RU2718162C1 - Aircraft onboard network architecture optimization method - Google Patents

Abstract

FIELD: computer engineering.

SUBSTANCE: invention relates to computer engineering. Technical result is achieved by setting a list of types of computers from an optimized onboard network with given parameters, a list of functions belonging to optimized onboard network systems, with a given resource consumption, interrelationships between given functions, a list of exceptions which determine prohibition of finding certain types of functions in one computer, formation of value of priority coefficients and limits of computer loading, successive comparison of the resource intensity of each function with parameters of each computer and selection of a calculator and functions which ensure the least comparison value when comparing parameters of the computer and resource intensity of functions assigned to it.

EFFECT: technical result consists in improvement of reliability and fail-safety of onboard network with minimization of number of computers onboard network and optimization of information flows in computer network of DME in terms of traffic minimization with observance of all requirements to information support of OBES.

5 cl

Description

The invention relates to means for optimizing the architecture of a peer-to-peer on-board computer network of an aircraft, in particular, to methods for reducing intra-network traffic due to the optimal redistribution of computational tasks between elements of an on-board network in avionics systems built on the basis of distributed modular electronics (RME). The invention can be used in the design or modernization of the architecture of the complex of on-board equipment (BWC) of an aircraft built on the basis of RME.

Avionics systems based on RMEs are characterized by an increase in the number of functions (hereinafter, a function is a task or a system of tasks united by a common logical goal performed by a special application installed on a computer) and an increase in the degree of integration, as well as being easier, safer and cheaper to maintain compared to previous generation systems. These advantages are achieved not only by developing new hardware devices for system configuration, but also by increasing the optimal planning of the architecture of avionics systems.

The optimization of the architecture of the on-board computer network during the design or modernization of avionics systems involves the analysis of the architectures under study using modern information computer technologies. Formalization of the necessary requirements for the system under analysis, for example, in terms of reliability and performance, should be reflected in subject-oriented models of the architecture of the avionics system. Designing or upgrading the architecture of avionics systems also often involves solving automation problems in evaluating the design goals (modernization) of such systems to create carefully designed projects.

Automation of the design (modernization) of the avionics system involves the automatic optimization of its architecture. The indicated approach is the only one possible at present, since the number of elements of the designed complex and the interconnections between them is so large that it does not allow us to solve the problem “manually”.

To develop a workable reliable fail-safe architecture of avionics, it is necessary to correctly implement the design (modernization) procedure. The design (modernization) of the system based on the RME should be based on the selection of the best, in the sense of a set of criteria, solutions and provide the optimal choice in the sense of these criteria.

The well-known "Method of optimizing parallel processing of information to minimize its cost", proposed in the patent of the Russian Federation No. 2191424 from 03.04.2000. The invention relates to parallel information processing systems. Its use in local area networks allows to obtain a technical result in the form of optimal use of resources of a local area network. The method is intended for use in local area networks containing workstations, among which there is one main workstation and remote, while the main workstation performs dispatching functions, decomposing the problem to be solved and distributing individual computing processes. The technical result is achieved through the distribution of individual computing processes with the least load in terms of processor time use, and the determination of the optimal number of parallel processes is carried out at the main workstation according to predetermined conditions. The proposed method only allows to optimize the use of computer time, but does not take into account a number of other important indicators that determine the resource consumption of the task, such as the occupied memory capacity of RAM and ROM, the number of used system clock cycles per unit time, the amount of non-network processor interfaces (such as SPI, I2C and etc.). Also, the proposed method is focused on optimizing the architecture of the client-server type and does not allow you to choose the composition of the network (its computing elements) from the a priori set, so that the distribution of the computing resource is optimal in terms of such indicators as the number of calculators used and the information load of the network. Thus, the described method is not applicable for optimizing the architecture of the peer-to-peer on-board network built on the basis of the RME, the design process (modernization), since it must take into account a number of important parameters (mentioned above) that are absent in the considered method.

