МЕТОД ДИНАМИЧЕСКОГО ФОРМИРОВАНИЯ ПУЛОВ В ИНФОРМАЦИОННО-ВЫЧИСЛИТЕЛЬНЫХ СИСТЕМАХ ВОЕННОГО НАЗНАЧЕНИЯ

Abstract

Purpose: Military computing systems can face situations when the amount of the tasks drastically increases, or when the systems degrades owing to natural failures and/or adversary's impact. In these situations, the computing system resources are usually insufficient and the functional environment is generally non-stochastic. This research is devoted to the problem of resource distribution for tasks in a non-stochastic environment using adaptive control theory methods. Methods: The informational and technical impact is formalized to specify adversary's opportunities in changing the computing system structure. This helps to create a computing system model in real time during the control process. The pool (assembled resources) is created during the solution of a modified bin packing problem. To speed it up, the set of alternatives is narrowed down step by step, and the tasks are distibuted over the system on a phased basis. Results: A method was proposed which adapts the current structure of a distributed heterogeneous computing system with stationary and mobile elements in order to assemble pools appropriate for the tasks. A pool is assembled from processors, links, inputoutput facilities and storage units. Originally, a pool has an entry-point for the task, i.e. the initial element for the task execution. Later the pool is expanded by adjacent elements. The expanding procedure repeats until all the system resources are used up or until a pool with necessary characteristics is assembled. The obtained pool is not optimal but its assembly time is close to real. Convergence and termination of the proposed method were proved. An example was given, and the conditions of using the method were shown. Practical relevance: The results of the research can be used for building an integrated informational environment of various governmental services using military automated systems.