculating the expected values 1. The method of reducing the entropy for this parameter is, for example, "roughening" the measurements of the ionization chamber in accordance with the required measurement accuracy for the selected The methodology that can be determined at the sensor design stage. Note that the relationship of these parameters can also be used to reduce the entropy of their readings, for example, if we supplement the methods for calculating the expected values for one of them with a method that will accurately calculate the readings of the other. Of course, in this example, their joint transmission rather demonstrates the redundancy of the data intended for transmission, so if you transmit only the dose, you can always calculate the increment by subtracting its adjacent values. Nevertheless, this example of supplementing the methods for calculating the expected value by methods that take into account the relationship, demonstrates the possibility of reducing the entropy of the behavior of interconnected, for example, some known law parameters.

The parameters "pressure" and "gas temperature of the remote control" (see frame 2 in figure 2 and the corresponding parameters in figure 5) in our example are shown to demonstrate the possibility of using the claimed method of creating a context for compressing measurement data, for compressing the data generated another measuring device. The compression method 1 assigned to them allows you to encode these parameters without loss in the form in which they are received from the measuring device. The method used to calculate the expected values of 1 (presumably with a zero bias) allows one to achieve a high compression effect with their smooth behavior, which is ensured by a high measurement frequency. Note that for these parameters it is also possible to reduce entropy, for example, by including a modulating device (similar to device 2 of FIG. 1) in the measuring circuit or by programmatically converting data frames containing these measurements. Another solution is the preliminary approval (in the form of requirements) of the data formats and the measurement accuracy of device 4 to limit data redundancy in accordance with the applicable measurement scheme for their subsequent compression.

The parameter "voltage (modulated)" (see figv) and a table of descriptions of the parameters in figure 5) demonstrates the ability to modulate the measurement data. In our example, this parameter, together with the two parameters “sign of shadow presence” and “BA status” described above, is contained in the measurement scheme of standard satellite telemetry (device 3 of FIG. 1). The value of the primary satellite power supply voltage is transmitted to the Earth in the spacecraft visibility zone, and its value can also be interrogated and used by other spacecraft systems and devices for its own purposes, including our measuring device 1 (see Fig. 1). If, for example, monitoring this parameter over the entire orbit of the spacecraft’s orbit and transferring this data at relatively short communication sessions with it is quite expensive and we are only interested in the voltage value that is outside the norm established for it, then we can do as follows : Firstly, according to the table of FIG. 5, assign a method for forming context 3 with a constant expected value corresponding to the middle of the interval of its normal behavior. And, secondly, on device 2, equate the values of this parameter that are within the normal range to the value of the middle of the normal range. Such modulation of the “voltage” parameter will make its behavior substantially more predictable, which, when this parameter is compressed, compression method 1 assigned to it in the table of parameter descriptions of FIG. 5 will give a significantly greater compression effect when the voltage is within normal limits. Note that if the voltage goes beyond the norm, its compression by method 1 can also significantly increase the volume required to transmit this parameter, which, in turn, can be used, for example, to signal a malfunction. So, if all the parameters in our measuring system will be modulated in the same way relative to the norm, then the sign of code growth becomes a universal way of signaling a malfunction, or an automated sign of abnormal measurements, proposed in the claimed method of taking measurements.

Finally, the CRC parameter in our example is used to control data integrity. The approach to its compression, depending on the implementation of the measuring circuit, can be different: For example, if it is necessary to control the integrity of the frames after extraction, we can choose for it any of the listed methods for calculating expected values and equal probability intervals for all differences, which will make the coefficient for it compression equal to one (one to one), but at the same time, it will significantly increase the degree of integrity control of the extracted code. In the case when the reliability of the code is not in doubt (or controlled by other methods), the best option would be to calculate the expected value for the CRC in accordance with, in fact, the CRC calculation algorithm. In our example, the possibility of this is provided by CRC coding after all the data from which it is calculated (at the end of the frame). With this approach, firstly, the probability of predicting the CRC value, and hence the compression coefficient for it, is significantly increased. Secondly, in case of CRC mismatch, the volume of the transmitted code increases significantly, which can be used as the mentioned alarm about the malfunction of the corresponding measuring equipment.

An advantage of the claimed methods of creating a context for compressing the measurement data and taking measurements is the possibility, on the basis of only the ID (without changing the compression algorithms and calculating the expected values), to select the compression methods and the composition of the measured parameters, to set the optimal data compression coefficients. So, for example, it is possible to replace one sensor with another when the first fails, or replace the parameterization coefficients and / or compression methods when changing the entropy of the parameter values. The disadvantages include the need for a clear definition for each parameter of the probabilistic picture of its behavior, which, in fact, can change. In practice, however, significant changes in the behavior of parameters occur extremely rarely, but the problem can be solved by using adaptive compression methods.

Claims (7)

1. A method of creating a context for compressing measurement data, in which information that is specific to devices that provide measurements and measurement conditions is constantly stored and synchronized between devices that encode and decode, characterized in that the information is specific to at least , one of the devices providing measurements, is set in the form of one or several interconnected arrays of data on the measurement parameters, in which the parameters provided on the corresponding device The measurement property is indexed and methods and settings for predicting and / or data compression and one or more arrays of data on the composition of the measurements are indexed and determined for them, in which the indices of the measurement parameters from the arrays of data on the measurement parameters are used to determine the composition and encoding order of the measurement data for cycles measurements specific to the corresponding device providing the measurements. 2. The method according to claim 1, characterized in that the current and / or starting expected values for the measuring parameters of the device are additionally stored in the data arrays of measurement parameters and / or in the associated data arrays, while the current expected values are generated in the compression process and extracting separately on the basis of data already encoded or already extracted from the code, respectively, encode the differences between the real and expected values of the measured parameters, and when decoding, the initial value is calculated as the sum Removing the difference and the expected value, starting the expected values and the current methods of calculating expected values agree between an encoder and a decoder. 3. The method according to claim 2, characterized in that the starting expected parameter values are used as a security key for the encoded data. 4. The method according to one (any) of claims 1 to 3, characterized in that the encoded information is divided into blocks limited by the length, measurement time or number of frames, in accordance with the capabilities, security requirements and / or usability considerations of the resulting code . 5. A measurement method in which the measurement data is encoded in order to limit the volume and / or to ensure the safety of the measurement data, and when encoding, prediction of the behavior of the measurement data and data compression methods with and without loss is used, characterized in that the encoding order of the measurement parameters as well as the methods for calculating the expected values and compression methods used for encoding the measurement parameters, are established in accordance with the context for compressing the measurement data created Wow according to claim 1. 6. The method of taking measurements according to claim 5, characterized in that all or part of the parameters are further modulated before encoding. 7. The method of taking measurements according to claim 5, characterized in that an increase in the compression ratio (increase in code size) is used as a sign to identify abnormal measurements and / or malfunctions.

Images