Closest to the claimed method is the "Resource Allocation System", proposed in the patent of the Russian Federation No. 2189073 from 13.10.2000. The technical result of the invention is to expand the range of tasks and simplify the algorithm of work. In the presented system, a sequential survey of the excess resources of each localized center (LC) takes place, the summation of homogeneous resources, the distribution of insufficient resources to each localized center. Each LC forms a surplus in resources, and also presents a list of resources that it (LC) needs. The distribution system out of the total amount of summed resources distributes according to the needs of each resource to each LC. The distribution of resources can be carried out both in the usual mode (without priorities), and with the establishment of priorities. The described analogue can be used for hardware and software implementation of the method of redistributing the computing resource of an onboard network during its operation, but is not applicable as a method of optimizing computing resources in the design (modernization) of onboard networks of aircraft built on the basis of RME, since this method requires a system distribution as a separate functional unit to which information is received about the excess or lack of resources from each LC.

The objective of the invention is to optimize the architecture of the peer-to-peer on-board computer network of the aircraft (in the sense of minimizing the number of on-board computer calculators, as well as minimizing intra-network traffic due to the optimal redistribution of computing tasks between the elements of the on-board network), taking into account the specified restrictions that determine the required characteristics in terms of functionality, reliability and fault tolerance optimized on-board network.

The technical result of the invention is to increase the reliability and fault tolerance of the designed (upgraded) on-board network while minimizing the number of on-board computer calculators and optimizing the information flows in the computer network of the RME in terms of minimizing traffic in compliance with all requirements for information support of the BWC.

The technical result is achieved using the proposed method, in which at the first stage, the composition and parameters of the analyzed on-board computer network of the aircraft are set in the electronic computer (computer): a list of types of computers from the optimized on-board network with specified parameters, a list of functions belonging to the on-board systems networks with a given resource intensity, the relationship between given functions, a list of exceptions that determine the prohibition of finding certain types of functions in one comput numerator, form the priority value ratios and limits load calculator.

At the second stage, the resource intensity of each function is alternately compared with the parameters of each calculator, and the calculator and the function are selected that provide the lowest comparison value when comparing the parameters of the calculator and the resource intensity of the functions assigned to it.

At the third stage, the function selected at the second stage is combined with each remaining function from the given list, taking into account the given list of exceptions, and the resource intensity of such a combination of functions is alternately compared with the parameters of each calculator from the given list, and the calculator is selected that provides the lowest value for matching the parameters of the calculator and the resource intensity of the combination of functions assigned to it.

At the fourth stage, the resulting combination of functions is combined in turn with each remaining function from the given list that is not included in the obtained combination, taking into account the given list of exceptions, and the resource intensity of such a combination of functions is alternately compared with the parameters of each calculator from the given list, and the calculator that provides the lowest value is selected comparing the parameters of the calculator and the resource intensity of the combination of functions assigned to it, and repeat the fourth step until, when setting the resource intensity of the combination of functions does not exceed the parameters of each calculator from a given list.

At the fifth stage, the last combination of functions and a calculator are used, when comparing the resource intensity and parameters of which the resource intensity of the last combination of functions did not exceed the parameters of each calculator from the given list, the calculator and a subset of the functions included in the resulting combination of functions are determined, marked as used, then the functions and calculators are used not marked as used in the previous step, and repeat steps two through five until each function is defined calculator. In accordance with the definition (Milner BZ, Theory of Organizations. M .: Infra-M, 2008. 480 s), the functional and structural organization of the aircraft's on-board network will be understood as the structure of the distribution of functions among the second-generation scalable RME computers. In the most general form, the functional blocks of modular avionics, as an information management system, include:

- input of information from external or internal sources;

- processing of input information and its presentation in a convenient form;

- output of information for presentation to consumers or transfer to another integrated circuit;

- feedback is information processed by flight crew members to correct input information.

