Evaluation of quality of digital signal filtration software in real time for critical systems

Abstract

The article is devoted to the development of methods and means of unified evaluation of quality of digital signal filtering software in real time for critical systems. The software class of automated flight control systems of aircraft, as typical representatives of critical systems that require certification to meet initial quality requirements, was studied.

Software control systems solve the general problems of reproduction of controlled parametric information (visualization of flight parameters), monitoring of exits of object parameters under restriction (tolerance control), quality control of object functioning and filtering of parametric information. Especially important, although small in size, is the filtration subsystem. The filtering subsystem performs the tasks of analysis of parametric flight information, removal of emissions, distortions, noise and approximation of areas of distorted sensor information to reliable values in order to further process this information by other subsystems that can not properly perform their functions without prior quality filtering. In the process of analysis it was found that to ensure the convenience of expert assessment, it is necessary to qualitatively perform the task of filtering flight information in real time.

The consequences of low quality and erroneous results of the basic subsystems that assess the condition of the object of control can lead to catastrophic consequences, and therefore the problem of improving the safety of objects controlled by automated control systems is always relevant. One of the ways to achieve this goal is to ensure the required level of quality of the main subsystems, which is especially important during the operation of critical systems.

Besides known methods of an estimation of quality of the filtering software use different axiomatics and different criteria in the process of selecting a set of quality indicators. They also do not substantiate the choice of metrics and measures. Therefore, for digital filters that process arrays of state vectors of the object of monitoring for each parameter, you should use a unified estimate with the mandatory definition of the basic characteristics and attributes of the filtering methods being compared. To do this, it is proposed to use a formalized assessment based on the quality model of the modern standard ISO/IEC 25010 from the group of standards SQuaRE.

The scientific problem investigated and solved in the article is to ensure the completeness and adequacy of the quality model of the filtering software subsystem to its requirements and the constructiveness and reliability of methods for evaluating the unified characteristics and quality properties included in the model.

The built quality model is adequate to the requirements of the filtering software and contains a set of selected characteristics, compliance criteria and metrics, which allows to obtain a unified method of quality assessment for different Kalman’s filters and other digital filters and choose a more efficient and accurate filter for the filtering subsystem of several competitors, which increases the reliability of software monitoring systems and safety of aircraft.

The quality model of filtering software created in the article was used to compare and obtain quality estimates of adaptive Kalman filter and αβ-filter, which are most often used in practice to solve the problems of filtering trajectory and navigation parameters of aircraft.

Unified quality model by analogy can be easily modified, adapted and used to assess the quality of software components of digital signal filtering in other related areas, especially in subject areas where critical software systems operate, for which the state of the object is monitored based on parametric information sensors