Simulation of resource distribution in large information systems based on multiagent approach

Abstract

Traditional planning theory considers the general problem of division of labor on computing devices. A fairly large class of planning tasks is included in the described task of resource allocation. The only type of resources in such tasks are computing resources. The article considers the tasks of resource allocation in multi-agent systems, options for their applicability and existing methods of solving these problems. The task of allocating resources is one of the fundamental tasks: from the efficient allocation of one's own time to the distribution between different activities and the tasks of allocating resources in large information systems. The article considers different approaches to resource allocation in large information systems: resource allocation within non-stationary tasks, stationary tasks and in multi-agent systems. These approaches help to solve various applications in real time. In the case of resource allocation in multi-agent systems, their decentralization must be taken into account: agents are directly responsible for themselves and do not have complete information about the system, which changes the very essence of the task. Thus, the development of multi-agent models is possible. The multi-agent scheduling methods used in the system can be used in real-time decentralized systems compared to the previously mentioned traditional methods. Their application provides management of planning and execution of tasks, they can be used to manage groups of objects consisting of a large number of devices and able to quickly process large-scale tasks. Thus, in a short time it is possible to design and commission based on the use of multi-agent technologies for planning new-generation software and hardware systems that can interact and work in a group and applicable to a wide range of tasks in various fields.