B., Thielecke F. A systems architecting framework for optimal distributed integrated modular avionics architectures // CEAS Aeronaut. J. 2015. Т. 6, № 3. С. 485-496).The functions of avionics are not standardized, their composition and structure depend on the choice of the aircraft manufacturer and the type of aircraft (

B., Thielecke F. A systems architecting framework for optimal distributed integrated modular avionics architectures // CEAS Aeronaut. J. 2015. T. 6, No. 3. S. 485-496).

The structure of functions can be represented in the form of a graph, either arrow or vertex (Taha H. Introduction to the study of operations: in 2 books. M: Mir, 1985). In the first case, the functions are indicated by arcs of the graph (arrows) located in a given logical sequence, while the vertices of the graph specify the start and end events of the function. In the second case, functions are denoted as vertices of the graph located in the sequence given by its arcs. Such a representation is advisable in connection with the connectedness of functions, due to the need for information exchange between them.

To plan a system based on RME, you need to know what functional blocks the systems consist of, their individual requirements and message flows. The main task of planning a system based on RME is to distribute the functional blocks of the system between the devices of the developed architecture.

The functions of avionics differ in static design and in dynamic design. Static design assumes an unchangeable device configuration, i.e. time-independent distribution of functional blocks between devices of the architecture of the RME system. Dynamic design involves the possibility of reconfiguration in order to increase the availability of the system in case of emergency situations.

To illustrate the composition of avionics functions in static design, you can use the generalized standard set of functions performed by the BWC of the aircraft (Butz H. Open integrated modular avionic (ima): State of the art and future development road map at airbus deutschland // Signal. 2010. T 10. 1000 pp., Degtyarev AR, Medvedev GV Algorithm for the distribution of tasks in multiprocessor complexes of integrated modular avionics // Automation of control processes. 2014. T. T.35, No. 1. P. 79-84. ):

- solving problems of navigation support;

- air navigation along the programmed route;

- the formation and display of flight navigation information (PNI);

- piloting and flight control;

- round-the-clock surveillance system;

- operational manual input of the set values of the flight parameters;

- countdown and display of current and flight time;

- manual and automatic tuning of radio navigation and landing systems;

- ensuring interaction with air traffic control radars;

- information transfer to ensure automatic dependent monitoring;

- monitoring the status of the power plant and on-board equipment;

- management of airborne equipment;

- management of the power supply system of on-board equipment;

- formation and registration of an array of flight information in the on-board recorder;

- the functions of the complex of communications;

- transmission of distress radio signals;

- flight automatic control of the health of the complex with the display of control results.

In dynamic design, functions are added that are associated with a change in the distribution of functional blocks between devices of the system architecture with a partial loss of functionality of the latter.

The choice of a variant of the functional and structural organization of the aircraft's on-board network is the most responsible task in the design of the RME, since it determines the effectiveness of solving problems of subsequent levels. This statement is well illustrated by the metamodel of the architecture of the RME system proposed in (Annighofer B., Thielecke F. A systems architecting framework for optimal distributed integrated modular avionics architectures // CEAS Aeronaut. J. 2015. T. 6, No. 3. C. 485- 496).

The architecture structure of the RME system is based on three levels - the system, equipment and installation. The upper level of the system reflects the functional structure. Each function corresponds to a software package, which, in turn, consists of elementary tasks, the structure of which is determined by the transmitted signals or (and) the logical sequence of tasks. The composition and structure of functions selected at the initial stage of architecture analysis determines their distribution among devices at the next level. Accordingly, the planning of communication lines between devices is determined by the signals of the first level. Further, the composition and structure of devices at the second level serves as initial information for the design of installation sites for devices and the laying of cable routes at the installation level. Thus, an irrational distribution of functions at the system level entails ineffective solutions at subsequent levels of the metamodel, since the mapping of software into RME devices and the mapping of devices to installation locations depends on the corresponding distribution. And vice versa, the optimal distribution of functions at the system level is a prerequisite for the optimal organization of the on-board network at the equipment and installation levels.

