Modified algorithm for achieving of consensus on the basis of local voting rule
DOI №______
Abstract
It is noted that decentralized networks are now widely used. There is an important task of balancing the network load, which is to maintain a uniform load of all agents in time. The possibility of using the local voting protocol for decentralized load balancing in the network is shown. This may change the topology and take into account interference and delays in the receipt of all tasks in the system. The article explores the possibility of creating a universal software for determining the parameters and conditions for achieving consensus based on a local voting rule. The conditions that must be taken into account when considering the load balancing problem are determined. The task of achieving consensus in a minimal time is formalized. A model of a network of agents (nodes) is presented that perform parallel tasks of the same type and in which the redistribution of tasks between agents on the basis of feedbacks is allowed. It is noted that with the distributed approach, load balancing algorithms are executed on all nodes. All nodes exchange status information with other nodes on the network. Redistribution occurs only between neighboring nodes. The stationary and nonstationary statements of the problem are considered with ensuring the uniform loading of all nodes of the network. The relevance of studying the properties of an algorithm of the stochastic approximation type under conditions of a randomly changing structure of bonds in a network under the action of interference is noted. The scheme of the developed control algorithm which is called the protocol of local voting is presented. The main conditions under which the protocol of local voting ensures the achievement of consensus is given. The characteristics and content of the stages of the developed algorithm are presented. The work of each stage is explained. It is noted that the proposed algorithm can be the basis for creating software that will allow to investigate the limiting parameters of agents when consensus is reached.
Keywords: achievement of consensus; decentralized systems; load balancing.
