UDC 004.75; 656.1
ON THE TASK OF BUILDING DECENTRALIZED INTELLIGENT TRANSPORT SYSTEM BASED ON RAFT PROTOCOL AND CLUSTERING BY NETWORK DISTANCE
M. G. Gorodnichev, Ph.D., Associate Professor, Dean of the Faculty of Information Technology, Head of the Department of Mathematical Cybernetics and Information Technology, MTUCI, Moscow, Russia;
orcid.org/0000-0003-1739-9831, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
E. A. Saksonov, Doctor in Technical Sciences, Professor, Professor of the Department of Mathematical Cybernetics and Information Technology, MTUCI, Moscow, Russia;
orcid.org/0000-0002-2015-6956, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
V. P. Kulagin, Doctor in Technical Sciences, Professor, Professor of the Department of Mathematical Cybernetics and Information Technology, MTUCI, Moscow, Russia;
orcid.org/0000-0001-8826-6724, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
A. F. Kaperko, Doctor in Technical Sciences, Professor, Professor, Department of Electronic Engineering, National Research University of Higher School of Economics, Moscow, Russia;
orcid.org/0000-0003-2253-1083, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The article is devoted to the development and experimental evaluation of decentralized architecture for intelligent transport system (ITS) based on Raft consensus protocol and network distance metric (RTT) server clustering method. Existing solutions are shown either to require manual configuration and centralized coordination, or not to be optimized for latency with geographical distribution of nodes, which reduces fault tolerance and increases system response time. A method for dynamic formation of subclusters with automatic selection of leaders, in which each server joins the leader closest to RTT is proposed. The leader is re-elected and reconfigured without administrator participation. A prototype of microservice system implementing access control with tokens and automatic reconfiguration is presented. Experiments demonstrate: acceleration of server preparation to be included in cluster by 87,5 %, reduction of loading time of 10 GB of data by 47,6 %, and leader switching time of about 730 ms in case of node failure.
Key words: : intelligent transport system, Raft, RTT, clustering, fault tolerance, microservices, RethinkDB, ZeroMQ.