The rules for choosing a rational structural organization of functions are the subject of study in the theory of distributed computing. Distributed computing is a special case of parallel computing, i.e. simultaneous solution of various parts of one computational task by several computing devices. In the study of parallel computing, the main emphasis is on the decomposition methods of the problem being solved into subtasks, which can be performed simultaneously, as a rule, to maximize the speed of calculations. The main feature in the organization of parallel computing is the need to take into account the difference in the characteristics of available computing devices and the presence of a significant time delay when exchanging data between them.

If we consider a distributed system from a hardware point of view as a set of interconnected autonomous computers or processors, from a software point of view - as a set of independent processes (executable software components or tasks) interacting by sending messages to exchange data and coordinate their actions, it becomes obvious that the structural organization of functions within the framework of the RME architecture is close to the organization of distributed computing. Therefore, the methods of organizing functional structures in avionics software applications can be based on well-known methods and decomposition algorithms in parallel computing (Gergel V.P., Fursov V.A. Lectures on parallel computing. Samara: Publishing House of Samara State Aerospace University, 2009) .

The goals of organizing functional structures may vary. When organizing parallel computing, the goal is usually considered to reduce the computation time for a given number of computing devices or to minimize the number of computing devices with a time limit. In avionics, when using multi-machine complexes with autonomous memory of each processor, it is necessary to provide high real performance (degree of load) of crates, which currently has a level significantly lower than the potential. From these positions, when solving multiply connected tasks and tasks similar to them in organizing the computing process, in which a large number of calculators with intensive interprocessor information exchange are involved, the use of multi-machine computing systems is impractical. There is an urgent task of developing mathematical methods for decomposing the functional tasks of computing systems into private independent subtasks, each of which can be implemented on its own computer in the common crate of the aircraft's onboard equipment complex.

The solution presented in the invention describes a method for optimizing the architecture of the on-board avionics computing network of the aircraft at the system level, regardless of the type of design (static or dynamic). The following function parameters are taken into account:

- type of function;

- the amount of RAM used;

- the amount of non-volatile memory used ROM;

- the frequency of the call T _call. ;

- the number of cycles of the calculator, for which the function N _T ;

- The size of the message sent each time DataSize is called.

and calculators:

- the amount of RAM RAM _CPU ;

- the amount of non-volatile memory ROM _CPU ;

- performance F _CPU ;

- bandwidth of the network interface V _AFDX .

We assume that the restrictions corresponding to the required level of functionality, reliability and fault tolerance of the on-board network are determined by the list of mutual exceptions for all functions of the system level, which can be presented in the form of a table. In other words, the list of mutual exceptions determines the prohibition of finding the specified types of functions in one calculator. We also note that by resource intensity of a subset of functions M we mean a combination of the following parameters:

; (1)- the amount of RAM used

; (one)

; (2)- amount of used non-volatile memory

; (2)

; (3)- share of used productivity

; (3)

(4)- share of network interface bandwidth

(4)

To solve this problem, the relationships between the functions of the level of the metamodel systems can be represented in the form of an oriented weighted graph (relationship graph). In this case, the vertices of the graph are functions with a given resource intensity. The edges of the interaction graph are determined by the relationships between functions at the system level, and the weight of the edges is determined by the intensity of data exchange between the related functions.

We assume that the relationship between the functions that were distributed between different calculators is carried out through a network interface (for avionics systems, this is the AFDX interface). In the optimization process, we will also take into account the minimization of traffic between all on-board network calculators (parameter V _AFDX ). Comparison of the parameters of the calculator and the resource consumption of a subset of the functions assigned to this calculator in the process of optimizing the architecture of the on-board network is carried out taking into account the constraints described by the following system of inequalities:

(5)

(5)

where L _RAM , L _ROM, and L _F are the load limits of the calculator for RAM, non-volatile memory, and performance, respectively (vary from 0.01 to 1).

