Distributed system of virtual machines for self-organized networks

Abstract

Introduction: Active data, being fragments of executable code transmitted between the nodes of an active network, are an effectivemechanism for the operation of software-reconfigurable distributed systems. Previously, in the works devoted to active data, not enoughattention was paid to the implementation of the runtime environment (the processor) for the executable code of active data, as well asto the issues of building hypervisors and load balancing in distributed systems. Purpose: Developing principles for the construction of virtual machines with active data, providing the reconfigurability of the target devices and network flexibility in general. Evaluatingthe possibility of using the existing approaches to load balancing for networks with active data. Methods: Our study uses the principlesof software-defined system development, the conception of active data, theoretical foundations and technology of virtualization.Results: Is has been proposed to use a distributed system of virtual machines as an active data execution environment, based on theobject-oriented approach to creating distributed applications. Each node of such a distributed system of virtual machines can act aseither a control or slave node during the object interaction. Based on the developed approach, we proposed to solve the problem ofbuilding a network of repeaters using active data, considering an unmanned aerial vehicle as an element of an active info-communicationnetwork which supports the active data technology. Since a distributed system of virtual machines enables asymmetric distribution ofdecentralized network nodes, a method has been developed for a distributed system whose nodes are unmanned aerial vehicles and acontrol node, to control the asymmetry value by creating objects of various decomposition levels. Practical relevance: The proposedmethods provide a way to control the resource consumption of the nodes of a distributed software-reconfigurable network and theamount of network data transmitted. For dynamic management of the load on the network nodes, a resource manager architecture anda resource allocation algorithm are developed.