Year of Graduation
Stochastic Models for Description of Modern Information Systems
Control Systems and Data Processing in Engineering
In thesis a stochastic model for distributed average consensus algorithm is considered. These algorithms are used in such applications as parallel processors load balancing, network synchronization and distributed coordination of mobile autonomous agents. In this model each node updates its state by using information obtained from its neighbors, and every new value is corrupted by an additive random noise with zero mean. Three types of consensus algorithms are considered: algorithm with linear and nonlinear protocol and load-balancing algorithm. Probabilistic characteristics and asymptotic properties of stochastic processes describing the given model are investigated analytically and by means of computer modeling. Approaches of the present thesis can be also applied to more wide classes of information systems.