In the process of optimizing the architecture of the on-board network using the present method, a comparison of the parameters of the calculator and the resource consumption of the subset of the functions of the level of metamodel systems assigned to it is used, which has the following form:

(6)

(6)

Where

(7)

(7)

and conditions are met

(8)

(eight)

Minimization of the matching value I _opt characterizes the process of minimizing the unused resources of the calculator by the set of parameters RAM _CPU , ROM _CPU , F _CPU , as well as traffic through the network interface of the calculator V _AFDX , which, in turn, minimizes the number of calculators in the optimized on-board network, and also the intensity of information flows between them. Note that the a priori requirements for the resources of the on-board network and the priorities of the optimization process can be set using the priority coefficients K _RAM , K _ROM , K _F , K _AFDX (vary from 1 to 10), which determine the importance (priority) of optimizing the parameters RAM _CPU , ROM _CPU , F _CPU and V _FDXA respectively.

Thus, the way to optimize the architecture of the aircraft’s onboard network is to minimize the comparison value I _opt (6), taking into account the fulfillment of conditions (8), as well as the conditions determined by the list of mutual exceptions.

The input data for the implementation of the method are a list of mutual exceptions, which can be presented in the form of a table, a list of calculators with specified parameters, a list of functions with a given resource consumption, and relationships between functions that can be represented using the relationship graph.

The invention is carried out as follows.

At the first stage, the computer is given a list of computers of the optimized on-board network with the given parameters: the amount of RAM RAM _CPU , the amount of volatile memory ROM _CPU , the capacity of the F _CPU , the throughput of the network interface V _AFDX . Next, the functions belonging to the systems of the optimized on-board network with the given parameters are set: the type of function, the amount of RAM used, the amount of non-volatile ROM used, the frequency of the call T _call. , the number of clock cycles of the calculator for which the function N _T is executed, the size of the message transmitted during each call to DataSize. Next, relations between functions are defined via the AFDX interface (a graph of relationships with many functions M, with a given resource intensity and edges defining relationships between given functions), and a list of mutual exceptions is formed that defines the prohibition of finding the given types of functions in one calculator. Next, set the load limits of the calculator for RAM, ROM and performance, respectively (vary from 0.01 to 1) L _RAM , L _ROM and L _F, respectively. Then the values of the priority coefficients K _RAM , K _ROM , K _F , K _{AFDX are formed} , which determine the importance of optimizing the parameters RAM _CPU , ROM _CPU , F _CPU , V _AFDX, respectively.

Next, they determine the possibility of placing all functions at the system level in one of the computers of the given list, and compare the resource intensity of the set of functions (the entire graph of the set of functions M) to each computer from the given list, which provides the smallest comparison of the parameters of the computer and the resource intensity of the Iopt functions assigned to it (6) .

If a calculator is found that provides the smallest value for comparing the parameters of the calculator and the resource consumption of the Iopt functions assigned to it (6), the result is written into the resulting tables and the design process of the on-board network ends. In this case, the synthesis of the onboard network without network interactions is ensured with the optimal ratio of the resource consumption of the computer and the set of functions placed in it.

If such a computer is not defined, then in the second step, the resource consumption of each free vertex of the graph is compared with each free computer from a given list, and the computer that provides the lowest value for comparing the parameters of the computer and the resource consumption of the functions I assigned to it is selected_opt (6).

At the third stage, the found vertex is combined in turn with each of the remaining free vertices (taking into account the list of mutual exclusions specified at the first stage) and each such combination is compared with each remaining free calculator from the given list so that the lowest value of the comparison of the parameters of the calculator and the resource consumption of the assigned functions l _opt (6).

At the fourth stage, the resulting combination of functions is combined in turn with each remaining function from the given list that is not included in the obtained combination, taking into account the given list of exceptions, and the resource intensity of such a combination of functions is alternately compared with the parameters of each calculator from the given list, and the calculator that provides the lowest value is selected mapping and resource parameters calculator assigns combining functions I _opt, and the fourth step is repeated up until when opostavlenii resource consumption exceeds combining functions of the calculator of the parameters of each predetermined list.

At the fifth stage, the last combination of functions and a calculator are used, when comparing the resource intensity and parameters of which the resource intensity of the last combination of functions did not exceed the parameters of each calculator from the given list, the calculator and a subset of the functions included in the resulting combination of functions are defined as used.

then use functions and calculators that are not marked as used in the previous step, and repeat steps two through five until each function defines a calculator.

Next, a check is made for the presence of free (not assigned to any calculator) vertices and if no free vertices are found, the results of the design of the on-board network are saved. In the case of finding free vertices go to the third stage.

The process stops when all functions are marked as used.

The output data of the method of optimizing the architecture of the aircraft’s onboard network are lists of the involved computers of the studied onboard network and lists of functions that belong to each computer of this list.

Thus, thanks to the application of this method, the problem of optimizing the architecture of the aircraft’s onboard network is solved by redistributing the computational tasks (functions) between the onboard network computers taking into account the specified restrictions that determine the required characteristics in terms of functionality and reliability, since when distributing all functions at the system level for calculators selected from a given list (with given parameters of each calculator), taking into account the table (list) of mutual exceptions, ensure The greatest load of each calculator according to the specified parameters is obtained, which leads to a decrease in the number of onboard network calculators, as well as information flows between network elements.

Claims (5)

1. The method of optimizing the architecture of the aircraft computer network, characterized in that at the first stage, a computer is given a list of types of calculators from the structure of the optimized on-board network with specified parameters, a list of functions belonging to the systems of the optimized on-board network, with a given resource consumption, relationships between given functions, a list of exceptions that determine the prohibition of finding certain types of functions in one computer, form the values of the priority coefficients and the load limits of the calculator, at the second stage, they alternately compare the resource consumption of each function with the parameters of each calculator the user, and select the calculator and the function that, when comparing the parameters of the calculator and the resource intensity of the functions assigned to it, the lowest comparison value, in the third stage, the function selected in the second stage is alternately combined with each remaining function from the given list, taking into account the given list of exceptions, and the resource intensity is compared such a combination of functions with the parameters of each calculator from a given list, while choosing a calculator that provides the lowest value of co-post of the parameters of the calculator and the resource intensity of the union of functions assigned to it, at the fourth stage, the resulting union of functions is combined alternately with each remaining function from the given list that is not included in the obtained union, taking into account the given list of exceptions, and the resource intensity of such a combination of functions is alternately compared with the parameters of each calculator from the given list, while choosing a calculator that provides the lowest value for comparing the parameters of the calculator and resource consumption the combining of functions assigned to it and repeat the fourth stage until, during the comparison, the resource intensity of the combination of functions exceeds the parameters of each calculator from the given list, at the fifth stage, the last combination of functions and the calculator are used, when comparing the resource intensity and parameters of which the resource intensity of the last combination of functions did not exceed the parameters each calculator from a given list, the calculator is determined and a subset of the functions included in the resulting combination of functions is marked to used ones, then they use functions and calculators that are not marked as used in the previous step, and repeat steps two through five until each function defines a calculator. 2. A method of optimizing the architecture of an onboard network of an aircraft according to claim 1, characterized in that the predetermined parameters of the computer are determined by the amount of RAM non-volatile memory capacity, performance. 3. The method of optimizing the architecture of the aircraft’s onboard network according to claim 1, characterized in that the resource consumption of the functions is determined by the parameters: the type of function, the amount of RAM used, the amount of non-volatile memory used, the frequency of the call, the number of clock cycles of the calculator for which the function is executed, and the message size transmitted on every call. 4. A method for optimizing the architecture of an aircraft’s onboard network according to claim 1, characterized in that the relationships between the given functions are specified using the relationship graph. 5. A method for optimizing the architecture of the aircraft’s onboard network according to claim 1, characterized in that the list of mutual exceptions is given in the form of a table.